Handbook This page intentionally

Document Sample
Handbook This page intentionally Powered By Docstoc
					This page intentionally left blank
Handbook of Biodiversity Methods

Biodiversity is recognised to be of global             (lower plants, fungi, vascular plants,
importance, yet species and habitats continue to be    invertebrates, fish, amphibians, reptiles, birds
under increasing pressure from human-induced           and mammals).
influences, whether in urban, rural or wilderness         The Handbook provides an invaluable
settings. Environmental concerns have never            compendium for ecologists, wildlife managers,
before been so high on the political agenda, driving   nature conservation professionals, local and
increased legislation which places major emphasis      national authorities, environmental managers,
on individual, public and corporate responsibility     corporate bodies and companies, government
for conserving biodiversity and for managing           conservation agencies and regulators involved in
development in an environmentally sensitive and        auditing ecological resources. It will enable
sustainable way. The starting point for assessing      practitioners to better monitor the condition of the
legal compliance is the requirement for a              biodiversity resource, resulting in improved data
comprehensive biodiversity audit. For those            upon which to base future conservation,
needing to undertake such audits, this Handbook        management, development and policy decisions
provides standard procedures for planning and          and actions.
conducting a survey of any terrestrial or freshwater      David Hill is Director of Ecology for RPS Group
species or habitat and for evaluating the data so as   plc, a leading environmental consultancy.
to determine its local, national and international        Matthew Fasham is a Principal Consultant for
significance.                                          RPS Group plc.
   Organised in three parts, the Handbook first           Graham Tucker is a freelance ecologist with
addresses planning, providing a pragmatic              Ecological Solutions.
approach to method selection, sampling strategy,          Michael Shewry is an environmental statistician
and data analysis and evaluation. The second part      with Scottish Natural Heritage.
is devoted to habitats, describing survey,                Philip Shaw is an environmental audit specialist
evaluation and monitoring methods for a broad          with the Advisory Services of Scottish Natural
range of habitats. Part III considers species and      Heritage.
provides information on general methods before            Any opinions expressed are those of the authors
addressing specific methods of survey and              and do not necessarily represent the views of RPS
monitoring for the major taxonomic groups              or Scottish Natural Heritage.
Handbook of Biodiversity Methods
      Survey, Evaluation and Monitoring

                                  Edited by
                                 David Hill
                           Matthew Fasham
                            Graham Tucker
                            Michael Shewry
                                Philip Shaw
  
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press
The Edinburgh Building, Cambridge  , UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521823685

© RPS Group plc and Scottish Natural Heritage 2005


This publication is in copyright. Subject to statutory exception and to the provision of
relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2005

-   ---- eBook (EBL)
-   --- eBook (EBL)

-   ---- hardback
-   --- hardback




Cambridge University Press has no responsibility for the persistence or accuracy of s
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
         In memory of Colin J. Bibby,
an outstanding conservation scientist
Contents


Preface                                             page xi    Part II Habitats
Acknowledgements                                        xiii   GRAHAM TUCKER , MATTHEW FASHAM , TIM RICH ,
                                                               MICK REBANE , GEORGE PETERKEN , FIONA
Part I Planning                                                MCMEECHAN AND DICK BIRNIE
GRAHAM TUCKER , MATTHEW FASHAM , DAVID HILL ,
MICHAEL SHEWRY , PHILIP SHAW AND MAX WADE
                                                               4 Introduction to habitat evaluation                   105
                                                               G . TUCKER , AND M . FASHAM
1 Introduction to planning
                                                                 4.1 How to use the Handbook: a recap                 105
G . TUCKER , D . HILL AND M . FASHAM
                                                                 4.2 Habitat survey and monitoring                    105
  1.1 The purpose of surveying and
          monitoring                                      3    5 Habitat requirements and issues                      107
                                                               T . RICH , G . PETERKEN , G . TUCKER , F . MCMEECHAN
2 Planning a programme                                    6
                                                               AND D . DOBSON
G . TUCKER , M . FASHAM , D . HILL , M . SHEWRY ,
P . SHAW AND M . WADE                                            5.1 Woodland and scrub                               107
                                                                 5.2 Lowland wood-pastures and parkland               114
  2.1 Setting the objectives for the monitoring
                                                                 5.3 Farmland boundary features                       117
          programme                                       6
                                                                 5.4 Grassland and herbaceous communities             121
  2.2 Selection of methods for monitoring
                                                                 5.5 Limestone pavement                               124
          each attribute                                 17
                                                                 5.6 Lowland and upland
  2.3 Designing a sampling strategy                      23
                                                                     heathland                                        126
  2.4 Reviewing the monitoring programme                 42
                                                                 5.7 Fens, carr, marsh, swamp and reedbed             128
  2.5 Data recording and storage                         46
                                                                 5.8 Lowland raised bog                               130
  2.6 Data analysis, interpretation and review           49
                                                                 5.9 Standing open water                              132
3 Biodiversity evaluation methods                        65      5.10 Rivers and streams                              136
G . TUCKER                                                       5.11 Montane habitats                                141
  3.1 Biodiversity values and evaluation                         5.12 Blanket bog                                     143
      purposes                                           65      5.13 Maritime boulders, rocks, cliffs and
  3.2 A framework for ecological evaluations             65           slopes                                          145
  3.3 Identification of valuable ecosystem                       5.14 Shingle above high tide                         147
          components                                     65      5.15 Sand dunes and strandline vegetation            148
  3.4 Principles underlying the setting of                       5.16 Saltmarsh                                       150
          conservation priorities                        68
                                                               6 Methods for surveying habitats                       154
  3.5 Species and habitat conservation priority
      lists                                              72      6.1 General habitat survey and monitoring
  3.6 Site evaluations and selection of                              methods                                          154
          protected areas                                81      R . V . BIRNIE , G . TUCKER AND M . FASHAM
  3.7 Site conservation designations                     88      6.2 Physical attributes                              195
  3.8 Site evaluations for management                            M . FASHAM AND G . TUCKER
          planning                                       95      6.3 River morphology and aquatic
  3.9 Site evaluations for Environmental                             vegetation composition                           197
          Impact Assessments (EIAs)                      96      G . TUCKER



                                                                                                                       vii
viii      CONTENTS



 6.4 Ground and shrub vegetation                        201   12 Lichens                                        279
   T . RICH , M . REBANE , M . FASHAM , F . MCMEECHAN         S . DAVEY , M . FASHAM AND D . DOBSON

   AND D . DOBSON
                                                                 12.1 Attributes for assessing condition        280
   6.5 Trees and woodland stands                        222      12.2 General methods                           280
   G . PETERKEN AND M . FASHAM
                                                                 12.3 Lichen conservation evaluation criteria   284

7 Surveying and monitoring management or                      13 Bryophytes                                     288
   environmental impacts                                237   G . ROTHERO , D . DOBSON AND M . FASHAM
M . REBANE , M . FASHAM AND G . TUCKER
                                                                 13.1 Attributes for assessing condition        288
   7.1 Grazing and browsing                             237      13.2 General methods                           289
   7.2 Burning                                          240      13.3 Bryophyte conservation evaluation
   7.3 Erosion                                          243            criteria                                 292
   7.4 Vegetation surveys in relation to
                                                              14 Aquatic macrophytes and algae                  295
       developments                                     244
                                                              N . STEWART AND M . WADE

8 Habitat conservation evaluation criteria              245      14.1 Attributes for assessing condition        296
G . TUCKER AND F . MCMEECHAN
                                                                 14.2 General methods                           296
   8.1 Key evaluation considerations                    245      14.3 Requirements for species of particular
   8.2 Protection status in the UK and EU               245            conservation importance                  300
   8.3 Conservation status in the UK                    246      14.4 Aquatic macrophyte conservation
                                                                       evaluation criteria                      301
Part III Species                                              15 Vascular plants                                303
9 Introduction to species assessment                    253   T . RICH , V . HACK AND   F . MCMEECHAN


   9.1 Species surveying and monitoring                 253      15.1 Attributes for assessing condition        305
                                                                 15.2 General methods                           307
10 General principles and methods for species           255      15.3 Vascular plant conservation
M . FASHAM AND S . MUSTOE
                                                                       evaluation criteria                      318
       10.1 Introduction                                255
                                                              16 Dragonflies and damselflies                    322
       10.2 Terminology                                 255   C . PLANT , R . SANDS AND M . FASHAM
       10.3 Total counts                                257
       10.4 Timed searches                              257      16.1 Attributes for assessing condition        322
       10.5 Quadrats                                    258      16.2 General methods                           322
       10.6 Distance sampling                           260      16.3 Odonata conservation evaluation
       10.7 Line and strip transects                    264            criteria                                 327
       10.8 Point counts                                266   17 Butterflies                                    328
       10.9 Trapping webs                               267   C . PLANT , R . SANDS AND M . FASHAM
       10.10 Removal method                             268
                                                                 17.1 Attributes for assessing condition        328
       10.11 Mark–recapture techniques                  268
                                                                 17.2 General methods                           329
11 Fungi                                                271      17.3 Butterfly conservation evaluation
R . WATLING , M . FASHAM AND D . DOBSON                               criteria                                  333

       11.1 Attributes for assessing condition          272   18 Moths                                          335
       11.2 General methods                             272   C . PLANT , R . SANDS AND M . FASHAM

       11.3 Fungus conservation evaluation                       18.1 Attributes for assessing condition        335
            criteria                                    276      18.2 General methods                           336
                                                                                                          Contents    ix



   18.3 Macromoth conservation evaluation                     24.6 Protection status in the UK and EU                427
           criteria                                 339
                                                          25 Bats                                                    433
19 Other terrestrial invertebrates                  341   R . STEBBINGS , H . MANSFIELD AND M . FASHAM
P . DENNIS , C . PLANT , R . SANDS AND M . FASHAM
                                                              25.1 Attributes for assessing condition                433
   19.1 Attributes for assessing condition          342       25.2 General methods                                   438
   19.2 General methods                             342       25.3 Bat conservation evaluation criteria              446
   19.3 Terrestrial invertebrate conservation
                                                          26 Other mammals                                           450
           evaluation criteria                      357
                                                          A . BENNETT , P . RATCLIFFE , E . JONES , H . MANSFIELD

20 Aquatic invertebrates                            359   AND R . SANDS
P . KERRISON , T . NORMAN AND M . FASHAM
                                                              26.1 Attributes for assessing condition                450
   20.1 Attributes for assessing condition          359       26.2 Indirect methods                                  450
   20.2 General methods                             360       26.3 General methods                                   455
   20.3 Requirements for species of particular                26.4 Direct methods                                    460
           conservation importance                  365       26.5 Requirements for species of particular
   20.4 Aquatic invertebrate conservation                            conservation importance                         464
           evaluation criteria                      365       26.6 Mammal conservation evaluation
                                                                     criteria                                        469
21 Fish                                             368
N . GILES , R . SANDS AND M . FASHAM                      Appendix 1 Monitoring and reporting obligations
   21.1 Attributes for assessing condition          368   under international conservation agreements                473
   21.2 General methods                             375   Appendix 2 Relationship between BAP Priority
   21.3 Freshwater fish conservation                      Habitat and Broad Habitat categories and Habitats
           evaluation criteria                      385   Directive nomenclature                                     478
22 Amphibians                                       387
                                                          Appendix 3 Annotated list of key references for plant
D . LATHAM , E . JONES AND M . FASHAM
                                                          identification                                             490
   22.1 Attributes for assessing condition          387
                                                             Lichens                                                 490
   22.2 General methods                             388
                                                             Bryophytes                                              490
   22.3 Amphibian conservation evaluation
                                                             Charophytes                                             491
           criteria                                 401
                                                             Ferns                                                   491
23 Reptiles                                         404      Vascular plants                                         491
D . LATHAM , E . JONES AND M . FASHAM
                                                          Appendix 4 Determining appropriate quadrat size for
   23.1 Attributes for assessing condition          404   vegetation sampling                                        493
   23.2 General methods                             404
   23.3 Reptile conservation evaluation criteria    410   Appendix 5 The relocation of permanent plots               495

24 Birds                                            412      Mapping                                                 495
S . MUSTOE , D . HILL , D . FROST AND G . TUCKER             Marker posts                                            495
                                                             Paint                                                   495
   24.1 Attributes for assessing condition          412
                                                             Buried metal markers                                    495
   24.2 General methods                             413
                                                             Photographs                                             495
   24.3 Some specific methods used in
                                                             Total Stations                                          496
           specialist EIA studies                   418
                                                             Global positioning systems (GPS)                        496
   24.4 Some key species regularly considered
        in EIA studies                              420   Appendix 6 Equipment required for undertaking
   24.5 Bird conservation evaluation criteria       422   different types of survey                                  497
x     CONTENTS



Recommended sources of further information         519      Aquatic invertebrates (Chapter 20)   524
                                                            Fish (Chapter 21)                    524
    Habitat requirements (Chapter 5)               519
                                                            Amphibians (Chapter 22)              525
    Survey methods (Chapter 6)                     519
                                                            Reptiles (Chapter 23)                526
    Methods for species assessment (Chapter 10)    519
                                                            Birds (Chapter 24)                   526
    Fungi (Chapter 11)                             520
                                                            Bats (Chapter 25)                    529
    Lichens (Chapter 12)                           520
                                                            Other mammals (Chapter 26)           529
    Bryophytes (Chapter 13)                        521
    Aquatic macrophytes and algae (Chapter 14)     521   References                              530
    Vascular plants (Chapter 15)                   522
                                                         Glossary                                551
    Dragonflies and damselflies (Chapter 16)       522
                                                            Monitoring terms and acronyms        551
    Butterflies (Chapter 17)                       523
                                                            Statistical terms                    552
    Moths (Chapter 18)                             523
    Other terrestrial invertebrates (Chapter 19)   523   Index                                   556
Preface


This generation is living at a time when the world’s     to expand their reserve network and extend new
biodiversity resources have never been so impover-       habitats near existing ones by means of novel tech-
ished. If we take the UK as an example of what has       niques based on scientific understanding. There is
happened across many parts of the planet, since          large-scale restoration of contaminated land sites.
1945, largely as a result of agricultural intensifica-   Coastal managed realignment offers opportunities
tion, we have lost over 50% of our ancient lowland       to create massive areas of wet grassland, saltmarsh
woodlands, 150 000 miles of hedgerow, 95% of             and reedbed habitat, which will provide substan-
traditional hay meadows, 80% of chalk downland           tial benefits to wildfowl and waders. Industry, too,
and 80% of wetland fens and mires. This has given        is working with organisations to create large-scale
rise to massive losses in some, once very common,        reserves in currently uninteresting farmland, a
farmland birds: in the past 30 years 40% of Song         prime example being the Great Fenland Project in
Thrushes, 54% of Yellowhammers, a staggering             the Cambridgeshire Fens of the UK.
87% of Starlings and 90% of Corn Buntings have              As biodiversity has dwindled in the past 50 years,
disappeared.                                             so policies and laws aimed at turning the tide have
   In addition to agricultural intensification, devel-   flourished. There are now over 200 legal instruments
opment pressure as a result of industrialisation,        aimed at protecting the environment and which
human population expansion and resultant                 have an impact on countries such as the UK. The
increases in the ‘ecological footprint’ of our own       greatest successes have been achieved where there
species through, for example, house building, air-       has been government regulation: we now have the
ports, seaports, road infrastructure, water supply,      best air and water quality in Britain for about 200
energy generation, waste management, freight             years, almost entirely as a result of regulation. Key
distribution and extraction of raw materials, has        instruments for biodiversity conservation in the UK
taken its toll on biodiversity. The UK government’s      are the Wildlife & Countryside Act, the Countryside
sustainable development commission recently              and Rights of Way Act, The Nature Conservation
announced that the country has a very long way to        (Scotland) Act EU Birds and Habitats Directives,
go before existing developments, and the way we          the Habitats Regulations, the EIA Directive and
manage environmental resources, can be deemed to         EIA Regulations, the Hedgerow Regulations, Bonn
be ‘sustainable’. This is without any consideration of   Convention, Ramsar Convention, Bern Convention,
the impending threat from climate change.                European and National Red Lists of species of con-
   But it would be wrong to focus entirely on the        servation concern, and Biodiversity Action Plans.
negatives. There are signs that our attitudes to our     A whole industry has developed to support biodiver-
environment are changing and there are a growing         sity conservation, to save what we have and improve
number of examples where the primary focus               upon it. In parallel there has been increased site-
of governments, companies and individuals is             based protection: the designation of local wildlife
towards the stitching back of the fabric of the          sites, green corridors, County Wildlife Sites, Sites of
environment and countryside. A range of agri-            Special Scientific Interest, National Nature Reserves,
environment schemes is attempting to redress the         Special Protection Areas, Special Areas of
damage caused to farmland biodiversity by the            Conservation, Biosphere Reserves and World
Common Agricultural Policy, reforming subsidies          Heritage Sites.
away from production and into environmental                 During this recent period we have moved from a
benefits. Organisations such as the RSPB continue        natural history mentality to an accountancy


                                                                                                              xi
xii   PREFACE



mentality, where numbers and targets are the order        survey and/or monitoring programme, sampling
of the day. Government has set out some ambitious         strategy and data analysis. There is then a section
targets for biodiversity: by 2010, for example, it        which describes generically how to evaluate the
wants 95% of all SSSIs in England to be in a              data collected: what does it mean at different spa-
Favourable Condition. We have a long way to go.           tial scales?
Currently about 42% of the one million or so hec-            Part II is devoted to habitat survey, evaluation
tares of SSSIs in England fail to make the grade of       and monitoring, describing approaches for the full
‘Favourable Condition’. The percentages in                range of habitats in the UK but with direct rele-
unfavourable condition in England, according to           vance to many countries. For each habitat type the
selected habitats, are: rivers and streams 69%, upland    potential attributes that indicate condition are
grasslands and heaths c. 65%, fen, marsh and swamp        defined, together with appropriate and commonly
35%, and lowland broadleaved woodland 33%. This           used methods for surveying them and establishing
gives an idea of the widespread losses in quality that    a monitoring scheme for the habitat concerned.
have taken place in addition to losses in habitat         Based on structural similarities the methods can
quantity. Changes to quality are being addressed by       be applied to the full range of habitat types found
a plethora of site or conservation management             in Europe and, indeed, in other parts of the world.
plans, and similar mechanisms are being used to           Evaluation criteria are developed and defined for
mitigate for development impacts, including Section       each habitat.
106 agreements, unilateral undertakings and mitiga-          Part III is devoted to the survey, evaluation and
tion plans.                                               monitoring of species. General methods applicable
   So, against this background of biodiversity            to a range of taxa are first described, such as total
decline and a commitment to rebuild it, there are         counts, timed searches, use of quadrats, distance
three observations I would make. First, ecology has       sampling, line transects, point counts, etc. Each
a vital part to play in delivering a better quality       taxonomic group is then addressed, from fungi to
environment and better quality of life for people.        mammals. For each group, the attributes for asses-
Second, environmental quality improvements are            sing condition are described, followed by survey
increasingly being seen as solutions rather than as       and monitoring methods that can be applied, and
costly problems at the levels of both the corporate       then details of particular methods for species of
entity and society at large. Third, there is a need for   conservation importance as appropriate. Finally,
high-quality information on which to base deci-           for each group there is a section that describes
sions. We have written this Handbook in order to          the currently applicable conservation evaluation
enable biodiversity data to be collected and evalu-       criteria.
ated according to standard procedures. Future                I hope that the approaches and methods
decisions on policy reforms, land management,             described in this Handbook will stand the test of
development impacts and biodiversity conserva-            time and enable us to better monitor the condition
tion initiatives at a range of spatial scales can then    of the biodiversity resource. We should then be able
be based on fact rather than on conjecture.               to plan improved biodiversity conservation and
   The Handbook consists of three parts. The first        measure how well we are doing towards meeting
(Part I) addresses planning and describes how to          targets in the years ahead.
set objectives, what is it you actually want to do,
selecting the appropriate method, how to design a                                                   David Hill
Acknowledgements




The writing of the Handbook has been a mammoth       Primrose, Deborah Procter, Geeta Puri, Rob
task. However, we have been very fortunate to        Raynor, Terry Rowell, Pamela Strachan, Chris
have been able to assemble a highly competent        Sydes, Neale Taylor, Gavin Tudor, Stephen Ward,
team of authors who not only eased the task but      Christine Welch, Peter Wortham and the late
were able to take the text to greater depths of      David Phillips.
detail than any one of the editors could possibly        RPS provided time and logistical support to the
have achieved. Their wisdom, knowledge and           whole project which enabled us to meet deadlines
experience shines through. We therefore thank        and to see the whole project through to a fruitful
our contributing authors for their superb support    conclusion. We are most grateful to them. Alan
and hard work.                                       Crowden of Cambridge University Press and
   We were also fortunate to have had to hand a      Michael Usher, then Chief Scientist for Scottish
long list of experts who kindly commented on the     Natural Heritage, were convinced that the project
original version. Our sincere thanks go to Helen     was worthwhile and gave much encouragement
Armstrong, Sally Blyth, Phil Boon, Mairi Cole,       and support.
Andrew Coupar, Louise Cox, Andy Douse, Kathy             Finally, we thank the many professionals who
Duncan, Willie Duncan, Lynne Farrell, Vin            are striving to ensure we stitch back together the
Fleming, Stuart Gardner, Martin Gaywood, Doug        fabric of the countryside, both in the UK and
Gilbert, Dave Horsfield, Keith Kirby, John Kupiec,   abroad, to secure a future environment in which
Kate Holl, Philip Immirzi, Ross Johnstone, Ed        it is worth living. We hope this book plays some
Mackey, Jane Mackintosh, Jill Matthews, Angus        small part in assessing how well we are doing in the
MacDonald, Ed Mountford, John Orr, Brigid            years to come.




                                                                                                      xiii
Part I   *   Planning
1       *    Introduction to planning




1.1         THE PURPOSE OF SURVEYING AND                 evaluated against set agreed criteria. Impacts are
            MONITORING                                   considered in respect of this resource and assessed
                                                         for significance. Parts II and III of this Handbook
The development of a successful programme is
                                                         describe specific survey methods for habitats and
dependent upon being clear about what you
                                                         the full range of species from lower plants to mam-
want to do and why, i.e. your objectives. It is there-
                                                         mals. However, for some studies, particularly
fore important to define what monitoring is and
                                                         in relation to testing the effects of macro-
how surveys relate to monitoring. Survey and
                                                         environmental policy changes at a large spatial
monitoring is undertaken for a wide range of
                                                         scale, actual monitoring is performed. The empha-
objectives: for example, to measure a site’s qual-
                                                         sis in Part I of this Handbook is the design of data
ity, or a species’ abundance, to assess species
                                                         collection and the analytical treatment of the data
and habitat trends, for Environmental Impact
                                                         collected. Much of Part I therefore considers the
Assessment (EIA) studies, for corporate reporting,
                                                         planning, design and implementation of survey
or to assess compliance with international conser-
                                                         and monitoring, the latter often comprising a
vation agreements. These operate at many differ-
                                                         series of replicated surveys using standard
ent spatial scales and therefore necessitate
                                                         methods.
targeted methods for different applications, objec-
                                                            Once the data have been collected they will need
tives and deliverables. The significance and global
                                                         to be used for a specific purpose. One of the most
importance of monitoring nature conservation is
                                                         important uses is to evaluate a site, species, com-
aptly summarised in Appendix 1, which describes
                                                         munity, habitat, region, etc. Part I therefore
the monitoring and reporting obligations under
                                                         includes a section on generic approaches to evalua-
international conservation agreements as an
                                                         tion of biodiversity data, with more specific treat-
example of the far-reaching implications of the
                                                         ment for habitats and species given in the relevant
need to use adequate methods.
                                                         sections of Parts II and III.
                                                            As with monitoring, it is essential at the outset
                                                         of a survey to define objectives. A project may not
1.1.1        General objectives of surveying and
                                                         meet its full potential unless the aims are properly
             monitoring
                                                         understood and researched before data collection
For the purposes of Environmental Impact                 begins. Before planning your survey methods, con-
Assessment (EIA) studies, the term ‘survey’ defines      sider the variety of possible scenarios that could
the collection of spatial and/or temporal data about     dictate your project’s fieldwork techniques. Do
a species, a community or a habitat. The informa-        the results need to apply to one site or to a wide
tion provides a snapshot of presence, absence and,       geographical area? Are many species involved
dependent on its design and sophistication, abund-       or just one? Are accurate counts needed (spatially
ance and spatial distribution. In EIA studies the        referenced) or will relative counts or presence–
survey data are used to evaluate the ecological          absence data suffice? Answers to these questions
resource on a site, which is then assessed or            will determine the time commitments required

                                                               # RPS Group plc and Scottish Natural Heritage 2005.
4     1 INTRODUCTION TO PLANNING



and hence cost. In general terms, surveys conducted       should not attempt to describe the general ecology
for EIA studies should aim to provide information         of a site. Unfortunately, monitoring schemes often
on the following.                                         resort to measuring a wide variety of variables,
                                                          which may or may not be related to the questions
*   What species and habitats occur ( ¼ the resource)?
                                                          that need to be addressed. As a result, resources
*   Where do they occur?
                                                          may be spent collecting unnecessary data. Even
*   How many of them are there or how much of the
                                                          worse, it may be found that key questions cannot
    habitat is there?
                                                          be answered with the information obtained. This is
*   How does this amount of the resource relate to
                                                          because monitoring is often planned backwards,
    that existing in the wider area/biogeographical
                                                          on a ‘collect-now (data), think-later (of a useful
    region?
                                                          question)’ basis (Roberts, 1991).
*   What are the seasonal changes and when is the
                                                             Strictly speaking, the minimum requirement of
    most susceptible or sensitive period for these spe-
                                                          monitoring is an assessment of adherence to, or
    cies/habitats?
                                                          deviation from, formulated standards. However, it
Monitoring is often loosely regarded as a pro-            is clearly desirable to collect data in such a way
gramme of repeated surveys in which qualitative           that gradual change can be detected to assist man-
or quantitative observations are made, usually by         agement decision-making. Management adjust-
means of a standardised procedure. However, by            ments (at both field and policy level) require
itself this is merely surveillance as there is no pre-    knowledge of the dynamic situation, i.e. whether
conception of what the findings ought to be.              the feature is moving towards or away from the
Monitoring can be more rigorously defined as              standard, from which direction, and whether the
‘intermittent (regular or irregular) surveillance         change is expected, acceptable or otherwise
undertaken to determine the extent of compliance          (Rowell, 1993).
with a predetermined standard or the degree of               Monitoring should not be confused with research
deviation from an expected norm’ (Hellawell,              aimed at investigating ecological processes.
1991) . In this context, a standard can be a baseline     Nevertheless, data collected for monitoring purposes
position (e.g. maintenance of the existing area of a      can sometimes also be used to examine possible
particular habitat or population of a particular spe-     causes of change and to investigate the relationship
cies) or a position set as an objective (e.g. mainten-    between features of interest and environmental vari-
ance of more than 200 ha of a desired habitat or          ables and pressures. Such information can then be
more than 200 individuals of a desired species).          used to formulate appropriate responses. For exam-
   Thus, whereas surveys and surveillance are to a        ple, comparison of sward composition with stocking
large extent open-ended, a monitoring programme           density may predict optimal management regimes.
has a specific purpose that requires the standard to      Further monitoring of the vegetation and stocking
be defined or formulated in advance. This requires        rates can then confirm whether management and
the identification of interest features (e.g. various     habitat objectives are being met.
habitats and species), their attributes (e.g. area,          Thus, in summary, monitoring can:
numbers, structure and reproductive success) and
their target state, i.e. the standard that is to be       *   establish whether standards are being met;
monitored (see Glossary for detailed definitions of       *   detect change and trigger responses if any of the
monitoring terms). Monitoring for conservation                changes are undesirable;
purposes should be closely linked to site manage-         *   contribute to the diagnosis of the causes of
ment and should test whether conservation and                 change; and
management objectives have been achieved, as              *   assess the success of actions taken to maintain
outlined in Figure 1.1.                                       standards or to reverse undesirable changes,
   The monitoring programme and methods cho-                  and, where necessary, contribute to their
sen must be focused and fit for their purpose and             improvement.
                                                               1.1 The purpose of surveying and monitoring     5




           Figure 1.1. A schematic representation of the relationship between site management and
           monitoring.

Monitoring should therefore be an integral part of       Designated Sites (JNCC, 1997) has been adopted by the
all conservation programmes.                             agencies and the Joint Nature Conservation
                                                         Committee (JNCC). This formalises the monitoring
                                                         principles outlined above and provides standards for
1.1.2     Common Standards Monitoring in
                                                         the setting of objectives, judging the condition of site
          the UK
                                                         features, recording activities and management mea-
The UK statutory conservation agencies (the              sures, and monitoring and reporting within an
Countryside Council for Wales, English Nature, the       agreed time-frame.
Environment and Heritage Service in Northern                For further information on the Common
Ireland, and Scottish Natural Heritage) have under-      Standards approach see Rowell (1993, 1997) and
taken to monitor statutory protected sites to deter-     Brown (1994). See Shaw & Wind (1997) for a discus-
mine whether the features of interest for which each     sion of monitoring European conservation sites.
site has been designated are being maintained in a       Detailed guidance on the interpretation and appli-
favourable condition. To provide a basic framework       cation of Common Standards Monitoring has been
that will ensure consistent monitoring throughout        prepared by the statutory agencies and is available
the UK, a Statement of Common Standards for Monitoring   from them.
2       *    Planning a programme




The major steps involved in planning and execut-              As there is clearly a link between habitat and
ing a monitoring programme are illustrated in              species features, there is often likely to be some
Figure 2.1. Many of the aspects are relevant to            overlap between their monitoring requirements.
planning and executing a survey. A list of key con-        Species, particularly plants, are often essential com-
siderations that must be addressed when planning           ponents that define a habitat (e.g. ericoid shrubs on
a monitoring programme is given in Box 2.1 with            heathlands). Individual species or species assem-
the relevant section numbers. All of these issues          blages may therefore often be monitored as attri-
should be carefully considered in a step-by-step           butes of a habitat feature.
process before any fieldwork is started.                      In addition to monitoring species for which sites
                                                           have been designated, it is important to monitor
                                                           the area and quality of suitable habitat for such
2.1         SETTING THE OBJECTIVES FOR THE
                                                           species. There may also be other species that,
            MONITORING PROGRAMME
                                                           although not necessarily of conservation concern
Clearly and explicitly defining your objectives is         in themselves, may require monitoring by virtue
probably the most important single step of any             of association with a species that is a feature of
monitoring programme. Failure to do so may render          interest (for example, the food plant of a particular
any results gained inappropriate to the question you       animal species). Monitoring such habitats and asso-
wished to address, and therefore useless. Carefully        ciated species can give extra information about the
defining your objectives will also allow you to select     condition of species features that may prove useful
the most appropriate methodology. In particular it         for formulating management options for the site.
is essential that you ask yourself: What do I really          Some sites may be important for the presence
need to know? The process of defining objectives           of a species assemblage (e.g. a diverse community
underpins good site management principles and the          of insects or a good example of a particular vegeta-
development of management plans (see, for exam-            tion community). For these assemblages, it may be
ple, CCW, 1996) of which monitoring should be an           possible to monitor one or more indicator species,
integral part (Figure 1.1). Guidance on establishing       which can be used to infer the presence or status
clearly defined objectives is provided below.              of other associated species, rather than monitor-
                                                           ing each individual species. However, the use of
                                                           indicator species should be approached with care,
2.1.1        What features of conservation                 and in particular should only be relied on when
             interest are to be monitored?                 the relationship between the condition of the indi-
The first step in defining the objectives of any eco-      cator and that of the interest feature has been
logical monitoring programme must be the identi-           proven and quantified. If this is not the case,
fication of features of interest on the site. Biological   then all relevant species will need to be moni-
features may be habitats, species or species               tored. See Rowell (1994) for further guidance on
assemblages.                                               the use of indicators.


# RPS Group plc and Scottish Natural Heritage 2005.
                                                                                     2.1 Setting objectives   7



                        Identify the features
                                                       Citations, Site Management
                      that should be monitored
                                                             Statements, etc.
                             on the site


                       Select attributes for             Habitats: Part II Chapter 5
                          each feature                 Species: Part III Chapters 11–26



                          Define limits or
                        targets for attribute


                                                              Part I Section 2.2
                       Select methods for
                                                         Habitats: Part II Chapter 6
                     monitoring each attribute
                                                       Species: Part III Chapters 11–26


                    Repeat for other attributes
                         of the feature


                     Devise sampling strategy
                        where necessary                       Part I Section 2.3



                           Collect data


                           Analyse data                       Part I Section 2.6


                        Determine whether
                        attributes achieve
                             targets set


                     Once all attributes have
                    been assessed, determine
                        feature condition


                     Repeat for other features


                     Act on findings if features
                    not in acceptable condition

           Figure 2.1. A schematic diagram of the steps involved in a monitoring programme.


  The monitoring of assemblages presents some          Assemblages can be assessed by using species rich-
problems. On a site important for its diverse beetle   ness or diversity indices; judgement will be required
community, for example, does the loss of one           to decide how to set limits for these.
species constitute serious damage, or do several          In general, an essential part of monitoring a
species need to decline before the assemblage is       species of conservation concern will be to monitor
considered to be in an unacceptable condition?         the area of suitable habitat, and an essential part of
8      2 PLANNING A PROGRAMME




    Box 2.1 A checklist of considerations
                                                              DESIGNING A SAMPLING STRATEGY (2.3)
    during the preparation of a monitoring
                                                              Has the method been thoroughly tested and are
    programme
                                                              preliminary field trials necessary? (2.3.1)
                                                              Is the method sufficiently precise? (2.3.2)
    SETTING OBJECTIVES FOR THE MONITORING                     Should sample locations be permanent or not? (2.3.3)
    PROGRAMME (2.1)                                           When should the data be collected? (2.3.6)
    What features of conservation interest are to be          How will consistency be assured? (2.3.7)
    monitored? (2.1.1)
    What is the objective for each feature? (2.1.2)           REVIEWING THE MONITORING PROGRAMME (2.4)
    What attributes define condition in these features and    Are there sufficient long-term resources available? (2.4.1)
    what are likely to be their acceptable limits? (2.1.2)    Are personnel sufficiently trained and experienced?
    How often should monitoring be carried out? (2.1.3)       (2.4.2)
    What are the operational and/or management objectives     Are licences required? (2.4.3)
    for the site? (2.1.4)                                     Is specialist equipment required and available? (2.4.4)
    Are there external factors that may have significant      Are there health and safety issues to consider? (2.4.5)
    impacts on the site? (2.1.5)
    What monitoring has been undertaken, and are baseline     DATA RECORDING AND STORAGE (2.5)
    surveys required? (2.1.6)                                 How will data be recorded in the field? (2.5.1)
    Should the site be subdivided into monitoring units?      How will the data be stored? (2.5.2)
    (2.1.7)                                                   Who will hold and manage the data? (2.5.3)


    SELECTION OF METHODS FOR MONITORING                       DATA ANALYSIS, INTERPRETATION AND
    EACH ATTRIBUTE (2.2)                                      REVIEW (2.6)
    Is the method likely to damage the environment? (2.2.1)   Who will carry out the analysis and when? (2.6.1)
    Are samples required? (2.2.2)                             How will the data be analysed? (2.6.2)
    Will the method provide the appropriate type of           What statistical tests are appropriate to analyse the
    measurement? (2.2.3)                                      data? (2.6.4)
    Can the method measure the attribute across an            Is transformation of the data necessary before statistical
    appropriate range of conditions? (2.2.4)                  analysis? (2.6.4)
    Is the method prone to substantial measurement            What statistical packages are available for the analysis of
    error? (2.2.5)                                            data? (2.6.6)


monitoring a habitat will involve the monitoring of
                                                              2.1.2           What is the objective for each
its constituent species.
                                                                              feature?
   Identifying notified features should be straight-
forward for Special Areas of Conservation (SACs)              For each interest feature to be monitored,
and Special Protection Areas (SPAs), as a list of fea-        an objective should be defined that identifies
tures is drawn up during the designation process.             appropriate attributes of the feature and, where
Identification of notified features may be more dif-          possible, sets a target for each one. Each target
ficult on Sites of Special Scientific Interest (SSSIs) for    may include an upper and a lower limit, within
which the citation may be imprecise or based on an            which the feature is considered to be in acceptable
early version of the selection guidelines. For clarifi-       condition.
cation, refer to the guidelines for the selection of            Attributes of a habitat may reflect a number of
biological SSSIs (NCC, 1989; Hodgetts, 1992; JNCC,            properties of the feature, including aspects of
1994) and contact the relevant country agency.                quantity (e.g. size or number of individuals),
                                                                                             2.1 Setting objectives      9




  Box 2.2 Examples of attributes that may be                 Quality: dynamics
  used to define the condition of habitats and               succession
  species                                                    reproduction or regeneration
                                                             cyclic change and patch dynamics
  HABITAT ATTRIBUTES
  Quantity                                                   Quality: function
  area                                                       physical and biochemical (e.g. soil stabilisation, carbon
  Quality: physical attributes                               sinks)
  geological (e.g. presence of bare rock or deep peat)       ecosystem (e.g. net producer)
  water (e.g. presence of open water or depth of water
  table)
                                                             SPECIES ATTRIBUTES
  Quality: composition                                       Quantity
  communities                                                presence/absence
  richness or diversity                                      range
  typical, keystone or indicator species                     population size
  presence–absence                                           frequency
  frequency                                                  number/density
  number or density                                          cover
  cover
                                                             Population dynamics
  biomass
                                                             recruitment
  Quality: structure                                         mortality
  inter-habitat (landscape) scale (e.g. fragmentation,       emigration
  habitat mosaics)                                           immigration
  intra-habitat scale
  macro-scale                                                Population structure
    horizontal (e.g. plant community mosaics)                age
    vertical (e.g. ground-, shrub- and tree-layer            sex ratio
    topography)                                              fragmentation or isolation
  micro-scale                                                genetic diversity
    horizontal (e.g. patches of short and tall vegetation)
    vertical (e.g. within-layer topography)                  Habitat requirements



composition (presence of particular species, over-           (e.g. breeding success and population structure) (see
all diversity, etc.), structure, function or dynamics        Box 2.2). In most cases, direct monitoring of species
(Box 2.2). These principles are outlined below.              will generally be targeted towards measuring range
There is further discussion of attributes that               and abundance; more detailed studies may be con-
define the condition of specific habitat types in            strained by a lack of resources or appropriate skills.
Chapter 5.                                                   The costs involved in monitoring population struc-
   Species attributes for which targets may be set           ture, for example, can be particularly high. It should
include range, abundance, population dynamics                be borne in mind that in some cases (for example,
and habitat requirements. Part III describes methods         monitoring bryophytes in fragile habitats), quanti-
for monitoring range (presence–absence across a              tative monitoring may damage the habitat and
site), abundance (population density) and dynamics           hence the species, and is therefore not feasible.
10    2 PLANNING A PROGRAMME



   The setting of targets and limits for attributes is     or abundance of typical species or vegetation
outside the scope of this Handbook as these are            communities.
dependent on local site conditions. The UK statu-            Typical species are hard to define, but Shaw &
tory agencies have produced guidance on this for           Wind (1997) suggest the following:
the purposes of Common Standards Monitoring.
                                                           *   species on which the identification of the habitat
                                                               is founded;
Habitat attributes
                                                           *   species that are inseparable from the habitat;
Quantity
                                                           *   characteristic species;
Quantity may be the simplest attribute of a habitat
                                                           *   species that are consistently present but not
in terms of indicating its condition. However, in
                                                               restricted;
many situations habitats and communities are not
                                                           *   species that are an integral part of the habitat; and
objectively or precisely definable and there is con-
                                                           *   keystone species (Jermy et al., 1996), which signifi-
sequently some doubt about where boundaries lie.
                                                               cantly influence the habitat’s structure and func-
This can make habitat quantification and interpre-
                                                               tion. (Note: such species may include animals as
tation of change difficult. None the less, especially
                                                               well as plants.)
for EIA studies, this is important if habitat area is to
be lost and needs to be replaced according to some         Diversity indices (Magurran, 1983) are not normally
criteria.                                                  recommended for habitat condition monitoring as
                                                           the setting of targets and interpretation of changes
Quality: physical attributes                               in these indices is difficult.
Certain physical attributes of a habitat can be               In some cases it may be appropriate to monitor
considered to be essential or desirable in their           ‘indicator species’. The presence and/or abundance
own right. For example, the presence of peat is an         of such species may be used to indicate favourable
essential attribute of blanket bog. Similarly, the         or unfavourable ecological conditions that may
presence of grikes is a characteristic attribute of        be difficult or costly to detect by other means. For
limestone pavements.                                       example, aquatic plants can be used as indicators of
   It is often difficult to decide whether physical        overall water quality (Palmer et al., 1992). Care
properties are direct attributes of a habitat or           should be taken with the use of indicator species,
factors that may influence it. For example, are the        however, as they may not always be reliable
chemical characteristics of river water (e.g. nutrient     (Rowell, 1994).
status and pH) attributes or factors that influence           There are a number of parameters that may be
other aspects of the habitat such as macrophytic           appropriate for target setting and measurement
communities? In principle, in habitats in which            when monitoring the abundance of typical (or
such distinctions are difficult, key factors that may      other) species. These are described below.
influence the habitat should be monitored.
                                                           Presence or absence
Quantity: composition                                      The simplest target for a species is that its presence
The composition of a habitat in terms of its com-          at the site, or at a defined location within it, is
munities and species is a fundamental attribute of         maintained. This is normally straightforward to
habitat condition. Many statutory sites are notified       monitor, but there are occasions when difficulties
because of the presence of particular vegetation           may arise: for example, for species that are incon-
communities and therefore monitoring should                spicuous, difficult to identify or rare, or those that
ensure that targets for these are being met.               inhabit inaccessible areas.
   Monitoring all species is clearly not feasible in          The distribution (range) of a species across a site
all but the simplest habitats. Therefore, the most         can be monitored by assessing presence–absence
commonly used species-based attributes of habitat          across a number of locations (e.g. grid squares),
composition are species richness and the presence          and distribution maps can be drawn up for such
                                                                                       2.1 Setting objectives   11



surveys. Repeat presence–absence surveys can indi-         monitoring because of the difficulties of defining
cate expansions or contractions in range.                  individuals of clonal or rhizomatous plants (White,
                                                           1979) and the amount of time required to count
Frequency                                                  numbers accurately in large sample sizes. However,
Frequency is the proportion of quadrats (or other          sub-plot frequency is often used as a quicker alter-
sample units) examined in which the species is             native. Densities depend on reproduction, disper-
present. Frequency is a simple, quantitative mea-          sal, population ages, etc., which may vary from
sure, and has been widely used to describe relative        year to year. These annual variations in population
abundance. With a large number of sampling                 sizes mean that samples have to be recorded regu-
units of sufficiently small size, frequency estimates      larly to separate normal fluctuations from direc-
of plant species can approximate to cover (see             tional change.
below). For plants, there are two measures of fre-            Density estimates can be converted to total
quency: shoot frequency (the presence of any foli-         population size estimates by multiplying the den-
age within the quadrat) and root frequency (the            sity by the area of similar habitat. Alternatively,
presence of rooted individuals only). Frequency            total population counts over an area may be used
estimates depend on the size of the quadrats and           to derive density. Extrapolating density estimates
of individual plant species (large plants may be           from a smaller area to a larger one is only mean-
over-represented compared with small plants) and           ingful if the larger area has the same characteristics
the spatial distribution of individuals of a species       as the area from which the density was originally
(clustered species may be under-represented com-           estimated. When making such extrapolations you
pared with more widely spaced ones). Frequency             need to be sure that all individuals are detected or
measures may also exaggerate the apparent bio-             that a detectability function can be estimated: see
mass of small species and hence overestimate               Section 10.6 for more details.
their functional significance.
   Changes in frequency are relatively insensitive         Cover
to seasonal or management changes, and therefore           Cover is a measure of the area covered by the above-
a large sample size is required to be effective for        ground stems and foliage of a plant species when
monitoring change in the short term. However,              viewed from above. Greig-Smith (1983) defined
frequency estimates are relatively free of observer        cover as ‘the proportion of ground occupied by a
error and hence are particularly useful for general        perpendicular projection onto it of the aerial parts
habitat condition monitoring purposes.                     of individuals of the species’. The sum of cover
   A useful extension of the simple frequency mea-         values from all species in layered vegetation often
sure is to record presence–absence within subdivi-         totals more than 100%. Cover is usually described
sions of each plot. For example, a plot may be divided     as a percentage, or by using one of the numerous
into a 5 Â 5 grid giving 25 subdivisions. The measure      categorical indices available (see Shimwell, 1971).
recorded is the proportion of subdivisions contain-        The most widely used of these is the Domin scale as
ing the species of interest. This will be more sensitive   used in the National Vegetation Classification
to change than simple frequency and is often quicker       (NVC) methodology (Rodwell, 1991 et seq.). (Box 2.3)
to record than cover. Within this Handbook, this           Cover estimates provide a good description of the
measure is referred to as sub-plot frequency.              contributions of each species to the vegetation; as
                                                           long as measurements are accurate, they are sensi-
Density                                                    tive to short-term fluctuations in season or man-
Density is the number of individuals per unit area         agement. However, cover estimates, whether
(e.g. plants within the habitat). Counts of numbers        percentages or scales, are prone to bias and con-
of individuals in quadrats have been widely used           siderable care is required to ensure accuracy and
for demographic studies, but less so for vegetation        consistency.
12        2 PLANNING A PROGRAMME




     Box 2.3 The Domin scale
                                                          5             11%–25%
     Domin scale   Equivalent percentage cover            4             4%–10%
     10            91%–100%                               3             <4% – frequent
     9             76%–90%                                2             <4% – occasional
     8             51%–75%                                1             <4% – rare
     7             34%–50%                                +             Insignificant: normally 1–2 individuals
     6             26%–33%                                              with no measurable cover




                                                          complexity increases with decreasing scale, the sett-
Biomass
                                                          ing of meaningful targets and their measurement
Biomass is the contribution of each species mea-
                                                          becomes increasingly difficult. The inclusion of
sured by weight (both fresh weight or dry weight
                                                          microstructural attributes of habitats is therefore
can be measured). Biomass estimates are useful
                                                          beyond the scope of most monitoring programmes.
for assessing productivity, but normally require
destructive sampling where vegetation is concerned.
                                                          Quantity: dynamics
Usually, only aerial parts of plants are collected.
                                                          Habitats are never static and therefore monitoring
                                                          must ensure that essential dynamic processes are
Quality: structure                                        functioning adequately. For example, vegetation
Structure is an important attribute of a habitat that     surveys can indicate insufficient regeneration in
can be measured at a variety of scales. Inter-habitat     individual species or whole vegetation commu-
or landscape-scale structure may be important             nities. Some habitats will require areas that are
where, for example, fragmentation alters habitat          temporarily altered, for example by fire or storms,
structure and ecological processes and thereby            to allow regrowth of young vegetation. This is
reduces the suitability of an area of habitat for         particularly important in woodlands, where gap
typical species. In contrast, some ecological pro-        dynamics affect the species composition and struc-
cesses and species may require mosaics of differing       ture of stands.
habitats.                                                    In some cases it is important to ensure that suc-
   Intra-habitat structure is also often a fundamen-      cessional habitats change in specific directions and
tal attribute of habitat condition, which itself occurs   at desirable rates.
at a variety of scales. At the larger scale, vegetation
community mosaics may be a distinctive feature of         Quantity: function
a habitat (e.g. bog pool and hummock communities          Processes (e.g. peat formation or dune formation) are
in some blanket bogs). Vertical structure is also         highly important ecosystem functions. However,
important, especially in woodlands, where three           such processes are difficult to define and even
or more layers of ground flora, shrubs and canopy         harder to monitor and assess. Thus, although these
vegetation may be found.                                  are important attributes, it is frequently impractic-
   There is probably no practical lower limit to the      able to use them for monitoring habitat condition.
size of significant structural variation within habi-
tats, as micro-scale variations in vegetation cover,      Species attributes
height and layering may be important components           Quantity
of habitat condition, especially for species with         Species may have minimum viable population sizes
specific habitat requirements. However, as the                 ´
                                                          (Soule, 1987): the number of individuals below
potential variety of structural attributes and their      which the population cannot persist. Minimum
                                                                                    2.1 Setting objectives   13



viable populations should therefore provide the         *   number or density
basis for establishing minimum quantities (lower        *   cover (for plants)
limits) for species. Unfortunately it is widely felt
                                                        These are discussed under the assessment of habi-
that there are usually too many unknowns (in
                                                        tat composition above. The choice of which mea-
terms of both theory and data) to make judgements
                                                        surement to use will depend on the characteristics
on what those lower limits might be. Furthermore,
                                                        of the target species (see Part III).
the use of the concept is complicated by the fact
                                                           It should be remembered when setting objec-
that sites will not normally hold discrete and iso-
                                                        tives for the abundance of a species at a site that
lated populations.
                                                        it is often very difficult and time-consuming to
   Minimum desirable populations (the minimum
                                                        establish absolute population numbers, whether
population that is considered to be safely above
                                                        by total counts or by sampling (see Section 2.3).
the minimum viable population) for species are
                                                        Objectives for abundance should be based on sim-
identified by judgement and consensus. Because
                                                        ple and efficient population index-based measures
they are normally based on estimates it is advisable
                                                        if such assessments are adequate for defining
to adopt the precautionary principle, so they should
                                                        condition.
always be well above the likely range of any mini-
mum viable population.
                                                        Population dynamics
   Trends in population size and range are there-
                                                        The dynamics of, and overall trends in, population
fore central to the conservation status of a species
                                                        size depend on the balance between recruit-
and are also relatively simple to monitor. Species
                                                        ment, mortality, emigration and immigration.
that are decreasing as a result of habitat changes
                                                        Recruitment and mortality are of particular impor-
may become rarer by two means:
                                                        tance to the condition of the population. These
1. restriction of their geographical range;             must at least balance if the population is to remain
2. reduction in their density.                          stable and hence in acceptable condition. However,
                                                        the population at a site may appear stable, yet may
In an ideal situation, a species’ condition on a        be dependent on immigration to offset poor pro-
site would be judged by determining whether it          ductivity. As such, if deaths are greater than pro-
had achieved targets set for its population size,       ductivity the population acts as a ‘sink’ and this can
rate of change, and levels of recruitment and mor-      be regarded as being in an unacceptable condition
tality (Davies & Yost, 1998). However, there are        (but see Figure 2.3). Conversely, a population may
generally insufficient data or resources to allow       act as a ‘source’ if more young are produced than
measurement of recruitment or mortality, and            are able to breed at the natal site. When consider-
too little information on ‘acceptable’ rates of popu-   ing the balance of recruitment and mortality,
lation change to be able to use this as part of an      source and sink populations therefore need to
objective. Natural variation in populations is also     be taken into account. To ensure a long-term
poorly documented and therefore difficult to incor-     underlying favourable population trend (i.e. stable
porate into definitions of condition for many spe-      or increasing) and to ensure that populations are
cies. One example in which population change            self-sustaining, it is desirable to measure recruit-
rates are being used at the site level, however, is     ment and mortality.
in the application of ‘alert limits’ for wildfowl          The following variables may have a bearing on
populations (Atkinson et al., 2000).                    the population dynamics of a species, and can be
   Measures for directly or indirectly assessing the    used as measurable attributes of species condition
population size of a species are:                       where appropriate:
*   presence–absence                                    *   number of offspring produced by parent(s),
*   range                                                   e.g. seedling germination for trees or the number
*   frequency                                               of young fledged per pair for birds;
14      2 PLANNING A PROGRAMME




                     Time 1: Eight areas of suitable habitat,
                     for a species, four of which are                                                               Sink
                     occupied (grey circles)                            Sink



N            N                                                                             Source
         X           Time 2: Two colonies have become extinct
                 X   (X); the species has colonised two more
                     areas (N), with the overall population
                     size remaining stable

                                                                         Sink                                 Source
                     If some of the habitat areas are lost
                     (black circles) and the species cannot
                     disperse over a sufficient distance, isolated
                     colonies cannot be recolonised if a             Figure 2.3. Source and sink populations. The
                     population becomes locally extinct.
                     The species is therefore at a greater           source–sink model, (Pulliam, 1988) is an elaboration
                     risk of extinction in all remaining areas.      of the metapopulation model, in which habitat
                     In the example here, only two of the
                     remaining populations are close enough          patches vary in their quality and hence their ability to
                     to allow dispersal. Populations at the other    support populations. Patches of good-quality habitat,
                     two areas cannot be replaced if they die out.
                                                                     known as sources, have a net positive population
Figure 2.2. Metapopulation structure and the effects                 growth. Patches of poorer habitat, known as sinks,
of habitat fragmentation.                                            have a net negative population growth. Source
                                                                     populations increase until their carrying capacity is
                                                                     reached, whereupon individuals disperse to sink
*    longevity; and                                                  habitats. Without this dispersal, the sink populations
*    mortality rates (these may vary at different life               would die out. However, sinks provide habitat for
     stages: mortality of young is often higher than                 surplus individuals from sources, and these
     that of adults).                                                individuals can recolonise sources in the event of
                                                                     extinctions of source populations. The existence of
Population structure                                                 sinks therefore increases total population size and the
Age and sex                                                          persistence of both source and sink populations.
The ratio of different age classes of a population
may be an attribute that defines condition. For                      habitat fragmentation. Such small sub-populations
example, if the mean age of a population is                          have a much greater chance of extinction because
increasing, it indicates that recruitment is prob-                   of random demographic accidents and local envir-
ably failing; if such a trend persists, the popula-                  onmental variations. Therefore, persistence of
tion may go into decline (if it is not already doing                 such sub-populations may be dependent on other
so). The sex ratio can also be important, particu-                   viable breeding sub-populations occurring within
larly if mortality rates vary between the sexes.                     the effective dispersal distances undertaken by
This is often the case for species that exhibit                      immigrating and emigrating individuals. Thus,
marked behavioural differences between males                         the population as a whole exhibits a metapopu-
and females. For example, female bats are more                       lation structure (see Figure 2.2).
at risk of catastrophic mortality when they gather                      Although some species such as the Marsh
in maternity roosts.                                                 Fritillary butterfly Eurodryas aurinia are thought
                                                                     to exhibit metapopulation structures, understand-
                                                                     ing of the processes within such populations is
Fragmentation or isolation                                           generally poor. It is therefore not normally possible
Many populations are composed of a number of                         to define condition reliably with respect to meta-
partly isolated sub-populations, often as a result of                population structure attributes in such species.
                                                                                       2.1 Setting objectives   15



Instead, the general aim should be to ensure that         availability of suitable micro-habitats is important
wherever feasible all sub-populations are ‘source’        for many invertebrates, lichens, fungi and other
populations, through provision of ‘optimal’ habitat       species. Guidance for monitoring micro-habitats
conditions, as opposed to ‘sink’ populations              is not specifically provided in Part II; however, pro-
(Figure 2.3).                                             vided the characteristics of the micro-habitat are
                                                          known, it should be possible to adapt the methods
Genetic diversity                                         in Part II for monitoring some aspects of these. The
Conservation strategies should be aimed at conser-        abundance and availability of prey species are also
ving genetic variability within species as well as        components of habitat quality: guidance on mon-
conserving the species themselves. The assessment         itoring these may be obtained in Part III.
of genetic diversity in detail is a scientifically
complex procedure, which is beyond the scope
                                                          2.1.3      How often should monitoring be
of a general species monitoring programme.
                                                                     carried out?
However, genetic diversity can be conserved by
ensuring that separate populations or races of spe-       The frequency with which monitoring should be
cies are conserved. For some species, morphologi-         carried out should be established at an early stage
cally distinct races can be identified without the        in the development of a monitoring programme.
need for complex analyses (e.g. the Fair Isle Wren        Although, within the conservation agencies’
Troglodytes troglodytes fridariensis); conservation and   Common Standards framework, notified features
monitoring can be targeted towards these. For the         need only be assessed once every six years, it is highly
majority of species, genetic diversity does not often     desirable that monitoring be carried out with suffi-
manifest itself physically between populations, but       cient frequency to detect changes before they result
conservation of separate populations will help to         in a feature’s condition becoming unacceptable. At
conserve within-species diversity. The introduction       the very least, it is clearly essential that monitoring
of individuals from outside the natural range of a        be frequent enough to ensure that changes are
species should generally be avoided, as this can          detected before they become irreversible.
reduce the genetic distinctions between popu-                The timescale over which changes are likely
lations (for example, many lowland chalk streams          to occur and be detectable will vary according to
formerly contained endemic races of Brown Trout           the feature in question. In particular, the intrinsic
Salmo trutta, most of which have been lost through        rate of change is of fundamental importance. For
the interbreeding of races following re-stocking          example, major structural changes may normally
with and competition from introduced Rainbow              be very slow in a woodland but potentially rapid
Trout Oncorhynchus mykiss). However, it could also        in sand dunes. Long-lived species (e.g. perennial
be argued that isolated and inbred populations may        plants such as trees) will exhibit changes in abun-
benefit from the introduction of genes from other         dance over a much longer period than will short-
populations.                                              lived species or species that live in ephemeral
                                                          habitats. Attributes also vary according to their
Habitat requirements                                      likely rate of change. For example, the extent of a
The EU Habitats Directive recognises that favour-         habitat may change very slowly, whereas typical
able conservation status is dependent on the avail-       species of conservation importance may decline
ability of sufficient habitat. Although these are         rapidly as a result of inappropriate management.
not strictly attributes of a species, habitat quality     General indications of the optimum timescale for
and quantity should therefore also be taken into          monitoring various attributes of specific habitats
account and monitored when defining condition             according to their intrinsic rate of change are pro-
of a species. General aspects of habitat condition        vided in Chapter 5. Recommended frequencies of
monitoring, such as physical vegetation type and          species monitoring are given in the sections on indi-
structural attributes, are covered in Part II. The        vidual species.
16      2 PLANNING A PROGRAMME



   Extrinsic factors may also influence a feature;            Habitats and species likely to change over time
monitoring programmes should therefore incorpo-            as a result of planned management actions should
rate sufficient flexibility to deal with unforeseen        be identified, as well as the likely timescale of
and potentially rapid and catastrophic events (e.g.        change in years. For example, it may be a manage-
storms and fires). Additional monitoring may be            ment objective to coppice a woodland area for
required to establish the condition of a site after        invertebrates and ground flora. Ideally, the two
such events. More extensive and detailed surveys to        groups should be sample-surveyed before and
establish new baseline conditions may be neces-            after management, and for a series of years up to
sary if damage has been extensive and features             canopy closure again. Monitoring of habitats and
have been partly or wholly destroyed.                      species should tie in directly with the management
   In addition to these biological considerations,         objectives and actions. This is important if the cost-
determination of the frequency of assessment should        effectiveness of management is to be maximised
also take into account the minimum required report-        and objectives achieved.
ing frequency and available financial resources.
Therefore, an appropriate procedure for determining
                                                           2.1.5      Are there external factors that may
surveillance frequency might be as follows,
                                                                      have significant impacts on the site?
1. Select an interval consistent with:
                                                           Habitat and species condition may also be affected
   * the intrinsic rate of change of the feature, taking
                                                           by external factors, such as airborne pollution or
     into account the precision with which that
                                                           climatic change. These may therefore also require
     change can be measured (see Sections 2.2.3 and
                                                           monitoring. However, because of the large scale of
     2.2.4);
                                                           some of these processes, monitoring may only be
   * the timescale dictated by reporting require-
                                                           feasible at a selection of sites. Data from existing
     ments; and
                                                           monitoring schemes (e.g. Meteorological Office
   * the availability of funds for surveillance.
                                                           weather stations) may also be suitable. The avail-
2. Aim to make a detailed assessment of the attribute at
                                                           ability of existing data on such factors should there-
   the required interval (for example, for woodland
                                                           fore be carefully investigated before including
   area, aerial photography may be required at inter-
                                                           them in monitoring programmes.
   vals of 10 years).
                                                              Detailed descriptions of methods for monitoring
3. Assess the risk of change from external factors.
                                                           management actions and external impacts are
Aim to make a basic inspection of the attribute more       beyond the scope of this Handbook. However, brief
frequently for signs of abrupt change due to extrin-       summaries of key management measures and
sic factors (for example, for woodland area, a basic       external factors influencing habitats are provided
inspection at intervals of three years may be              in Chapter 5, together with sources of further
appropriate).                                              information.


2.1.4      What are the management                         2.1.6      What monitoring has been
           objectives for the site?                                   undertaken and are baseline surveys
                                                                      required?
In addition to directly assessing the condition of
features of interest, monitoring should, where             A baseline survey is carried out to determine the
resources permit, establish whether management             habitats and species present on a site and their
objectives are being met. Ongoing management               current condition. If a baseline survey of features
objectives and associated actions should therefore         and their attributes has not been undertaken, this
be identified and appropriate monitoring methods           will be required before a detailed monitoring pro-
selected. Monitoring of management impacts is              gramme can be planned. It is necessary to establish
briefly covered in Chapter 5.                              the baseline levels of the various attributes so that
                                                                                       2.2 Selection of methods   17



any subsequent changes in these levels can be                *   one part of a feature is in particularly poor condi-
identified.                                                      tion and you wish to track its recovery.
   It is therefore important to establish from the
                                                             Unit boundaries must not cross feature boundaries;
outset whether monitoring has previously been
                                                             each unit should only encompass part of one fea-
undertaken at the site, including the attributes cov-
                                                             ture and each objective must apply to the feature as
ered, the methods used, the timescale and fre-
                                                             a whole.
quency over which it took place and whether or
                                                                A general discussion of monitoring complex
not it is ongoing. It is not unusual for the results of
                                                             sites is provided by Stone (1997).
monitoring studies to be forgotten, especially if they
are unpublished and the members of staff responsi-
ble have moved on. A careful and detailed investiga-
                                                             2.2       SELECTION OF METHODS FOR
tion may therefore often be worthwhile.
                                                                       MONITORING EACH ATTRIBUTE
   Available data from previous monitoring pro-
grammes or ad hoc surveys should be used to review           Once you have defined your objectives for the mon-
the appropriateness of methods and sampling stra-            itoring programme and decided which feature
tegies employed (see Section 2.3.1). Previous mon-           attributes are important, you should then decide
itoring programmes should not be followed                    on the most appropriate methods for monitoring
without careful consideration of their suitability,          each attribute that defines condition for each feature.
as they may have been established to meet differ-            Monitoring methods for habitat attributes are
ent objectives. However, if existing monitoring              listed in the tables for each habitat in Chapter 5
programmes are likely to contribute to current               and described in Chapter 6. Survey and monitoring
monitoring objectives they should be continued.              methods for species attributes are listed in the
Where appropriate, existing methodologies should             general methods tables for each species in
also be followed to maintain the validity of long-           Chapters 11–26. The methods described in these
term datasets. It may also be useful to use existing         sections are often specific to one particular group
fixed marker systems or permanent quadrats.                  of species. Chapter 10 gives a general introduction
   Take care over the use and interpretation of              to population monitoring and describes the theory
historical data that have not been properly docu-            behind the sampling methods most commonly
mented. Grid references may be unreliable, and               used. Although not comprehensive, the methods
maps sometimes differ between editions and                   given for both habitats and species are likely to be
scales. Round numbers (e.g. 100 plants) are usually          the most appropriate and efficient tried and tested
highly indicative of estimates. Similarly, national          methods currently available.
distribution maps are very poor indicators of the               Clearly, to maximise the efficient use of moni-
real status or change in a species (see Rich & Smith         toring resources the most cost-effective method
(1996) for a detailed review).                               appropriate to the monitoring objective should be
                                                             used. Thus, quick and cheap subjective methods
                                                             should be used if they are adequate. It is unaccep-
2.1.7        Should the site be subdivided into              table to use cheap methods, however, if they can-
             monitoring units?                               not detect all degrees of undesirable change. Such
                                                             methods may be a false economy, as in the long
For ease of assessment it may sometimes be advan-
                                                             term the financial cost of repairing damage to a
tageous to divide habitat features into ‘monitoring
                                                             feature is likely to exceed by far the costs of moni-
units’. This may be useful if:
                                                             toring and of early management intervention. It
*   features are too extensive or too fragmented to be       should be remembered that the closer an attribute
    surveyed adequately in one visit;                        is to the limits that define condition, the more
*   you wish to assess the effects of management prac-       precise the chosen method must be to determine
    tices that apply only to certain parts of the feature;   whether it is above or below the limit.
     FOR EACH ATTRIBUTE TO BE MONITORED
   CONSIDER THE MOST COST-EFFECTIVE METHOD
      (See Section 2.2 and appropriate sections in
               Chapter 5 (Habitats) and
              Chapters 11–26 (Species))



                                                                                 Consider the next most
                        Is the method:                                            cost-effective method
Unlikely to damage the species or environment ∗ (2.2.1)?

      Able to provide a type of measurement consistent
      with the target objectives for the species (2.2.3)?
    Able to measure the attribute across an appropriate                           NO
               range of conditions (2.2.4)?

      Able to provide sufficiently precise observations
      to detect appropriate scales of change (2.3.2)?




                           YES




       Is the method subject to significant bias (2.2.5)?             YES



                                                            Does the bias matter for
                                                             monitoring purposes
                          NO
                                                              if it is consistent?

                                                                       YES

                                                                Can the bias be
                                                YES          measured or controlled?          NO

             Are samples required (2.2.2)?                  Take direct measurements
                                                                of entire attribute


                          YES



               Design a sampling scheme
            (see Section 2.3 and Figure 2.6)

     Figure 2.4. Selection of methods for monitoring each attribute (see the relevant chapter
     or section for further information.)
     *This may depend on whether sampling uses permanent or temporary plots (see
     Section 2.3.3).
                                                                                  2.2 Selection of methods   19



   Selecting the most appropriate method is there-
fore an important step and needs to take
                                                           2.2.2      Are samples required?
into account the key considerations described              It is sometimes possible to make complete assess-
below (summarised in Figure 2.4). If the most cost-        ments of an attribute of a feature of interest at a
effective method fails any of these considerations,        site (e.g. by aerial photography if complete site
the next most cost-effective method should be              coverage exists, by Phase I mapping, or by counting
assessed.                                                  the total number of an easily detectable species). In
                                                           this case, complete measurements can be taken,
                                                           although the accuracy of the method will still
2.2.1       Is the method likely to damage
                                                           need to be considered when presenting the results.
            the environment?
                                                           However, it is seldom possible, or even necessary,
It is clearly essential that monitoring activities do      to do a complete count. Unless species are very
not damage features of conservation interest on a          rare, very conspicuous, or very localised, total
site. However, there are unfortunately numerous            counts (Section 10.1) can prove too expensive or
examples in which research and monitoring pro-             be prone to under- or overcounting. This is particu-
grammes have merely measured and recorded the              larly true when counting mobile species such as
damage caused by their own activities. Therefore,          birds, or animals and plants that have a large
great care should be taken to ensure that the meth-        range and inhabit remote sites where counting is
ods chosen will not cause any damage. In particular        either impossible or impracticable.
the following precautions should be strictly                   More commonly, it is only practicable to study
observed:                                                  a sample (i.e. part of a feature), and to generalise
                                                           from observations made in the sample to the
*   Ensure that the target attributes are not damaged      whole feature. Using sampling methods allows
    during sampling (e.g. by trampling) and that distur-   the researcher to invest more time in avoiding
    bance to other habitats and species is minimised.      the problems with measurement error (see
*   Be aware of other species or groups of conserva-       Section 2.2.5). Although there are a number of stan-
    tion importance in the area and take care not to       dard methods for certain species, when designing a
    cause disturbance to them.                             new study it is advisable to tailor monitoring to
*   Do not use vehicles on the site unless particular      achieve the most efficient and appropriate means
    tracks have been identified and impacts can be         of data collection. Some methods are designed to
    avoided.                                               standardise data collection by time (i.e. counting
*   Do not use destructive sampling methods unless         species for a set length of time) and some by space
    absolutely necessary. If they are used, minimise       (i.e. counting within a set area such as a quadrat).
    impacts and understand the extent and impor-           Choosing a method will also rely heavily on the
    tance of the impact to the ecological community        ecology of the species or habitat concerned.
    and the attribute being monitored.                     Sampling is covered in some detail in the next
*   Be able to relocate fixed quadrats and sampling        section.
    locations easily in successive years or sampling
    periods to minimise the need for walking over
    the site and potentially damaging the habitats
                                                           2.2.3      Will the method provide the
    you are going to monitor (see Section 2.3.3 and
                                                                      appropriate type of measurement?
    Appendix 5 for further discussions).
*   Position fixed sampling locations sensitively          The type of measurement produced by a method is
    and avoid or minimise damage during their              particularly important and must be consistent with
    establishment.                                         the objective for each attribute. For example, the
*   Avoid excessive revisiting of sites and sampling       target for dead wood volume may be expressed as
    locations.                                             ‘one or two large (>50 cm diameter) fallen trees or
20    2 PLANNING A PROGRAMME



trunks visible with plenty of 5–50 cm pieces in view      that have to be visited. Presence–absence data are
at each sample point’. This would require only            quick to collect but provide little information
that a subjective assessment is made at surveyed          about each location. Thus, large numbers of loca-
locations. If, however, the target is expressed           tions may have to be visited to build up a picture of
as ‘a mean of at least 10 m3 ha–1 of dead wood            the quantity and condition of a feature at a site.
across the site’, quantitative estimates based on         Quantitative data can provide much more informa-
density measurements at survey locations would            tion and usually require fewer locations to be vis-
be required.                                              ited. However each measurement will take rather
   The type of data being collected will also have a      longer to collect. Section 2.3 provides more infor-
significant impact on the survey design and range         mation on the design requirements for different
of analyses that can be carried out. The most com-        measurement types.
monly collected types of data are as follows.                Data collected for monitoring will be either a
                                                          direct or an indirect measure of the attribute. A
Nominal
                                                          direct measure involves making measurements of
Each survey location is assigned to a predefined
                                                          the attribute itself. An indirect measure involves
category. For example, a species may be recorded as
                                                          measuring a related variable, which is used to
being present or absent at a location, or a habitat may
                                                          infer the status of the attribute being monitored
be classified as a particular type of grassland.
                                                          (for example, using counts of otter spraints as an
Ordinal                                                   index of the number of otters present on a site).
An extension of nominal data in which the cate-           Such measurements are described as population
gories are ordered. Thus, for example, the                indices. An index of population size is also
abundance of a plant species at a location may            obtained from direct sampling of a subset of a
be classified in an ordered scale such as ‘rare’,         total population. For example, male moth popula-
‘occasional’, ‘frequent’, ‘abundant’ or ‘dominant’        tion size can be estimated by using pheromone
(the so-called DAFOR scale). This clearly provides        traps. These data are treated as an index of total
more information than a nominal measure such              population size, since one cannot be sure of the
as presence–absence, but may take longer to               numbers of females.
assess.

Quantitative                                              2.2.4      Can the method measure
To measure abundance we may instead actually                         the attribute across an appropriate
count the number of plants present or,                               range of conditions?
alternatively, what area of ground they cover. This       The method needs to be able to measure the attri-
provides a quantitative measure. Other examples are       bute fully across the range of states over which
the height of vegetation, the mass of an animal           condition is defined. It is essential that the method
and the number of species at a site. These provide        has appropriate limits of detection (i.e. the level
finer, more sensitive measures than ordinal data but      beyond which it is not possible to measure or dis-
may take longer to collect and may be prone to            tinguish between presence or absence). This is
measurement error. If necessary, quantitative data        most relevant to measurements of chemical con-
can always be converted to ordinal data by                centrations, but may also affect other attributes of
grouping the data into categories, e.g. 0–2 plants,       habitats and species. For example, the use of satel-
3–5 plants, etc.                                          lite-based remote sensing may be inappropriate
Further information on data types can be found in         for measuring changes in habitat at the site
Fowler et al. (1998).                                     level because of the limited spatial resolution
   If the site is to be sampled, there is usually a       obtainable.
trade-off between the type of data collected at              As another example, grapnel trawl surveys can
each sample location and the number of locations          be used to detect presence–absence of most aquatic
                                                                                 2.2 Selection of methods   21




           Figure 2.5. Bias in measurement. (a) Measurements are unbiased: estimates are
           distributed around the true value. (b) Measurements are biased: they tend to
           overestimate the true population.

macrophyte species but cannot be certain of detect-      this, it is important to know something about how
ing some small, fine-leaved or rare species. If          much error can be expected.
data are required for such species, an alternative          In the second set of results (Figure 2.5b) the
method may have to be used.                              observers tend to overestimate the true popu-
   On the other hand, for simple monitoring pur-         lation: these estimates are said to be biased. Bias is
poses there is little point in using a technique that    clearly of concern if we think the population is
measures an attribute well beyond its range of           much larger than it really is; if such error is pre-
acceptable condition. For example, it is not neces-      sent, we need to know about it. Bias is a systematic
sary to measure sub-surface water levels (e.g. by        source of error that results in under- or overestima-
dipwells) if condition is merely dependent on            tion of the attribute being measured. It causes esti-
water levels always being well above the surface.        mates to be inaccurate. Methods free of bias are
                                                         said to be accurate.
                                                            Of course, some observers may produce unbiased
2.2.5      Is the method prone to substantial
                                                         results in a given situation, whereas others may be
           measurement error?
                                                         biased. This possibility needs to be considered
Some measurement error is almost always unavoid-         if successive surveys are likely to be carried out
able in ecological studies and it is important to con-   by different people and accurate measurement of
sider whether such error is likely to affect the value   change is important.
and validity of the study. An example should help           Bias may arise from several sources in a study,
make this clear.                                         including:
    Suppose the population of Capercaillie Tetrao           Observer
urogallus at a site is estimated by several observers    * Incorrect identification of species.

independently. Each observer comes up with a dif-        * Failing to detect and count all individuals of a

ferent estimate because it is impossible to count the       particular species being surveyed.
numbers without error. Figure 2.5 illustrates two        * Different observers recording identical observa-

possible sets of results. In the first (Figure 2.5a),       tions in dissimilar ways.
the counts are all spread fairly evenly around the       * Different observers having expertise in different

true population and so on average the observers get         areas, which may affect their interpretation of the
it about right. The error in each observer’s estimate       observations they record.
may or may not be important, depending on how            * Variation    in observer effort (e.g. speed of
good an estimate is required. To be able to judge           assessment).
22      2 PLANNING A PROGRAMME



     Location                                              Hatton, 1990; Rich & Smith, 1996). These investiga-
                                                           tions show that:
*    Studying a species only where it is common; if
     areas in which it is rare have been ignored the       *   recorders are better at repeating their own work
     full dispersion of the species will not be under-         than that of others;
     stood. Estimates of total extent across the whole     *   results from small areas intensively searched are
     site cannot therefore be made.                            more repeatable than those from larger areas less
*    Using a small subjectively selected sample area           intensively searched;
     when the site being studied is not homogeneous.       *   large, broad-leaved or clumped taxa are better
     Habitat differences                                       recorded than small, well-dispersed or fine-leaved
                                                               taxa (Clymo, 1980; Sykes et al., 1983); and
*    Species of plant may be more easily detected in       *   quantitative work involving fully objective mea-
     some habitats than in others. For example, soil           surements is more repeatable than work that uses
     type and moisture content can affect the growth           subjective qualitative or semi-quantitative mea-
     patterns of plants.                                       surements; visual cover estimates, in particular,
*    Some habitats may be more accessible to survey            are often inconsistent.
     than others.
                                                           Whether bias matters to your monitoring pro-
     Species differences                                   gramme depends on the accuracy of the estimates
*    Some species may be more easily identifiable than     required. If the bias remains consistent and an accu-
     others.                                               rate estimate is not required then the bias is less
                                                           important, because it will not affect your ability to
     Temporal sources                                      detect change. However, determining whether the
*    The time of year (or day) when a survey is carried    bias remains consistent or not is likely to be very
     out can affect the results. Season is particularly    difficult. There are three ways of combating bias:
     important for vegetation monitoring; survey times
     must be standardised.                                 1. If the likely sources are anticipated, steps can be
                                                              taken to minimise bias for a particular project. Bias
     Weather                                                  can be reduced or controlled in a number of ways:
                                                              * Always record as much detail in your monitor-
*    Inclement weather affects observers’ concentra-
                                                                 ing as possible and use the same methods,
     tion and the time they will willingly spend in the
                                                                 approach, analysis, etc., across years, observers
     field. Variation between different observers’ capa-
                                                                 and sites; that way any observer bias can be kept
     city for working under difficult conditions can
                                                                 fairly consistent. If effort cannot be kept con-
     introduce bias.
                                                                 stant, the next best thing is to measure it.
*    It may be more difficult to distinguish species in       * When choosing a monitoring method, check
     wet vegetation than in dry vegetation.
                                                                 whether its assumptions (see individual habitat
*    Many species are less active under inclement
                                                                 and species chapters in Parts II and III) will hold
     weather conditions.
                                                                 for the habitat or species you wish to study and
By being aware of potential sources of bias these                for the period of time over which it is to be
problems can be reduced, measured or otherwise                   studied.
taken into account. For example, each time vegeta-            * Record relevant weather conditions when sur-

tion is measured, it is inevitable that there will be            veying. Agree and record beforehand at what
some recording bias, even with trained observers;                point weather conditions should postpone work.
numerous studies document this (for example, see              * Agree and record definitions (e.g. sample size,

Hope-Simpson, 1940; Smith, 1944; Lamacraft,                      type, population unit, etc.) beforehand.
1978; Sykes et al., 1983; Nilsson & Nilsson, 1985;            * Calibrate observers against each other before,

Kirby et al., 1986; Rich & Woodruff, 1990; West &                and during, monitoring. Introduce a system for
                                                                                        2.3 Sampling strategy     23



      quality assurance to verify the data (perhaps by      measurements are taken. For example, this may be
      using a person unconnected with the study or          a quadrat in a habitat survey or a bird’s nest in a study
      by observers checking each other’s work).             of breeding success. The term population is used to
2. With careful design it is possible to avoid the          mean the set of all possible sample units across the
   problem by confining comparisons of results to           site under study. Thus the population of 1 m  1 m
   attributes that have the same bias.                      quadrats is the total area of the site divided up into a
3. It is possible, though difficult, to measure the         1 m  1 m grid. A sample is a particular set of sampling
   bias. Measuring bias can be done only if the true        units and associated measurements.
   value can be occasionally ascertained; normally             The object of sampling is to avoid having to study
   this is unachievable. A separate experiment may          an attribute across the whole site but to still be able
   be helpful: for example, you could compare the           to estimate what is happening across the whole site.
   results obtained by different observers measuring        As Samuel Johnson said, ‘You don’t have to eat the
   the same population.                                     whole ox to know that it is tough.’ Thus we make an
                                                            inference about the whole based on an examination
If the bias may affect the monitoring and cannot be
                                                            of only a part. For these inferences to be valid, sam-
adequately measured, controlled for or reduced,
                                                            pling must follow certain principles:
then an alternative method should be used. If it
is not possible to find one method that provides            *   samples must be as representative of the whole as
an apparently unbiased estimate, use a number of                possible;
different methods and compare the results, or               *   more than one sampling unit is required. This is
change the objectives to match what is achievable.              known as replication.
                                                            The first principle enables reliable estimation of
                                                            an attribute’s value for the population and is
2.3       DESIGNING A SAMPLING STRATEGY
                                                            usually achieved by randomly locating the sample
First and foremost the design of a sampling strat-          units. The second principle enables the variation
egy must take into account the objectives of the            between samples to be calculated and can be used
monitoring study, to ensure these are met. These            to estimate the uncertainty in the estimate due to
objectives might be defined in terms of                     having only studied part of the population. For
                                                            example, the area of a site covered by Heather
*   how ‘good’ the estimate of the attribute, for the
                                                            Calluna vulgaris may be estimated by calculating
    whole site, needs to be;
                                                            the mean area of Heather in a sample of
*   what level of change between surveys needs to be
                                                            1 m  1 m quadrats and multiplying this figure
    detected; and
                                                            by the size of the site in square metres. The uncer-
*   which sections of the site are of particular interest
                                                            tainty in this estimate may be measured by the
    or most likely to change.
                                                            standard deviation of the estimate. An account of
Some other factors influencing the design are               these important statistics is given in Box 2.4,
                                                            below. See also the Glossary for further definitions
*   the type of attribute being measured;
                                                            of statistical terms.
*   the method being used;
                                                               A key feature of sampling is that as the number
*   the variability of the attribute across the site (if
                                                            of samples taken increases, so our uncertainty over
    known); and
                                                            how closely the sample estimate reflects the true
*   the time and costs of sampling.
                                                            population value decreases. The greater the sample
Figure 2.6 summarises the steps that should be              size, the greater the amount of survey time
considered. The steps are described in greater              required, and so a balance is needed between
detail in the following sections.                           ensuring that the estimate is ‘good enough’ and
   In this account of sampling design we use the            not expending unnecessary effort. Often, defining
term sampling unit to mean the unit from which              what is ‘good enough’ is not easy but will depend on
24     2 PLANNING A PROGRAMME



                                 FOR EACH METHOD AND ATTRIBUTE
                                     TO BE MEASURED (see 2.2)


                                   Are any data available on the size and
                                      variation of the attribute (2.3.1)?         NO                 Consider a
                                                                                                 preliminary survey
                                                      YES

                                 Decide whether sample locations should
                                   be permanent or temporary (2.3.3)



     Consider judgemental             Is a representative sample of the
       sampling (2.3.4)       NO            attribute required? (2.3)

                                                      YES

                               Decide whether samples should be selected
                                    randomly or systematically (2.3.4)



                                   Is there likely to be substantial variation    YES      Consider stratified
                                        in the attribute across the site?                   sampling (2.3.4)

                                                      NO

                         Will travel time between sample sites be high or                          Consider multi-stage
                     does the attribute vary substantially at small spatial scales      YES          sampling (2.3.4)
                                         or along transects?

                                                      NO

                                      Estimate minimum sample size
                                    required for desired precision and/or
                                   probability of detecting change (2.3.5)


                                     Estimate time and cost of sampling


                                       Is sampling programme feasible        NO      Reconsider sampling strategy
                                      taking into account requirements
                                      for other attributes and species?                       OR
                                                      YES
                                                                                        Seek more resources
                                     Document methods and sampling
                               strategy as a Standard Operating Procedure

            Figure 2.6. Designing a sampling scheme.


the quality of information required for the particu-            precise, population estimate is likely to be required.
lar feature under study and the use to which the                In other situations only a quick check may be
results will be put. For example, if a species of inter-        needed to confirm that a population is still doing
est is important and its population is believed to be           well. Further guidance on establishing the number
close to the limit of what is viable, then a good, or           of sample units required is given in Section 2.3.5.
                                                                                      2.3 Sampling strategy   25



     (a)        Estimate mean                    (b)


                                                                    Estimate mean




                   True size                                            True size
                 of population                                        of population

           Figure 2.7. Precision of measurements taken from a sample. In (a) the measurements are
           fairly precise: they are closely spread around the mean value. In (b), however,
           measurements are imprecise: they are widely spread around the mean.


   The size of sample unit chosen will depend on the   *   the surveyor is unfamiliar with the site and/or
species or habitat being sampled, the type of mea-         method; or
surements being made and the method used for           *   there are no existing data available for the site
sampling. This aspect of a sampling scheme is there-       that may help formulate a good sampling design.
fore considered in the individual method sections in
                                                       Data from preliminary trials can provide an initial
Part II, Chapter 6, and Part III, Chapters 11–26.
                                                       assessment of how close feature attributes are to
   Sample units for monitoring habitats and many
                                                       their targets and limits, and an estimate of the
species, particularly plants, will usually be quad-
                                                       variation in these attributes between sampling
rats (Section 6.4.2) or transects (Section 6.4.6).
                                                       units. This information can be an invaluable aid
Appropriate quadrat size for habitat monitoring is
                                                       when deciding how best to distribute the sampling
discussed in the sections on the use of the National
                                                       units across the site and how many samples are
Vegetation Classification (NVC) for monitoring
                                                       required. Although such a preliminary trial may
(Section 6.1.6) and is treated in more detail in
                                                       be time-consuming, it is likely to save time and
Appendix 4. Transect length for habitat monitoring
                                                       resources in the long term, particularly where
is discussed in Section 6.4.6.
                                                       sites and their features are poorly known.
   Sample units for species can be varied. The use
                                                          Preliminary field trials also enable the surveyor to:
of total counts, timed counts, quadrats and trans-
ects, and some other generic monitoring methods        *   become familiar with the characteristics of habi-
for species are discussed in Chapter 10. Other             tat or study species on the site;
methods for particular species groups are dis-         *   become familiar with the geography of the site; and
cussed in Part III (Chapters 11–26).                   *   iron out any problems applying the method.

                                                       Larger sites tend to be more complex, with more
2.3.1      Has the method been thoroughly              variables influencing the habitats, and so the larger
           tested and are preliminary field trials     the site, the greater will be the benefits of using a
           necessary?                                  field trial.

Preliminary field trials can be extremely valuable
and are often overlooked. They are particularly        2.3.2       Will the appropriate level
important if:                                                      of precision be achieved?
*   the methodology has not been used before in        Precision is a measure of the closeness of repeated
    similar circumstances;                             measurements to each other and provides a
26       2 PLANNING A PROGRAMME



measure of the uncertainty due to sampling (see                                 variable and we have less confidence that the sam-
Figure 2.7). Intuitively, if the sample measurements                            ple has pinned down the population mean.
are all very close to each other then it is likely that                             Because the sample mean is our estimate of the
the site is very uniform and so, provided there is no                           population mean, we are usually more interested in
bias in the measurements, the sample mean is likely                             its precision rather than that of the individual mea-
to be close to the population mean. Conversely, if                              surements. Regardless of the precision of individual
the measurements vary a lot then the site is very                               measurements, we would expect that, as we increase


     Box 2.4 Descriptive statistics: some important                             Variance: the square of the sample standard deviation
     definitions                                                                (s2). This is another commonly used measure of data
                                                                                variability but, unlike the standard deviation, it is not
     Mean: The sum (Æ) of all individual values (x) divided by                  measured in the same units as the observations.
     the number (n) of observations: (Æx)/n. See figure below.                  Standard error (SE): the standard deviation of the sam-
     Median: The middle observation in a set of observations                    ple mean, given by:
     that have been ranked in magnitude.
                                                                                                             s
     Mode: The most common value of a set of observations.                                             SE ¼ pffiffiffi :
                                                                                                              n
     Standard deviation: A measure of the variability
     of a dataset in terms of the deviation of observations,                    This is a more informative statistic than the standard
     xi (i ¼ 1 to n), from the mean. When monitoring, we are                    deviation when the main interest is in the sample
     generally sampling a subset of the population. In this                     mean. It will decrease as the sample size increases.
     case, the sample standard deviation, s, is given by:                       Coefficient of variation: Another useful measure of the
                                                                                relative variability in the data, which can be compared
                                 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi                       between different attributes regardless of the units
                                  P
                                       ðxi À xÞ2                                in which they are measured. Relative variability is
                              s¼                          ;
                                        nÀ1                                     measured by calculating the coefficient of variation
                                                                                (%cv), which is the standard deviation expressed as a
     where " is the sample mean.
           x                                                                    percentage of the sample mean:

                                                                                                    %cv ¼ 100s=x:

                                           Mode = 5
                        35                            Median = 6

                        30                                    Mean = 7.2
                        25
            Frequency




                        20

                        15

                        10

                         5

                         0
                             1 2     3    4     5 6       7     8   9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
                                                                              x

                        Figure B2.4. An example of a distribution showing mean, median and mode.
                                                                                         2.3 Sampling strategy   27



the number of measurements, so the sample means             2. Detecting changes between one survey and the
from repeated samples will tend to become closer               next. If small changes need to be detected then a
and closer to one another and so each sample will              precise estimate of change will be needed, which
tend to be closer to the population mean.                      will usually depend on having a precise estimate
   Precision in the sample mean is often measured              of the attribute’s value from each survey.
by its standard error (Box 2.4), which becomes smal-
                                                            Further guidance on establishing the number of
ler as the sample size is increased (Figure 2.8). Thus
                                                            sample units required is given in Section 2.3.5.
precision improves as the standard error decreases.
                                                               Finally, it should be remembered that simple
   This is not quite the whole story. As the sample
                                                            monitoring methods may actually be extremely
size increases, so we would expect that the sample
                                                            time-consuming overall because they produce
means from repeat samples will tend to become
                                                            relatively imprecise results and therefore require
closer and closer to one another. However, the
                                                            more intensive sampling. On the other hand,
sample size (n) cannot be larger than the popula-
                                                            although some sampling methods appear daunting
tion size (N). When this limit is reached, the sample
                                                            because they use a complicated methodology, the
is of the whole population, and there is then
                                                            quality of data collected per unit time may be
no uncertainty remaining as to the value of the
                                                            much higher.
population mean. So for the standard error to be
a good measure of precision it should be zero when
n = N, which is not the case for the formula given          2.3.3       Should sample locations be
in Box 2.4. For a finite population a more exact                        permanent or not?
formula for the standard error of the sample
                                                            Permanent sample locations can provide a good
mean is
                                                            approach for improving precision when detecting
                         rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi           changes over time is of prime importance. There
                          s2             n
                  SE ¼           1À               :         are, however, a number of significant disadvantages
                           n              N
                                                            to using permanent locations. See Appendix 5 for
                                                            further information on establishing permanent
The quantity n/N is often termed the finite                 sample locations.
population correction (fpc) and it is only when its value      Permanent locations should only be used if:
is greater than around 0.1 that it has much effect on
the value of the standard error. If less than 10% of a      *   Maximising change detection is of prime import-
feature is to be sampled, as will often be the case, the        ance and some consistency is expected in how the
fpc can be safely ignored.                                      attribute changes across the site; or information is
   Larger sample sizes require more time; preci-                needed on turnover and species dynamics; or the
sion, as measured by the standard error, typically              feature being monitored is a rare sessile species,
increases only in proportion to the square root of              which is confined to precisely known location(s).
the sample size. Hence, to reduce by half the stan-         *   There is sufficient fieldwork time available for
dard error obtained from 10 sample units requires               marking and relocating permanent sampling
about 40 units.                                                 locations, and this time cannot be more efficiently
   What precision is needed depends on the use to               used for collecting data from a greater number of
which the results will be put. Two key uses for                 temporary sample locations.
survey data are as follows.                                 *   Sample locations are representative of the site (see
                                                                Section 2.3.4 for further discussion) and sufficient
1. Determining whether the population value of the              samples are taken to minimise the risk of chance
   attribute being measured is above or below a                 events reducing representativeness.
   limit. If this value is close to this limit it may       *   Provision has been made for the unexpected loss
   require a large sample size to be certain which              of sample locations.
   side of the limit it really is.
28   2 PLANNING A PROGRAMME



                            25
                                 (a)
                            20
             No. of plots
                            15

                            10

                             5

                             0

                            25
                                 (b)
                            20
             No. of plots




                            15

                            10

                             5

                             0

                            25
                                 (c)
                            20
             No. of plots




                            15

                            10

                             5

                             0
                                 75    80   85   90   95   100   105    110   115   120   125   130
                                                      No. orchids per plot

        Figure 2.8. The effect of increased sample size on the precision of sample means. The data for the graphs
        above are taken from a hypothetical population of orchids distributed in a uniform habitat. The true
        population density is 100 orchids per plot (our hypothetical sampling unit), with a standard deviation
        (SD) of 10. This density is estimated by the mean (or average) number of orchids in the sample plots.
        (a) Three randomly selected plots give the mean number of orchids per plot as 94.3, with a high standard
        error (SE) and no clear picture of how the data are distributed. Estimates of means based on only three
        samples are likely to vary considerably and would therefore be imprecise. (b) The mean of 20 random
        plots gives a closer approximation (96.4 orchids per plot) to the true density, and has a lower standard
        error (2.65). (c) The mean of 100 plots (99.4 orchids per plot) is very close to the true density, with a
        correspondingly lower standard error (1.07). Thus estimates of means based on this high number of
        samples are likely to be very close to the true mean, i.e. precise. However, the cost and effort of taking 100
        samples may not be worth the extra increase in precision. This will depend on available resources. Note
        also that the number of orchids per plot is now revealed to be approximately normally distributed
        (see Box 2.7).
                                                                                      2.3 Sampling strategy   29



*   The feature being monitored and the surrounding         conservation reasons, it may also cause the samples
    environment will not be significantly altered or        to become unrepresentative of the site as a whole.
    damaged.                                                   Permanent sample locations may become unre-
                                                            presentative of the whole study area (assuming
                                                            that they were representative initially) as a result
Advantages of permanent locations
                                                            of chance events that affect the locations dispro-
If permanent locations tend to change with some
                                                            portionately. Such events may also have perma-
consistency, we are more likely to be able to detect
                                                            nent or long-lasting effects, as successive changes
change than when using the same number of tem-
                                                            at one point tend to be correlated. Therefore, any
porary locations. This is because the estimate of
                                                            recorded changes will not reflect the true pattern
change is based on the mean change within sam-
                                                            of change over the site and may be significantly
pling units rather than the change in the mean of
                                                            biased. This difficulty can usually be overcome by
different sampling units. The standard error of this
                                                            avoiding small sample sizes. Alternatively, record a
mean change will tend to be small when the units
                                                            second set of samples at the end of the first mon-
tend to change by the same amount.
                                                            itoring period, which are used to estimate changes
   For example, suppose that mean species richness
                                                            in the second period and so on, i.e. samples A are
over 20 quadrats is 15 in one year and 10 in
                                                            enumerated on the first occasion, samples A and B
a subsequent survey and that the decline is fairly
                                                            on the second, samples B and C on the third, and so
consistent across the site. If permanent quadrats are
                                                            on (Greig-Smith, 1983).
used this consistency of change is apparent from the
                                                               Permanent sample locations may be effectively
way the quadrats change. However, if new quadrats
                                                            lost as a result of unforeseeable events, such as
are selected for the second survey we only have the
                                                            permanent or long-term flooding of part of the site,
change in mean richness in the two sets of quadrats
                                                            or the growth of trees over long time periods. This
to go on, and there is a greater possibility that the
                                                            problem can be alleviated by recording ‘spare’
change is due to the chance location of quadrats in
                                                            samples.
the first survey being in richer parts of the site. Thus,
with temporary quadrats, a change would need to
be larger for us to be confident it indicated a real        2.3.4      Should the samples be located
change across the whole site, than if permanent                        randomly, systematically or by
quadrats were used.                                                    judgement?
   Permanent plots can also be particularly useful
                                                            The arrangement, number, and size of samples has
for monitoring sparse sessile species, such as some
                                                            a critical influence on the results obtained and how
lichens, which may be confined to a small part of a
                                                            they can be interpreted. Not surprisingly, there
site and do not spread. In this case randomly
                                                            have been numerous investigations into this issue
located plots would be very inefficient as most
                                                            (see Greig-Smith (1983) and Shimwell (1971) for
would miss the species altogether.
                                                            reviews of plant surveying techniques). Various
                                                            techniques have been used to position the samples.
Disadvantages of permanent locations                        These are summarised in Table 2.1, with various
Marking and relocating permanent sample loca-               options relating to random and systematic meth-
tions can be difficult and time-consuming. This             ods described further in Table 2.2.
may offset any advantage from additional precision             An illustration of the different random and sys-
if observations from non-permanent samples can              tematic sampling strategies described in this
be obtained much more quickly.                              Handbook is given in Figure 2.9.
   Surveying at permanent locations may alter
or damage the attribute being monitored or its              Locating samples by judgement
surroundings, e.g. by trampling. Apart from the             Sampling units that are located by judgement can-
potential unacceptability of such damage for                not reliably be regarded as being representative
30    2 PLANNING A PROGRAMME



Table 2.1. Summary of the advantages and disadvantages of different approaches to sample selection

Sample
location method         Advantages                                     Disadvantages

Judgement               Can be quick and simple if knowledge of        Extrapolation of results to the whole
                        habitat/species is sufficient                  feature or site is not valid without strong
                                                                       justification
                        Samples can be deliberately taken              Comprehensive knowledge of the site is
                        around e.g. a rare species or feature of       required
                        particular importance; useful when all         Statistical analysis is not valid and errors
                        locations of a rare species are known          cannot be quantified
Random                  Requires minimum knowledge of the              Collection of sample observations can be
                        population                                     time-consuming
                        Easy to analyse data and compute               Can result in larger errors for a given
                        variability                                    sample size compared with systematic
                                                                       sampling
Systematic              If the attribute is ordered spatially, there   If sampling interval matches a periodic
(regular)               is a stratification effect, which reduces      feature in the habitat (e.g. regular
                        variability compared with random               ditches), significant bias may be
                        sampling                                       introduced
                        Determining sample locations is easy           Strictly speaking, statistical tests are not
                        Provides an efficient means of mapping         valid, although in most cases conclu-
                        distribution and calculating abundance         sions are unlikely to be substantially
                        at the same time                               affected


of the entire study area. Consequently, observa-           for demonstrating typical changes. However, for
tions cannot be extrapolated across the site as a          the reasons described above, such data should
whole without strong justification, and it is not          generally not be extrapolated to the whole feature
valid to calculate summary statistics or perform           or site.
statistical tests on such data. Such samples there-           Judgement sampling can, however, be informa-
fore only yield information on their own particular        tive if carried out by surveyors who use a thorough
location. This may not matter when selecting sites         knowledge of the site and of the processes acting
for monitoring rare and/or sessile species with            on the site to describe the site in an intelligent way.
known locations or habitat preferences.                    The usefulness of the results can be increased if the
   Temporary sampling locations are rarely placed          surveyors record as much detail as possible about
by judgement except during NVC surveys (Section            what they did, and take photographs. If this is done
6.1.6), in which quadrats are placed on ‘representa-       well it may be possible to repeat a survey quite
tive’ stands of vegetation to assist with the identi-      closely. However, it is preferable not to rely on
fication of NVC types. Data from such samples              subjective techniques such as this.
should not be used for monitoring purposes.
   It is fairly common practice to locate perma-           Random sampling
nent plots by judgement, particularly when mon-            When the goal of sampling is to provide an indica-
itoring rare species that are likely to be missed by       tion of what is happening across the whole site,
random or systematic surveys. Data collected               random sampling designs are generally recom-
from such plots can be informative and useful              mended. Random sampling is usually designed to
                                                                                      2.3 Sampling strategy    31



Table 2.2. Summary of the advantages and disadvantages of different random sampling designs

Type of structure      Advantages                                   Disadvantages

Simple random          Selecting sample units is quicker and        Estimates will be less precise on hetero-
                       easier than for other random designs         geneous sites than with stratified
                                                                    sampling
                       Statistical analysis of data is              Travel time between sample units can be
                       straightforward                              high
Stratified             Ensures that all the main habitat types      If strata have not been identified prior to
                       present on a site will be sampled (if        monitoring, preparation can be time-
                       defined as strata)                           consuming
                       Characteristics of each stratum can be       The most appropriate stratification for a
                       measured and comparisons between             site at one time may have changed when
                       them can be made                             repeat surveys are carried out; monitor-
                                                                    ing efficiency may therefore also change
                       Greater precision is obtained for each
                       stratum and for overall mean estimates if
                       strata are homogeneous
Multi-stage or         Can reduce sampling times, thus              When calculating overall means, etc.,
cluster                increasing efficiency                        larger errors are obtained than with a
                       Useful for sites that are heterogeneous at   simple random sample of comparable
                       small spatial scales and for studying gra-   size if sample units within major units
                       dients along transects                       are highly correlated




ensure that each of the population of sampling           quadrats, and many of the same considerations
units has an equal chance of being selected.             apply. The direction of fixed-length transect lines
Standard statistical methods can then be used to         should usually be randomly allocated. However, it
analyse the data (see Section 2.6). Plot location        may be desirable to select a direction that allows
should not in any way be influenced by any prior         samples to be taken along a perceived environmen-
knowledge. Randomly located plots are picked             tal gradient (e.g. a transition from acid to calcare-
from a numbered list of all plots that could be sur-     ous grassland). This has the effect of reducing
veyed, by using random numbers generated by a            variation between transects, thereby improving
computer or from tables. Locating plots by eye does      precision.
not yield randomness, because samples are usually           Sometimes it is impossible not to deviate from
spaced too evenly. Throwing quadrats to obtain loca-     randomness when sampling, for instance if access
tions, although better than locating by eye, does not    to a particular area is not possible. If the inaccessible
achieve true randomness either (this is known as         area is small this may not matter, but if significant
haphazard sampling). Random samples can, how-            bias is possible, the issue should be documented and
ever, be time-consuming to locate in the field.          population estimates may need adjustment.
   Figure 2.10 shows a method for choosing sam-
pling units randomly. Units that are found to fall       Systematic sampling
outside the area are ignored.                            It is often convenient to take samples at regular inter-
   Transect lines may also be located by utilising       vals, for instance at fixed distances along a river.
these points. Transects are essentially long, thin       However, this method creates one main problem: if
32   2 PLANNING A PROGRAMME



          (a)            Random sampling              (b)            Stratified random sampling




          (c)          Systematic sampling            (d) Stratified systematic unaligned sampling

                                                                         a q             a r
                                                                             b                                   b
                                                              b                              b
                                                                     p               q                               s
                                                                                                     r
                                                                             c                               c
                                                             c                               c
                                                                                     q                               s       c
                                                                     p                               r                               t
                                                                                 d                               d
                                                                 d                               d                       s
                                                                         p               q               r                       d       t
                                                                             e                               e                               e
                                                                                             e
                                                                                 q                               s           e t             u
                                                                                                 r




          (e)       Cluster sampling (two types)      (f)                    Two-stage sampling




                                               B
                A




         Figure 2.9. Different types of sampling strategy. (a) Random sampling: samples taken randomly
         from the whole study area. (b) Stratified random sampling: study area divided into strata and random
         samples taken in each stratum. (c) Systematic sampling: samples taken at regular intervals.
         (d) Stratified systematic unaligned sampling: study area subdivided into equal blocks and x
         co-ordinates (a–e) and y co-ordinates (p–u) generated randomly. Distance a is used in every block in the
         first row, b in the second row, etc. Distance p is used in every block in the first column, q in the second
         column, etc. (e) Two types of cluster sampling. In type A cluster areas (large squares) are chosen
         randomly and all sample units (x) sampled in each. In type B points are chosen randomly and samples
         taken in a fixed pattern relative to each point. (f) Two-stage sampling: major units (large squares)
         chosen randomly and minor units (x) sampled randomly from each. Major units may also be transects.
                                                                                       2.3 Sampling strategy   33



                                                            et al., 1990; Watt, 1997) without causing substantial
       8         X                                          problems, unless a systematic bias such as that out-
       7
                                                            lined above occurs. Sophisticated statistical techni-
                                                            ques have been developed for spatial analysis of
       6                                                    both systematic and random samples (Cressie, 1993;
       5                   S                                Webster & Oliver, 2001), which enable distribution
                                                            and density maps to be developed as well as providing
       4                                  S                 alternative estimation methods.
       3                                                       Systematic sampling can be useful because sam-
                                                            ple sites are relatively easy to select and relocate,
       2                                                    and the approach is often more appealing and
       1         X                                          straightforward to surveyors. A particular use may
                                                            be when trying to map both distribution and total
            1    2    3    4    5    6    7    8
                                                            abundance of an organism across a study area. The
Figure 2.10. Choosing sample units from a                   advantages of a regular distribution of sample sites
two-dimensional map. As an example, consider                might then outweigh the population estimation
choosing 1 ha sample units from a woodland. The             disadvantages, for example, if a distribution map
shaded area in the diagram represents the woodland;         based on a regular grid were the objective of the
the lines superimposed on it are 100 m apart. Number        study. Grid surveys repeated regularly can provide
the grid rows and columns (as above). Select pairs of       excellent comparative data to identify potential
random numbers by using a random number generator.          causes and influences of change.
The first number defines the column in the grid; the
second defines the row. Reject any grid squares that fall   What are the advantages of stratified, cluster
outwith the study area. This procedure can be carried       and multi-stage sampling?
out within a spreadsheet or Geographical Information        Stratified sampling
System (GIS). An obvious advantage of using a GIS is that   Stratified sampling is very commonly used in
it can produce a map of the sample locations. On the        environmental monitoring as a way of improving
ground, sample units are most easily located by using       the precision of estimates. Very often there is
a Global Positioning System (GPS). Small errors inherent    substantial variation across the site in the feature
in GPS readings are not important provided these are        attributes being measured. This may be due to
random.                                                     environmental gradients or differences in man-
                                                            agement, for example. In this situation it makes
                                                            sense to divide the site into sub-units (strata) that
the sampling interval constantly coincides with a           relate to the different values of the attributes
particular regularity in a species or habitat being         being monitored (e.g. different densities of a par-
monitored, the results will be biased. For example, if      ticular species) and sample each sub-unit sepa-
you are sampling vegetation at 10 m intervals, and          rately (see Figure 2.9). Separate estimates are
this interval coincides with the raised parts of a hum-     then made for each stratum, which are then
mock–hollow microtopography (perhaps stretching             combined to provide an estimate for the whole
the example!), the vegetation in the hollows (which         site. Stratification has a number of potential
may be different) will not be sampled. The results will     advantages:
therefore give a biased picture of the vegetation.
Systematic samples are not placed independently of          *   An attribute can be estimated with greater preci-
each other (unlike random samples) so, strictly, sta-           sion, provided that the value of the attribute
tistical analysis is not valid. However, if a large num-        differs substantially between strata and there is
ber of samples are taken, systematic samples can                more variation in the attribute between strata
usually be treated as random samples (Schaeffer                 than within strata.
34        2 PLANNING A PROGRAMME




     Box 2.5   Optimal allocation of sampling effort                 The proportion of sampling effort that should
     for stratified sampling                                      optimally be made in the hth stratum (considering
                                                                  variability and cost) is given by:
     First we require:                                                                                 pffiffiffiffi
                                                                                    nhðoptÞ    Nh sh = ch
     n = total number of sampling units required                                            ¼ PÀ        pffiffiffiffiÁ :
                                                                                      n          Nh sh = ch
     (e.g. quadrats).
        For each stratum (h) we require:                          If the measurement is a proportion (e.g. the proportion
     nh(opt) = optimum number of units to be sampled in           of quadrats containing a species), this formula can be
     stratum h;                                                   written as
     Nh = total number of possible sampling units in                                         pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
                                                                                nhðoptÞ    Nh ph qh =ch
     stratum h (the stratum area can be used instead);                                  ¼ P pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
                                                                                  n         Nh ph qh =ch
     sh = estimated standard deviation of measured variable
     per sampling unit in stratum h;                              where
     ch = relative cost of taking one sample in stratum h (e.g.      ph = estimate of the proportion being measured, in
     c1 = 1, c2 = 1.5, c3 = 2)                                    stratum h;
        It is important to note that an estimate of                  qh = 1 – ph.
     standard deviation of the variable being measured               As a rule of thumb, for quantitative data no fewer
     must be made in advance for each stratum. This can be        than five samples should be allocated to each strata.
     done either by making an educated guess or by                For proportions rather more are advisable as,
     conducting a preliminary survey. In the absence of           otherwise, the estimation of the standard deviation
     better information, costs are often all assumed to           becomes very unreliable.
     be the same.                                                    For further detail consult Cochran (1977).




*    Separate estimates can be made for each stratum                 If the cost of sampling varies, or the within-
     if these are of interest in their own right.                 stratum variance in each stratum differs, sampling
*    Stratification slightly reduces the time taken to            should be more intensive in the strata in which
     randomly locate samples.                                     the costs of sampling are lower or which are more
                                                                  variable. Sample size should be proportional to
To maximise the benefits of stratification the site               the size of the strata if the costs and variances of
should be subdivided in such a way that it mini-                  each stratum are similar, or in the absence of
mises the within-stratum variability in the attri-                such information. A formula for the optimal
bute being measured (i.e. strata should be as                     allocation of sampling effort between strata is pro-
uniform, or homogeneous, as possible). This nor-                  vided in Box 2.5. Methods for calculating overall
mally requires previous survey data or a prelimin-                means and confidence intervals are provided in
ary survey to be carried out.                                     Box 2.13.
   Alternatively, you can stratify according to
known site variations in habitat or ecological                    Multi-stage and cluster sampling
factors, which are believed to influence the fea-                 In many situations a site may be so large that a high
ture attributes (e.g. a sudden change in soil type).              proportion of time is spent travelling between sam-
Although these divisions are not going to be as                   ple sites. In this instance cluster or multi-stage
accurate, as long as there is lower variability                   sampling could be considered as a means of increas-
within strata this sampling method will provide                   ing sampling efficiency and in some instances can
better results than simple sampling across the                    improve precision for a given sample size. Multi-
whole site.                                                       stage sampling is also known as multi-level sampling
                                                                                         2.3 Sampling strategy    35



or subsampling. With multi-stage and cluster sam-            relative cost of sampling at the two stages is required.
pling a major sample unit is selected, which is divided      This may be obtained through a preliminary survey,
up into minor units. Data are then collected from            or estimated based on available knowledge of the
some or all of the minor units (see Figure 2.9). With        habitat in question. A preliminary survey may also
cluster sampling all the minor units are sampled, but        be designed to investigate the optimal size of the
with multi-stage sampling a random or systematic             major units as there will be a trade-off between the
sample of minor units is selected. In some cases the         benefit of having a large sample of major units and
minor units are themselves sampled (three-stage              increasing their size to reduce between-unit varia-
sampling) but two-stage sampling is the most com-            tion. Formulae for estimating the optimal number
mon technique.                                               of minor and major units are provided in Box 2.6.
   A common example is one in which the major                   Methods for calculating means and confidence
units are transects and the minor units are quad-            limits for two-stage sampling are given in Box 2.13.
rats along each transect. If all quadrats are sampled        These methods assume that all minor units are of
this is known as a belt transect.                            equal size and that each major unit contains the
   The main consideration with this technique                same number of minor units. Table 2.2 summarises
is that sample units within each major unit are              the advantages and disadvantages of the different
unlikely to be independent of one another since              random sampling designs. For further information
spatial correlation may occur (i.e. sample units are         and detail on these see, for example, Cochran
likely to be more similar the closer they are to each        (1977) or Yates (1981).
other). Unless the minor units are sufficiently far
apart to avoid this, overall precision is likely to be       Some other approaches
mainly determined by the variation between the               Stratified systematic unaligned sampling
major units. In cluster sampling, the minor units            This is a variation of stratified sampling that com-
are usually combined and analysis is reduced to              bines the advantages of random and systematic
simple random sampling of the major units. This              sampling. The area to be sampled is first stratified
may still be advantageous, compared with simple              into equally sized blocks (not strata based on habitat
random sampling of minor units, if there is a sig-           characteristics as in stratified random sampling).
nificant reduction in the variation between sam-             Samples are placed in each block by using different
pling units as these units get larger.                       x co-ordinates for each column of blocks but the
   Thus, cluster and multi-stage sampling are likely         same x co-ordinate within one column, and differ-
to be most useful when the area being sampled is             ent y co-ordinates for each row of blocks but the
relatively uniform at large spatial scales and most of       same y co-ordinate within one row (see Figure 2.9).
the variance occurs at small spatial scales (but at scales   This technique can be an improvement on stratified
larger than the size of the sample unit). Transects will     random sampling because the systematic misalign-
be most effective if oriented along a gradient in the        ment is not subject to localised clustering. This tech-
attribute being measured. For example, in a study of         nique does not appear to have been widely used. The
tree regeneration around woodland, the transects             time taken to position samples is similar to that for
may be oriented away from the woodland, assuming             stratified random sampling.
regeneration will decline with distance.                        Smartt & Grainger (1974) compared the techni-
   The precision of the overall estimate is primarily        ques discussed above. They found that stratified
affected by the variance between the mean values for         techniques exhibited greater overall comparative
major units and, to a lesser extent, by the variance         precision than random or systematic techniques,
between minor units within each major unit.                  especially at low sample densities with clustered
Precision is also affected by the number of units            distributions. In this situation, the sampling strate-
sampled at each level. In order to determine the             gies ranked in increasing order of relative precision
optimum number of major and minor units to sam-              were: random, systematic (regular), stratified ran-
ple, some knowledge of the two variances and of the          dom, stratified systematic unaligned.
36       2 PLANNING A PROGRAMME




     Box 2.6 Optimal allocation of sampling effort            It can be seen from this equation that uopt increases
     for two-stage sampling                                   as tm/tu increases and as the variance within units (ms )
                                                              increases relative to the variance between units (s2 ).
                                                                                                                 x
     The following quantities need to be determined during       If either uopt > U or s2 ms < 1/u0 , then you should
                                                                                         x
     the preliminary survey:                                  sample all of the minor units within each major unit,
        N = number of major units available for sampling;     thereby simplifying the sampling strategy to basic
        n = number of major units sampled;                    cluster sampling.
        U = number of minor units within each major unit;        If we assume that the mean approximately follows a
        u = number of minor units sampled within each         normal distribution (Box 2.7), the number of major
            major unit;                                       units to be sampled, np, required to give a confidence
        u0 = number of minor units in preliminary study;      interval that extends no more than P% either side of
        tm= time taken to locate each major unit;             the mean (Section 2.3.5), is roughly calculated by using
        tu = time taken to sample each minor unit.            the following equation:
        We then require the means and variances of each
     major unit. These are used to calculate the following:                             s2
                                                                                         x
       " = overall mean, i.e. the mean of the major unit
       x                                                               np ¼ À        À1 1 ÁÁ À " Á2 :
                                                                           1 2                 x
                                                                           N sx þ m s u À U þ 200 P
           means;
        s2 = variance of the major unit means;
         x
        ms = mean of the variances for each major unit.       These two equations can therefore be used to calculate
        The optimum number of minor units to be sampled,      the optimum number of minor units to sample within
     uopt, is calculated thus:                                each major unit and the optimum number of major
                              vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi             units to sample for a required precision, once initial
                              u t =t
                              u m u                          estimates have been made of the variability of the data
                       uopt ¼ t 2                :
                                  sx        1
                                  ms   À u0                   and the relative costs of sampling major and minor
                                                              units.




Adaptive sampling                                             One disadvantage is that the sample size cannot be
Another approach to consider for features that have           determined in advance; it will depend on what is
very clustered distributions is adaptive sampling             encountered in the initial sample.
(Thompson and Seber, 1996). This involves selecting              Some other sampling techniques for species, such
an initial random or systematic sample. If the target         as distance sampling, are discussed in Chapter 10.
species is found in a given sampling unit, the adja-
cent sampling units are also included on the basis            2.3.5       How many samples will be required?
that there is a good chance these will also contain
the species. Potential advantages include:                    As previously mentioned, increasing sample size
                                                              increases precision (see Figure 2.8) as well as the
*    although specialised formulae are required for           cost of monitoring. Preliminary field trials or pilot
     estimation, adaptive sampling can provide better         surveys enable the distribution of the species or
     precision for a given amount of effort than simple       habitat to be assessed and the amount of variation
     random sampling;                                         in each attribute to be estimated. This can be of
*    the method is more satisfying for surveyors, as          enormous value in helping to optimise the sam-
     they do not have to ignore sightings that fall just      pling design and in establishing the number of
     outside a sampling unit;                                 samples required. Without a pilot survey or some
*    a better picture of the species’ spatial distribution    data from a previous survey, sample size estimates
     is obtained.                                             are usually down to guesswork.
                                                                                                2.3 Sampling strategy         37




Box 2.7 Probability distributions                              being interested in the distribution of the
                                                               measurements themselves, we want to be able to say
The probability distribution for a particular                  something about the mean of those measurements.
measurement shows the probability that an individual           This mean also has a probability distribution. An
drawn at random from a population takes a particular           important result, the Central Limit theorem, states
value, or lies within a range of values. The true              that as sample size increases the distribution of means
distribution of a measurement across a study area is           of samples will almost always converge towards a
usually unknown and has to be approximated by using            normal distribution, regardless of the shape of the
measurements taken from a sample. For quantitative             distribution of the original measurements. Of course,
measures, such as vegetation height or counts of plants        the more non-normal (e.g. asymmetrical) the parent
in a quadrat, the distribution can be illustrated by           distribution is, the larger the sample size will have to
plotting a histogram, such as those shown in                   be for this to hold.
Figure 2.8, but unless sample sizes are fairly large this         In addition, although the normal distribution is
may be a poor representation of the population distri-         continuous it can often be used to approximate the
bution. For presence–absence data the distribution is          distribution of count and binary data. This simplifies
defined by the probability that a sample unit will             the calculations of probability for these data.
contain the species of interest.                                  In some cases a transformation, such as log or
   A lot of statistical analysis relies on assuming that the   square root, can be used to make data fit a normal
distribution of a measure can be described, at least           distribution more closely and therefore facilitate the
approximately, by a particular mathematical function.          use of statistical methods that rely on this assumption
This enables us to estimate the probability that the           (see Section 2.6.4).
measure is less than, or greater than, some value of              For data that are normally distributed, the
interest.                                                      probability that a measurement is less than a
   By far the most common theoretical distributions            specified value does not have a simple mathematical
that arise in monitoring are the normal and binomial           formula. However, it can be found in statistical tables
distributions. These are described below.                      or is readily calculated in most spreadsheet or
                                                               statistical software.
                                                                  As mentioned above we are often more interested in
THE NORMAL DISTRIBUTION
                                                               the distribution of the sample mean rather than that of
The normal distribution is the most commonly used
                                                               the original data. The mean of normally distributed
function to describe the distribution of continuous                                        pffiffiffi
                                                               data has parameters  and = n, where n is the sample
data. It is one of the easiest distributions to handle in
                                                               size. It is convenient to standardise normally distributed
terms of statistical computation. This distribution is
                                                               data so that the standardised data follow a normal
determined by two parameters, , the mean value of
                                                               distribution with a mean of zero and a variance of one.
the population, and , the standard deviation. Because
                                                               The standardised mean is calculated as
 and  relate to the population we must estimate their
values by using the sample mean and standard                                            ðx À Þ
                                                                                                :
deviation.                                                                                 pffiffi
                                                                                           
                                                                                            n
   A normal curve is symmetrical, with the axis
of symmetry passing through the mean. About 68%                   Usually, we do not know what the value of  is: it
of all observations drawn at random from a normal              has to be estimated by s, from the sample. If we replace
distribution will fall within one standard deviation of         by s in the above equation the distribution ceases
the mean; about 95% will fall within two standard              to be exactly normal. In fact the standardised mean
deviations. See Figure B2.7.                                   follows what is called the t-distribution. This distribution
   The normal distribution has some very desirable             is dependent on the sample size n, and is referred to as
mathematical properties. Very often, rather than               having n-1 degrees of freedom. For sample sizes greater
38       2 PLANNING A PROGRAMME




     than about 30 the t-distribution is almost the same as           certain character with the probability p or fail to
     the equivalent normal distribution and this can be used          exhibit it with the probability 1 À p = q. The probability
     as an approximation.                                             P that, in a sample of size n, x individuals possess the
        The t-distribution is widely used in calculating              character is given by:
     confidence intervals (Box 2.8) and in statistical tests
                                                                                                n!
     (see Section 2.6). For t-distributed data, as with                                P¼              px qðnÀxÞ ;
     normally distributed data, the probability that a
                                                                                            x!ðn À xÞ!
     measurement is less than a specified value can be                where ! = factorial (e.g. 5! = 5 Â 4 Â 3 Â 2 Â 1).
     found in statistical tables or is readily calculated in             The mean and variance of the distribution are given
     most spreadsheet or statistical software.                        by np and np (1 À p), respectively. These quantities can
                                                                      be used to approximate the binomial distribution by an
     THE BINOMIAL DISTRIBUTION                                        equivalent normal distribution provided n is large and
     Observations that are binary (can take one of two                p is not close to 0 or 1. This can facilitate the calculation
     values) such as presence–absence often conform to the            of confidence intervals for p (see Box 2.8) and the use of
     binomial distribution. Individuals may possess a                 standard statistical tests (see Section 2.6).




                                                                68%



                                               –2σ        –1σ    µ        1σ      2σ

               Figure B2.7. The normal distribution curve showing one (shaded) and two (unshaded)
               standard deviations.


   How many samples are required depends very                         This will typically be in terms of the amount of
much on the individual sampling method, the type                      error or uncertainty we are prepared to tolerate.
of attribute being measured, the degree of variabi-                   For example, suppose we are estimating the per-
lity in the attribute being sampled and the desired                   centage of a raised bog that is covered by sphag-
precision or the degree of change that must be                        num moss, and this is thought to be around 70%.
detected. A method for establishing the number                        We might be content for our sample of the bog to
of samples required to provide a desired level of                     tell us that this percentage is somewhere between
precision is outlined below and methods for detect-                   60% and 80%. So, in this case, an uncertainty of 10%
ing change in Box 2.10.                                               is acceptable. What error is acceptable will depend
   Both approaches require some understanding                         on the objectives for the monitoring. In some cases
of probability distributions and how these are                        a rough and ready estimate may be acceptable if a
applied to different types of measurement. This                       feature is believed to be well within acceptable
is outlined in Box 2.7, but a fuller account can                      limits, but where the feature is of particular impor-
be found in most introductory statistics texts                        tance or is the subject of some concern, a very
(see, for example, Fowler et al., 1998; Sokal &                       precise estimate may be desirable.
Rohlf, 1996).                                                            This approach suggests choosing a sample size
                                                                      that will give us enough measurements so we can
Achieving a desired level of precision                                be confident that the value of the attribute lies
The first question to be answered is how we specify                   between l and u , where the range u À l is sufficiently
how good we want our population estimate to be.                       small for our purposes. The range u À l is termed a
                                                                                           2.3 Sampling strategy         39



                                                                  pffiffiffi
Box 2.8 Confidence intervals                                 4:08= 10 ¼ 1:29: Assuming the mean count is
                                                             approximately normally distributed, the 95%
If a population parameter such as the mean is                confidence limits are
estimated from a sample, there will always be uncer-
tainty as to the whole-population value of the mean.                  l ¼ 6:2 À ð2:26 Â 1:29Þ ¼ 3:3 and
Confidence intervals derived from our estimates indicate             u ¼ 6:2 þ ð2:26 Â 1:29Þ ¼ 9:1 ;
a range of values within which we have some confi-
dence the true mean lies.                                    since, from statistical tables, for a t-distribution with
   A 95% confidence interval that ranges from 10 to 30       10 À 1 ¼ 9 degrees of freedom, p(T < 2.26) ¼ 0.025. That
indicates that we are 95% confident that the true            is, we are 95% confident that the true mean density of
population mean lies between 10 and 30. Suppose              bramble shoots lies between 3.3 and 9.1 shoots per
the sample has yet to be selected, then we define upper      quadrat.
and lower 95% confidence limits, L and U, for the               Formulae for calculating standard errors for
population mean, , so that                                  stratified and two-stage sampling are provided in
            p (L <  < U ) ¼ 0.95,                           Box 2.13.
                                                                The value of 95% is by far the most common
or equivalently
                                                             confidence level used, although in some circumstances
            p( < L) þ p ( > U ) ¼ 0.05,
                                                             a higher or lower level of confidence may be
where p stands for probability. L and U are functions of     appropriate.
the mean of the sample to be selected. Once we have             Presence–absence data generally conform to the
our sample data we can calculate particular values of L      binomial distribution (see Box 2.7). In this case, the
and U; let us call them l and u. If it were possible to      number of samples required to provide a given level of
select many different samples then each would result         precision depends solely on the proportion of all
in different values of l and u, but 95% of the confidence    quadrats containing the species, which is unknown.
intervals calculated from these samples would contain        This will also depend on the size of quadrat chosen
the true mean.                                               (Appendix 4). For large sample sizes and proportions
   To calculate l and u for a particular sample we need      greater than 0.1 and less than 0.9 the methods
to know something about its probability distribution.        described above for normally distributed data are
For example, for a mean that is normally distributed         probably sufficient (see Table 2.5). Otherwise, exact
the standardised mean will follow a t-distribution           binomial confidence intervals can be calculated in a
with nÀ1 degrees of freedom, where n is the sample           spreadsheet by using the method described at, for
size, and 95% confidence limits are calculated as            example, www.itl.nist.gov/div898/handbook/prc/
follows:                                                     section2/prc241.htm.
                                                               For quantitative measurements that have a very
   l ¼ x À ðtnÀ1 Â SEÞ and u ¼ x þ ðtnÀ1 Â SEÞ;
                                                             non-normal distribution it is probably best to use
where tnÀ1 is the value of a t-distribution, T, with n À 1   a data transformation so that the transformed data
degrees of freedom such that p(T < tnÀ1) ¼ 0.025. This       are approximately normally distributed. Common
value is used because we want a confidence interval          transformations are covered in Section 2.6.4, but it is
that is symmetric around the mean and so we set both         important to note here that, if a transformation is
p ( < L) and p ( > U) to be 0.025.                         used, the mean of the transformed data will not
   For example, suppose we have counts of the number         usually be the same as the transformed mean of the
of bramble shoots for a simple random sample of ten          original data. Thus the resulting confidence interval
quadrats. The sample mean is 6.2 and the standard            will be for the reverse transformed mean of
deviation is 4.08. Assuming the population of potential      the transformed data. See Section 2.6.4 for an
quadrats is large, the standard error is therefore           example of this.
40      2 PLANNING A PROGRAMME



confidence interval for the attribute we are estimating   birds or juvenile population size of newts after
and its calculation depends on the nature of the          breeding). The best times to survey particular vege-
measurements collected and their probability distr-       tation types are shown in Box 6.13. It is inevitable
ibution (see Box 2.8 for more information).               that some surveys will have to be carried out outside
   Being able to estimate the confidence interval         these times, but it is particularly important that
one would expect requires some knowledge about            repeat sampling of habitats for monitoring pur-
the mean and standard deviation of the measure-           poses is carried out within about two weeks of the
ment of interest. This is usually obtained from a         day of the year that the original survey was carried
pilot survey or from a previous, similarly designed,      out. Serious bias may occur if surveys are carried out
survey. Suppose the pilot study suggests the mean         at different times of year.
is " and the standard deviation is s. If the data are
   x                                                         The timing also depends on what is being
approximately normally distributed then, for the          monitored; the best time to monitor floristic
main study, we would expect the 95% confidence            components may differ from the best time to moni-
interval that would result from a sample size of n        tor grazing impact. For example, in heathlands, the
to be:                                                    best time to assess grazing impact is March–May,
                               pffiffiffi                       but this is probably not a good time to assess flor-
                    x Æ tnÀ1 s= n:                        istic composition. The effect of detectability on
                                                          data is important. In some cases, plants are best
Thus if we want the main study to provide a               counted after flowering; this avoids the possibility
confidence interval that extends no more than p%          of being attracted to flowering plants only.
either side of the estimated mean, n should be
such that
                                                          2.3.7      How will consistency be assured?
                                      2
                 n > ð100 tnÀ1 s=pxÞ :                    For reliable monitoring it is essential that the
                                                          methods used for assessing attributes are constant
For sample sizes greater than about 30, tnÀ1 can be       between surveys. Therefore, before the first survey
replaced by 2.                                            is carried out a standard operating procedure (SOP),
                                                          otherwise known as a monitoring protocol, should
Power analysis                                            be written, describing in detail the methods to
Power analysis is a method for establishing the num-      be used, so that everyone understands what is
ber of samples required to detect a given change          required and the methods are kept consistent
with a given level of statistical certainty. Before       between observers and years. Any modifications
continuing we need to introduce the idea of hypoth-       that are found to be necessary during the execution
esis testing, which, among other things, provides a       of the survey should be noted down immediately
framework for deciding whether a change is likely         afterwards and the SOP amended accordingly. The
to have taken place. Box 2.9 provides the necessary       SOP should then be followed as closely as possible
introduction, an understanding of which will also         in all subsequent surveys. However, if deviations
be important for the account of data analysis in          from the SOP are necessary, then these should be
Section 2.6.                                              recorded. Monitoring reports should ensure that the
                                                          SOPs are written out in full in the methods section or
2.3.6      When should data be collected?                 placed in an appendix. Deviations from SOPs should
                                                          also be reported in the monitoring report, and the
It is important that repeat sampling is carried out       implications for the results and interpretation of the
at approximately the same time of year each year          monitoring should be discussed.
unless seasonal cycles are being investigated. The           When designing a SOP, consultation with the
time of sampling will depend on the attribute being       appropriate government agency specialist is
measured (e.g. winter populations of migratory            important in case a standard procedure is already
                                                                                              2.3 Sampling strategy           41




Box 2.9 Hypothesis testing                                       In a monitoring study a Type II error amounts to
                                                              concluding that no change is taking place when in fact
Once data have been collected we are typically                it is. In many situations it is preferable to err on the side
interested in using them to test some hypothesis about        of caution and try to limit Type I errors. However, Type II
the population under study: for example, that the             errors may have profound consequences in monitoring
population of birds at a site is above some preset level,     studies because real changes in the condition of a feature
or that the percentage of a raised bog covered by             may not be detected. For monitoring studies it may
sphagnum moss is higher in year 2 than in year 1.             therefore be prudent to follow the precautionary princi-
Because our measurements are from a sample of the             ple and specify significance levels above 5% at least as a
site or population, we cannot test these hypotheses           trigger for further studies.
with certainty, but provided measurements are from
a statistically representative sample we can estimate         CHOOSING A TEST
how confident we are that the hypothesis holds.               Performing a test involves calculating an appropriate
   Hypothesis testing involves comparing two hypoth-          test statistic and deciding whether its value is extreme
eses, a null and an alternative hypothesis. Typically, we     enough to warrant rejecting the null hypothesis. The
want to test whether the data from the sample provide         type of test chosen will depend on the type of data
any evidence for rejecting the null hypothesis (usually       being analysed and the assumptions that we make
denoted H0) and accepting the alternative hypothesis          about the distribution of the data (see Section 2.6.4 and
(usually denoted H1). Thus for example, H0 might be           Figure 2.11). For example, one class of statistical test,
that sphagnum cover has not changed across the site           known as parametric tests, assumes that the data are
and H1 that it has. Has there been a sufficiently large       drawn from a particular distribution. The normal dis-
change in the sample measurements of cover that it is         tribution is the one most commonly used; if continu-
unlikely that H0 is true and that it can be rejected?         ous data are not normally distributed, they can often
   To see whether the collected data are consistent           be transformed into a closer approximation of a nor-
with the null hypothesis of no change, we need to             mal distribution (see Section 2.6.4). Alternatively, non-
estimate the likelihood of observing such a large             parametric or resampling methods can be employed,
change in a sample given that the null hypothesis is          which do not make assumptions about the underlying
true. If this likelihood is small then the null hypothesis    distribution of the data.
is unlikely to be true, and it is likely that a real             The test statistic is compared with tables of values
change has taken place. This likelihood is termed the         derived from the appropriate statistical distribution
significance level of the test and is frequently set at 5%.   for the required level of significance. If the test statistic
This would mean that if the null hypothesis is true only      is greater (or, for a few tests, smaller) than the tabulated
1 in 20 samples would be expected to show such a large        value for the chosen significance level, then we can
change, suggesting that the null hypothesis is false.         conclude that the null hypothesis (no difference) can be
                                                              rejected. By convention, 5% is frequently used as the
TYPES OF ERROR                                                significance level, but it is useful to present the actual
Whatever the significance level used there is always a        level of significance at which the null hypothesis would
chance of incorrectly rejecting the null hypothesis, and      be rejected. A very small value would give strong
this is termed a Type I error. In addition, there is always   evidence for rejection, but a value around, say, 10% would
a chance of incorrectly accepting the null hypothesis         still raise doubts that the null hypothesis is correct.
when it is false, and this is called a Type II error. After      For example, suppose that the null hypothesis is true
all, just because a test is not significant, it does not      and the test statistic is normally distributed. If its value
necessarily follow that the null hypothesis is true. For      under this hypothesis lies in either of the 2.5% tails, there
example, it is possible that the change in sphagnum           is a 95% chance that the data come from a different
cover is quite small and that insufficient data have          distribution, and so there is evidence for rejecting the
been collected to detect it.                                  null hypothesis. See Figure B2.9.
42       2 PLANNING A PROGRAMME




     ONE- OR TWO-SIDED TESTS                                             The power of a statistical hypothesis test is the
     In the example above, the statistical test is two-sided: H0      probability of rejecting the null hypothesis given that
     is rejected if the test statistic falls in either tail of the    the alternative hypothesis is true, and, in the case of
     distribution. These are the norm and arise, for exam-            monitoring, is therefore a measure of the likelihood of
     ple, if the alternative hypothesis is that there has been        correctly deciding that a change has taken place.
     a change, without specifying the direction of change.            Conversely, the significance level of a test is the
     This is normal because we do not usually know the                probability of rejecting the null hypothesis when it is
     direction of change in advance. A one-sided test is              true. Thus it measures the likelihood of incorrectly
     appropriate if a one-sided hypothesis is specified in            deciding that a change has taken place. This is illu-
     advance. This arises, for example, if we want to test            strated, with an example for normally distributed data,
     whether an attribute is below a specified value.                 in Box 2.10.
                                         Reject Ho           Accept Ho         Reject Ho



                                              2.5%              95%            2.5%




                Figure B2.9. Accepting and rejecting the null hypothesis. See text for details.




used for the species or habitat and to check the                      two repeats. Alternatively, the site may have to be
suitability of the proposed monitoring method.                        resurveyed for a third time, or one set of results
   It is important to check the repeatability of the                  may have to be discounted. The final decision
method used. This can be tested by having one                         should be accurately recorded so that the same
observer repeat a survey immediately after another                    remedy can be applied in the event of the problem
observer, or by the same observer conducting                          recurring (see p. 21).
duplicate counts. The results of quality control are                      There is no doubt that the accuracy of interpre-
useful for several reasons. Apart from highlighting                   tation is considerably enhanced when one recorder
the occurrence of any differences that may be                         is involved in repeating work on one site for a long
present due to the ability of a surveyor, such as in                  period of time.
species identification, plant cover estimates,
interpreting maps or using a compass, it may be
                                                                      2.4       REVIEWING THE MONITORING
possible to incorporate the results into statistical
                                                                                PROGRAMME
tests. Confidence limits and standard errors can be
calculated based on the variation in total counts or
                                                                      2.4.1       Are there sufficient long-term
mean values in order to break down the variation
                                                                                  resources available?
caused by observer bias compared with the varia-
tion due to other reasons. If major discrepancies                     It is important that the long-term future of the mon-
are found between two surveys, the underlying                         itoring programme be considered from the outset;
cause should be identified and corrected if possi-                    monitoring of some species and habitats has to be a
ble. If this is not achievable, the results for the                   long-term undertaking. However, many projects
survey on which quality control was carried out                       start off being too ambitious. Therefore, before
may need to be taken as the average of the                            embarking on the detailed design of a monitoring
                                                                                         2.4 Reviewing the programme             43




Box 2.10 Power analysis                                               ¼ P(À z/2 < T < z/2 \ H1 is true) ¼ chance of Type II
                                                                error;
When testing a hypothesis we calculate the appropriate             1 À  ¼ P(T > z/2 when H1 is true) ¼ chance of
test statistic T, and if that statistic exceeds a critical      correctly deciding a change has occurred (the power
value t, the null hypothesis, that no change has occurred       of the test);
between time 1 and time 2 (H0: 2 À 1 ¼ 0) is rejected         where T is a test statistic that follows a standard normal
in favour of the alternative hypothesis (H1: 2 À 1 6¼ 0),     distribution and will be calculated after the data have
where 1 and 2 are the population values at times 1            been collected, and z/2 is the value from tables from
and 2, respectively.                                            the normal distribution for the chosen significance
   The ‘power’ of the test is therefore                         level, . For example, if  ¼ 0.05 then z ¼ 1.645 and
p(T > t \ 2 À 1 6¼ 0), where ‘\’ means ‘given that’. Thus     z/2 ¼ 1.96.
the power is the probability that the test statistic                 The power, 1À, for a given sample size can be
exceeds the critical value when the means of samples            found from the following equation:
from time 1 and time 2 are different (the test reaches
the correct conclusion).
                                                                                     z ¼ d=sd À z=2 ;
   Plotting this probability against 2 À 1 shows how
                                                                where sd is the estimated standard error of a level of
the power increases as 2 and 1 become further apart,
                                                                change, d, for the given sample size. For
i.e. we are much more likely to detect a change if that
                                                                non-permanent plots, this is usually calculated by first
change is large (see Figure B2.10).
                                                                estimating the sample standard
   When 2 À 1 ¼ 0 the probability that T > t is
                                                                deviation, s, from the pilot data. For reasonably large
the probability  of rejecting H0 when H0 is true: the
                                                                sample sizes, sd is then usually estimated as
significance level, or the probability of a Type I error.
   When 2 À 1 6¼ 0 the probability that T          t is the                                  p
                                                                                        sd ¼       ð2s2 =nÞ :
probability of not rejecting H0 when H1 is true () À a
Type II error À and the power is 1 À .                         Trying different values of n will give an indication of
    A power analysis consists of calculating the number         the sample size required to achieve a given power.
of samples required to detect a given level of change for       Permanent plots are more problematic in that sd will
chosen values of  and .                                       depend on how consistently the plots change, which is
                                                                difficult to predict.
NORMALLY DISTRIBUTED DATA                                       If plots are non-permanent and simple random
The example presented here is suitable for equally              sampling has been used (i.e. without stratification,
sized, normally distributed samples (software is                etc.), these formulae can be solved for n:
available for this and other distributions; see below).
A pilot survey is required to obtain an initial estimate                                  2ðz=2 þ z Þ2 s2
                                                                                    n¼
of the variance of the data.                                                                    d2
   We need to decide what we consider to be the
                                                                or
acceptable probability of concluding that no change
is taking place when in fact it is, and of concluding that
                                                                                         2ðz=2 þ z Þ2 P 2
                                                                                                          cov
change is taking place when in fact it is not. The null                           n¼
                                                                                                C2
hypothesis, H0, is that no change has occurred. The
alternative, H1, is that change has occurred.                   where s2 is the sample variance from the preliminary
   We define:                                                   survey, Pcov is the percentage coefficient of variation
 ¼ P(T > z/2 or T < À z/2 \ H0 is true) ¼ chance of Type     (see Glossary) of the preliminary sample, and C is the
I error (the significance level);                               change to be detected expressed as a percentage.
44       2 PLANNING A PROGRAMME



     It can be seen from these equations that the number of                Statistical power depends on the type of statistical
     samples required to detect change will increase as the            test chosen. One test is said to be more powerful than
     significance level of the test increases (i.e. as the             another if it is more likely to reject the null hypothesis
     chance of making a Type I error decreases), the power             when the null hypothesis is false.
     of the test increases (i.e. the test is more likely to be             Power analysis is most useful when planning
     correct) and the variance of the data increases. The              a study, at which point it is used to calculate the
     number of samples required decreases as the level of              number of samples needed to detect a given change
     change required to be detected increases.                         with a predetermined power and significance level
                                                                       (e.g. to detect a 10% change in cover with a power of
     EXAMPLE                                                           80% at the 5% significance level). When selecting a
     Let us assume that we are monitoring the number                   suitable significance level and statistical power, the
     of orchids flowering in a meadow by using simple                  precautionary principle should be considered (i.e. is it
     random sampling. The change to be detected is set at a            better to conclude that change is taking place when in
     mean decrease of flowers per 4 m2 quadrat of 10%. We              fact it is not than to conclude that no change is taking
     will use the conventional significance level of 5% for           place when in fact it is?).
     and we will arbitrarily set the power of the test as 60%              A retrospective power analysis can be carried out
     (i.e. there is a 40% chance that the test will wrongly            after the study, which can be useful if a non-significant
     accept the null hypothesis).                                      result is obtained (see, for example, Thomas & Juanes,
        A preliminary study was carried out, and the                   1996). In this case, sample size and significance level
     percentage coefficient of variation of the number of              are known; these, and the estimate of variance
     orchids in the quadrats was estimated to be 36.21. Using          obtained from the study, can be used to calculate the
     tables we obtain z0.2 ¼ z0.025 ¼ 1.96 and z ¼ z0.4 ¼ 0.25.       size of change that was detectable with the desired
     The number of samples required is therefore:                      level of statistical power.
                                                                           Power analysis is useful because it can provide an
              2ð1:96 þ 0:25Þ2 36:212 12807:7                           indication of what is achievable for a given amount of
           n¼                       ¼        ;                         effort. Thus it may become apparent that only very large
                       102             100
                                                                       changes are detectable with current resources and that
            t ¼ 128:08 % 128:                                          to detect small amounts of change, particularly in
        So roughly 128 quadrats would be needed to detect              variable populations, requires a lot of effort. If this is
     a change of 10% in orchid numbers per quadrat at the              considered at the planning stage of a monitoring
     95% significance level with a power of 60%.

                                                              P (T>t \ µ 2 – µ1 ≠ 0)

                                                                                       β




                                                                              α
                            µ 2 – µ1
                                                                       0
                                              H1 true                                      H1 true


                                                                    H0 true

               Figure B2.10. The increase in the power of detecting change as the change increases
               in extent.
                                                                                  2.4 Reviewing the programme           45




  programme it will help to avoid the possibility that the   powcase/monitor.html, but some care is needed to
  monitoring will fail to achieve its objectives.            ensure that the assumptions made by MONITOR
    Performing a power analysis requires data from a         are likely to be met. For an introduction to power
  pilot or previous survey that provides an indication of    analysis, a software review and details of where
  variability in the measurement across the population.      to obtain power analysis programs see the
  Software is usually needed to do the necessary             Internet site www.interchg.ubo.ca/cacb/
  calculations and there are a number of programs            power and the US Geological Survey web page
  freely available. An example is DSTPLAN, which             www.mp1-pwrc.usgs.gov/powcase/powcase.html;
  is available at odin.mdacc.tmc.edu/anonftp/. The           see also Thomas & Krebs (1997).
  program MONITOR can be used to estimate the number            Further information on this subject is also available
  and intensity of surveys needed to achieve a given         in Rotenberry & Wiens (1985), Lipsey (1990), Peterman
  power for detecting trends over time. This program can     (1990a,b), Muller & Benignus (1992) and Taylor &
  be downloaded from www.mp1-pwrc.usgs.gov/                  Gerrodette (1993).


programme, it is vital to assess the resources avail-        excluding attributes for which monitoring is dis-
able, including funding, staff time, staff expertise         cretionary. However, it should be remembered that
and existing equipment. This should then be taken            excluding other features and attributes from mon-
into account in the selection of features and attri-         itoring may be a false economy. In the long term,
butes to be monitored (see Sections 2.1.1 and 2.1.2)         the costs of restoring habitats or species popula-
and the frequency of monitoring (see Section 2.1.3).         tions may far exceed the costs of monitoring and
However, as a minimum, resources should be suffi-            early management intervention.
cient to provide an adequate standard and fre-                  At the outset it is important to work out how
quency of monitoring for all the features and their          much time the optimum monitoring protocol will
attributes for which monitoring is mandatory.                take to achieve. Then determine whether sufficient
   Following this initial assessment, the full study         resources (especially suitably qualified staff) will be
requirements should be assessed after establishing           consistently available when required.
the optimum methods. Monitoring costs should be
based on the most cost-effective method that meets
the objectives for monitoring each attribute and
                                                             2.4.2       Are personnel sufficiently trained
the required standards of precision and accuracy,
                                                                         and experienced?
etc., as described in Section 2.2.
   The assessment should take a long-term view of            Consideration of staff resources available for mon-
the requirements for monitoring and available                itoring must include an assessment of the expertise
resources, including likely year-to-year variations          and experience necessary for the chosen methodol-
in monitoring needs and budgets. A poor monitor-             ogy and, if necessary, the acquisition of a licence (see
ing design is one in which the monitoring effort             below). As a minimum, it is essential to be familiar
changes from year to year, or in which monitor-              with the habitat, study species and survey methods
ing is dropped in one year because of a lack of              required. The correct identification of target species
resources. This variability introduces yet another           may require specialist personnel even if the meth-
confounding factor, which will cloud the interpre-           ods themselves are straightforward. Alternatively,
tation of the results obtained.                              the method itself (e.g. electrofishing or bird ringing)
   If the resources needed for a full monitoring             may require specialist training and/or licensing. If
programme exceed those available, the two                    the monitoring involves several people they should
options are: (i) to seek more funds; or (ii) to trim         all be trained to a minimum standard and recording
the monitoring programme in the least damaging               techniques should be standardised; this can be done
way, e.g. by monitoring less frequently or by                as part of a preliminary study.
46      2 PLANNING A PROGRAMME



   Monitoring work may well be contracted out to         valuable in determining what equipment is neces-
staff from outside agencies. These people must also      sary. Going into the field and realising that a piece
be suitably trained and experienced to carry out the     of equipment is required halfway through getting to
work to a sufficient standard.                           the site, or once on it, is disorganised and may waste
                                                         considerable time and money. On the other hand
                                                         going into the field laden with excessive gear will
2.4.3      Are licences required, and are there
                                                         slow you down and is unnecessary.
           animal welfare issues to consider?
An important consideration when surveying and
monitoring species (particularly for rare species)       2.4.5      Are there health and safety issues
is that staff may need to hold a licence. For exam-                 to consider?
ple, under the 1981 Wildlife & Countryside Act,          Fieldwork can be a dangerous activity and so
licences are required from the relevant government       before carrying out any such work, a careful risk
department or agency to enter bat roosts, trap pro-      assessment should be undertaken to identify
tected species such as Great Crested Newts, and          potential risks and minimise these by ensuring
survey many rare breeding birds. Invasive mark–          that safety precautions are strictly followed.
recapture methods such as toe-clipping for amphi-        Recommended precautions for general fieldwork
bians may require a Home Office licence under the        are given in Box 2.11, but these are not intended
Animals (Scientific Procedure) Act 1986. It is there-    to be comprehensive. Fieldwork involving the
fore necessary before a method is selected and used      use of specialist equipment or activities, such as
that the need for a licence be investigated. If one is   diving, will certainly need additional safety meas-
required, staff should obtain the licence and any        ures and may well require staff to be suitably
necessary training in advance.                           qualified.
   Animal welfare issues may need to be considered:
certain survey and monitoring techniques may have
unacceptable effects on the animals being surveyed,      2.5     DATA RECORDING AND STORAGE
or on other groups (for example, small mammals
                                                         2.5.1      How will data be recorded
may be killed in invertebrate or amphibian pitfall
                                                                    in the field?
traps). Apart from the obvious point that humans
are morally obliged not to cause unnecessary suffer-     Once the sampling protocol has been defined,
ing to wildlife in the cause of surveying, there are     the field data sheets can be designed. Specially
other reasons for considering animal welfare. Many       designed forms encourage consistency and reduce
scientific journals such as the Journal of Zoology are   unnecessary writing. Where lots of data are being
now asking authors and referees specifically to          recorded relatively quickly it may be advantageous
address whether animal welfare issues have been          to type the data directly into a hand-held datalog-
taken into consideration. In addition, public (and       ger. A database structure should be written, which
political) support for monitoring activities may be      prompts the observer to enter the appropriate
affected by the impact of survey methods on wildlife     record. The advantage of this method is that a
and the environment in general.                          large dataset can be downloaded directly to a
                                                         computer.
                                                            Some remote sampling in which continuous
2.4.4      Is specialist equipment required
                                                         recording of environmental variables is required
           and available?
                                                         as part of the habitat condition assessment can
All equipment needed for the monitoring study            also be achieved with automatic dataloggers. It is
should be made available for its duration so that        unlikely that automatic datalogging will be essen-
standardised methods are employed. A scoping exer-       tial or cost-effective for the majority of methods of
cise prior to starting the formal monitoring may be      assessing habitat condition.
                                                                                  2.4 Reviewing the programme             47




Box 2.11  General health and safety considera-                  items such as flares, electronic devices and air/- or
tions for working in the field                                  gas-pressured alarms should be considered.
                                                            *   The International Alpine Distress Signal is six long
*   Before undertaking monitoring, survey work, etc.,           whistle blasts or torch flashes in succession,
    discuss the proposed activity and terrain with your         repeated at 1 min intervals. The reply is three long
    line manager and others with relevant knowledge             whistle blasts or torch flashes repeated at 1 min
    and experience. This will help in deciding the              intervals.
    relevance of the items below and those elsewhere in     *   Always carry a first aid kit and know how to use it.
    this Handbook.                                              Emergency first aid training is available for those
*   Lone working procedures should be followed. The             not in possession of full certificates.
    minimum requirements, whether alone or as a             *   Inoculation against tetanus is strongly
    party, are that you leave details of your itinerary         recommended for all staff engaged in fieldwork.
    with a responsible person; you make arrangements        *   Staff receiving special medical treatment, such as
    to contact a responsible person at least every eight        a course of injections, or suffering from medical
    hours and at the end of the working day; and you            conditions, such as diabetes, allergies, rare blood
    ensure that your contact knows what to do if you fail       groups, etc., are reminded of the advisability of
    to make scheduled contact.                                  carrying a card or some other indication of special
*   Always have suitable clothing for the activity,             medical requirements.
    terrain and weather conditions. Principles of good      *   Where applicable, sufficient additional medicines,
    clothing concern insulation, and protection from            etc., should also be carried on field trips to ensure
    precipitation and wind. Although it should be               that no medical complications arise owing to lack of
    recognised that survival in exposed winter                  treatment. In an emergency the carrying of such
    mountain environments can be extremely difficult            items can save a lot of time and perhaps save
    without improvising an effective shelter, a fair test       your life.
    of your clothing and equipment is the answer to the     *   Staff visiting hazardous areas should inform those
    question: could I survive, be it very uncomfortably,        based at the location of any special medical
    if I were immobilised for 24 hours? High-visibility         condition, e.g. diabetes. This, in the event of an
    clothing is desirable in many situations, both to           accident or the person becoming lost, is of great
    prevent accidental injury and, more importantly, to         value to the rescue services. If a party is well
    be located in an emergency. Boots provide                   equipped and it is overdue, no great concern may be
    protection and grip. In general, choose the lightest        shown for several hours. However, if a member of
    pair that will do the job: the requirement for a            the group requires regular treatment, a search may
    rigid-soled heavier boot increases if you will be           be speeded up; for example, instead of a preliminary
    travelling in steep, rocky and winter terrain.              foot search being undertaken, a helicopter could be
*   Carry a map and compass. Know how to use the                called for immediately.
    compass to take a bearing, set a course and walk        *   If you are new to an area, ask the area staff about any
    on a compass bearing.                                       hazards.
*   Consider whether a survival bag might be necessary      *   Do not fail to inform visiting members of staff of any
    in remote and/or upland or/mountainous situations.          dangers in the area they intend to work in.
    Spare blankets are not recommended.                     *   Staff about to embark on a rigorous period of
*   Take spare warm clothing.                                   fieldwork, especially after a period of relative
*   Have with you some high-sustenance food such as             inactivity, are reminded that some attention given
    sweets, chocolate, glucose tablets or biscuits.             to physical fitness beforehand can make the job
*   Always carry some means of raising alarm.                   more enjoyable as well as being a positive
    A whistle and torch are essential items and other           contribution to safety.
48        2 PLANNING A PROGRAMME




     *   Always move carefully over rough, rocky or             *   Take care to avoid hazardous substances such as
         vegetation-covered ground, avoiding any loose              herbicides and pesticides.
         boulders, etc. Care should be taken on wet ground      *   Exercise extreme caution in areas of landfill, tips
         such as bogs, mud or fens.                                 and spoil heaps, which could be unstable,
     *   Never run down scree slopes or steep hills, and take       especially in wet weather. Look out for weakness
         care not to dislodge loose rocks or other objects.         resulting from underground combustion and for
     *   Before setting out on a field trip check the local         any toxic substance, including gas, that may be
         weather forecast. This could save a wasted journey         present.
         or prevent you or your party encountering adverse      *   Identify areas where game shooting may take place.
         weather conditions, which could put your lives             Find out when and where this is taking place and
         at risk. Be aware of weather conditions around             take appropriate measures, including wearing
         you while outside, for example distant storms,             high-visibility clothing.
         which may change direction and come towards you.       *   Note that care of general health while doing
     *   Avoid machinery, whether in use or not.                    fieldwork is essential, as exhaustion can lead to
     *   Enquire about and avoid potentially dangerous              careless mistakes and lower resistance to diseases.
         animals.                                               Source: extract from SNH Health and Safety Manual.



                                                                   Back-ups of all data files should be kept on disks
2.5.2          How will the data be stored?                     or different computers, preferably in different build-
Storage directly on to a computer or interface in the           ings. Logs of existing data, with descriptive details
field saves time but machines are prone to breaking             and locations, should be kept for all sites. Hard copies
down and may be expensive to back up. Work out a                of all data should also be kept. Although much of this
standard procedure for storing the data with under-             is common sense and generally accepted good prac-
standable file names if on a computer, or filed by              tice, it is surprisingly often ignored.
project or site name if in map form. Maps can also be
scanned in and stored on computer hard drives, or
                                                                2.5.3          Who will hold and manage the data?
CD-ROM as a security measure. Collecting data in
the field is laborious, expensive and difficult to              It is usually valuable to make one person respon-
repeat, so good computer and digital storage of                 sible for databases and for managing them, i.e.
information is sensible. The information storage                updating, upgrading, producing reports from them,
needs to take account of software and hardware                  and so forth. Some databases, such as Microsoft
obsolescence: data will need to be retrieved several            Access1, allow the manager to design standard
decades from now. Consider also the requirements                reporting forms and outputs, which anyone
of data analysis software, which may only read data             responsible for a site, or an aspect of the habitat,
organised in a particular format.                               or, for example, policy factors affecting it can use
   Databases, such as Microsoft Access1, enable easy            to produce standard outputs. This limits individual
manipulation of data for various methods of analysis.           bias in interpretation and presentation. Such data-
They are also useful for holding textual data, such as          bases can be made read-only prior to entering a
descriptions of sites, changes to vegetation, etc.,             password, which prevents data from being chan-
which are non-numeric. However, spreadsheets                    ged by unauthorised personnel. However, the data
offer better capacity for analysing numerical data              should be made available for use throughout an
and are easier to use, especially for beginners. If             organisation and beyond, depending on commer-
the data are entered into a spreadsheet they can                cial confidentiality or other constraints, so it is
usually be imported into a variety of statistical               important to make sure that people know of its
packages for analysis.                                          existence and the name of the contact person or
                                                                    2.6 Data analysis and interpretation   49



data manager so that they can gain access to it        analysis used to demonstrate that such a change is
should they need to.                                   occurring.


2.5.4     Will the data be integrated with
          other datasets and if so, how?               2.6       DATA ANALYSIS, INTERPRETATION
                                                                 AND REVIEW
If data are held in a program suite, such as that
containing Microsoft Access1 or Excel1, they can       2.6.1     Who will carry out the analysis
be integrated with data collected by someone else                and when?
quite easily. If stored on a compartment basis (i.e.   Data should be collected, stored and filed in such a
a management compartment for a site, as used in        way that anyone with the required skills should be
the standardised Countryside Management System         able to analyse it. It is always important to describe
(CMS)), integration with species-based data can        through written SOPs:
be achieved in relation to compartment-based
management projects. This is an essential part
                                                       *   how and when the data were collected;
of project planning as part of the Conservation
                                                       *   what problems and/or issues arose and how they
Management Plan for a site. For example, data              might affect the interpretation of the data;
for a site may be held on a compartment-by-
                                                       *   the sampling design together with clearly labelled
compartment basis, corresponding to manage-                maps of site and stratum boundaries;
ment units listed as projects in the Conservation
                                                       *   the notation and codes for species; and
                                                       *   the format, location and file names of computer
Management Plan for the site. As such, both textual
and numerical data can be held in the same file.           datasets or hard copies.
In addition, there could be a number of fields that    In addition it is important that monitoring pro-
describe other data held for the site, which have      grammes should identify the resources required
been collected elsewhere. This information should      for data analysis and the writing of reports, who
consist of type of information, e.g. habitat survey    should be responsible for this, and when it should
data, year(s), compartments, whether material          be undertaken. Often this is overlooked and data
has been published, and perhaps compatibility          accumulate that are never properly analysed and
of these data with those held in the monitoring        presented.
database. Links to research projects could similarly      The data analysis should be carried out by some-
be made.                                               one with a good understanding of statistics and, in
   Spatially referenced data can be integrated into    particular, an awareness of when particular analy-
a geographical information system (GIS) such as        tical methods are appropriate and the potential
ArcGIS1. GIS systems are becoming increasingly         pitfalls associated with their use. Misapplied tests
widely used; they can add value to the analysis of     or poorly presented data can lead to misinterpreta-
spatially referenced data by enabling other datasets   tion and poor management decisions.
held for the site to be overlaid and compared (for
example, data on soil type can be overlaid and
                                                       2.6.2       What are the steps in analysing
correlated with data on vegetation communities).
                                                                   data?
The examination of spatial trends in the range of
a species can also be carried out with a GIS pro-      A comprehensive account of statistical methods for
gram. In addition, the ability to generate visual      data analysis would take up most of this book and
representations of your data can greatly enhance       there are already numerous books devoted to this
the ease with which it can be interpreted and          (see the suggested references at the end of this
understood. For example, a map of a site showing       section). The sections that follow simply outline
changes in the extent of a particular habitat type     the approach to take with some common methods
will lend weight to a description of the statistical   and the pitfalls to look out for.
50       2 PLANNING A PROGRAMME




     Further information on statistical techniques                Sokal & Rohlf (1996)
                                                                  Young & Young (1998)
     BOOKS                                                        Zar (1984)
     See References for full details.
        Bailey (1981)                                         INTERNET SITES
        Dytham (2003)                                         www.ltsn.gla.ac.uk (a good starting point for online
        Fowler et al. (1998)                                  statistical resources)
        Kent & Coker (1992)                                   www.anu.edu.au/nceph/surfstat/surfstat-home/surf-
        Krebs (1999)                                          stat.html (an online text in introductory statistics)
        Manly (1997)                                          www.stats.gla.ac.uk/steps/glossary/index.html
        Mead et al. (1993)                                    (detailed statistical glossary)




  Three distinct stages in the analysis of survey                 data values or how different measurements are
and monitoring data can be identified, as follows.                related. These can affect how the data are then
                                                                  analysed and interpreted. For example, are the
1. Description and presentation of data
                                                                  measurements of vegetation height normally dis-
2. Making inferences about the site or population
                                                                  tributed or are there distinct vegetation types on
3. Interpretation and presentation of findings
                                                                  the site, resulting in a more complex distribution?
Each of these stages is discussed in turn and                     Alternatively, is there an association between the
together they provide a framework for ensuring                    abundance of one species and another? Spatial
that appropriate methods are used and that the                    patterns are only likely to be revealed by map-
findings are communicated successfully.                           ping; if these can be combined with other local
                                                                  datasets, additional relationships may become
2.6.3         Description and presentation of data                apparent.
                                                              *   The main features of the data are poorly presented.
The importance of exploring and summarising                       Interpretation and presentation of findings is cov-
data, before launching into anything more com-                    ered in Section 2.6.5, but it is worth noting here
plex, cannot be overstated. The dangers of missing                that graphical displays are often the most powerful
out this step include the following.                              way of communicating what the data show and
                                                                  it is worth taking time to find the best way of
*    Inappropriate analyses are used or the assump-
                                                                  achieving this.
     tions for these do not hold.
*    Peculiar or erroneous data values, which may exert       The best way of summarising and displaying data
     a strong influence on how the data are interpreted,      depends on the following.
     are not detected. These ‘outliers’ may be caused by
     measurement or recording error or mistakes dur-          *   The type of data available. Whether the data are
     ing data entry. Alternatively, they may be valid             nominal, ordinal or quantitative will affect what
     measurements that just happen to be rather                   descriptive statistics and displays make sense.
     extreme. In the latter case, it may be decided to        *   The amount of data available. With a sample size
     include these values in the analysis, but it is impor-       of 10 it will not be possible to say much about the
     tant to be aware of the extent to which conclusions          underlying probability distribution the data fol-
     are influenced by one or two outliers.                       low, but with 100, a histogram, or similar, should
*    Clear patterns and other features of the data are            be informative.
     missed. Graphical and tabular display of data can        *   The objectives for the analysis. If interest centres
     reveal important aspects of the distribution of              on whether a measurement has changed over
                                                                       2.6 Data analysis and interpretation   51



  time, examination of changes in some summary              In addition, the mean is sensitive to outliers, so
  measure, such as the sample mean, is appropriate.      if, for example, the variable has a distribution
  If relationships between measures are of interest a    where most of the values are fairly small but a
  scatterplot or cross-tabulation can be helpful.        few are fairly large, changes in the mean over
                                                         time will be very sensitive to the values from a
Nominal and ordinal data                                 small number of sampling units. In this case the
Nominal and ordinal data are summarised by cal-          median, which does not suffer from this problem,
culating the proportion of sampling units that fall      may be a better summary statistic.
within each category: for example, the proportion           Changes between surveys can be viewed by
for which the species was present or the proportion      plotting summary measures, such as means, as
of quadrats in a particular vegetation height class.     a time series. The addition of error bars to such
Such data can be displayed in bar charts, and shifts     charts, which may show confidence intervals or
in the distribution of values in each class may be       standard errors, can illustrate changes in the vari-
apparent by plotting the results of two or more          ables’ variability. Displaying the individual values,
surveys together. For ordinal data this might take       perhaps as a series of box plots or dot plots, can
the form of a series of stacked bar charts.              reveal changes in the distribution as well as in the
   Relations between two such categorical variables      mean or median.
can be investigated through cross-tabulation or bar         Scatter plots can help explore relations between
charts with one variable grouped within the other.       quantitative variables; and the extent of linear asso-
                                                         ciation can be measured by calculating correlation
Quantitative data                                        coefficients.
A much greater range of possibilities is available
for quantitative variables. If data from sufficient
sampling units are available, the distribution of        2.6.4        Making inferences about the site
each variable can be investigated by using graphs                     or population
such as histograms, box plots and dot plots. The
                                                         In general, we want to use data collected from a
latter two are particularly useful for revealing out-
                                                         sample of a population to be able to say something
liers. Histograms, box plots and normal-probability
                                                         about the population as a whole. That is, we want
plots can reveal peculiarities in the shape of the
                                                         to make an inference about the population. For
distribution (e.g. skewness) and indicate whether
                                                         example: has sphagnum cover changed and by
a data transformation (Section 2.6.4) might be
                                                         how much; has the abundance of species x
required prior to using hypothesis tests, as for
                                                         declined; or is the breeding success rate of species
example in Box 2.12.
                                                         y at a satisfactory level?
    Descriptive statistics such as the mean and stan-
                                                            Thus the reason for analysing survey and mon-
dard deviation of the dataset (Box 2.4) can be useful
                                                         itoring data will usually be either:
summaries, providing a central, middle value and a
measure of variability, at least for data that are not
                                                         *   to compare data from a single sampling occasion
too non-normal. Formulae for calculating mean and
                                                             against a pre-defined limit (for example, the limit
standard deviations for data collected by using stra-
                                                             below which a population should not fall); or
tified or two-stage sampling are provided in Box 2.13.
                                                         *   to compare data from two or more sampling
    The mean will not be a very informative sum-
                                                             occasions to determine what changes have
mary if the variable has a distribution with a cluster
                                                             occurred.
of low values and another of high values, for exam-
ple. In this case the distribution is said to be bimo-   Both of these objectives will usually involve the use
dal and this could happen if the site covers two very    of hypothesis tests (Box 2.9) and or confidence
different habitats or the population contains            intervals (Box 2.8). Hypothesis tests enable us to
distinct sub-populations.                                say, for example, whether there is evidence that
52           2 PLANNING A PROGRAMME




     Box 2.12 Histograms, box plots and normal                       (B)      BOX PLOT
     probability plots                                               The box (Figure B2.12b) extends to the lower and upper
                                                                     quartiles, with a line indicating the median. The lines
                                                                     outwith the box extend 1.5 times the width of the
     The data presented are percentage cover
                                                                     interquartile range or to the lowest or highest value.
     measurements for Heather Calluna vulgaris from 120
                                                                     Values outside this range are indicated as outliers.
     quadrats.
                                                                     For normally distributed data we would expect a fairly
                                                                     symmetrical plot around the median. This is not the
     (A)              HISTOGRAM                                      case here; a number of outliers are indicated.
     The area of each bar represents the proportion of
     measurements falling in the interval shown by the               (C)      NORMAL PROBABILITY PLOT
     horizontal axis. In most cases, where all bars have the         This plots the cumulative proportion of data values
     same width, the height of each bar is the number of             against that expected for normally distributed data
     measurements in that interval. In this example                  (Figure B2.12c). Departures from the straight line
     (Figure B2.12a) the distribution is clearly skewed              indicate departures from normality. Again the data are
     and a transformation is likely to be needed prior to            clearly non-normal and transformation is required if
     analysis.                                                       parametric methods are to be applied.

     (a)                                        (b)
                                                                                     (c)

                                                           100                                             1.00
             30
                                                            80
                                                                                     expected cum. prob.

                                                                                                           0.75
                                                            60
             20
                                                 % cover
     count




                                                            40                                             0.50

                                                            20
             10
                                                                                                           0.25
                                                            0

             0                                             –20                                             0.00
                  0   10 20 30 40 50 60 70 80                                                                 0.00   0.25   0.50    0.75   1.00
                             % cover                                                                                 observed cum. prob.



                         Figure B2.12. Examples of (a) histogram, (b) box plot and (c) normal probability plot.
                         See text for details.


change in some population measure has taken                             Statistical methods will usually make some
place. However, this of itself may be of limited                     assumptions about the distribution of the data, or
interest. Often we know that some change is likely                   more specifically about the test statistic. This is the
because populations and habitats are naturally                       summary measure for which we want to calculate
dynamic. Rather, we want to know how much                            confidence intervals or form hypotheses. Most
change there has been so that the ecological sig-                    commonly this will be the mean value or the
nificance of this change can be evaluated. This is                   proportion falling in a particular class. This sum-
where confidence intervals, or similar, are useful                   mary measure will also have a sampling distribu-
in providing a range in which we have some con-                      tion, because different samples will result in a
fidence that it contains the true level of change.                   different value for the summary measure, but its
See Eberhardt (2003) for a discussion of this and                    distribution may be very different from that of the
other issues surrounding hypothesis testing.                         original data.
                                                                                          2.6 Data analysis and interpretation                          53




  Box 2.13 Estimating means and standard                               way (Box 2.8). The number of degrees of freedom (df)
  deviations for stratified and two-stage sampling                     for calculating the required t-statistic is

                                                                                                    P
  STRATIFIED SAMPLING                                                                              ð gh s2 Þ2
                                                                                           df ¼ P 2 4 h            ;
  For each stratum (h) and for a measurement x, we                                               ðgh sh =ðnh À 1ÞÞ
  define:
     n ¼ total number of units (e.g. quadrats,                         where gh¼ Nh(NhÀ nh)/nh.
  individuals, etc.) sampled in all strata;
     nh¼ number of units sampled in stratum h;
                                                                       TWO-STAGE SAMPLING
     Nh¼ total number of possible sampling units in
                                                                       We assume that the same number of minor units are
  stratum h;
                                                                       sampled within each major unit. If we define the
     xh ¼ mean of x in stratum h;
                                                                       following:
     sh¼ standard deviation of x in stratum h.
                                                                          N ¼ number of major units available for sampling;
     Then if we calculate
                                                                          n ¼ number of major units sampled;
     Wh¼ stratum weight for stratum h ¼ Nh/Ntotal where
          P                                                               U ¼ number of minor units within each major unit;
  Ntotal ¼ Nh,
                                                                          u ¼ number of minor units sampled within each
     the estimate of the overall mean is calculated as
                                                                       major unit;
                              X
                                                                         then the overall mean is estimated as
                                   Wh xh :
                                                                         "¼ mean over the major units of the minor unit
                                                                         x
  If we define                                                         means.
                                                                          Defining
                            fh ¼ nh =Nh                                   s2 ¼ variance of the minor unit means;
                                                                           x

                                                                          ms¼ mean of the variances within minor units;
  then the standard error of the overall mean is
                                                                          the standard error for " is calculated as
                                                                                                 x
                     rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
                      X                                                          sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
                                                                                                        
             SEx ¼                 2
                              Wh s2 ð1 À fh Þ=nh :
                                       h
                                                                                               n  s2                            u ms 
                                                                                                           x
                                                                             SEx ¼     1À                        þ 1À                               :
                                                                                               N          n                       U un
  If this mean is approximately normally distributed
  then we can calculate confidence intervals in the usual




   As introduced in Box 2.7, a key result (the Central                 will require a large sample size before their mean
Limit theorem) states, in essence, that as the sample                  can be regarded as normal.
size increases the distribution of the mean of a                          An outline of the stages of data analysis is
variable from a random sample will converge to                         shown in Figure 2.11. This illustrates that selecting
being normally distributed. As many of the stan-                       an appropriate statistical method depends cru-
dard methods in statistical inference assume that                      cially on the distribution of the test statistic.
the test statistic is normally distributed, this result                The sections that follow describe the various meth-
is of huge significance. It means that, provided                       ods available. Methods primarily aimed at quanti-
the number of samples is sufficiently large, we can                    tative data are split into three broad classes:
assume that the mean of the variable of interest                       parametric methods, methods based on ranks and
is approximately normal. What we mean by suffi-                        resampling methods. The analysis of category
ciently large will depend on the distribution of the                   data is considered separately. A rather different
underlying data. For example, data that exhibit a                      approach, Bayesian inference, is also briefly
very skewed or otherwise non-normal distribution                       described.
54   2 PLANNING A PROGRAMME



                                        Summarise and                                         Correct data errors, if
                                        display variable                                           necessary




                                                 Determine approximate
                                                distribution of test statistic




                                                 Test statistic non-normal,
                    Test statistic is roughly           e.g. skewed         Distribution unknown (e.g.
                     normally distributed                                     because data too few)




         Apply tests and/or                                 Apply                                    Use distribution-specific,
       calculate confidence         Successful          transformation              Not                non-parametric or
        intervals for normal                                                     successful            resampling method
                data




                                                   Interpret and present
                                                          findings

           Figure 2.11. Flow diagram outlining the steps involved in data analysis, for each variable
           of interest.


   The choice of test also depends on whether                      analysis (see p. 61) or to use a non-parametric or
data for more than one survey are paired or                        resampling method.
unpaired. Paired data arise from permanent plots                      For quantitative data that are not very skewed in
in which the measurement taken in a given plot                     distribution, 25–30 samples are usually sufficient
during one survey can be compared directly with                    for assuming that the sample mean is approxi-
the measurement taken during another survey.                       mately normal. If the distribution is very skewed,
If a new sample is selected for each survey the                    for example because of a large number of zero
two samples cannot be paired in this way. Tests                    counts, rather more samples are needed. An alter-
for paired samples are usually more power-                         native is outlined on p. 60.
ful than those for unpaired, or independent,                          Table 2.3 lists some parametric tests to consider
samples.                                                           for different situations. Although these cannot be
                                                                   detailed here, accounts are readily found in most
                                                                   statistical textbooks or in the documentation for
Parametric methods                                                 statistical software.
Parametric methods assume that the test statistic
follows a particular distribution, usually the nor-
mal distribution. If the underlying data are very                  T-tests
non-normal and/or sample sizes are small it will                   The t-test compares the mean values from two
be necessary either to transform the data prior to                 groups and is by far the most commonly applied
                                                                         2.6 Data analysis and interpretation   55



                Table 2.3. Some parametric statistical tests appropriate for analysing survey and
                monitoring data

                Number of             Paired or unpaired            Some tests
                samples               data                          to consider

                One                   n/a                           One-sample t-test
                Two                   Paired                        Paired t-test
                                      Unpaired                      Independent sample t-test
                More than two         Paired                        Repeated measures
                                                                      analysis of variance
                                                                    Route regression
                                                                    Generalised additive models
                                      Unpaired                      Analysis of variance
                                                                    Linear or polynomial
                                                                      regression
                                                                    Generalised additive models
                                                                    Time-series analysis


test. It is worth summarising a few of the consid-          When data from more than two surveys are avail-
erations needed for using it appropriately.                 able, the class of methods called Analysis of
                                                            Variance (ANOVA) can be used to look for differ-
*   T-tests are reasonably robust to minor departures       ences across the surveys. Repeated measures
    from the assumption that the data are normally          ANOVA is used for data from permanent
    distributed.                                            plots. ANOVA can be used to look for trends,
*   The independent sample t-test would be used to          but as more data become available the methods
    compare results from two surveys using non-per-         for detecting trends outlined on p. 59 can be
    manent plots. This test assumes that the data from      considered.
    the two samples have similar variances. Although
    the test is also robust to small departures from        Methods based on ranks
    this assumption, provided the two samples are of        When there are insufficient data to be able to apply
    similar size, the variances should be compared.         parametric methods with confidence, or when
    Most statistical packages include a test for com-       there are other concerns over the applicability
    paring variances. Some packages will provide a          of such methods, then non-parametric methods
    modified t-test to be used when the variances are       based on ranking the data provide an alternative.
    different.                                              If the measurement of interest is ordinal then
*   A paired t-test should be used for data from            methods based on ranks are often appropriate. In
    permanent plots and will be more powerful than          general, such non-parametric methods are less
    an independent sample test provided there is            powerful than the parametric equivalent, so the
    some correlation between the data from the two          applicability of parametric or resampling methods
    surveys.                                                should be considered first.
*   Tests can be one-sided or two-sided. Use two-sided         The two most commonly used tests are the
    tests unless a one-sided hypothesis test has been       Mann–Whitney rank-sum test and the equivalent
    specified in advance of the survey or there is an a     for paired data, the Wilcoxon signed rank test.
    priori reason for change being in only one              These and some other rank-based tests are listed
    direction.                                              in Table 2.4.
56    2 PLANNING A PROGRAMME



                   Table 2.4. Some rank-based tests appropriate for analysing survey and
                   monitoring data

                   Number of             Paired or unpaired           Some tests to
                   samples               data                         consider

                   One                   n/a                          Sign test
                                                                      Wilcoxon test
                   Two                   Paired                       Wilcoxon signed
                                                                        rank test
                                                                      Sign test
                                         Unpaired                     Mann–Whitney test
                   More than two         Paired                       Friedman test
                                         Unpaired                     Kruskal–Wallis test



Mann–Whitney test                                        positive differences are calculated separately. The
Sometimes called the Mann–Whitney U-test or              test statistic is the smaller of these two sums; parti-
Wilcoxon’s rank-sum test, this compares the distri-      cularly small or large values will indicate that one
butions of two independent samples. Unlike the           or other group tends to take larger values than the
t-test, which specifically compares the mean from        other. As for the Mann–Whitney test, the Wilcoxon
the two samples, the Mann–Whitney test simply            test is usually performed by using statistical soft-
tests whether the two distributions are identical or     ware, which will also take account of ties.
whether one tends to have larger values than the
other.                                                   Resampling methods
    The two samples are combined and numbered            The advent of fast computers has made possible
according to their rank, from smallest to largest.       the development of another class of methods
The sum of these ranks for one of the samples is         that derive distributional properties of summary
then selected as the test statistic with particularly    statistics by generating large numbers of new sam-
small or large values indicating that the selected       ples from the original data. Primarily used for quan-
sample comes from a distribution that is shifted to      titative data, such methods enable the calculation of
the left or right of the other one. Whether the test     confidence intervals and the use of hypothesis tests
provides evidence for rejecting the hypothesis that      without making assumptions about the distribution
the groups have the same distribution is deter-          of the data and are usually more powerful than non-
mined from statistical tables or, more commonly,         parametric methods that make use of data rankings
by the software performing the test. Such software       rather than the data values themselves. Two parti-
will also make allowance for any ties in the             cular methods in common use are the bootstrap and
rankings.                                                randomisation tests. A good reference for further
                                                         detail is Manly (1997).
Wilcoxon’s signed rank test
For paired data this tests whether one group tends       Bootstrapping
to have larger, or smaller, values than the other.       The idea behind bootstrapping is that if it is diffi-
For each pair the difference in values is calculated     cult to make distributional assumptions about a
and then these differences are ranked from smal-         summary measure then the data themselves are
lest to largest without regard for sign. The sum of      the best guide to what that distribution is. To
the ranks for the negative differences and for the       approximate what would happen if new samples
                                                                     2.6 Data analysis and interpretation      57



were selected from the population under study,          reasonable to suppose that the observed difference
bootstrapping involves selecting new samples            will not be particularly large when compared with
(resamples) from the sample data themselves.            this distribution. The null hypothesis is rejected if
Such resamples are selected with replacement,           less than 2.5% of the randomised differences are
that is each sample value can occur more than           greater than the observed difference (two-sided
once in each resample. Large numbers of resamples       test). In practice it is usually impracticable to gener-
(typically around 1000) are drawn and for each one      ate all possible allocations and so a large random
a new estimate of the summary measure is calcu-         sample of allocations is more commonly used.
lated. In its simplest form a bootstrap 95% confi-         Randomisation tests are used for detecting
dence interval is then estimated by reordering the      change and trends where the data are extremely
resampled estimates from smallest to largest and        non-normal. They are also used in multivariate ana-
selecting the 2.5 and 97.5 percentile values as the     lysis, for example to test the significance of relation-
interval limits. This simple form of the bootstrap is   ships between species and environmental variables.
relatively easy to implement in a spreadsheet,
although many general-purpose statistics packages       Categorical data
provide a range of bootstrapping methods.               For presence–absence data or where interest cen-
   Bootstrapping can also be used for hypothesis        tres on the proportion of samples falling into a
testing but the above method can be adapted if          particular category, parametric methods can some-
interest centres on the likelihood of change in a       times be applied but rather more samples are likely
mean value between two surveys. If permanent            to be needed than for quantitative data. Table 2.5,
plots were used then simply calculate the change        adapted from Cochran (1977), gives minimum sam-
for each plot and bootstrap these change values         ple sizes for a confidence interval based on a nor-
using the mean as the summary measure. If the           mal approximation to be applicable.
resulting confidence interval does not extend              Calculation of confidence intervals is outlined in
across zero we can be reasonably confident that         Box 2.8. For proportions, the formula to use for the
there has been a change. For non-permanent plots        standard error is
the two samples have to be bootstrapped sepa-                              rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
                                                                             pð1 À pÞ 
                                                                             ^           ^                n
rately. Each resample from the two samples is                         SE ¼                       1À              ;
                                                                               nÀ1                       N
paired and the difference in means calculated.
These can then be used to estimate a confidence                 ˆ
                                                        where p is the sample estimate of the proportion
interval for the difference in means.                   of interest.
   Bootstrapping has gained in popularity in recent
years and will often be recommended by journal
referees if there is any doubt over the distribution
                                                        Table 2.5. Smallest sample sizes needed to use a normal
for quantitative data.
                                                        approximation when calculating confidence intervals for
                                                        proportions
Randomisation tests
Another class of methods that are primarily used for
                                                        Proportion of plots                     Required
testing differences between two or more groups are
                                                        in category of interest                 sample size
randomisation tests. This is essentially a resampling
method without replacement where the observed                         0.5                              30
difference between groups is compared with what                       0.4                              50
would be obtained by randomly allocating the data                     0.3                              80
to the groups. In theory all possible allocations                     0.2                             200
could be considered and a distribution of possible                    0.1                             600
differences generated. Under a null hypothesis that                   0.05                           1400
there is no difference between the groups it seems
58   2 PLANNING A PROGRAMME



               Table 2.6. Some tests appropriate for analysing categorical data.

               Number of                 Paired or unpaired              Some tests
               samples                   data                            to consider

               One a                     n/a                             Exact binomial
                                                                           confidence interval
                                                                         Normal approximation
                                                                         Binomial test
               Two b                     Paired                          McNemar’s test
                                         Unpaired                        Chi-squared test
                                                                         Fisher’s exact test
                                                                         Normal approximation
               More than two             Paired                          Cochran Q test
                                         Unpaired                        Chi-squared test

               a
                 For ordinal data the ranking methods described in Methods based on ranks
               (p. 56) may be appropriate.
               b
                 Note that the parametric regression and modelling methods may also be
               appropriate.


               Table 2.7. Example data for chi-squared test

                           No of quadrats            Expected number               Total number
                           with species              of quadrats with              of quadrats
               Year        present (O)               species (E)                   taken

               1                   40                          25                       50
               2                   30                          25                       50
               3                   50                          50                      100
               4                   30                          50                      100
               Total              150                         150                      300


   To compare two samples to see, for example,             plots containing a species has changed. The data
whether a change has taken place in the propor-            can be presented in the form of a table (Table 2.7)
tion of the site falling within a category interest,       whose cells show the number of sample plots for
the most commonly used tests are chi-squared (2)          which the species was present and absent for each
tests and, for paired data, McNemar’s test. These          survey. In this example new plots have been used
and some other tests for categorical data are listed       for each survey.
in Table 2.6.                                                 To test the null hypothesis that the proportion
                                                           of quadrats in which a species is present is the same
Chi-squared tests                                          in each year, we compare the observed data with
These are a class of tests for examining hypo-             that which would be expected if no change had
theses for category data. For example, suppose pre-        taken place. The chi-squared test is then used to
sence–absence data are available from four surveys         see whether the observed and expected values are
and interest centres on whether the proportion of          sufficiently different for it to be unlikely that no
                                                                      2.6 Data analysis and interpretation   59



change has occurred. The chi-squared statistic is            The chief advantages in this approach are that
calculated from the equation:
                                                         *   all available information is made use of;
                       X ðOi À Ei Þ2                     *   the interpretation of credible intervals and
                2 ¼                   ;                     Bayesian hypothesis tests is more straightforward
                              Ei
                                                             than for classical methods; and
where Oi is the observed frequency of the species in     *   recent developments mean that complex models,
question in a given year i and Ei is the expected            including spatial structure, can be analysed.
frequency if the species is not changing.
                                                         The disadvantage, for some, is that specification of
   Expected values are calculated by:
                                                         the prior distribution is inevitably partly subjec-
                                                         tive. However the effect the prior has on the final
Eðone yearÞ                                              estimates can be controlled: vague prior distribu-
         total quadrats in which species present         tion will have relatively little effect and the more
       ¼
             total number of quadrats overall            data that are collected, the greater will be the rela-
         Â total quadrats taken in that year:            tive influence of the data compared to the prior.
                                                            The basic principles behind Bayesian inference
   This value is compared with values of 2 from         are straightforward, but its implementation can
statistical tables. We need the degrees of freedom,      quickly become complicated. Although fast com-
which is given by the number of years minus 1; in        puters have made complex Bayesian models feasi-
this case 3.                                             ble, keeping things relatively simple depends on
                                                         careful choice of prior and data distributions.
                                                            The availability of software tools, such as
Bayesian inference                                       WinBUGS (www.mrc-bsu.cam.ac.uk/bugs), have
Bayesian inference differs from the classical methods    raised the profile and popularity of Bayesian meth-
described so far in that it makes use of prior infor-    ods. For an introduction see Lee (1987) or Marin
mation about the population measure of interest.         et al. (2003).
Rather than treating this measure as being fixed,
Bayesian methods give it a probability distribution,     Detecting trends
which is determined by the nature of the measure         When a monitoring scheme has been running for
and the extent of prior knowledge. This prior distri-    some years the question is likely to arise as to
bution is then combined with information provided        whether there are discernible trends in the size of
by the survey data, using Bayes’ theorem, to derive a    a population of interest or in the extent of a habi-
posterior distribution for the measure. This posterior   tat, for example. It is unlikely to be worth investi-
distribution tells us what we know about the popu-       gating this until five or more repeat surveys have
lation measure given the data and our prior knowl-       been carried out.
edge and can be used to provide an estimate of the          The first step, as always, is to plot the data as
measure and a Bayesian equivalent of the confi-          a time series, i.e. the summary measure on the
dence interval, often called the credible interval.      vertical axis against time on the horizontal axis.
   Prior information comes from expert knowl-            Are any trends apparent? Are they linear or more
edge about the site or population, from previous         complex? Are cyclical patterns apparent?
surveys and/or from surveys on similar sites or             Testing for trends is most straightforward if the
species. The prior distribution is defined according     measure used comes from complete counts rather
to the quality of this information, so, for example, a   than from sample surveys and the trend appears to
normally distributed prior will have a large vari-       be reasonably linear. The most common approach
ance if the prior information is rather vague and        is to then fit a regression line through the values,
uncertain, and a small variance if the population        with time as the explanatory variable. The gradient
measure is fairly well known.                            in the regression line is then tested to see whether
60    2 PLANNING A PROGRAMME



it is significantly different from zero. If so, a trend    concerned with modelling, studying autocorrela-
is indicated.                                              tion structure, detecting cyclical behaviour and
    There is one complication with this method.            forecasting. A good introduction is provided by
Standard regression analysis assumes that                  Chatfield (1996).
the values of the measurement variable are inde-
pendent of each other. In effect this means they are       Some particular issues
uncorrelated. However, the results of successive           The following sections provide a discussion of two
surveys are very commonly correlated because               common scenarios that arise.
the size of a population or habitat in Year 1 will
have an effect on its size in Year 2. The effect of this   Data with many zeros
autocorrelation is that the statistical significance of    This situation frequently arises when the species of
the gradient will be overestimated. This is unlikely       interest is often absent from sample plots.
to be an issue if the gradient is very highly signifi-     Measures such as counts of individuals or percen-
cant, but in some cases positively autocorrelated          tage cover often exhibit a preponderance of zeros.
data can give the appearance of a trend.                   For example, Figure 2.12 shows the distribution
    There is no easy way of getting around this pro-       of counts of occupied Manx Shearwater Puffinus
blem. One approach is to use bootstrapping of the          puffinus burrows from a sample of 20 m2 plots.
regression model parameters. An alternative is to          Because this distribution is so skewed the distribu-
include a term in the regression for the previous          tion of any summary statistic is also likely to be
year’s count, to remove some of the autocorrela-           skewed, despite the relatively large sample size.
tion effect. However, the correct method to use            For this example, either of two approaches will
will depend on the nature of the autocorrelation;          probably work well.
a fairly large number of surveys are likely to be
needed before it can be studied in detail. Perhaps         1. Given the large sample size it is reasonable to
the simplest way forward is to consider whether               assume that the mean count per plot will approxi-
autocorrelation is likely given the ecology of the            mately follow a log-normal distribution (a skewed
species or habitat under study and, if so, to inter-
pret borderline trends with caution.
    If the trend is clearly non-linear, and sufficient
data are available, more complex models such
as polynomial regression or generalised additive                               300

models can be fitted.
                                                             Number of plots




    Where the survey data comprise sample mea-
surements then this should be allowed for in the
                                                                               200
trend analysis. The simplest way is to include each
measurement in the model so that uncertainty in
the true population mean is taken into account.
One alternative for permanent plots is to model                                100
each plot separately and combine the resulting
trend estimates to get a picture of the overall
trend. This is the approach taken by route regres-
                                                                                 0
sion (Geissler and Sauer, 1990) which is very widely                                 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
used in North America. Generalised additive mod-
                                                                                              Number of occupied burrows
els have been proposed as a more flexible alterna-
tive (see, for example, Fewster et al., 2000).             Figure 2.12. Bar chart showing the number of
    Long time series are fairly uncommon in ecology        occupied Manx Shearwater burrows. See text for
but there is a substantial literature for methods          futher details.
                                                                        2.6 Data analysis and interpretation   61



   distribution whose logarithm would be normally          (as it does for the Poisson distribution, for example).
   distributed). Confidence intervals for the mean can     Most parametric tests assume that the data are nor-
   then be calculated by using the methods described       mally distributed, which also implies that the var-
   in the next section. Change in this mean could          iance is independent of the mean. Applying a
   probably be assumed to be approximately nor-            transformation to the data can often help to rectify
   mally distributed.                                      these problems by ‘stabilising’ the variance and
2. To avoid any distributional assumptions, boot-          making the distribution more symmetrical.
   strapping of the counts can be used to generate            If the distribution of the transformed variable is
   confidence intervals, etc. In this particular exam-     not exactly normal, this is probably not critical,
   ple this was complicated by the fact that a strati-     provided that the sample size is moderately large.
   fied random sampling scheme was used and so             Often it is more important to choose a transforma-
   each stratum had to be resampled separately.            tion that stabilises the variance. However, transfor-
   Results were very similar to those obtained by          mations may not work for data with a sophisticated
   using the first approach.                               or complex structure.
                                                              The effect of different transformations should
                                                           be examined to see which gives the best approxi-
Small sample sizes
                                                           mation to the normal curve. Histograms or normal
When sample sizes are small it is likely to be diffi-
                                                           probability plots can be used for this purpose and
cult to determine the distribution of the data with
                                                           the goodness-of-fit of your data to a normal distri-
any confidence. See Figure 2.8 for an example of
                                                           bution can be tested by using a chi-squared test
how the true distribution only emerges as sample
                                                           (see above, p. 58).
size increases. In this situation it may be difficult to
                                                              Data are usually transformed to make para-
justify applying parametric methods. Exceptions
                                                           metric analysis possible. Any confidence interval,
occur where there are theoretical grounds for
                                                           or similar, obtained through transformed data
assuming a particular distribution or there is evi-
                                                           should be back-transformed into the original
dence from other, similar, data.
                                                           units. This is because an answer expressed in
   An example of the former is presence–absence
                                                           terms of angular degree units or square roots will
data, where the proportion of plots where the spe-
                                                           not be intuitively meaningful when considering
cies is present may be assumed to follow a binomial
                                                           estimates of counts of species, etc. Note, however,
distribution. In this case the sample size may be too
                                                           that the back-transformed mean of the trans-
small to be able to use a normal approximation
                                                           formed data will not usually be the same as the
but exact confidence intervals can be calculated
                                                           mean of the untransformed data, and so the back-
for binomial data. Many statistical packages can
                                                           transformed confidence interval will be for a dif-
do this, but an Internet search should also reveal
                                                           ferent summary measure. For example, if the log
a number of relevant tools and methods.
                                                           transformation is used, so that
   In many cases non-parametric methods may
                                                              yi ¼ log(xi) where xi are the original measurements
be the only alternative. Resampling may also be
                                                           then
suspect if there are insufficient data to adequately
                                                              " ¼ Ælog(xi)/n ¼ log(product of the xi)/n
                                                              y
regenerate the underlying distribution.
                                                                             ¼ log (geometric mean of the xi).
                                                              Thus the reversed transformed value of " is the
                                                                                                           y
Is transformation of the data necessary before             geometric mean of the xi and reverse transformed
statistical analysis?                                      confidence intervals will be for the geometric
Many examples of count or frequency data are               mean, not the usual arithmetic mean.
drawn from distributions that are strongly skewed             For the above example there is an alternative
(i.e. asymmetrical) and therefore do not nearly            approach that provides confidence intervals for
approximate to a normal distribution. In addition,         the arithmetic mean. If the logarithm of the mean
the distribution’s variance may depend on the mean         of a measurement, x, is normally distributed, x
62       2 PLANNING A PROGRAMME



itself is said to follow a log-normal distribution. An         First, take the square root of each observation.
approximate 95% confidence interval for " is x                 Then find the angle in degrees whose sine equals
                                                               this value. Percentage observations should first
                                     
                             x                                 be converted to proportions (divide by 100). The
                               ; xK
                             K                                 arcsin transformation is also useful for cover data
                                                               that have been converted to proportions.
where

                                 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi    Analysing more than one variable at a time
              K ¼ exp½1:96       varðloge xފ               More complex statistical analyses, such as multi-
                                                             variate techniques (e.g. principal components ana-
and                                                          lysis and correspondence analysis) for more than
                                                   !         one variable, can be employed to examine com-
                                            varðxÞ
             varðloge xÞ ¼ loge 1 þ                  :       munity composition or the relation between
                                              x2
                                                             community composition and environmental vari-
                                                             ables. These techniques are mainly exploratory in
   A transformed distribution may not look vastly            nature. There are also various analytical techni-
different from the original; the transformation is           ques for examining relations between variables,
also acting on the variance of the data, which may           such as correlation and multiple regression,
be more important. For example, the mean of a                which might be appropriate for more in-depth
Poisson distribution is equal to the variance and            analysis of data. These are beyond the scope of
the variance thus increases with the mean; the               this Handbook but see, for example, Jongman et al.
variance is dependent on the mean. In this case a            (1995) for more information.
square root transformation helps make the vari-
ance independent of the mean, allowing tests
                                                             2.6.5      Interpretation and presentation
based on the normal distribution to be used.
                                                                        of findings
   Some commonly used transformations are given
below.                                                       Once the data have been described and analysed,
*    log x: This is appropriate for clumped count data       the results have to be interpreted and presented.
     in which the variance of the sample is greater than     This is often the longest and most difficult part of
                                                             the process. Great care should be taken to ensure
     the mean, or for variables that always take positive
                                                             that appropriate conclusions are drawn and that
     values such as area (such distributions are often
     skewed to the left). Each observation is replaced by    results are successfully communicated. A study is
     the logarithm of itself.                                only as good as the ability of people to understand
*    log (x þ 1): This is appropriate for count data con-    its findings. After all, the key aim of such presenta-
     taining zeros. Log (x þ 1) is used because log (0) is   tions is to influence the management of a site or
     undefined and hence meaningless. Adding 1 to            species and effect change where needed.
     each observation avoids this problem.
     p
*       x: The square root transformation for count          Interpreting analyses
     data, which follows a Poisson distribution (ran-        Analyses should be geared towards satisfying the
     domly distributed), or for regularly distributed        objectives for which a survey or monitoring study
     data, is appropriate when the variance of a sample      was set up. Typically this involves determining the
     is roughly equal to the mean.                           status of a site or species and, possibly, whether
             p
*    arcsin( x): The arcsin (sinÀ1 or inverse sine) trans-   the site or species is in acceptable condition. In
     formation is appropriate for observations that are      addition, it is often desirable to examine whether
     proportions or percentages, or for frequency mea-       existing survey work is adequate and what
     sures (e.g. presence–absence within sub-quadrats).      improvements are needed. For example:
                                                                         2.6 Data analysis and interpretation    63



*   Are sample sizes sufficient to give adequate preci-       analysis methods as well as the rationale for these.
    sion and/or to detect small enough changes? If            If necessary, some technical detail may be con-
    not, then either more effort is required for future       signed to an annex.
    surveys or expectations will have to be reduced.       4. The results of the study. Full data tables, if
*   Are the measurements adequate and can they                required, may be consigned to an annex.
    be taken with sufficient accuracy? It may, for         5. A discussion of the findings together with the
    example, become apparent that measurements                management implications.
    are too error-prone for analysis to be reliable.       6. An assessment of the study and its adequacy
*   Is the sample design adequate or could improve-           together with recommendations for future
    ments to the stratification, for example, be made?        improvement.
Some other points to bear in mind when drawing             Most results will be presented in the form of tables
conclusions from analyses include the following.           and/or charts. Charts can provide a very powerful
*   If the statistical test is non-significant this does   way of conveying information, and appropriate
    not mean that the null hypothesis is true, just        displays should be considered wherever possible.
    that there is insufficient evidence to reject it.      However, they are open to misuse. In particular:
*   If a significance level of 5% is used then bear in
                                                           *   Ensure charts are clearly labelled with units of
    mind that 1 result in 20 will be significant purely
                                                               measurement and avoid unnecessary clutter (e.g.
    by chance. This may not be important for single
                                                               gridlines and non-essential annotations). Clutter
    tests, but if many tests are performed there is an
                                                               can detract from the message a chart is intended
    increasing likelihood that one will be significant
                                                               to convey.
    by chance. This frequently arises in ANOVA when
                                                           *   Avoid exaggerating trends, for example by only
    multiple comparisons between groups are being
                                                               displaying the observed range of data. The axes
    made. Most statistical textbooks will suggest stra-
                                                               for measurements that can take any positive
    tegies for dealing with this problem.
                                                               value should normally start at zero. It is easy to
*   Where possible, check that a test’s assumptions
                                                               make a trend look very substantial by starting axes
    are satisfied. Many tests are fairly insensitive
                                                               at the minimum observed value.
    to mild departures from their assumptions,
                                                           *   Include error bars, such as confidence intervals,
    e.g. t-tests and the assumption that the data are
                                                               around graph values that are derived from a sam-
    normally distributed.
                                                               ple. This avoids giving the false impression that
                                                               the exact population values are shown and hence
Presenting results                                             that any change is a real change.
The key to successful presentation is to decide
what are the main messages you what to get across          Tables should also be clearly labelled and unclut-
and how to convey them bearing in mind the                 tered. For clarity:
nature of the audience you are aiming at. The
                                                           *   Data values should be right-justified and sepa-
type of audience will affect the level of detail
                                                               rated for ease of reading. Comma separators can
included and the level of technical expertise that
                                                               be used to make large values clearer.
can be assumed. In most cases a survey or monitor-
                                                           *   When the results for statistical tests are presented,
ing report would be expected to include the
                                                               show p-values rather than just whether or not the
following.
                                                               result was significant. Levels for determining sig-
1. The rationale for the study together with any               nificance are to some extent arbitrary, so results
   required ecological background.                             that do not quite achieve significance may still be
2. A statement of the study objectives.                        of interest.
3. An account of the methodology used: the sam-            *   Show the criteria used to determine the signifi-
   pling design, field methods, measurements and               cance or otherwise of tests.
64      2 PLANNING A PROGRAMME



*   Confidence intervals for estimates are also often      (www.sas.com) and Genstat (www.nag.co.uk) are
    useful, not only to convey uncertainty but to show     all examples that can cope with most of the statisti-
    the range within which the true value is likely to     cal tests and models in common use and rather more
    lie in relation to a target value, for example.        besides. STATA and SPSS are specifically able to
                                                           analyse data from stratified and other survey
2.6.6        What statistical software is                  designs.
             available for the analysis of data?              Some resampling methods are available in many
                                                           general-purpose statistics packages. An example
Most spreadsheet programs, such as Microsoft
                                                           of a dedicated package can be found at www.
Excel, have functions for simple statistics and a
                                                           resample.com.
reasonable range of tailor-made analytical routines
                                                              More specialised software may be required for
and graphics. However the range of statistical
                                                           certain types of ecological data and analyses.
analysis that can be carried is usually limited;
                                                           CANOCO, for example, provides a good range of
non-parametric tests, for example, are generally
                                                           ordination and clustering techniques for multivari-
absent. Statistical add-ons for Excel are available
                                                           ate environmental data (www.canoco.com; ter
and these provide an inexpensive way of gaining
                                                           Braak & Smilauer, 1998). DECORANA is another
access to most of the commonly used tests.
                                                           well-established program for ordination and
Examples of these are Berk & Carey (2000),
                                                           TWINSPAN is widely used for the classification
Analyse-it (www. analyse-it.com), and XLStat
                                                           of species and sites according to similarity (both
(www.xlstat.com).
                                                           available from the Centre for Ecology and
   However, dedicated statistics programs are best
                                                           Hydrology in the UK: www.ceh.ac.uk). MVSP is
for most statistical analyses, not only for their analy-
                                                           another popular multivariate analysis package that
tical capabilities but for the ease with which they can
                                                           also enables calculation of a range of diversity
be used to present data and check assumptions.
                                                           indices (www.kovcomp.co.uk/mvsp). Finally, a good
These are recommended if a regular requirement
                                                           online resource with links to many free
for statistical analysis can be identified. Systat
                                                           and commercial software sites can be found at
(www.systat.com), Minitab (www.minitab.com),
                                                           www.statistics.com.
SPSS (www.spss.com), STATA (www.stata.com), SAS
3     *    Biodiversity evaluation methods




3.1       BIODIVERSITY VALUES AND                        texts such as Usher (1986), Smith & Theberge
          EVALUATION PURPOSES                            (1986), Spellerberg (1991) and Treweek (1999).

In general terms, biodiversity evaluation is the pro-
cess of measuring the value (ideally quantitatively)
                                                         3.2      A FRAMEWORK FOR ECOLOGICAL
of biodiversity components, such as the number of                 EVALUATIONS
species present, the population of a species, a habi-    Appropriate approaches and criteria for biodiver-
tat (usually meaning a vegetation community) or          sity evaluations vary considerably depending upon
the sum of all such components within a given area       their purpose, their scale and the biodiversity
or site. Such evaluations may be carried out for a       components in question. As stated previously, it is
variety of reasons, e.g. for conservation priority       vital that objectives be clearly defined and the
setting, as part of Biodiversity Action Plan (BAP)       work planned through to its conclusion.
development, for the selection of Protected Areas,          Spellerberg (1991) identifies the following six
for the identification of a site’s features of conser-   general best practices that should be included in
vation interest, as part of conservation objective       any evaluation framework.
setting, management planning and monitoring
processes, and as part of an EIA or other statement      1. Evaluation objectives should be defined.
to comply with planning procedures for a proposed        2. Criteria should be quantifiable, rather than
development.                                                subjective.
   Evaluations may be carried out on various com-        3. Evaluations should be repeatable.
ponents of biodiversity (i.e. from genetic variation     4. Evaluations should be based on biological
within species, to individual species, species              principles.
assemblages, biotopes and biomes) and at a variety       5. The methods, results and analysis should be
of scales, from specific sites, to counties, regions,       explained so that they can be understood by every-
countries, biogeographical areas (although these            one who has an interest in the area being evaluated.
may be smaller than countries) and global. A wide        6. Costs in time and money should take into account
range of potential biodiversity values may be con-          the depth and integrity of underlying surveys.
sidered, including intrinsic and socio-cultural          We have incorporated such concepts into a pro-
values (Daily, 1997; Posey, 2000), and more direct       posed generic framework for conducting site-based
socio-economic benefits (Daily, 1997), such as food,     biodiversity evaluations, as outlined in Figure 3.1.
building resources, medicines and waste decompo-
sition, etc. (Spellerberg & Hardes, 1992).
                                                         3.3     IDENTIFICATION OF VALUABLE
   As this Handbook concentrates on site surveys
                                                                 ECOSYSTEM COMPONENTS
and monitoring, rather than on regional- or
national-scale studies, this chapter focuses on site     A key step in any evaluation is the identification of
evaluations. We do not consider socio-economic           biodiversity components or functions that are con-
and socio-cultural values; the reader is referred to     sidered to be important or valuable. These are


                                                                # RPS Group plc and Scottish Natural Heritage 2005.
66    3 BIODIVERSITY EVALUATION METHODS



                                               Define scope and objectives


                                              Consult with interested parties
                                                   and local experts

       Carry out preliminary survey                                                Review existing biodiversity
       (scoping survey for EIAs)                                                                   information



                                 Identify habitats and main taxonomic groups present
                                                                                                  For EIAs etc,
                                                                                        give special regard to
                                                                                     legally protected species




       Check existing                                                                           Check lists of
       designation status                                                                   species / habitats
       (e.g. SSSI, SPA, SAC, local Wildlife Site)                                     of conservation concern



                        Identify Valuable Ecosystem Components (VECs) potentially present



                                                      Design survey



                                          Survey and confirm presence of VECs
                                          and quantify if necessary and possible



                                  Carry out evaluations against appropriate criteria
                                            with respect to objectives for:



         Protected area selection                                                  Impact assessment in EIAs
         (against designation criteria)




                                             Setting conservation objectives
                                               e.g. for features of interest

           Figure 3.1. A generic framework for site biodiversity evaluation.

called Valuable Ecosystem Components (VECs) by                  conservation agency terminology, VECs would at
Treweek (1999) and are sometimes referred to else-              least include notified features of interest.
where as Valued Ecological Receptors or Valued                     The identification of VECs has a major and
Ecological Resources. In SSSI and UK statutory                  obvious bearing on the outcome of any evaluation
                                                       3.3 Identification of Valuable Ecosystem Components     67



exercise, as an ecosystem, habitat or site will not be     *   species known to be sensitive to specific land use
regarded as important if interest features are over-           actions that may serve as ‘early warning’ or indi-
looked. Indeed, one of the main underlying causes              cator species for an affected wildlife community;
of biodiversity loss is the lack of appreciation of the    *   species that perform a key role in a community
value of its properties and functions.                         because, for example, of their role in nutrient
    It may not be practical to identify and use all VECs       cycling or energy flows; and
in an evaluation, even in the most simple habitats,        *   species that represent groups of species that uti-
as too many species and other components and                   lise a common environmental resource (guilds).
functions will have some value. It is thus normal
to base evaluations on a sub-set of selected VECs of       In the UK and elsewhere in Europe, the presence of
particular value. For EIAs, however, it is necessary       species or habitats of high conservation priority is
to identify all VECs that are of more than negligible      one of the most commonly used criteria for pro-
value and which will receive impacts. The criteria         tected area designation and consideration in EIAs.
used to select or identify such VECs should be objec-      The presence of particularly high numbers or high
tive, consistent, transparent and defensible               proportions of species (irrespective of their conser-
(Treweek, 1999). Ecosystem attributes that may be          vation status) is also a frequently used criterion for
selected as VECs at a site may include distinct            the selection of sites for protection for nature con-
genetic populations of a species, species popula-          servation purposes. For example, the internation-
tions, species assemblages, vegetation communities,        ally recognised criteria for the designation of
habitats and ecosystem functions.                          Ramsar Sites (see Section 3.7.6) include thresholds
    In practice, species and habitats of particular        for the proportion of biogeographic waterfowl
conservation importance are the most commonly              populations (e.g. more than 1% of the flyway popu-
identified VECs as these are easiest to define objec-      lation) and total waterfowl numbers (e.g. more
tively and to measure quantitatively. In contrast,         than 20 000 individuals).
ecosystem functions, though perhaps of as great an             The evaluation framework (See Figure 3.1) iden-
importance as VECs, are difficult to define and            tifies a number of activities that will assist in iden-
describe in terms that allow objective evaluations         tifying VECs. These include reviewing existing
to be made of their importance. Nevertheless, this         biodiversity information, consulting with local
should be attempted as far as possible when it is          experts (e.g. county recorders, biological records
considered that a site is likely to provide an impor-      centres, Wildlife Trusts) and conducting prelimin-
tant ecological function.                                  ary surveys. Even brief surveys are likely to be
    There are a variety of species attributes that         valuable. They can establish the range of VECs
may be used as criteria for their selection as VECs.       that may be present, and this can help considerably
These include commercial value, rarity, endanger-          in the subsequent design of full surveys (or mon-
ment, their role as flagship or umbrella species           itoring). Although preliminary surveys will not be
(i.e. ability to provide benefit to others through         able to adequately establish the presence of all
their conservation), their importance for ecosystem        species VECs, they should identify habitats that
function (i.e. keystone species) and their value as        are present. This information may be used to iden-
indicator species (See, for example, Eberhardt,            tify potential species VECs, which may then be
1976; Treweek, 1999).                                      verified by subsequent surveys. For example, a
    The US Fish and Wildlife Service Habitat               brief site visit might establish that ponds are pre-
Evaluation Procedure (HEP) (USFWS, 1980), for              sent that are suitable for, and within the vicinity
example, identifies four categories of ‘evaluation         of known populations of, Great Crested Newts
species’:                                                  Triturus cristatus. A specific newt survey may then
                                                           be planned and carried out at a suitable time.
*   species with public interest, economic value or            Evaluation of sites on the basis of species- and
    both;                                                  habitat-based VECs requires some assessment to
68      3 BIODIVERSITY EVALUATION METHODS



have been made of the conservation status of the        are not immediately threatened with extinction. As
individual habitats and species in question. The        a result, a suite of common but rapidly declining
following sections therefore describe key princi-       farmland birds, such as Turtle Dove Streptopelia
ples underlying the assessment of species and habi-     turtur, Skylark Alauda arvensis, Starling Sturnus vul-
tat conservation priorities. Specific details and       garis and Yellowhammer Emberiza citrinella, are
guidance on assessments and legislation affecting       now on the UK Red List (Gregory et al., 2002) and
current species and habitat conservation priorities     are a focus of considerable conservation action.
within the UK are then provided in Section 3.5.         Priority-setting under the UK Biodiversity Action
                                                        Plan (BAP) process has also used broader criteria
                                                        than just extinction risk in its selection of Priority
3.4      PRINCIPLES UNDERLYING THE                      Species.
         SETTING OF CONSERVATION
         PRIORITIES
                                                        3.4.2      The importance of rarity
3.4.1      Conservation objectives
                                                        Rarity has often been considered to be one of the
Conservation priorities depend in the first instance    most important factors influencing the risk of
on conservation objectives. In terms of global          extinction of a species, and many Red Data lists
objectives, there is reasonable agreement that the      have focused on this. Rarity has also often been
prevention of global extinction should be the focus     used as a secondary criterion whereby, for example,
of activity, in which case the degree of threat (i.e.   a declining species is not considered to be threa-
risk of extinction) is of primary concern in setting    tened unless it is has also crossed a rarity threshold.
priorities. This is reflected in the production of      However, rarity is not a straightforward concept:
IUCN Red Lists of species that are considered to be     there may be a variety of circumstances under
at risk of global extinction according to various       which species may be rare (Rabinowitz, 1981).
categories of threat. Beyond this, there are many       Species may have small (or large) total ranges,
different views on global biodiversity conservation     occupy few (or many) habitat types, and be scarce
priorities; such diversity of opinion is not surpris-   or abundant where they do occur. As indicated in
ing as there is no single feasible way of measuring     the brief examples in Table 3.1, seven of the com-
or valuing biodiversity overall (Purvis & Hector,       binations of these factors (the shaded boxes) would
2000).                                                  qualify as rare within the possible range of mean-
   The risk of extinction at national level is also     ings of the term. It is therefore evident that rarity
probably the commonest basis for national species       embraces both a spatial and a numerical dimen-
conservation priority setting. However, at national     sion. For any particular species some aspects of
or sub-national levels biodiversity conservation is     rarity may be an evolutionary property, such as
increasingly incorporating broader multiple objec-      habitat specificity, small natural range or low nat-
tives. In the UK, for example, bird conservation        ural densities. Such species may always be rare and
objectives have traditionally focused on rare spe-      therefore unlikely to respond to conservation mea-
cies, but in recent years greater attention has been    sures. On the other hand, small range or low den-
given to species that occur in internationally          sities may be the result of human impact, which
important numbers, despite many of these being          may be reversible.
highly abundant (e.g. many species of wintering            Inclusion of rarity factors in an evaluation
waterbird). There is also increasing concern for        requires data on the range or number of indivi-
species that are common and widespread but              duals of a species (or habitats, communities, or
declining rapidly, as rapid declines of common          abiotic features), not only at the site in question,
species may involve the loss of many millions of        but at wider scales. Important elements of rarity
individuals from the environment. This is clearly a     are also scale-dependent. A locally rare species may
substantial biodiversity impact even if the species     also be regionally or globally rare, or it may simply
     Table 3.1. The seven forms of rarity based on a species’ geographical range, habitat specificity and local population size
     Indicative bird species have been added for each category in relation to range, population size and habitat use in the UK.

                                                                            Geographical range

                                                        Large                                                     Small

     Local population
     size                    Wide habitat use              Narrow habitat use           Wide habitat use              Narrow habitat use

     Large, dominant         Common and locally abun-       Locally abundant over a      Locally abundant in several Locally abundant over a
     somewhere               dant over a large range in     large range in a specific    habitats over a small range small range in a specific
                             several habitats               habitat                                                  habitat
     Example birds           Blackbird Turdus merula        Reed Warbler Acrocephalus    Pink-footed Goose Anser     Great Skua Stercorarius
                                                            scirpaceous                  brachyrhynchus (wintering) skua
     Small non-              Constantly sparse over a       Constantly sparse over a     Constantly sparse over a    Constantly sparse over a
     dominant                large range in several         large range in a specific    small range in several      small range in a specific
     Example birds           habitats                       habitat                      habitats                    habitat
                             Long-eared Owl Asio otus       Garganey Anas querquedula    Montagu’s Harrier Circus    Capercaillie Tetrao
                                                                                         pygargus                    urogallus

     Source: Rabinowitz (1981).




69
70      3 BIODIVERSITY EVALUATION METHODS



be rare because it is at the edge of its range (e.g.
breeding Golden Orioles Oriolus oriolus or Redwings
Turdus iliacus in the UK). Normally, increased impor-                        Stone-curlew        White-tailed
tance should be given to species that are rare on a                                                Eagle




                                                        Risk of extinction
global scale. Some locally abundant species may
                                                                                                                         Scottish
also be of high conservation importance if the spe-                                                                      Crossbill
                                                                                                 Increasing
cies in question is rare at a global or wide geogra-                                            conservation
phical scale (e.g. Great Skua Stercorarius skua and                              Turtle Dove       priority
Great Crested Newt).
                                                                                                                    Manx
                                                                                                                  Shearwater
                                                                                   Blackcap
3.4.3      Levels and scales of threat and
           population importance                                                            Global importance of the
                                                                                             population / resource
Whichever criteria are used for threat evaluations,
a hierarchical level of importance should be estab-     Figure 3.2. Key factors defining the conservation
lished according to the scale of the assessment, so     status of a species’ population or area of habitat.
that the highest priority for conservation and/or       (Scientific names of species: Stone-curlew Burhinus
protection is given to species or habitats that are     oedicnemus,White-tailed Eagle Haliaeetus albicilla, Turtle
globally threatened. However, it is also necessary      Dove Streptopelia turtur, Blackcap Sylvia atricapilla,
to take into account their local status to assess the   Manx Shearwater Puffinus puffinus, Scottish Crossbill
necessity for taking action at a local scale. This      Loxia scotica.)
enables the principle of ‘thinking globally and act-
ing locally’ to be put into practice. The highest       the importance of the population or resource being
priority should be given to species and habitats        considered. Thus, the evaluation of a species’ popu-
that are both globally and locally threatened.          lation conservation status should consider two key
   Assessments below global scales should also          independent factors: the risk of extinction of the
refer to appropriate biogeographical populations.       population in question (i.e. its threat status) and its
In practice, however, assessments of populations        biogeographical importance, i.e. the proportion it
are more often based on national or regional (e.g.      represents of the appropriate biogeographical (or
European) populations for political and adminis-        national or regional) population (Figure 3.2). The
trative reasons. This is because some species could     same conservation evaluation principles may be
otherwise have more than one conservation status        applied to an area of habitat.
within a country, which would send confusing and            Thus, for example, a very high priority should be
mixed messages to policy makers and the general         given to a species’ population that is endemic and
public. Some steps towards defining conservation        is at a high risk of extinction. However, it is import-
status on the basis of biogeographic populations        ant to note that a population may be a high priority
have, however, been made for migratory water            nationally because the species is highly threatened
birds. Different waterfowl flyway populations           nationally, irrespective of its numbers in relation
have been defined (Rose & Scott, 1997) to enable        to international or global populations (e.g. Stone-
identification of important waterbird populations       curlew in Figure 3.2). This is because the mainte-
under the Ramsar Convention. These flyway popu-         nance of a species’ range (and potential genetic
lations have in turn been used to define threatened     variation associated with this) can also be an impor-
waterbird populations for the African–Eurasian          tant conservation aim after prevention of complete
Waterfowl Agreement, an agreement under the             extinction. On the other hand, a population of a
Convention on Migratory Species of Wild Animals.        species may be very important because it is a large
   Any evaluation of conservation priorities for a      proportion of the biogeographical population, irre-
species (or habitat) should also take into account      spective of its conservation status (e.g. Manx
                                              3.4 Principles underlying the setting of conservation priorities       71



Shearwater in Figure 3.2). In these circumstances a         National boundary:         Regional boundary:
country has a particular responsibility for the spe-
cies and should at least take appropriate measures                        A    A   A    A    A   A    A     A    A
to monitor the status of the species and guard                            A    A   A    A    A   A    A     A    A
against potential events (e.g. an oil spill) that                         A    A   A    A    A   A    A     A    A
could affect the population suddenly and catastro-                             A   A    A    A   A    A     A    A
phically, or gradually over a longer period of time.                           A   A    A    A   A    A     A    A
Such species are often the subject of national and              D    D         A   A    A    A   A    A     A    A
local Biodiversity Action Plans (BAPs).                         D    D             A    A    A   A    A     A    A
   This concept of assessing both the risk of extinc-                                   A    A   A    A     A    A
tion and the importance of the population can be                C    C         E
applied at a variety of scales. For example, for bird
                                                                C    C    C                                 E
species, the status and importance of a population
                                                                C    C    C        B    B             E     E
on a site can be compared with that of the county,
                                                                               B   B    B
country or biogeographic region (e.g. flyway). A
                                                                               B   B    B
hypothetical example of national priority setting
according to a species’ biogeographical range is         Figure 3.3. National priority setting, based on a species’
depicted in Figure 3.3.                                  regional (e.g. European) range size and the proportion of
   Although consideration of the biogeographical         its range occurring within the country in question. Each
importance of populations is not normally expli-         species’ geographical range is represented by an array of
citly carried out in the preparation of Red Data         letters. Species A is widespread within the region, but
Books (RDBs), this approach was developed by             rare within the country. Species B and E are equally rare
BirdLife International in its assessment of the con-     within the country, but much less widespread within the
servation status of European birds (Tucker & Heath,      region as a whole. Species D has half of its range within
1994; BirdLife International, 2004). The highest of      the country, whereas species C is endemic (and relatively
four categories of Species of Conservation Concern       widespread) within the country. If conservation status
(SPECs) was given to species that were globally          were to be assessed solely on the basis of national range
threatened, irrespective of the proportion present       size, species A, B and E would be of high priority for
in Europe, because it was felt that these species        conservation action or protection, and species D and C
should be a high priority wherever they occur reg-       less so. Alternatively, if regional range size is the sole
ularly. However, the second highest priority (SPEC       criterion used, species D and E would be accorded the
2) was given to those species that were considered       highest priority, followed by B and C, then A. Finally, a
to have an Unfavourable Conservation Status in           national assessment based on the proportion of each
Europe and populations that are concentrated (i.e.       species’ range occurring within the country would select
more than 50%) in Europe. Other species with an          C, then D, followed by E, B and A. Ideally, national
Unfavourable Conservation Status were placed in          priority setting should attempt to balance all three
the SPEC 3 category.                                     measures: national conservation status, wider
   This approach has been taken further in the           conservation status (regional or global) and degree of
UK. The first RDB for birds in the UK (Batten et al.,    endemism.
1990) included the international importance of
populations as one of its qualifying criteria.
However, this concept was expanded with revised
criteria in the subsequent reassessments and              combined NGO and Agency Red and Amber-listed
publication of Birds of Conservation Concern by UK        birds (Gregory et al., 2002). These lists included
non-governmental organisations (NGOs) (Gibbons            the BirdLife International SPEC categories 1–3 as
et al., 1996), Birds of Conservation Importance by the    well as species with internationally important
UK Statutory agencies (JNCC, 1996) and the                populations.
72      3 BIODIVERSITY EVALUATION METHODS



Table 3.2. Species traits other than population status that have been used for ranking between-species conservation
priorities

Trait                     Priority given to                                   Example references

Evolutionary              Species with most unique characters                 Vane-Wright et al. (1991)
uniqueness                Species with greatest genetic diversity             Crozier (1992, Nee and May (1997)
                          Species in clades a undergoing evolutionary         Erwin (1991)
                          radiations
Phenotypic traits         Maximising diversity of phenotypic traits           Owens & Bennett (2000)
Protection status         Species poorly represented in protected             Scott et al. (1993), Cassidy et al. (2001)
                          areas
Land use change           Species in areas susceptible to destruction         Menon et al. (2001)
Ecosystem role            Species important in ecosystems (e.g.               Allen-Wardell et al. (1998)
                          pollinators)
Multi-species             Maximal phylogenetic diversity within a set         Witting et al. (2000)
interactions              of interacting species

a
 A group of species (in this instance) sharing a closer common ancestry with one another than with members
of any other clade.
Source: (Mace & Collar, 2002).




3.4.4       Other factors affecting species                 3.5       SPECIES AND HABITAT
            conservation priorities                                   CONSERVATION PRIORITY LISTS
Factors other than population status or range size          There are a large number of conservation assess-
may influence the overall conservation priority             ments and legislative instruments that should
ranking for a species or habitat, examples of which         be taken into account in any evaluation of
are listed in Table 3.2. Most of these have rarely been     a species’ or habitat’s conservation priority. In
applied. However, genetic diversity is increasingly         the UK ecological evaluations should take special
being incorporated into decision-making in relation         note of:
to rare and/or threatened species. A species distribu-
ted across a number of isolated sites, e.g. Pollan          *   IUCN global Red Lists
Coregonus autumnalis pollan in lakes in Ireland, has        *   Convention on Migratory Species Appendices
a potentially high genetic biodiversity; and each           *   Bern Convention Appendices 1, 2 and 3
isolated population contributes to the diversity            *   European Red Lists or lists of species of conserva-
within the species and hence its overall ability to             tion concern
survive. In the case of Floating Water-plantain             *   Wild Birds Directive Annex I
Luronium natans, much of the population in the UK           *   Habitats Directive Annex I and 2
is found in the canal network and is thought to             *   Wildlife & Countryside Act Schedules 1, 5 and 8
derive from vegetative reproduction from a single           *   CROW Act 2000 list of habitats and species under
population in Wales. If this is the case, the canal             Article 74 (for England and Wales only)
plants have relatively low genetic diversity com-           *   Nature Conservation Act 2004 (Scotland)
pared with that of isolated lake populations and            *   UK Red Lists and birds of conservation concern
would rank low in an evaluation.                            *   UK BAP listed species and habitats
                                                           3.5 Species and habitat conservation priority lists   73



Background information on the derivation of each           assessment of extinction risk, which can also be
of these lists is outlined below. Further informa-         consistently applied by different people across the
tion and accounts of other lists referring to specific     full range of taxa. All new assessments and reas-
species groups are provided in each species                sessments of IUCN Red Lists use this system. Some
chapter.                                                   assessments from 1996 to 2000 have also been
                                                           converted to follow the revised categories and cri-
                                                           teria. It is now intended that SSC will leave this
3.5.1      IUCN Red Lists
                                                           system unchanged for a sufficient period to allow
The IUCN Red Lists and Red Data Books (RDBs) were          genuine changes in conservation status to be
first conceived in 1963 to draw attention to the           monitored.
conservation needs of globally endangered species.            The current categories of threat are listed below
In particular, the identification of endangered spe-       and a diagrammatic summary of the relationships
cies was carried out to assist with defining conser-       between these categories is shown in Figure 3.4.
vation priorities and the drafting of species protection   The assessments may be made either by relating
legislation. The Red Lists were prepared under the         simple population status attributes to numerical
auspices of the Species Survival Commission (SSC),         thresholds or by a more complex Population
one of the commissions of IUCN (The International          Viability Analysis (PVA). PVAs use demographic
Union for the Conservation of Nature). The selection       models to predict the probability that a given popu-
of species for inclusion was carried out by using          lation will become extinct (or decline to a specified
standard data sheets and largely subjective assess-        level) within a given time period (see Beissinger &
ments. Species were categorized according to threat:       Westphal (1998) for review).
Endangered, Vulnerable, Rare, Out Of Danger or
Indeterminate.                                             *   Extinct (EX): A taxon is Extinct when there is no
   This simple priority classification set a global            reasonable doubt that the last individual has died.
standard for conservation assessment for more                  A taxon is presumed Extinct when exhaustive
than 30 years. By the late 1980s discussions were              surveys in known and/or expected habitat, at
taking place on how the criteria could be quanti-              appropriate times (diurnal, seasonal, annual),
fied to make the selection process more objective              throughout its historic range have failed to record
(Fitter & Fitter, 1987). After an extensive period of          an individual. Surveys should be over a time frame
preparation and consultation, IUCN adopted more                appropriate to the taxon’s life cycle and life form.
precise and quantitative Red List Categories in 1994       *   Extinct in the wild (EW): A taxon is Extinct in the
(IUCN, 1994). These criteria (referred to as Version           Wild when it is known to survive only in cultiva-
2.3) were used for the 1996 IUCN Red List of Threatened        tion, in captivity or as a naturalised population
Animals (Baillie & Groombridge, 1996), The World List          (or populations) well outside the past range. A
of Threatened Trees (Oldfield et al., 1998) and the 2000       taxon is presumed Extinct in the Wild when
IUCN Red List of Threatened Species (Hilton-Taylor,            exhaustive surveys in known and/or expected
2000).                                                         habitat, at appropriate times (diurnal, seasonal,
   Then, in 1996, IUCN members called for a                    annual), throughout its historic range have failed
further review to ensure that the criteria were                to record an individual. Surveys should be over a
applicable to a wide range of organisms, especially            time frame appropriate to the taxon’s life cycle
long-lived species and species under intensive                 and life form.
management. As a result a further revised set of           *   Critically Endangered (CR): A taxon is Critically
threat categories and criteria was adopted by IUCN             Endangered when the best available evidence
Council in 2000 and published in 2001 as Criteria              indicates that it meets any of the criteria A to E
Version 3.1 following further refinement (IUCN,                for Critically Endangered, and it is therefore con-
2001). The aim has been to develop a method and                sidered to be facing an extremely high risk of
set of criteria that provide a more objective                  extinction in the wild. Criteria A to D relate to
74      3 BIODIVERSITY EVALUATION METHODS




                                                                              Extinct (EX)

                                                                              Extinct in the Wild (EW)
                                                                              Critically Endangered (CR)
                                     (Adequate data) (Threatened)
                                                                              Endangered (EN)

                                                                              Vulnerable (VU)

                       (Evaluated)                                            Near Threatened (NT)

                                                                              Least Concern (LC)


                                                     Data Deficient (DD)


                                     Not Evaluated (NE)



              Figure 3.4. Current IUCN Red List threat categories (IUCN, 2001).



     numerical thresholds for species in rapid decline,         qualifying or likely to qualify for a threatened
     with small, fragmented, declining or fluctuating           category in the near future.
     ranges, or with very small populations or ranges.      *   Least Concern (LC): A taxon is Least Concern
     Criterion E is an unfavourable PVA indicating a            when it has been evaluated against the criteria
     probability of extinction of more than 50%                 and does not qualify for Critically Endangered,
     within ten years or three generations (whichever           Endangered, Vulnerable or Near Threatened.
     is longer).                                                Widespread and abundant taxa are included in
*    Endangered (EN): A taxon is Endangered when                this category.
     the best available evidence indicates that it          *   Data Deficient (DD): A taxon is Data Deficient
     meets any of the criteria A to D for Endangered,           when there is inadequate information to make a
     and it is therefore considered to be facing a very         direct or indirect assessment of its risk of extinc-
     high risk of extinction in the wild, or if under           tion based on its distribution and/or population
     Criterion E it has a PVA indicating a probability          status. A taxon in this category may be well stu-
     of extinction of more than 20% within 20 years or          died, and its biology well known, but appropriate
     five generations.                                          data on abundance and/or distribution are lack-
*    Vulnerable (VU): A taxon is Vulnerable when the            ing. Data Deficient is therefore not a category of
     best available evidence indicates that it meets any        threat. Listing of taxa in this category indicates
     of the criteria A to D for Vulnerable, and it is           that more information is required and acknowl-
     therefore considered to be facing a high risk of           edges the possibility that future research will
     extinction in the wild, or if under Criterion E it         show that threatened classification is appropriate.
     has a PVA indicating a probability of extinction of        It is important to make positive use of what-
     more than 10% within 100 years.                            ever data are available. In many cases great care
*    Near Threatened (NT): A taxon is Near Threatened           should be exercised in choosing between DD and a
     when it has been evaluated against the criteria but        threatened status (CR, EN or VU). If the range of a
     does not qualify for Critically Endangered,                taxon is suspected to be relatively circumscribed,
     Endangered or Vulnerable now, but is close to              and a considerable period of time has elapsed
                                                             3.5 Species and habitat conservation priority lists   75



    since the last record of the taxon, threatened sta-     species listed in Appendix I of the Convention and
    tus may well be justified.                              by concluding multilateral Agreements for the con-
*   Not Evaluated (NE): A taxon is Not Evaluated            servation and management of migratory species
    when it has not yet been evaluated against the          listed in Appendix II. The Bonn Agreements of
    criteria.                                               direct relevance to terrestrial and freshwater habi-
                                                            tats and species in the UK at the moment are the
Full details of the current IUCN Red List Categories
                                                            African-Eurasian Migratory Waterbird Agreement
and criteria are provided in IUCN (2001). They can also
                                                            and the Agreement on the Conservation of
be obtained together with guidelines on their use at
                                                            Populations of European Bats (EUROBATS).
http://www.redlist.org/info/categories_criteria.html.
                                                               The CMS is typically implemented legally
   The most recent published list of globally threa-
                                                            through the legislation of any given country and/
tened species is in 2004 IUCN Red List of Threatened
                                                            or the European Union, for example in the UK, the
Species (Baillie et al., 2004). An updated list of threa-
                                                            Wildlife & Countryside Act, the EU Birds Directive
tened taxa is maintained in a searchable database
                                                            (79/409/EEC) and the EU Habitats Directive
by the SSC Red List Programme accessible at
                                                            (92/43/EEC).
www.redlist.org/. However, the only taxonomic
groups that have been comprehensively assessed
are the birds and mammals. The vast majority of             3.5.3       Bern Convention
plant taxa listed in the 1997 IUCN Red List of
                                                            The Convention on the Conservation of European
Threatened Plants (Walter & Gillett 1998) have not
                                                            Wildlife and Natural Habitats, also known as the
yet been evaluated against the revised Red List
                                                            Bern Convention, was adopted on September 1979
Criteria and are therefore not included. Instead,
                                                            in Bern (Switzerland) and came into force on 1 June
the conservation status of plants may be ascer-
                                                            1982. It now has 45 Contracting Parties including
tained by searching the SSC database and the
                                                            39 Member States of the Council of Europe as well
UNEP-WCMC Threatened Plants database at http://
                                                            as the European Community, Monaco and four
www.wcmc.org.uk/species/plants/red_list.htm.
                                                            African states.
   Most Red List assessments are carried out by
                                                               The aim of the Convention is to:
the members of the IUCN Species Survival
Commission. All the birds are assessed by BirdLife          *   conserve wild flora and fauna and their natural
International and its partners. Other assessments               habitats;
and much taxonomic and distribution information             *   promote co-operation between states; and
have been provided by various partner                       *   give particular emphasis to endangered and
organisations.                                                  vulnerable species, including endangered and
                                                                vulnerable migratory species.

3.5.2       The Bonn Convention                             The contracting parties have undertaken, inter alia,
                                                            to protect the habitats of wild flora and fauna spe-
The Convention on the Conservation of Migratory             cies, and to give special attention to the conserva-
Species of Wild Animals, more often known as                tion of the species listed in:
the Bonn Convention (or CMS), aims to conserve
                                                            *   Appendix I: strictly protected flora species.
terrestrial, marine and avian migratory species
                                                            *   Appendix II: strictly protected fauna species.
throughout their range. It is one of a small number
                                                            *   Appendix III: protected fauna species.
of intergovernmental treaties concerned with the
conservation of wildlife and wildlife habitats on a         This Convention has greatly influenced the devel-
global scale.                                               opment of EU nature conservation legislation,
   Parties to the CMS work together to conserve             being the inspiration for the EU Birds and
migratory species and their habitats by providing           Habitats Directives. It has also had an important
strict protection for the endangered migratory              influence on the UK’s main conservation
76      3 BIODIVERSITY EVALUATION METHODS



legislation, the Wildlife & Countryside Act                   In 1992 the then European Community adopted
1981. See http://www.nature.coe.int/english/cadres         Council Directive 92/43/EEC on the Conservation
/bern.htm for more information on the Bern                 of Natural Habitats and of Wild Fauna and Flora,
Convention and lists of species on the various             known as the Habitats Directive. This international
Appendices.                                                wildlife legislation is intended to provide EU
                                                           Member States with a mechanism to meet their
                                                           obligations under the 1979 Bern Convention (see
3.5.4      European Red Lists and lists of
                                                           above) and to complement the provisions of the
           Species of Conservation Concern
                                                           1979 Birds Directive. The main aim of the
European Red Lists or their equivalent have been           Habitats Directive is:
produced for some taxa. For example, BirdLife
                                                            . . . to contribute towards ensuring biodiversity through the
International has produced lists of Species of
                                                           conservation of natural habitats and of wild fauna and flora
Conservation Concern (SPECs) in Europe (Tucker &
                                                           in the European territory of the Member States to which the
Heath, 1994; BirdLife International, 2004). As
                                                           Treaty applies (Article 2).
described above, this does not use IUCN criteria but
develops these and includes assessments of the                The 24 articles of the Directive specify a range of
importance of European populations as well as              measures, including conservation of features in the
their threat status. Further information on the cate-      landscape that are important for wildlife, the pro-
gories and criteria used for SPECs is given in Part III,   tection of species listed in the annexes from
Chapter 24 on birds.                                       damage, destruction or over-exploitation, the sur-
   A European Red List has also been produced by           veillance of natural habitats and species, and
the Council of Europe for butterflies (http://             ensuring that introductions of non-native species
www.vlinderstichting.nl/en/randc/rdb.htm).                 are not detrimental to naturally occurring habitats
                                                           and species. One of the most stringent obligations
                                                           (under Article 3) is to select, designate and protect a
3.5.5      European Union Birds and Habitats               series of sites, to be called Special Areas of
           Directives                                      Conservation (SACs), for 169 natural Habitats of
In 1979, the European Community adopted Council            Community Interest listed in Annex I of the
Directive 79/409/EEC on the conservation of wild           Directive and 623 Species of Community Interest
birds in response to the 1979 Bern Convention.             listed in Annex II1.
This Directive, usually referred to as the Birds              Habitat types of Community Interest are, within
Directive, provides for the protection, management         the EU territory2:
and control of naturally occurring wild birds within       1. in danger of disappearance in their natural range;
the European Union through a range of mechan-                 or
isms. One of the key provisions (under Article 4) is       2. have a small natural range following their regres-
the establishment of an internationally co-ordi-              sion or by reason of their intrinsically restricted
nated network of protected areas, known as                    area; or
Special Protection Areas (SPAs) for 182 species listed     3. present outstanding examples of typical charac-
in Annex I of the Directive. These are species that are       teristics of one or more of the five following
considered to be in danger of extinction, vulnerable          biogeographical regions: Alpine, Atlantic,
to specific changes in their habitat, rare, or requir-        Continental, Macronesian and Mediterranean.
ing particular attention by reason of the specific
nature of their habitat. Within SPAs, Member
                                                           1
                                                             Birds are not included because they are listed in Annex I of
States are obliged to take necessary steps to avoid
                                                           the Birds Directive.
deterioration of natural habitats and disturbance of       2
                                                             Within EU territory: i.e. within the European territory of the
the species, where this disturbance would be signif-       Member States to which the Treaty establishing the European
icant in terms of the objectives of the Directive.         Economic Community applies.
                                                         3.5 Species and habitat conservation priority lists   77



Some of these Habitats of Community Interest are         species, the liverwort Western Rustwort
given priority status because the Community has a        Marsupella profunda. Further information on their
particular responsibility for their conservation in      occurrence is given in McLeod et al. (2002).
view of the proportion of their natural range which          Together the SACs and SPAs are known as the
falls within the EU territory. The importance of         Natura 2000 network. This network will provide
these Priority Habitat types is emphasised at sev-       the most stringent protection mechanism for
eral places in the Directive (Articles 4 and 5 and       many habitats and species with restricted ranges
Annex III), not only in terms of the selection of        or small populations. However, for other more dis-
sites, but also in the measures required for site        persed species (e.g. those associated with many
protection (Article 6) and surveillance (Article 11).    farmland habitats), site designation is unlikely to
Definitions and interpretations of the Habitats of       protect more than a small portion of the total
Community Interest have been provided by the             resource. The Habitats Directive therefore also spe-
European Commission Environment Directorate              cifies that the conservation status of flora and
(European Commission, 1999) and further informa-         fauna should be maintained throughout their
tion on their occurrence in the UK is given by           range.
McLeod et al. (2002). Of these Habitats of                   In the UK, the Directives have been transposed into
Community Interest, 76 are believed to occur in          legislation by The Conservation (Natural Habitats, &
the UK, of which 22 are Priority Habitat types.          c.) Regulations 1994 and The Conservation (Natural
   Species of Community Interest (listed in Annex II     Habitats, &c.) (Northern Ireland) Regulations 1995,
of the Directive) are those that, within the EU ter-     as amended (informally known as ‘The Habitats
ritory are:                                              Regulations’). Under British law all SACs and SPAs
                                                         will be underpinned by SSSI designations.
1. endangered, except those species whose natural
                                                             The UK submitted its first report to the European
   range is marginal in that territory and which are
                                                         Commission summarising the implementation of
   not endangered or vulnerable in the Western
                                                         the Habitats Directive in the UK from 1994 to
   Palaearctic region;
                                                         December 2000 (Salmon, 2001). As of June 2003
2. vulnerable, i.e. believed likely to move into the
                                                         there were 242 SPAs in the UK, covering some
   endangered category in the near future if the
                                                         1 470 000 ha. As of January 2004, 605 sites covering
   causal factors continue operating; or
                                                         some 2 500 000 ha had been proposed as candidate
3. rare, i.e. with small populations that are not at
                                                         SACs, and a further ten as proposed SACs covering
   present endangered or vulnerable, but are at risk.
                                                         some 290 000 ha. As a matter of policy for planning
   The species are located within restricted geogra-
                                                         and all other consent regimes, the UK
   phical areas or are thinly scattered over a more
                                                         Government and the devolved administrations
   extensive range; or
                                                         already treat candidate SACs as though they were
4. endemic and requiring particular attention by
                                                         fully designated.
   reason of the specific nature of their habitat and/
   or the potential impact of their exploitation on
   their habitat and/or the potential impact of their    3.5.6      Wildlife & Countryside Act
   exploitation on their conservation status.
                                                         The Wildlife & Countryside Act 1981 was intro-
A number of Species of Community Interest are            duced as the principal mechanism for the legisla-
also given priority status because the Community         tive protection of wildlife in Great Britain. It does
has particular responsibility in view of the propor-     not extend to Northern Ireland, the Channel
tion of their natural range which falls within the       Islands or the Isle of Man. It has been subsequently
EU territory.                                            amended with significant changes relating specifi-
   In recent times 51 Species of Community               cally to Scotland and England and Wales. This leg-
Interest have been recorded in the UK, but only          islation is the chief means by which the
one Priority Species currently occurs as a native        Convention on the Conservation of European
78      3 BIODIVERSITY EVALUATION METHODS



Wildlife and Natural Habitats (the ‘Bern                    Countryside Act 1981 in England and Wales up to
Convention’) and the European Union Directives              date. The importance of biodiversity conservation
on the Conservation of Wild Birds and Natural               is also given a statutory basis, requiring govern-
Habitats and Wild Fauna and Flora are implemen-             ment departments to have regard for biodiversity
ted in Great Britain. The Act is divided into four          in carrying out their functions, and to take positive
parts.                                                      steps to further the conservation of listed species
                                                            and habitats.
*    Part I is concerned with the protection of wildlife.
                                                                Section 74 of the CROW Act requires the
*    Part II relates to the countryside and national
                                                            Secretary of State for England and the National
     parks (and the designation of protected areas).
                                                            Assembly for Wales each to publish a list of species
*    Part III covers public rights of way.
                                                            and habitat types that are of principal importance
*    Part IV deals with miscellaneous provisions of
                                                            for the conservation of biological diversity in
     the Act.
                                                            England and Wales, respectively. The Section 74
Sections 1–8 of Part 1 relate to the protection of          list for England can be viewed on the DEFRA web
birds. Section 1 prohibits the intentional killing,         page http://www.defra.gov.uk/wildlife-countryside/
injuring or taking of any wild bird and the taking,         cl/habitats/habitats-list.pdf. The equivalent list for
damaging or destroying of the nest (while being             Wales can be viewed on the National Assembly for
built or in use) or eggs. It prohibits possession of        Wales website: http://www.wales.gov.uk/subienvir-
wild birds (dead or alive) or their eggs. There are         onment/content/guidance/species-statement-e.htm.
additional penalties for offences relating to birds         These two lists are based on UK Biodiversity Action
on Schedule 1, and it is also an offence to disturb         Plan (UK BAP) Priority Habitats and Species lists.
Schedule 1 birds at the nest or the dependent                   In England, The England Biodiversity Strategy
young of Schedule 1 birds. Section 2 outlines excep-        (DEFRA, 2002), developed under the UK BAP
tions to Section 1: notably, it identifies quarry and       process, is the principal means by which the
pest species.                                               Government complies with its duties to conserve,
   Section 9 prohibits the intentional killing, injur-      and promote the conservation of, habitats and
ing or taking of, the possession of and the trade in        species listed under Section 74 of the Act. The list
wild animals listed on Schedule 5. In addition,             will be kept under review and a report on any
places used for shelter and protection are safe-            necessary revisions will be made as part of the
guarded against intentional damage, destruction             first report on progress on the Biodiversity
and obstruction; animals protected under the rele-          Strategy for England.
vant part of Section 9 must not intentionally be
disturbed whilst occupying those places.
   Section 13 identifies measures for the protection
                                                            3.5.8      Nature Conservation (Scotland)
of wild plants. It prohibits the unauthorised inten-
                                                                       Act 2004
tional uprooting of any wild plant species and for-
bids any picking, uprooting or destruction of plants        In a similar way the Nature Conservation (Scotland)
listed on Schedule 8. It also prohibits the sale, or        Act 2004 overhauls the current legislation concern-
possession for the purpose of sale, of any plants on        ing SSSIs and the protection of wildlife in Scotland
Schedule 8 or parts or derivatives of Schedule 8            and gives a statutory basis for the Scottish
plants.                                                     Biodiversity Strategy (SBS). It aims to make the
                                                            SSSI system more adaptable and efficient and
                                                            strengthens the legal protection of specified
3.5.7        The CROW Act 2000
                                                            species of plant and animal. The Act imposes a
The Countryside and Rights of Way Act 2000                  new duty on public bodies ‘in exercising any
(CROW Act) strengthens the legal protection for             functions, to further the conservation of biodiver-
threatened species and brings the Wildlife &                sity so far as is consistent with the proper
                                                           3.5 Species and habitat conservation priority lists   79



exercise of those functions’. The Act refers them to       extinction risks within the region. Less often,
the SBS, which elaborates on what this duty may            upgrading may occur where the population within
involve.                                                   the region is a demographic sink, such that it is
   The SBS is underpinned by a series of five imple-       unable to sustain itself, and where the extra-regio-
mentation plans, a research strategy and a set of          nal source is expected to decrease. To date these
biodiversity indicators. Details can be found at           guidelines and the IUCN (2001) criteria have not
www.scotland.gov.uk/biodiversity. The Scottish             been applied in any UK Red Lists, but have been
Executive is also publishing lists of those habitats       used elsewhere, e.g. for birds in Sweden, Finland
and species considered to be of principal impor-           and Switzerland.
tance in respect of the new responsibilities placed           However, the IUCN criteria still only focus on
on public bodies.                                          establishing extinction probabilities and do not
                                                           prioritise species according to the importance of
                                                           the biogeographical populations in question. This
3.5.9      National Red Lists and lists of
                                                           lack of a ‘big-picture’ view can result in some
           species of conservation concern
                                                           important priorities being missed (Mace & Collar,
A wealth of Red Lists have been produced in the UK,        2002). In the UK, and other developed countries
and elsewhere, on species that are considered to be        with relatively low levels of biodiversity, there is
at risk of national extinction (as described in Part III   also justification for giving attention to species that
of this Handbook). These have typically adapted the        remain relatively common (and thus are currently
early IUCN criteria and used the same largely sub-         at a very low risk of extinction) but are nevertheless
jective categories of threat (see Section 3.5.1). This     declining. Consequently, some recent assessments
has tended to produce lists that are dominated by          of the conservation status of species in the UK have
rare species, many of which are likely to be at the        moved away from the narrower Red Listing of
edge of their range, and hence these lists under-          threatened species to more inclusive lists of species
value global priorities. Application of the new            of conservation concern, which, for example,
quantitative IUCN (2001) criteria is also problema-        include species with internationally important
tical at regional or national levels, as conspecific       populations and common but declining species.
populations (i.e. populations of the same species)         Such an approach has been taken for birds in the
may support the population of interest. IUCN               UK (Gregory et al., 2002) and all species under the
(2003) have therefore produced guidelines for the          UK BAP.
regional and national application of the IUCN                 Red Lists and other lists of species of conserva-
(2001) Red List criteria that go some way to over-         tion concern should therefore be taken into
coming the problems described above. It should             account in evaluations as appropriate to their
also lead to greater standardisation of criteria,          underlying objectives and particular assessment
which will aid comparisons between countries.              criteria. Detailed descriptions of such UK lists for
   According to the IUCN guidelines, regional and          each species group are given later in the appropri-
national assessments should be carried out in a            ate chapters of this Handbook.
two-step process, which differs slightly for breed-
ing and non-breeding populations. In Step 1, the
                                                           3.5.10      UK Biodiversity Action Plan listed
IUCN 2001 Red List criteria are applied to the popu-
                                                                       habitats and species
lation in question, resulting in a preliminary cate-
gorisation. In Step 2, the existence and status of any     The Biodiversity Convention was ratified by the UK
conspecific populations outside the region that            Government in June 1994. However, even before
may affect extinction risks is taken into account.         this, the Government had committed itself to
For example, preliminary categories should be              produce a consultative national action plan,
downgraded (i.e. to a lower threat status) if immi-        Biodiversity: the UK Action Plan (Anon., 1994)
gration from outside the region is likely to reduce        based on the principles of the Biodiversity
80      3 BIODIVERSITY EVALUATION METHODS



Convention. This plan was launched with the over-            habitats of community importance, is provided in
all goal: ‘to conserve and enhance biodiversity              Appendix 2.
within the UK and to contribute to the conserva-                Species that qualify under one or more of the
tion of global biodiversity through all appropriate          following criteria should be considered as Species
mechanisms’.                                                 of Conservation Concern (SoCC) :
   The plan stated that this is to be achieved
                                                             *   Threatened endemic and other globally threa-
through the conservation and, where practicable,
                                                                 tened species;
enhancement of:
                                                             *   Species where the UK has more than 25% of the
*   the overall populations and natural ranges of                world or appropriate biogeographical population;
    native species and the quality and range of wild-        *   Species where numbers or ranges have declined by
    life habitats and ecosystems;                                more than 25% in the last 25 years;
*   internationally important and threatened species,        *   In some instances, where a species is found in
    habitats and ecosystems;                                     fewer than fifteen 10 km squares in the UK; and
*   species, habitats and natural and managed ecosys-        *   Species listed in the EU Birds or Habitats
    tems that are characteristic of local areas;                 Directives, the Bern, Bonn or CITES Conventions,
*   the biodiversity of natural and semi-natural habi-           or under the Wildlife & Countryside Act 1981 and
    tats where this has been diminished over recent              the Wildlife Order (Northern Ireland) 1985.
    decades.
                                                                Species that qualify for one or both of the follow-
By October 1999, three hundred and ninety-one                ing categories should be considered as Priority
Species Action Plans (SAPs) and 45 Habitat Action            Species:
Plans (HAPs) had been published in reports pro-
                                                             *   Species that are globally threatened;
duced by the UK Steering Group (UKSG, 1995a,b;
                                                             *   Species that are rapidly declining in the UK, i.e. by
UKBG, 1998a,b,c, 1999a,b,c).
                                                                 more than 50% in the past 25 years.
   As part of the development of the UK BAP, lists
were produced of Priority Habitats and Species               The intention is that all Priority Species should be
requiring conservation actions. Priority Habitats            the subject of conservation action through the
were defined3 as:                                            development of SAPs. A full list of UK BAP Priority
                                                             Habitats, SoCCs, further notes on their selection
*   Habitats for which the UK has international
                                                             criteria and current versions of HAPs and SAPs are
    obligations;
                                                             available from the UK BAP website library at
*   Habitats at risk, such as those with a high rate of
                                                             www.ukbap.org.uk/Library.
    decline, especially over the past 20 years, or which
                                                                The UK BAP listing of species is primarily
    are rare;
                                                             for guiding strategic conservation priorities
*   Habitats that may be functionally critical (i.e.
                                                             and therefore reference to SoCC lists with respect
    areas that are part of a wider ecosystem but pro-
                                                             to site evaluations is not always appropriate.
    vide reproductive or feeding areas for particular
                                                             Implementation of necessary actions for SoCC
    species); and
                                                             species will be largely through the SAPs and coun-
*   Habitats that are important for UK BAP Priority
                                                             try strategies such as the England and Scottish
    Species (see below).
                                                             Biodiversity Strategies. Many SoCC species are
All EU Habitats of Community Interest (Annex I)              also common and widespread, such as the Song
occurring within the UK are included as UK BAP               Thrush Turdus philomelos (a Priority Species),
Priority Habitats. A full list of UK BAP Priority            and site-based actions may not be of significant
Habitats, indicating their relationship to EU                benefit for such species. Reference should there-
                                                             fore be made to individual SAPs to assess the
3
 Two additional categories were identified and adopted for   importance of site-based measures for SoCC
marine habitats.                                             species.
                                                          3.6 Site evaluations and selection of protected areas   81



                                                            incorporating complementarity considerations.
3.6       SITE EVALUATIONS AND SELECTION
                                                            Simple criterion-led approaches typically define a
          OF PROTECTED AREAS
                                                            standard (e.g. area of a particular habitat of conser-
3.6.1       General principles and criteria                 vation importance, or threshold number of indivi-
                                                            duals of a species of conservation importance) such
One of the commonest reasons for undertaking                that all sites exceeding this standard are included
ecological evaluations is to assess the importance          in the protected area series.
of a site in relation to potential designation as some          An important advantage of this approach is that it
form of protected area. Evaluations may also be             is simple and can be applied gradually, with sites
carried out at a later stage for management plan-           added sequentially to the series as data become avail-
ning or related purposes to identify, or confirm, the       able. The conservation importance of features, habi-
important features (such as species or habitats) that       tats and species needs to be reasonably well known,
are present and that qualify the site for a particular      but the location of all habitats and species of con-
designation status.                                         servation importance within the territory being con-
   The approaches and methods used for identify-            sidered does not need to be known in advance.
ing areas that should receive some form of protec-              One of the problems with a criterion-led
tion vary widely and depend on the overall                  approach is that it is open-ended, such that sites
objectives for individual sites and the series of pro-      are added no matter how much of a resource has
tected areas within a given territory (Williams,            been given protection. Thus, overall representation
1998; Margules & Pressey, 2000). However, an over-          targets for species or habitats in a protected area
all set of principles for protected area selection is       series are not explicitly stated beforehand. This
given by Ratcliffe (1977), who suggests that priority       may be acceptable if the aim is to protect all sites
should be given to sites and features that:                 above a defined value, but may result in problems if
*   are intrinsically most fragile and sensitive to         criteria turn out to be too inclusive or too restrictive.
    human impact;                                           To overcome such problems, the idea of setting
*   have already been reduced in area or quality            objectives for a protected area network as a whole
    through human impact;                                   has emerged and is receiving increasing support.
*   are predictably most vulnerable to further                  Target-setting can also explicitly ensure that the
    damage and loss through a combination of fragi-         series of sites contains adequate representation of
    lity, sensitivity and probable expansion of impact;     the total territory range of habitats, vegetation
*   would represent the greatest loss to nature con-        communities, species assemblages and individual
    servation if they were damaged or destroyed; and        species that are considered to require protection.
*   would be the most difficult to restore or re-create     However, this needs to be carried out against a
    if they were damaged or destroyed.                      classification of the range of variation in habitats,
                                                            communities and species, which the series of sites
A variety of approaches have been developed for             is intended to represent. The minimum aim of the
identifying sites that should be included in a pro-         representative principle would therefore be to
tected area series, most of which focus on the              select a series of sites that complement one
fourth point above, i.e. evaluations of a site’s eco-       another in terms of the habitats or species present.
logical importance.                                         Ideally, each habitat or species should be repre-
                                                            sented by at least one, and preferably the best,
Threshold criteria versus target-led approaches             example. However, in the face of existing threats
The selection of protected areas is essentially based       to sites this approach is unlikely to be sufficient to
on a process of comparison, usually with certain            maintain the representative set of habitats and
selection criteria. These can be broadly categorised        their characteristic features. An important princi-
as simple criterion-led approaches, selection meth-         ple of site selection should therefore be that as
ods based on targets, and selection strategies              rarity, threat or other ecological values increase,
82    3 BIODIVERSITY EVALUATION METHODS



then so does the need to ensure that a larger pro-         the degree of flexibility in the series and irreplace-
portion of the habitat or species’ population is           ability of individual sites to be evaluated.
under protection.                                          Irreplaceability is particularly important because
   In practice, most protected area series have been       the loss of the more irreplaceable sites in a series
selected by using simple criterion-based systems           closes options of ever achieving a representative
within an overall representation target framework.         protected area system. Irreplaceability provides a
This is probably because it is easier to make relative     single measure of conservation value of a site that
judgements about the conservation value of habi-           is defensible (Bibby, 1998). Thus, some algorithms
tats and species than it is to make difficult a priori     give a high priority to sites that contain species or
decisions on how much should be protected.                 habitats that are not found elsewhere. Some algo-
Protection targets also often tend to be arbitrary         rithms also check back to ensure that early choices
and unless widespread consultation is undertaken           remain appropriate after the inclusion of others.
they will have little ownership or acceptance out-            Using such an approach, Williams et al. (1996)
side the conservation community.                           identified a set of twenty-seven 10 km  10 km grid
                                                           squares in which all British breeding birds were
Complementarity considerations                             represented at least once. Similarly, Hacker et al.
Criterion-led approaches (which, as described              (1998) showed that all African primate species
below, often focus on species rarity and diversity)        occur within a set of grid squares that cover just
tend to result in selection of sites that are similar in   3.8% of the area sampled.
terms of their range of habitats and species. Such            However, despite a great deal of research being
approaches are therefore inefficient in the selec-         carried out into protected area selection techni-
tion of sites where objectives focus on representa-        ques, selection algorithms and other computerized
tion across a network of sites. Instead, an                approaches have been used only infrequently by
evaluation of the degree to which species commu-           those involved in making decisions on the estab-
nities complement each other may help identify a           lishment of protected area networks (Prendergast
minimum set of sites or area of land required to           et al., 1999). This appears to be mainly because
safeguard all species within a particular group            decision-makers are unaware of the sophisticated
(Vane-Wright et al., 1991; Howard et al., 1997;            computer tools for reserve selection. Where this is
Balmford & Gaston, 1999). This involves identifying        not the case, there have been problems with fund-
a series of sites whose habitats or species lists lar-     ing and with understanding of the tools, and a
gely complement each other.                                general antipathy towards what is seen as a pre-
   The majority of protected area selection techni-        scriptive approach to conservation. There are also
ques incorporating complementarity do so by                common methodological limitations to its applica-
using iterative steps; at each step all sites are com-     tion, including the following:
pared to test how well they complement areas that
have already been chosen. The most straightfor-            *   Inadequate data on species and habitat distributions.
ward approach, which uses a so-called ‘greedy’                 Selection algorithms can only be fully applied
algorithm (i.e. rule-based calculation), first selects         when species distributions are well known and
the area that is richest in the selected feature of            the contents of potential nature reserves deduced.
interest (e.g. threatened species), then selects the           Consequently, as noted by Sutherland (2000),
area that adds the highest number of features that             where there are sufficient data for carrying out
are not in the first. It then selects the area that adds       such analyses, protected area networks are usually
the highest number of features that are not present            already well established.
in either of the first two, and so on.                     *   Differences in scale. There is often a substantial mis-
   Alternatives include various weighted algo-                 match between the resolution at which distribu-
rithms, including ones that consider a site’s level            tion data are available and the scale at which
of irreplaceability (Pressey et al., 1994). This allows        protected areas are designated. Indeed, data
                                                             3.6 Site evaluations and selection of protected areas   83



    resolution is often close to one order of magnitude        in their ecological characteristics. Thresholds of
    greater than the average size of protected areas           significance, for instance size or species diversity,
    (Hopkins et al., 2000).                                    might differ significantly between habitat types or
*   Dependence on presence–absence data. A species’ pre-       biogeographical regions. It is therefore advisable to
    sence at a site does not necessarily mean that it          group sites for evaluation into similar types before
    has a viable population there. Areas selected by           comparisons are carried out.
    using complementarity may miss the best examples              More detailed consideration is given below to
    of certain species’ populations, particularly of           five of the most widely used criteria.
    those associated with species-poor areas. Some
    selection algorithms have therefore been refined           Rare or abundant species
    to take account of species abundance rather than           Many protected areas have been selected and desig-
    mere presence–absence (Turpie, 1995).                      nated purely on the basis of the occurrence of spe-
*   Political and economic considerations. Sites selected      cies. This is largely because species provide the
    through complementarity analysis may not neces-            simplest quantifiable and most objective currency
    sarily offer the best opportunities for successful         of conservation value. Most species-based selection
    long-term conservation. In practice, the selection         criteria assess a site’s importance for particular
    of protected areas is usually strongly influenced          rare species or other species of conservation impor-
    by political considerations, opportunities, land           tance and/or the abundance of species.
    prices (Ando et al., 1998), the threat of local devel-        Such approaches have been frequently used for
    opment (Margules & Usher, 1981), or proximity to           the selection of protected areas for birds. For example,
    existing reserves. Accordingly, reserve selection          the Ramsar Convention includes a criterion for
    algorithms have recently been developed that               designating Ramsar sites on the basis of the pre-
    incorporate rules for including mandatory areas,           sence of 20 000 waterfowl or more than 1% of a
    forcing adjacency and excluding undesirable                flyway population (see Section 3.7.6 below).
    areas (Lombard et al., 1997).                              BirdLife International has taken this type of
                                                               approach and developed it to create a standard set
                                                               of global criteria and regionally specific criteria for
3.6.2        Evaluation and selection criteria
                                                               identifying Important Bird Areas (IBAs), which
Although no standard set of criteria has emerged               BirdLife recommend should receive appropriate
for the purpose of site evaluations, assessment cri-           statutory protection (Heath & Evans, 2000).
teria have commonly focused on the presence of                    The species occurrence approach is also increas-
certain rare (or otherwise threatened) species or              ingly being applied to other taxa. For example,
habitats, diversity, size and naturalness. Other cri-          PlantLife International has developed criteria for
teria have included fragility, degree of threat, edu-          Important Plant Areas for Europe (Anderson,
cational, scientific, recreational and cultural value,         2002), which include thresholds for the presence
ecological or geographical location, the presence of           of rare species (as well as rare habitats).
potential buffer areas, shape, accessibility and
potential conservation effectiveness (see Smith &              Diversity
Theberge (1986) for review).                                   Species and habitat diversity has been one of the
   Whichever criteria are used, to be defensible they          most frequently used evaluation measures for bio-
need to be objective, explicit, based on widely                tic communities. Selection of diverse sites tends to
accepted ecological scientific principles and the              be favoured because they are believed to contain
best available data, and (ideally) quantifiable.               more of the variety of natural resources within a
However, criteria have varied widely in this respect.          given area; however, problems may arise if comple-
   It is also important to remember that criteria              mentarity is not taken into account. There are also
will not be meaningful if applied across a wide                less tangible aesthetic values in diversity and it has
range of sites and habitats that differ considerably           often been suggested that diversity is a factor
84    3 BIODIVERSITY EVALUATION METHODS



promoting ecosystem stability, but this idea is not    Consequently, habitat richness indicators need to
well founded in observation or theory (Bibby,          be related to expected levels of species richness for
1998).                                                 habitats and, ideally, ecological regions.
   Species or habitat richness (measured in num-       Degradation of natural habitats can also actually
bers of species occurring) are the most intuitively    result in an increase in overall biodiversity even
simple measures of diversity. However, if popula-      though species of conservation importance may
tion sizes or areas are measured, they can be          decline. Thus, species richness and diversity mea-
combined into more sophisticated diversity             surements should also focus on those species that
indices (Magurran, 1983). Two commonly used            are characteristic of the habitat, and not on the
                                   P
indices are Shannon’s Index,         (pi ln pi), and   overall richness of the habitat.
                       P 2
Simpson’s index, 1 À pi . Here, pi is the propor-         In practice, diversity scores have been little used
tion of individuals in the population belonging to     for protected-area selection. However, some eva-
species i. In each case, a high value indicates a      luation scoring systems have been developed that
large number of species with similar abundances,       combine abundance, species conservation value
whereas a low number indicates domination by a         and species richness data. For example, in the UK
few species. The requirement for abundance             this approach has been used for selecting SSSIs
information makes the measurement of diversity         (NCC, 1989) for birds and amphibians (see Part III).
more time-consuming, however, and indices are             The Environment Agency (EA) have also devel-
more difficult to interpret than species or habitat    oped a Community Conservation Index (CCI) for
numbers.                                               evaluating the conservation importance of riverine
   Species richness measurements also depend on        macro-invertebrate communities. Although this
the scale at which they are measured; richness         has not been introduced into full use yet by EA, it
invariably increases with the increasing size of       has been used in a number of Environmental
the area surveyed (MacArthur & Wilson, 1967;           Impact Assessments. A CCI score is calculated by
Rosenzweig, 1995). Such species–area relations         first assigning tabulated Conservation Scores to
are likely to occur as a result of environmental       each species in the sample based on their national
heterogeneity. Increasing the area will include        conservation status. An overall CCI score is then
additional habitat types, or variations within         derived by calculating the mean Conservation
these, depending on how habitats are defined.          Score and multiplying this by a tabulated
However, at large scales biogeographic and historic    Community Score, which reflects either species
factors may be important contributors. Species–area    richness or the rarest taxon present (depending
effects can be examined and corrected by regres-       on whichever gives the highest score). A high CCI
sion analysis. Richness estimates may also vary        score indicates a site with high conservation value
according to sampling effort (for instance the num-    for its rare taxa or species richness. CCI scores can
ber of quadrats measured). This can also be cor-       vary from 1.0 (no conservation interest) to 30.0 þ
rected for (although it is more complex) by            (high conservation interest).
rarefaction, which resamples the data to estimate
the number of species for a fixed (smaller) sam-       Size (extent)
pling effort (Simberloff, 1972).                       The size of a site, typically its area but sometimes
   Particular care must be taken in interpreting the   its length (linear habitats), depth or volume (e.g.
results of richness and diversity assessments. For     lakes), is important in modified environments
example, some semi-natural habitats of high ecolo-     where natural communities are fragmented and
gical value are characteristically species poor.       isolated. In more extensive habitats, it is less
Indeed, Bibby (1998) suggests that the appeal of       obvious because the boundaries may be difficult
diversity as an indicator has almost certainly led     to define.
to under-representation of inherently less diverse        Of foremost importance is the attainment of a
habitats in many protected area systems.               minimum size, which will vary according to the
                                                             3.6 Site evaluations and selection of protected areas   85



protection objectives. In a fragmented landscape               Naturalness
large blocks may be necessary to contain viable                Naturalness is a criterion that is difficult to define
populations of particular species (especially wide-            objectively, yet it is highly valued in conservation
ranging predators, such as large raptors or, outside           assessment. The principal reason for valuing natur-
the UK, Wolves Canis lupus, Lynx Lynx lynx, etc.).             alness as a criterion for identifying sites for protec-
Large sites may also be more likely to support inter-          tion is that there is often a close relationship
ior species, which are intolerant of habitat edges.            between the naturalness of a habitat and its biodi-
   Larger sites may also be valued beyond the mini-            versity value. For example, research on river inver-
mum viable size, because large areas of habitat                tebrate communities in England has starkly shown
typically contain a greater diversity of habitats              that diverse physical river reaches, as found in
and species. Size may also influence the manage-               more natural rivers, support invertebrate commu-
ment options available, such as the ability to                 nities of much greater diversity than highly mod-
resolve usage conflicts or to enable natural vegeta-           ified or uniform ones (Smith et al., 1991). Perhaps
tion succession and landscape-scale dynamics.                  most importantly, a high proportion of rare species
   Related to the issue of size is that of shape, and a        are often associated with natural or near-natural
site’s proximity and linkage to other similar habi-            habitats, primarily because such habitats now
tat blocks, as these factors may influence species             tend to be rare in developed countries.
occurrences through emigration and immigration.                   The application of the naturalness criterion is
Based on biogeographical studies, some basic rules             particularly relevant to habitats such as rivers
of thumb have been proposed, and widely referred               (Boon et al., 1996a) and other wetlands. It is also
to, for the selection of nature reserves (Diamond,             consistent with developments under the Water
1975). These are simply that:                                  Framework Directive where the condition of river-
                                                               ine habitats will be related to definitions of pristine
*   larger areas are better than smaller areas;                habitat. Application of the naturalness criterion to
*   one large area is better than separated areas of the       other habitats needs some care and qualification as
    same total area;                                           many that are of conservation value (such as moor-
*   adjacent areas are better than isolated areas;             lands, heathlands, and flower-rich grasslands) are
*   linkages (‘corridors’) between areas are better            plagioclimax habitats that have been maintained
    than completely isolated areas;                            by centuries of human intervention. Determining
*   clusters of areas are better than areas in a line; and     naturalness therefore requires considerable knowl-
*   compact areas are better than linear areas.                edge of individual habitats, the effects of long-term
                                                               natural process on them, and their response to
However, there have been many erroneous and                    human intervention.
inappropriate applications of these island-biogeo-                Other reasons for favouring natural habitats and
graphy-based theories to site evaluation and the               ecosystems have included the scientific need for sites
selection of protected areas. Other factors need to            at which to study natural ecological processes. There
be taken into account in considering the optimal               are also emotional and aesthetic factors at play in the
size, distribution and shape of nature reserves.               greater perceived value of wilderness and ecosystems
Indeed, in many cases these other factors will be              less tainted by humans (McCloskey & Spalding, 1989).
more important. For example, in some circum-
stances, very small reserves might be appropriate,             Representativeness
for instance in safeguarding plants with minute                This criterion (also occasionally referred to as typical-
ranges. In linear habitats, such as rivers, other              ness) aims to select sites that best represent a parti-
properties such as length and continuity become                cular habitat of interest and which possess as many
important. These typically relate to their corridor            desirable habitat characteristics and special features
function, their value rising with the number or                as possible. On closer analysis, however, it is clear
areas of sites they are able to connect.                       that this selection criterion involves a mixture of
86      3 BIODIVERSITY EVALUATION METHODS



desirable attributes that are best separated. In other      indices of habitat or species assemblage value, in
words, the concept of representativeness is made up         an attempt to provide more quantifiable and objec-
of other separate commonly used ecological evalua-          tive assessments.
tion criteria of diversity in particular, but also to          Several authors have proposed the use of multi-
some extent of size and rarity. It may therefore be         ple evaluation criteria, which are then given
more appropriate to regard representativeness as an         weights or priorities by using scoring systems.
underlying principle, which site selection processes        Scoring enables abstract evaluation criteria to be
aim to satisfy, both within a single site and through       expressed numerically and hence they can be used
the totality of all sites, and then to satisfy this prin-   more readily in decision-making. Various scoring
ciple by the application of criteria relating to it.        procedures have been developed and have been
                                                            reviewed by Margules & Usher (1981), Smith &
                                                            Theberge (1986) and Usher (1986). Westman
3.6.3      Nature Conservation Review criteria
                                                            (1985) summarises the four main types of scale
The use of multiple criteria that take into account         and associated permissible mathematical and sta-
the presence of species, habitat quality and other          tistical operations associated with scoring
ecological factors gives a better integrated assess-        procedures.
ment of the overall value of a site than does concen-          Although a wide variety of systems have been
tration on selected attributes or species groups.           developed for evaluating habitats, few have been
However, the broad range of information that must           widely accepted or used for protected area selec-
be taken into account makes evaluation more com-            tion in the UK or internationally. However, one
plex and subjective. One set of criteria that have been     system that has been widely used is SERCON
particularly frequently used are those developed by         (System for Evaluating Rivers for CONservation),
Ratcliffe (1977) for the UK Nature Conservation             which was developed in the mid-1990s as a techni-
Review (NCR). These are summarized in Table 3.3.            que for assessing the conservation value of rivers in
   Although now over 25 years old, the NCR criteria         the UK (Boon et al., 1996a,b). It aims to provide a
have been widely adopted and adapted in the UK              more comprehensive, rigorous, and repeatable
and abroad. In particular, they formed the basis for        method for conservation evaluation of rivers than
the UK SSSI selection criteria (NCC, 1989). They            had hitherto been available. SERCON evaluations
have also been frequently used as the basis for             first involve gathering all available information
ecological evaluations for management planning              about the physical character of a river system,
purposes and many ecological impact assessments             from Environment Agency River Habitat Corridor
in the UK. However, Treweek (1999) points out that          Surveys (Environment Agency, 2003), which then
these criteria have their drawbacks; notably the            guides the user into determining the size of the
lack of criteria relating to recoverability or the          assessment unit (the Evaluated Corridor Site).
replaceability of natural resources of particular           Data on other physical, chemical and biological
importance when assessing impacts and mitigating            features of the river corridor are then gathered
for them. In addition, not all of the criteria (e.g.        together from all relevant sources. Finally the col-
intrinsic appeal) are measurable by using defensi-          lated data are translated into a series of scores for
ble, consistent and objective techniques.                   specific attributes on a scale of 0–5. Scores are
                                                            weighted and combined to give separate indices
                                                            of conservation value for six criteria (Physical
3.6.4       Scoring systems
                                                            Diversity; Naturalness; Representativeness; Rarity;
A frequently cited problem with ecological evalua-          Species Richness; Special Features). Scores can be
tions is that assessments are subjective and there-         combined to give an overall assessment on an A–E
fore comparisons among sites may be difficult               scale, with A representing the highest-quality band.
or misleading. Some multiple-criterion scoring sys-            A scoring procedure may also be used for
tems have therefore been established to produce             selecting SSSIs on the basis of their breeding bird
                                                          3.6 Site evaluations and selection of protected areas   87



Table 3.3. The NCR criteria developed by Ratcliffe (1977) for evaluating nature conservation importance

                                                   Application / notes

Primary criteria
Size                                               Including both area of vegetation types and population sizes
                                                   for individual species.
Diversity                                          Applied either as simple species richness, or by giving differ-
                                                   ent weightings to species according to their ‘interest’.
Rarity                                             Applied either to habitats or to species. The latter most
                                                   commonly tested by comparisons with national or county
                                                   population size or distribution by 10 km squares.
Naturalness                                        Habitats that are least intensively modified by humans are
                                                   generally more highly regarded.
Representativeness or typicalness                  A measure of how well the study area represents habitats or
                                                   vegetation types on a wider scale.
Fragility                                          Some habitats or species are especially vulnerable or sensi-
                                                   tive to anthropogenic change. Those with restricted area or
                                                   ranges are generally held to be more vulnerable.
Secondary criteria
Recorded history                                   Can be useful in confirming that a site has been ‘important’
                                                   for some time. Sites with a long history of study may con-
                                                   tribute significantly to our understanding of ecological
                                                   processes.
Potential value                                    Relates to the likelihood that appropriate management
                                                   could restore or enhance an area’s ecological value.
Position in geographical or ecologi-               Some areas of fairly low ‘intrinsic value’ may be more
cal unit                                           important because they form successional stages between
                                                   more important areas. In addition, nationally common
                                                   habitats or species might be very rare locally.
Intrinsic appeal                                   Habitats or species with public appeal promote the cause of
                                                   nature conservation and can attract funds. This criterion can
                                                   also be interpreted to include estimates of public use, access
                                                   and amenity value.

Source: Treweek (1999).


assemblage (NCC, 1989). This ‘BTO index’ is largely         are primarily associated with the habitat, or which
derived from a study of bird communities in differ-         are particularly threatened by habitat change. All
ent habitats in Britain (Fuller, 1982). The index for a     species with a British population of more than one
site is calculated by summing the scores across all         million birds are excluded. Each species score is
species that are regularly present according to a           based on its British population size, such that species
habitat-specific list. Each habitat list includes all       with fewer than 10 pairs score 6, whereas species
the characteristic species of the habitat that have a       with a population of 100 000–1 000 000 pairs score 1.
total British population of fewer than 1 000 pairs (at      The site index may then be compared with the stated
the time) and all other more abundant species that          site selection threshold score, which is based on the
88    3 BIODIVERSITY EVALUATION METHODS



theoretical maximum score for each habitat for spe-       Area categories      and   associated   designation
cies with more than 100 pairs. For most habitats, the     criteria.
threshold for qualification as an SSSI is taken as 50%
                                                          *   World Heritage Sites
of the theoretical maximum.
                                                          *   Biosphere Reserves
   Treweek (1999) notes that there has been some
                                                          *   Biogenetic Sites
debate over the efficiency and validity of scoring
                                                          *   Wild Birds Directive SPAs
methods for ecological evaluations (van der Ploeg &
                                                          *   Habitats Directive SACs
Vlijm, 1978; Gotmark et al., 1986; Usher, 1986;
                                                          *   Ramsar sites
Anselin et al., 1989). In particular, difficulties have
                                                          *   National Nature Reserves
arisen over the combination of quantitative and
                                                          *   Site of Special Scientific Interest (Wildlife &
qualitative criteria.
                                                              Countryside Act, CROW Act, Nature Conservation
   As in the examples described above, weighting is
                                                              (Scotland) Act)
often used to convert different systems of measure-
                                                          *   Local Nature Reserves
ment to common formats and to increase the influ-
                                                          *   Wildlife Sites
ence of criteria related to particularly important
ecological factors. However, there is rarely any biolo-   The background to these designations and, where
gical rationale for weightings. For example, there is     appropriate, their selection criteria are provided
no clear basis for deciding that birds with British       below.
populations of 1–10 pairs should be given a weight-
ing six times that of a species with a population of
                                                          3.7.1      World Heritage Sites
100 000 – 1, 000 000 pairs. Weightings therefore
often tend to be arbitrary and based on the subjective    World Heritage Sites are areas of global natural
opinion of a few experts. To solve this problem some      and/or cultural significance, and are nominated
methods use consultation approaches to provide            by the state within which they are situated. The
weighting values. For example, the Environmental          nominations are then considered by a World
Evaluation System (Dee et al., 1973) uses a combined      Heritage Committee of Party States. Sites that are
scoring and weighting method that attempts to com-        accepted are placed on the World Heritage List.
bine scientific measurements of ‘value function’          World Heritage Sites must have strict legal protec-
with selected indicator variables with weightings         tion and any management of the site must ensure
based on values allocated by members of the public        that this continues. Further information on the
chosen to represent relevant interest groups.             Convention and a list of World Heritage Sites can
   When scoring systems are used they should              be found at http://whc.unesco.org/nwhc/pages/
ensure that all criteria and weightings are expli-        home/pages/homepage.htm.
citly defined. Care should also be taken to ensure
that the use of single, numerical indices based on
                                                          3.7.2       Biosphere Reserves
combined scoring and weighting systems does not
create a false impression of precision and conceal        Biosphere Reserves represent globally significant
uncertainties in the underlying evaluation data.          examples of biomes (biological communities) for
                                                          both terrestrial and coastal environments. They
                                                          have particular value as benchmarks or standards
3.7      SITE CONSERVATION DESIGNATIONS
                                                          for the measurement of long-term changes in the
As indicated in Figure 3.1, any evaluation of a site’s    biosphere as a whole. They were devised by
conservation priority should establish its protected      UNESCO under Project No. 8 of their Man and the
area status and, whether or not it is designated as       Biosphere (MAB) programme, and were launched
such, assess its compliance with appropriate desig-       in 1970. Criteria and guidelines for selection of
nation criteria. Ecological evaluations in the UK         sites were produced by a UNESCO task force in
should take into account the following Protected          1974. Although Biosphere Reserves are not always
                                                                           3.7 Site conservation designations     89



statutorily protected areas, all British sites are also    of the Directive) or areas important for the con-
National Nature Reserves. Further information on           tinued well-being or survival of selected non-bird
the MAB Programme and a list of Biosphere                  species (listed in Annex II) in a European context.
Reserves can be found at http://www.unesco.org/            The process for the selection and designation of
mab/index.htm.                                             SACs is set out in Article 4 and its application in
                                                           the UK is summarised in McLeod et al. (2002). SACs
                                                           are identified, initially as Sites of Community
3.7.3       Biogenetic Reserves
                                                           Importance (SCIs), by a two-stage process accord-
In 1973, the European Ministerial Conference on            ing to criteria provided principally in Annex III of
the Environment recommended that a European                the Directive. Sites that are adopted by the
network of reserves to conserve representative             Commission as SCIs must then be designated by
examples of European flora, fauna and natural              the member state as SACs as soon as possible, and
areas be established. Their selection is generally         at the latest within six years.
based on two criteria.                                        Stage 1 is an assessment of the relative impor-
                                                           tance of sites containing examples of the individual
*   Their value in terms of nature conservation: they
                                                           Annex I habitat types and Annex II species in each
    must contain specimens of flora or fauna that are
                                                           member state, against the following summarised
    typical, unique, rare or endangered.
                                                           Annex III criteria:
*   The effectiveness of their protective status: this
    must be sufficient to ensure the long-term conser-     *   Habitats:
    vation or management of a site according to the            (a) degree of representativeness;
    objectives set, as defined in Council of Europe            (b) area;
    Resolution (76) 17.                                        (c) degree of conservation of habitat structure and
                                                                   functions and restoration possibilities; and
All sites in the UK are existing Sites of Special
                                                               (d) global assessment of conservation value (i.e. an
Scientific Interest (SSSIs), and most are also
                                                                    overall assessment, based on (a–c) above).
National Nature Reserves (NNRs).
                                                           *   Species:
                                                                (a) population size and density;
3.7.4       Special Protection Areas (SPAs)                     (b) degree of conservation of the features of the
                                                                    habitat that are important for the species, and
The 1979 EC Directive on the Conservation of Wild
                                                                    restoration possibilities;
Birds requires member states to take conservation
                                                                (c) degree of isolation of the population in rela-
measures particularly for certain rare or vulnerable
                                                                    tion to the species’ natural range; and
species and for regularly occurring migratory species
                                                                (d) global assessment of conservation value (i.e.
of bird. In part this is achieved through the designa-
                                                                     an overall assessment, based on (a–c) above).
tion of statutory Special Protection Areas (SPAs) by the
UK government on the advice of the statutory con-          Further guidance on the assessment of the Annex
servation agencies. All SPAs, apart from those that are    III Stage 1 criteria is given in the EC guidance docu-
proposed for designation at sea, have first to be noti-    ment for the Natura 2000 Standard Data Form
fied as Sites of Special Scientific Interest (SSSIs).      (European Commission DGXI, 1995).
   Further information on SPA designation proce-               The global assessment referred to in the Annex III
dures and criteria is provided in Chapter 24.              criteria is an assessment of the overall value of the site
                                                           for the conservation of the relevant Annex I habitat or
                                                           Annex II species. Particular attention is paid to the
3.7.5       Special Areas of Conservation (SACs)
                                                           global assessment as an overall index of a site’s con-
SACs are designated under the EU Habitats                  servation value. Following the European Commission
Directive and are defined as areas with outstanding        DGXI (1995) guidance, sites are graded A, B or C,
examples of selected habitat types (listed in Annex I      which in the UK has been interpreted as follows.
90      3 BIODIVERSITY EVALUATION METHODS



(A) Sites holding outstanding examples of the habi-                need to be taken into account in assessments of
    tat or populations of the species in a European                sites against the Annex III criteria. These include:
    context.
                                                                   1. restrictions on the site selection obligations in
(B) Sites holding excellent stands of the habitat, or
                                                                      respect of widely dispersed and aquatic species
    populations of the species significantly above the
                                                                      (Article 4.1);
    threshold for SSSI or ASSI4 notification but of
                                                                   2. the requirement to contribute towards the main-
    somewhat lower value than grade A sites.
                                                                      tenance of Favourable Conservation Status
(C) Examples of the habitat or populations of the
                                                                      (Article 2.2 and Article 3.1); and
    species that are of at least national interest (i.e.
                                                                   3. the obligation on each Member State to select a
    usually above the threshold for SSSI or ASSI
                                                                      series of sites that reflects the proportion of the
    notification on terrestrial sites) but not signifi-
                                                                      EU resource of a given habitat or species within
    cantly above this. These habitats or species are
                                                                      their national territory (Article 3.2).
    not the primary reason for SACs being selected.
                                                                   Further selection principles and guidance on the
Although there is a distinction between the princi-
                                                                   interpretation of the Annex III Stage 1 criteria
pal features for which sites have been selected
                                                                   were produced at a meeting between Member
(those graded A or B) and those that are only of
                                                                   States of the Atlantic Biogeographical region and
secondary interest (those graded C), it is important
                                                                   the European Commission in 1994 (Hopkins &
to note that all three grades are qualifying SAC
                                                                   Buck, 1995), as follows.
interest features and hence all such sites are
afforded protection at the European level.                         Provision of information
   Stage 2, which is also known as the ‘moderation’                Acknowledging that the quality and extent of infor-
stage, is an assessment of the overall importance of               mation about habitat types and species varies within
the sites in the context of the appropriate biogeo-                the Region, Member States will provide information
graphical region and the EU as a whole. All the sites              to the Commission in the Natura 2000 data entry
identified by member states in Stage 1 which con-                  form using the best scientific information available
tain Priority Habitat types and/or Priority Species                at the time according to the format agreed by the
are adopted as SCIs. Other sites listed by Member                  Habitats Committee.
States are assessed in relation to their contribution
                                                                   Balancing the national lists
to maintaining or re-establishing, at a Favourable
                                                                   1. Acknowledging that outstanding single interest
Conservation Status, a natural habitat in Annex I or
                                                                      sites in terms of quality, extent or range make
a species in Annex II and/or to the coherence of
                                                                      an important contribution to the Natura 2000
Natura 2000 according to the following sum-
                                                                      network, special emphasis will be given to identi-
marised Annex III criteria.
                                                                      fying and delimiting sites containing complexes
*   the relative value of the site at a national level;               of interests on Annexes I and II as valuable ecolo-
*   the relationship of the site to migration routes;                 gical functional units.
*   the total area of the site;                                    2. Member States will give significant additional
*   the diversity of habitats and species present on the              emphasis in number and area to sites containing
    site; and                                                         priority habitat types and species.
*   the overall quality of the site in the context of the          3. In considering the degree of representativeness of
    biogeographical region and/or the European Union.                 Annex I habitat types on individual sites, Member
                                                                      States will take account of the best examples in
The text of the Directive also includes other site
                                                                      extent and quality of the main type (which is most
selection requirements or qualifications and these
                                                                      characteristic of the Member State) and its main
                                                                      variants, having regard to geographical range.
4
  Areas of Special Scientific Interest as designated in Northern   4. Acknowledging that sites containing Annex I habi-
Ireland.                                                              tat types and Annex II species at the centre of their
                                                                             3.7 Site conservation designations   91



   range will make an important contribution to               been carried out in the UK by expert judgement
   Natura 2000, Member States will take responsibil-          (McLeod et al., 2002), which is in accordance with
   ity for proposing sites containing habitats and spe-       the European Commission view that ‘best expert
   cies that are particularly rare in that Member State,      judgement is an appropriate means of ranking sites’
   with a view to preserving the range.                       (European Commission DGXI, 1995).
5. It is acknowledged that certain habitat types                 This does, however, mean that it is difficult for
   and species listed in Annexes I and II are relatively      others outside the expert group of assessors to
   common and extensive in certain Member States.             establish whether or not a particular site should
   These Member States will have particular respon-           qualify as an SAC. Evaluations of sites with respect
   sibility for proposing a proportion of the resource        to their potential qualification as SACs can there-
   that is sufficient to contribute significantly to the      fore only be judged against the general principles
   maintenance of the habitat types and species at a          described above, their interpretation in the UK (as
   favourable conservation status.                            documented in McLeod et al. 2002) and the prece-
6. Where Annex II species’ populations are too small to       dent set by sites that have been proposed as candi-
   be naturally viable, or where the species occur only       date SACs (i.e. SCIs).
   as vagrants or reintroductions, Member States may             All Annex I habitats and Annex II species consid-
   exclude them from consideration for site selection.        ered to be of European importance that occur in
7. Artificial areas need not be excluded from site            candidate SACs are identified as qualifying fea-
   selection if they have spontaneously given rise to         tures. However, habitat fragments, small popula-
   Annex I habitat types or host Annex II species and if      tions of species, and habitats and species occurring
   it is considered that they have exceptional value.         outside their natural range are generally treated as
                                                              ‘non-significant presences’. Although these habi-
Defining boundaries                                           tats and species are listed on the Natura 2000 stan-
It is acknowledged that different Member States will          dard data forms, they do not require conservation
have different approaches to the definition of bound-         objectives and are not protected under the
aries (e.g. the inclusion of buffer zones within the site),   Directive (as stated in the EC guidance document
according to the habitat type or species concerned and        Managing Natura 2000 sites (European Commission,
the legal and management measures necessary to pro-           2000)). In the UK, the criteria and associated thresh-
tect and extend the landscape context.                        olds used for SSSI selection have generally been
                                                              used to distinguish between non-significant pre-
   A summary list of the selection criteria and addi-         sences and qualifying interest features.
tional principles used for site selection in the UK              For further information on the SAC selection pro-
are shown in Table 3.4.                                       cess see McLeod et al. (2002) and www.jncc.gov.uk/
   The various Annex III criteria and additional guide-       ProtectedSites/SACselection/default.htm. This site
lines listed above are principles that should be taken        also includes up-to-date details of candidate SACs
into account when assessing sites for potential inclu-        and the Annex I habitats and Annex II species
sion in the SAC network. They do not include quali-           represented within them.
tative thresholds or standards against which site
attributes (e.g. area) can be measured. Measurable
thresholds could be set for some criteria, such as            3.7.6      The Convention on Wetlands
area, but SAC selection requires an assessment of                        (Ramsar Convention)
multiple criteria, many of which (e.g. conservation           The Convention on Wetlands of International
structure and function) cannot easily be quantified.          Importance especially as Waterfowl Habitat, more
There is also no straightforward or non-arbitrary way         popularly known as the ‘Ramsar Convention’, was
of combining and weighting the relative importance            the first of the modern global intergovernmental
of the various criteria. The assessment of sites against      treaties on conservation and wise use of natural
the criteria and principles in Table 3.4 has therefore        resources, and came into force in 1975. Compared
92     3 BIODIVERSITY EVALUATION METHODS



Table 3.4. Summary of criteria and additional principles used for SAC selection in the UK

                                                           Reference

Site assessment criteria: Annex II species
(i) Representativeness                                     Annex III Stage 1A(a); Article 1e; Conclusions of 1994
                                                           Atlantic Biogeographical Region Meeting (para. 4).
(ii) Relative surface area of habitat                      Annex III Stage 1A(b); Article 1e; Conclusions of 1994
                                                           Atlantic Biogeographical Region Meeting (para. 4).
(iii) Conservation of structure and function               Annex III Stage 1A(c); Article 1e.
(iv) Global assessment                                     Annex III Stage 1A(d).
Site assessment criteria: Annex II species
(v) Proportion of UK population                            Annex III Stage 1B(a); Article 1l; Conclusions of 1994
                                                           Atlantic Biogeographical Region Meeting (para. 7).
(vi) Conservation of features important for                Annex III Stage 1B(b); Article 1i.
       species survival
(vii) Isolation of species populations                     Annex III Stage 1B(c); Conclusions of 1994 Atlantic
                                                           Biogeographical Region Meeting (para. 7).
(viii) Global assessment                                   Annex III Stage 1B(d).
Additional principles
(ix) Priority/non-priority status                          Annex III Stage 1D; Article 1d; Conclusions of 1994
                                                           Atlantic Biogeographical Region Meeting (para. 3).
(x) Geographical range                                     Article 1e.
(xi) Special UK responsibilities                           Article 3.2; Conclusions of 1994 Atlantic
                                                           Biogeographical Region (para. 6).
(xii) Multiple interest                                    Annex III Stage 2.2(d); Conclusions of 1994 Atlantic
                                                           Biogeographical Region Meeting (para. 2).
(xiii) Rarity                                              Conclusions of 1994 Atlantic Biogeographical Region
                                                           Meeting (para. 5).

Source: McLeod et al. (2002).


with more recent ones, its provisions are relatively        conservation of wetlands. The first obligation
straightforward and general. The official name of           under the Convention is to designate at least one
the treaty reflects its original emphasis on the con-       wetland for inclusion in the List of Wetlands of
servation and wise use of wetlands primarily to             International Importance (the ‘Ramsar List’) and
provide habitat for waterbirds. Over the years,             to promote its conservation. Selection for the
however, the Convention has broadened its scope             Ramsar List should be based on the wetland’s
to cover all aspects of wetland conservation and            significance in terms of ecology, botany, zoology,
wise use, recognising wetlands as ecosystems that           limnology or hydrology. As of March 2004 the
are extremely important for biodiversity conserva-          Convention had 138 Contracting Parties, who
tion and for the well-being of human communities.           have designated 1369 wetlands (amounting to
   The treaty requires Contracting Parties to desig-        nearly 120 million hectares) as Ramsar sites.
nate Ramsar sites, promote the wise use of wet-                The Contracting Parties have adopted specific
lands, establish Nature Reserves, initiate training         criteria for identifying Ramsar sites, the most
and undertake international co-operation in the             recent of which were adopted by the 7th
                                                                                  3.7 Site conservation designations      93



Conference of Parties (CoP7) in 1999. However,                       fishes, spawning ground, nursery and/or migration path
these should be used in conjunction with the                         on which fish stocks, either within the wetland or else-
strategic Framework and Guidelines for the Future                    where, depend.
Development of the List of Wetlands of International
Importance, as also adopted by CoP7. The current
criteria are listed below and can be obtained,                     3.7.7      National Nature Reserves
together with the Strategic Framework document,
                                                                   National Nature Reserves (NNRs) are nationally
from the Ramsar website at www.ramsar.org.
                                                                   important sites that are protected and managed
Group A of the Criteria. Sites containing representative, rare     for wildlife. They were initially established under
or unique wetland types                                            Sections 16–29 of the National Parks and Access to
   Criterion 1: A wetland should be considered internation-        the Countryside Act (1949) to protect the most
   ally important if it contains a representative, rare, or        important areas of wildlife habitat and geological
   unique example of a natural or near-natural wetland type        formations in Britain, and as places for scientific
   found within the appropriate biogeographic region.              research. All are therefore nationally important.
Group B of the Criteria. Sites of international importance for     These provisions were strengthened by the
conserving biological diversity                                    Wildlife & Countryside Act (1981). NNRs are either
   Criteria based on species and ecological communities            owned or managed by the UK Statutory Agencies
   Criterion 2: A wetland should be considered internation-        or by approved bodies such as Wildlife Trusts.
   ally important if it supports vulnerable, endangered, or
   critically endangered species or threatened ecological
   communities.
                                                                   3.7.8      Sites of Special Scientific Interest
   Criterion 3: A wetland should be considered internation-        The first SSSIs were designated under The
   ally important if it supports populations of plant and/or       National Parks and Access to the Countryside Act
   animal species important for maintaining the biological         (1949). Further SSSIs were subsequently desig-
   diversity of a particular biogeographic region.                 nated and SSSI protection measures enhanced
   Criterion 4: A wetland should be considered internation-        under the Wildlife & Countryside Act (1981) and
   ally important if it supports plant and/or animal species       the Wildlife & Countryside Amendment Act
   at a critical stage in their life cycles, or provides refuge    (1985). The protection of SSSIs was further
   during adverse conditions.                                      strengthened by the Nature Conservation
Specific criteria based on waterbirds                              (Scotland) Act 2004 and, in England and Wales, by
   Criterion 5: A wetland should be considered internation-        the Countryside and Rights of Way Act 2000. Both
   ally important if it regularly supports 20,000 or more          Acts amend the 1981 Act’s provisions; they con-
   waterbirds.                                                     tain measures to increase protection of SSSIs and
   Criterion 6: A wetland should be considered internation-        to provide additional powers for the prosecution
   ally important if it regularly supports 1% of the indivi-       of perpetrators of wildlife crime. This includes
   duals in a population of one species or subspecies of           enabling the conservation agencies to refuse con-
   waterbird.                                                      sent for damaging activities; providing new
Specific criteria based on fish                                    powers to combat neglect; increasing penalties
   Criterion 7: A wetland should be considered internation-        for deliberate damage; a new court power to
   ally important if it supports a significant proportion of       order restoration; improving powers to act against
   indigenous fish subspecies, species or families, life-history   cases of third-party damage; and placing a duty on
   stages, species interactions and/or populations that are        public bodies to further the conservation and
   representative of wetland benefits and/or values and            enhancement of SSSIs. The new measures for
   thereby contributes to global biological diversity.             SSSIs came into force on 30 January 2001.
   Criterion 8: A wetland should be considered internation-           SSSIs form the basic unit of UK protected
   ally important if it is an important source of food for         area legislation; most higher designations are
94    3 BIODIVERSITY EVALUATION METHODS



superimposed onto existing SSSIs. As legally
defined, an SSSI is an areas which, in the opinion
                                                         3.7.9        Local Nature Reserves
of the Nature Conservancy Council (NCC) and its          A local authority can declare a site that it owns,
successor bodies, is ‘of special interest by reason of   leases or of which it controls the management as a
any of its flora, fauna, or geological or physiogra-     Local Nature Reserve (LNR) under Section 21 of the
phical interest’. Coastal SSSIs do not extend beyond     National Parks and Access to the Countryside Act
the mean low water mark and therefore do not             1949. Sites are of local interest but not necessarily
cover marine habitats.                                   national interest. Under the Act local authorities
    The criteria for selection have been steadily        have the power to issue bylaws to protect their
refined over the years. The rationale for site selec-    LNRs; although there is no obligation to manage
tion was originally based primarily on habitat           an LNR to any set standard, management agree-
types, recognizing six major habitat ‘formations’:       ments are often put in place.
(1) coastlands, (2) woodlands, (3) lowland grassland,
heath and scrub, (4) open water, (5) peatlands and
(6) upland grassland, moor and mountaintops. In
                                                         3.7.10         Wildlife Sites
each case, the best known examples were selected         In an attempt to give further protection to wildlife,
according to a range of criteria, as described pre-      particularly at a local level, the Wildlife Trusts and
viously. In the 1960s and 1970s the sites were           others have designated a series of some 40 000
graded as of International, National, Regional and       Wildlife Sites5 across Britain over the past 25
County importance, and the first two categories          years (Everitt et al., 2002). The designation of
became known as ‘key sites’.                             Wildlife Sites complements the protection
    In 1989 NCC published a standard set of              afforded by statutory sites (SSSIs), primarily by
Guidelines for selection of biological SSSIs (NCC,       identifying other areas that have substantive wild-
1989). However, the criteria have been reviewed          life interest and therefore merit some form of pro-
at intervals, and refined in the light of new sur-       tection. They may also play an important role in
veys. The approaches and criteria used to select         protecting and enhancing the value of SSSIs, by
SSSIs differ considerably between habitats and           maintaining wildlife corridors (and thereby linking
species groups. Details of the 1989 criteria and         sites) and by providing buffer areas (which may
updates are therefore not discussed here, but are        reduce impacts on SSSIs).
provided in Parts II and III for habitats and spe-          The Wildlife Sites Handbook (The Wildlife Trusts,
cies, respectively. Since 1991 the Joint Nature          1997) states that:
Conservation Committee (JNCC) has been the
focus for the production and revision of the guide-      Wildlife Sites, identified by locally-developed criteria, are the
lines, which can be viewed on the following              most important places for wildlife outside legally protected
JNCC webpage: http://www.jncc.gov.uk/Publications/       land such as SSSIs and ASSIs.
sssi/sssi_content.htm.                                     The DETR Local Sites Review Group defined the
    As described previously, the condition of SSSIs      purpose of Wildlife Sites6 in March 2000 as follows:
in the UK is monitored under a Common Standards
                                                         The series of non-statutory Wildlife Sites seek to ensure, in the
Monitoring (CSM) framework agreed between the
                                                         public interest, the conservation, maintenance and enhance-
UK statutory conservation agencies and the JNCC
                                                         ment of species, habitats, geological and geomorphological
(JNCC, 1998). The key element of CSM is that the
condition of each site is assessed with respect to
                                                         5
site-specific conservation objectives for the inter-       The term ‘Wildlife Sites’ is used here in accordance with
                                                         national recommendations of the Wildlife Trusts and equates
est feature(s) for which the site was notified or, in
                                                         to County Wildlife Sites, Local Wildlife Sites, Sites of County
the case of SPAs, cSACs (Natura 2000 sites) and          Biological Importance, Sites of Nature Conservation
Ramsar sites, the features for which the site was        Importance (SINCs) and other similar terms.
                                                         6
designated.                                                Originally referred to as Local Sites by the Group.
                                                                      3.8 Site evaluations for management planning    95



features of substantive nature conservation value. Wildlife     habitats. This is often by setting a threshold for
Site systems should select all areas of substantive value       richness of listed indicator species (i.e. characteristic
including both the most important and the most distinctive      species of the habitat that are associated with habi-
species, habitats, geological and geomorphological features     tats that are considered to be of high ecological
within a national, regional and local context. Sites within     quality). Some criteria systems allocate scores to
the series may also have an important role in contributing to   the indicator species to reflect the quality of the
the public enjoyment of nature conservation.                    habitat with which they are associated, and/or the
                                                                conservation importance of the indicator species.
  Wildlife Sites are seen as complementary to                   However, it is not always clear how these scores
SSSIs, while differing in two key respects.                     are set. Habitat quality may also be defined by phy-
                                                                sical or chemical properties or features, such as
*   Wildlife Sites are not statutorily designated, but
                                                                water quality, or the presence of riffle and pool
    many receive statutory protection through the
                                                                systems on rivers. The advantage of using such cri-
    planning system.
                                                                teria is that they can be applied where comprehen-
*   The Wildlife Site system aims to select all sites that
                                                                sive species data are unavailable. However, they are
    meet the given selection criteria, not just a sample
                                                                indirect measures of wildlife value and should there-
    of these sites.
                                                                fore be based on good scientific evidence of associa-
It has been recognised by the Wildlife Trusts and               tions between the features and biodiversity value.
DETR (now DEFRA) that a wide range of approaches                   Species-based criteria for Wildlife Sites tend to
and criteria have been used to identify Wildlife Sites.         be simpler, with sites qualifying if they are species-
For example, an assessment of Wildlife Site identifica-         rich, or if they hold viable or significant popula-
tion systems found that although some 68% used                  tions of particular species of conservation impor-
quantifiable written criteria, 7% selected sites on pro-        tance, sometimes referred to as notable species.
fessional judgements alone (Everitt et al., 2002). In           Such notable species typically include species that
most cases Wildlife Site criteria are based on the              are protected under the Wildlife & Countryside
Ratcliffe (1977) criteria, or in some cases adapted             Act, the Wild Birds Directive or the Habitats
SSSI criteria (NCC, 1989). These criteria are typically         Directive, or listed as a UK BAP Priority species, or
used to review habitat types in each county and to              nationally rare, or scarce or rare at a county level.
identify those that are of substantive value and require           Evaluations of particular sites with respect to
conservation. Criteria are then listed which define             county Wildlife Site standards need to be made in
habitat types that qualify as Wildlife Sites. These may         relation to the specific criteria for the county in
be very simple, e.g. ‘semi-improved grasslands which            question. These can normally be readily obtained
retain a significant element of unimproved grassland’.          from the appropriate Wildlife Trust.
However, most criteria provide some form of quantity
threshold for each habitat, sometimes in relation to
                                                                3.8        SITE EVALUATIONS FOR
NVC communities, e.g. ‘neutral grasslands supporting
                                                                           MANAGEMENT PLANNING
good examples of at least 0.2 ha in size (either in a
block or as a number of smaller areas) of one of the            An ecological evaluation is usually an essential
following NVC communities: (i) MG4 (Meadow Foxtail              component of any site management plan, as this
– Great Burnet flood meadow); (ii) MG5 (Crested Dog’s-          identifies, or confirms (if a prior evaluation was
tail – Common Knapweed meadow and pasture)’.                    conducted), the VECs, or features of interest, that
   However, one of the limitations of such criteria is          are present and assesses their overall importance.
that they do not define ‘significant elements’ or               This assessment forms the basis for setting the over-
‘good examples’, and this can lead to inconsisten-              all objectives for the site and conservation objec-
cies in designation and difficulties with recording             tives for each feature, the most important function
justifications for designation. More sophisticated              of the management planning process. As stated in
criteria therefore also define quality thresholds for           the Ramsar Management Planning Guidelines:
96    3 BIODIVERSITY EVALUATION METHODS



It is essential that management objectives be defined      of significance of the impact on these species. The
for each important feature of the ecological charac-       protection afforded to legally designated sites is in
ter of the site and for all other important features       some cases specified with respect to ecological
related to the functions and values of the site,           impacts (e.g. SACs). In these instances, legal and
including socio-economic, cultural and educational         other guidance should be followed to determine
values. In other words, those responsible for devel-       whether a proposal will cause any contravention
oping the management plan must be clear about              of legal status or protection, or have a significant
what they are trying to achieve.                           effect on the integrity of a system, resource or
                                                           feature. However, more often the requirements
   A variety of formats and structures have been
                                                           relating to legally protected sites are not so clear-
developed and adopted for site management plan-
                                                           cut, as the law enables decision-makers to permit
ning purposes, but many of these are similar and
                                                           development on such sites if a good case can be
recommend the Ratcliffe (1977) criteria as a basis
                                                           made or where impacts can be successfully miti-
for evaluations, e.g. Ramsar Management Planning
                                                           gated or compensated. To inform this decision-
Guidelines (Ramsar Bureau, 2002), the RSPB
                                                           making process, it is therefore necessary to demon-
(Hirons et al., 1995) and the Countryside Council
                                                           strate the degree of significance of impacts.
for Wales (CCW, 1996). Further information on
                                                              The second purpose is to identify, document and
ecological evaluations for management planning
                                                           quantify as far as possible all potentially valuable
is provided in these publications.
                                                           ecological components (VECs) that may be affected
                                                           by the development. VECs should include those that
3.9      SITE EVALUATIONS FOR                              may be affected by off-site impacts such as those
         ENVIRONMENTAL IMPACT                              from emissions or effluents, waste material dump-
         ASSESSMENTS (EIAs)                                ing, production of material to be used on site, road
                                                           construction, water supplies and building material.
Biodiversity evaluations for an EIA can be consid-         Thus it is essential that all species of conservation
ered to have two purposes. The first is to establish       concern (see Section 3.5), important habitats (3.6),
whether a proposed development will have any               designated sites (3.7) and other potential VECs are
legal biodiversity protection requirements. This           identified within an appropriate zone of potential
requirement applies to all developments that               impact for the species or habitat in question. For
could affect legally protected sites (e.g. SSSIs), habi-   example, a zone of impact for an area of heathland
tats (e.g. ancient hedgerows), species (i.e. those pro-    that may suffer from atmospheric pollution from a
tected under the Wildlife & Countryside Act), or           proposed road development may be much larger
other biological features (e.g. individual trees), irre-   than, and entirely separate from, the road footprint.
spective of the size, scale or location of the devel-         Potential impacts should first be identified as
opment. It is also worth noting that legal                 part of a scoping exercise, involving consultations
protection does not necessarily reflect biodiversity       with interested parties (e.g. statutory conservation
value (e.g. Badgers Meles meles under the 1992             agencies and local conservation NGOs), interrogat-
Protection of Badgers Act, Foxes Vulpes vulpes and         ing biological records centres, an appraisal of exist-
other mammals under the Wild Mammals                       ing information and a preliminary site visit by
Protection Act, trees protected under Tree                 experienced ecologists to identify the main habi-
Preservation Orders and hedges protected under             tats and species groups that are present. This
the Hedgerow Regulations 1997).                            should then be followed up by full surveys of appro-
   With respect to legal protection of species, the        priate groups to verify and quantify as necessary
EIA must set out what steps will be taken to ensure        the presence of VECs identified in the scoping exer-
that the law is not contravened. In this sense, the        cise. Further information on these aspects of the
results of the assessment must be ‘absolute’: there        EIA process can be found in PPG9, the UK govern-
is no requirement for an assessment of the degree          ment’s planning policy guidance on nature
                                                                   3.9 Environmental Impact Assessments       97



conservation currently being reviewed as Strategic        co-operation with the Department for Transport.
Planning Guidance (www.defra.gov.uk).                     This approach is used to assess which biodiversity
                                                          VECs (termed ‘features’ in GOMMMS) matter, why
                                                          they are important now and how that may change
3.9.1       EIA evaluation criteria
                                                          over time in the absence of the proposal. This pro-
The collected information on each valued ecologi-         vides a base level of environmental capital against
cal component within its potential impact zone is         which the impact of the proposal on that level of
then typically assessed against a range of biodiver-      capital can be appraised. VECs are categorized
sity evaluation criteria, such as the Nature              according to five levels of value: negligible, lower,
Conservation Review (Ratcliffe, 1977) criteria            medium, high and very high. However, interpreta-
described above (see Section 3.6.3). However, a           tion of these categories is difficult as the assessment
wide range of approaches and criteria have been           criteria are not clearly explained. Furthermore,
developed and used, some of which are reviewed            there are insufficient categories to allow for distinc-
by English Nature (1994), DoE (1995b), Treweek            tion of biodiversity components of regional or local
(1999) and Byron (2000).                                  values. This results in rather subjective broad-brush
   One important strategic-level appraisal frame-         assessments that can be acceptable for initial strate-
work in the UK is contained within the New                gic assessments but are inappropriate for EIAs.
Approach to Appraisal (NATA) developed by the                Nevertheless, the approach can be used as a basis
Department for Transport and the Regions (DETR,           for a more detailed and clearly defined assessment
1998). This was initially formulated to provide a         framework, as for example outlined in Table 3.5,
clear and open framework for the appraisal and            where VECs are classified into six categories from
prioritisation of trunk road investment proposals.        Parish/Neighbourhood Importance to International
Subsequently the NATA approach has been further           Importance. As with the application of other frame-
developed into Guidance on the Methodology for Multi-     works, where a VEC can be valued at a number of
Modal Studies (GOMMMS) for use in general transport       levels, the highest value should be used.
planning (DETR, 2000). The GOMMMS approach                   Use of this framework will unavoidably require
continues to be seen by the Department for                some decisions to be made by judgement, for exam-
Transport as the primary source of guidance for           ple on what constitutes a viable area of habitat and
the development and appraisal of strategies and           significant populations etc. These decisions should
plans for surface transport. In 2003 the advice           be made by informed expert judgement, based on
originally set out in GOMMMS and its key                  the best available data and in close consultation
supporting documents was fully incorporated               with statutory conservation bodies and other rele-
into the Transport Analysis Guidance (TAG) website        vant consultees. Further guidance may also be
(www.webtag.org.uk).                                      obtained from The Institute of Ecology and
   The approach has four stages:                          Environmental Management (IEEM) guidelines for
                                                          ecological assessment (www.ieem.org.uk) (currently
*   Stage 1: Describing biodiversity features.
                                                          under revision at time of going to press).
*   Stage 2: Appraisal of environmental capital.
                                                             It is also important to note that a distinction
*   Stage 3: Appraisal of the proposal’s impact.
                                                          needs to be made between significant and impor-
*   Stage 4: Derivation of an overall assessment score:
                                                          tant populations or habitat areas. Significant popu-
    large/moderate/slight beneficial and adverse,
                                                          lations relate to species that are in some way
    neutral.
                                                          threatened or otherwise of conservation impor-
The appraisal stage (2) uses the concept of               tance and that are substantial enough to warrant
defining environmental capital, which has been            conservation. For example, six pairs of Hen Harrier
developed by the statutory environmental bodies           Circus cyaneus would be significant, but six pairs of a
(Countryside Agency, English Nature, English              common and widespread species such as Song
Heritage and the Environment Agency) in                   Thrush would not; both of these species are Red
98    3 BIODIVERSITY EVALUATION METHODS



Table 3.5. A potential framework for defining the ecological value of Valued Ecosystem Components (VECs)

Level of value                       Examples of qualifying VECs

International                        *   Valuable biological features within sites of international importance,
                                         i.e. World Heritage Sites, Biosphere Reserves and Biogenetic Reserves.
                                     *   Designated or qualifying features within a Ramsar site or site of EU
                                         importance, i.e. designated or candidate Natura 2000 site (SAC or SPA),
                                         or features that qualify an area for such designations.
                                     *   Internationally significant and viable areas of a habitat type listed in
                                         Annex I of the Habitats Directive.
                                     *   Regularly occurring globally threatened species (i.e. IUCN Red Listed) or
                                         species listed on Annex I of the Bonn Convention.
                                     *   Internationally important populations of a species (e.g. more than 1% of
                                         a flyway population of birds).
                                     *   Nationally significant populations of an internationally important
                                         species (i.e. listed on Annex II of the Habitats Directive, or Annex I of the
                                         Birds Directive, or with an unfavourable status in Europe).
                                     *   Regularly occurring populations of internationally important species
                                         that are threatened or rare in the UK or of uncertain conservation
                                         status.
National                             *   Designated or qualifying features within nationally designated sites
                                         (SSSIs, ASSIs, NNRs, Marine Nature Reserves), or features that meet the
                                         published selection criteria for national designation.
                                     *   Nationally significant and viable areas of UK BAP Priority Habitats
                                         identified as requiring site protection (see HAPs).
                                     *   Nationally important populations of a species (e.g more than 1% of
                                         national population for birds).
                                     *   Significant populations of nationally important species, i.e. listed on
                                         Schedules 5 and 8 of the 1981 Wildlife & Countryside Act (as amended)
                                         and UK Red Data Book species (excluding scarce species) or, if not a non-
                                         Red Data Book species, listed as occurring in 15 or fewer 10 km squares
                                         in the UK.
                                     *   UK BAP Priority Species requiring protection of all nationally
                                         important sites.
                                     *   Any regularly occurring population of a nationally important species
                                         that is threatened or rare in the region or county.
Regional (i.e. government            *   Regionally significant and viable areas of key habitat identified in a
regions)                                 Regional BAP.
                                     *   Regionally significant and viable areas of key habitat identified as being
                                         of regional value in the appropriate English Nature Natural Area.
                                     *   Regionally important populations of a species.
                                     *   Significant populations of a regionally important species.
                                     *   Regularly occurring, locally significant populations of species listed as
                                         being nationally scarce (i.e. which occur in 16–100 10 km squares in the
                                         UK), or in a Regional BAP or relevant Natural Area on account of their
                                         regional rarity or localisation.
                                                                         3.9 Environmental Impact Assessments         99



    Table 3.5. (cont)

    Level of value                      Examples of qualifying VECs

    County/ Metropolitan               *   Designated or qualifying features within Local Nature Reserves or
                                           Wildlife Sites, selected on county/metropolitan criteria, or features that
                                           meet the published selection criteria for designation.
                                       *   Semi-natural ancient woodland greater than 0.25 ha in area.
                                       *   Significant and viable areas of habitat identified in County BAPs as
                                           requiring site protection.
                                       *   Species populations of county/metropolitan importance.
                                       *   Significant populations of a county/metropolitan important species (i.e.
                                           listed in a County/Metropolitan Red Data Book or BAP on account of
                                           their regional rarity or localisation).
*   District/Borough                   *   Biological features within Local Nature Reserves, etc., selected on
                                           District/Borough ecological criteria.
                                       *   Areas of habitat identified in a sub-County (District/Borough) BAP or in
                                           the relevant Natural Area profile, and other features that are scarce
                                           within the District/Borough or that appreciably enrich the District/
                                           Borough habitat resource.
                                       *   Diverse and/or ecologically valuable hedgerow networks.
                                       *   Semi-natural ancient woodland smaller than 0.25 ha in area.
                                       *   Species populations of District/Borough importance.
                                       *   Significant populations of a District/Borough important species
                                           (i.e. listed in a local BAP on account of their local rarity or localisation).
    Parish/Neighbourhood                Areas of habitat considered to appreciably enrich the habitat resource
                                        within the context of the Parish or Neighbourhood, e.g. species-rich
                                        hedgerows.
                                        Valuable biological features within Local Nature Reserves selected on
                                        Parish ecological criteria.

    Source: Based on draft IEEM guidelines for EIAs, currently under review at the time of going to press.
    Notes: See Section 3.7 for details on designations.
    The viability of an area of habitat is defined by NCC (1989), paragraph 2.10.3, as ‘Given that the intrinsic
    vegetational quality of the habitat is acceptable, its area must be big enough to be viable, in respect of the
    resistance of the habitat and its flora and fauna to edge effects, loss of species and colonisation by unwanted
    species.’
    Unless defined in appropriate publications, significant numbers of species or habitat areas (internationally,
    nationally, etc.) should be agreed in consultation with statutory conservation agencies and other relevant
    consultees.

    Listed in the UK (Gregory et al., 2002). Important         fuligula that is more than 1% of its flyway
    populations are those that are sufficiently abun-          population).
    dant to warrant conservation irrespective of the               Care should also be taken in assessing the import-
    conservation status of the species in question             ance of individual sites for species, particularly with
    (e.g. a wintering population of Tufted Duck Aythya         regard to widespread Red List and BAP listed species.
100    3 BIODIVERSITY EVALUATION METHODS




  Box 3.1 An evaluation of Valued Ecosystem                 mixed farmland and a small brackish lagoon. The
  Components (VECs) potentially affected by a               development site partially overlaps with an SAC, SPA,
  hypothetical development                                  SSSI and a County Wildlife Site. Example VECs are listed
                                                            in the table (according to their highest ecological value),
  This example relates to the proposed development of a     but for simplicity many are omitted, and VECs that may
  marina on an estuary in East Anglia, which may directly   be affected by off-site impacts are not considered. HD,
  affect some 120 ha of mudflats, sand dunes, saltmarsh,    Habitats Directive; WBD, Wild Birds Directive.



  Value
  (see Table 3.5)           Example VECs                                 Rationale

  International             108 ha of estuary habitat                    Designated SAC feature (HD Annex I
                                                                         habitat)
                            80 ha mudflat                                Qualifying SAC feature (HD Annex I
                                                                         habitat)
                            22 ha Atlantic salt meadows                  Qualifying SAC feature (HD Annex I
                            (Glauco-Puccinellietalia maritimae)          habitat)
                            Important wintering population of            Designated SPA feature
                            Wigeon Anas penelope (>2% of NW
                            European flyway population)
                            Petalwort Petalophyllum ralfsii              Qualifying SAC feature (Annex II
                                                                         species)
  National                  1.8 ha coastal lagoon                        SSSI Notified feature (HD Annex I
                                                                         habitat, but not SAC qualifying
                                                                         feature)
                            Nationally important wintering               Notified SSSI feature
                            population of Lapwing Vanellus vanellus
                            Small wintering population of           WBD Annex I species, population not
                            Short-eared Owl Asio fammeus            nationally significant, but species is
                                                                    regionally rare
                            Stinking Goosefoot Chenopodium          Schedule 8 of Wildlife and
                            vulvaria                                Countryside Act 1981 as amended
                                                                    1988, Nationally vulnerable
  Regional                  Otter Lutra lutra: regular reports,          HD Annex II spp, but dispersed
                            population unknown                           species and population not of inter-
                                                                         national or national significance, but
                                                                         spp rare in East England
                            Regular breeding population of               Regionally significant population of
                            Redshank Tringa totanus                      Amber listed species (see Chapter 24)
                            Small-flowered Catchfly Silene gallica       Nationally scarce, UK BAP priority
                                                                         species and County BAP species
  County /                  3 ha of mobile foredune and yellow           Small area of degraded dunes does
  Metropolitan              dune                                         not conform with HD Annex I
                                                                         habitats or SSSI criteria; qualifies as
                                                                         Wildlife Site (but not designated)
                                                                  3.9 Environmental Impact Assessments        101




  Value
  (see Table 3.5)           Example VECs                             Rationale

                            0.8 ha reedbed                           County Wildlife Site feature
                            Breeding Reed Bunting Emberica           UK BAP Priority Species and Red List
                            schoeniceus                              species but dispersed and site protec-
                                                                     tion not proposed in SAP. Listed as
                                                                     target species in County BAP
                            Sharp Rush Juncus acutus                 Nationally scarce, locally rare and
                                                                     diminishing
  Parish                    200 m of ancient hedgerow,               Rare habitat in parish
                            with some mature oaks
                            Farm pond                                Artificial pond, but frog spawning
                                                                     site. Would be an issue if lost and
                                                                     contained Great Crested Newt
                            Breeding Linnet Carduelis cannabina      UK BAP Priority species and Red List
                                                                     species but dispersed and site protec-
                                                                     tion not proposed in SAP.
                                                                     Only breeding colony in parish



It is inappropriate to give all UK BAP Priority Species   regional BAP identifies a requirement to protect all
a ‘national’ level of importance within EIA evalua-       populations of a particular species (where there is no
tions. Instead, evaluation levels should be dependent     similar recommendation in the UK BAP) each popu-
on the ‘Policy and Legislation’ and ‘Site Safeguard’      lation should be considered to be of regional import-
measures that are given in the BAP or corresponding       ance. Similarly, if a sub-national BAP identifies the
Species Action Plan (SAP) for the species in question.    need to protect all populations above a certain size,
Thus, species for which the national SAP recom-           these populations should be considered to be of
mends the safeguard of all sites are afforded national    importance at the relevant scale.
importance. The level of importance that is attached         Some example evaluations of VECs according to
to other species should then reflect the recommen-        the framework and categories outlined in Table 3.5
dations in sub-national BAPs. For example, if a           are presented in Box 3.1 for a hypothetical site.
Part II   *   Habitats
4     *    Introduction to habitat evaluation




4.1       HOW TO USE THE HANDBOOK:                       acceptance as a cost-effective way of assessing
          A RECAP                                        the quality of designated sites. Depending on the
                                                         objectives of the study, the range of methods
Part II of the Handbook is intended as a general-
                                                         described in this chapter will be useful for different
purpose source of detailed, practical information
                                                         situations.
on study design, sampling and analysis as well as
                                                            Chapter 5 provides detailed information on the
on the most commonly used methods for surveying
                                                         most commonly used habitat survey and monitor-
and monitoring terrestrial and freshwater habitats.
                                                         ing methods. Qualitative as well as quantitative
   The development and successful implementation
                                                         methods are included to cover situations in which
of a survey and monitoring programme involves
                                                         resources are limited or sophisticated methods
making a series of crucial decisions. Part II of the
                                                         are unnecessary. The descriptions also aim to pro-
Handbook is therefore designed to provide a step-
                                                         vide sufficient practical information for most of
by-step guide through the process of planning and
                                                         the techniques to be applied in the field. More
executing a survey and monitoring programme.
                                                         specialised techniques are summarised and key
However, the design and implementation of a pro-
                                                         sources of further information listed. Specialised
gramme is not a linear process, but often involves
                                                         monitoring techniques carried out in collaboration
iterative steps that depend on the outcome of other
                                                         with other bodies are also summarised and
decisions. Because monitoring is largely defined
                                                         the reader is referred to appropriate specialist
by a series of surveys, the term ‘survey’ is usually
                                                         documentation.
also implied where the term ‘monitoring’ is used
                                                            Each section on methods, dealt with fully in
throughout this Handbook. The main topics covered
                                                         Chapter 6, starts with a table summarising the
in Part II are listed below.
                                                         information covered in that section. The following
                                                         points are considered:
4.2       HABITAT SURVEY AND MONITORING                  *   the recommended uses of the method;
                                                         *   the efficiency of the method, i.e. the combined
This chapter identifies the attributes of major
                                                             quantity and quality of data produced in relation
habitat types that provide an indication of their
                                                             to cost and effort;
condition. These should be the focus of habitat con-
                                                         *   the objectivity of the method;
dition monitoring programmes. For each habitat a
                                                         *   the precision obtainable;
summary table lists these attributes and provides
                                                         *   the likely nature of any inherent bias;
cross-references to descriptions of the recommen-
                                                         *   the expertise and equipment required;
ded methods for monitoring them (provided in
                                                         *   field methods; and
Chapter 5). Reference should also be made to gen-
                                                         *   data storage and analysis.
eric guidelines on defining Condition Objectives for
statutory sites that are being developed by the UK       Chapter 7 briefly describes monitoring methods
conservation agencies (visit their websites for latest   for management practices (such as grazing and
information). Condition monitoring has gained            burning) and other environmental processes, such

                                                                 # RPS Group plc and Scottish Natural Heritage 2005.
106    4 INTRODUCTION TO HABITAT EVALUATION




                                                                                          Where to find it

                                                                                                    Chapter
                                                                                                    or
                             Topic                                                Part              Section

Planning a                  General monitoring theory                                I                 1.1
monitoring                  Setting objectives for monitoring                        I                 2.1
programme                   Selection of monitoring methods                          I                 2.2
                            Designing a sampling strategy                            I                 2.3
                            Reviewing a monitoring programme                         I                 2.4
                            Data recording and storage                               I                 2.5
                            Data analysis                                            I                 2.6
Habitat                     Attributes of major habitat types                        II                5
monitoring                  Habitat monitoring methods                               II                6
                            Management monitoring methods                            II                7
Species                     General species monitoring theory                        III              10
monitoring                  Species attributes and monitoring methods                III              11–26


as erosion, all of which may influence the features      sampling schemes, the reader should first look
of interest on a site. It is important that monitoring   up the habitat type that is to be monitored in
programmes, where necessary, include monitoring          Chapter 5. The habitat tables in each of the sections
of management practices; although it is beyond the       in that chapter list the appropriate methods for
scope of this Handbook to deal with these in detail,     monitoring each attribute of the habitat. Consult
further sources of information on these subjects         the relevant sections in Chapter 6, and then return
are listed.                                              to Part I to follow the remaining steps for designing
   When using this Handbook as a guide for select-       a monitoring programme and sampling scheme
ing habitat monitoring methods and designing             (if the appropriate method requires it).
5     *    Habitat requirements and issues




The primary purpose of this section is to identify, for    programmes for habitats on designated sites, as
each broad habitat type, the potential attributes that     there may be additional attributes that require
indicate the condition of the habitat and to recom-        monitoring listed in the JNCC guidelines which
mend methods that may be used for monitoring               are not covered in the habitats sections in this
each of these. These recommended methods are               Handbook.
described individually in Chapter 6, or in Part III
(species) for methods that are more often applicable
to surveying and monitoring individual species. The        5.1      WOODLAND AND SCRUB
section also identifies key management actions and
                                                           5.1.1      Survey and monitoring requirements
other environmental factors that may have impacts
                                                                      and methods
on the habitat and may therefore require monitor-
ing. Finally, any specific monitoring issues (e.g. prac-   Attributes for assessing habitat condition
tical implementation, health and safety and the            Woodlands can exist under a great range of envir-
frequency of monitoring) that may influence the            onmental conditions, from tree line to floodplain,
design of a survey or monitoring programme in              on virtually all soil and rock types, from dry rock
the habitat are briefly described. Habitats have           outcrops to permanently waterlogged marshland,
been divided according to UK Biodiversity Action           and from coastal dunes to inland mesic loams.
Plan (BAP) Broad Habitat types. Based on structural        Inevitably, when presenting guidance for the full
similarities the methods can be applied to the full        range of woodland types, there will be a large num-
range of habitat types found in Europe and, indeed,        ber of methodologies, some of which will be
other parts of the world.                                  required in only one woodland type.
   Within the UK, JNCC have recently published                Woodlands also vary greatly in structure.
online guidelines on Common Standards Moni-                Although they are generally defined by the preva-
toring. This provides guidance on setting and asses-       lence of trees, the trees can vary in stature and
sing conservation objectives for the range of              degree of cover. For example, acid oak woodland
species and habitat features which occur on UK             may occur as tree-line scrub, and coastal scrub
protected sites. The process is now well advanced,         woodland may be no taller than a person; on the
with guidance available on conservation objectives         other hand, towering groves of single-stemmed
and assessment methodologies for about 75% of the          trees, spreading pollards from old parkland and
features of designated sites. At the time of writing,      multi-stemmed coppices growing out from past
guidance was available for coastal, lowland grass-         felling are all included in the term ‘woodlands’.
land, lowland heathland, upland and woodland               Furthermore, the aims of treatment vary within
habitat features. Guidance on lowland wetlands             any one structure, so that, for example, a formerly
and freshwater habitats is being developed.                coppiced stand may be re-coppiced, allowed to
   These guidelines are available online at www.           grow naturally, or converted to wood-pasture. The
jncc.gov.uk/csm/guidance/default.htmintroductory           attributes that should be monitored within one
and should be consulted when devising monitoring           particular structure may not be appropriate for

                                                                   # RPS Group plc and Scottish Natural Heritage 2005.
108     5 HABITAT REQUIREMENTS AND ISSUES



another. The implications of what is observed will         *   Subjective observations by an experienced obser-
vary according to management intentions.                       ver, interpreted in terms of trends and factors. For
   Change within woodland varies in kind, spatial              example, the adequacy of regeneration can be
scale and timescale. It can be broadly classified as           assessed in a single, short visit by an observer,
follows:                                                       who can make experienced judgements on the
                                                               opportunities for, and state of, present regenera-
1. Natural expression of the growth, mortality and
                                                               tion, and on its future prospects. The results are
   regeneration of trees and shrubs, which changes
                                                               not normally valid statistically, but assessments
   stand structure and composition and influences
                                                               are immediate, are useful for management plan-
   other components of the woodland. This is
                                                               ning, and help to identify issues for more detailed
   usually measurable over decades: significant trends
                                                               investigation.
   are rarely detectable in less than 10 years.            *   Detailed, precise observation, which enables any
2. Disturbances created by natural forces or by man-
                                                               change to be quantified and validated by statis-
   agers. These may be abrupt events (e.g. fire, blow-
                                                               tical tests. This requires careful design, long-term
   down, felling) or slow changes (e.g. as a result of
                                                               data storage and a capacity to analyse the data. It
   drought, change in deer culling regime), the effects
                                                               thus requires substantial resources and effort, and
   of which will require monitoring over many years.
                                                               must be carried out to a high standard.
3. Responses by ground vegetation (open-space vege-
   tation and field layer under trees) to 1 and 2,         The principal questions for most woodland stand
   combined with the internal dynamics of these            monitoring are (i) is it regenerating? (ii) is it chan-
   assemblages. Response rates, and thus monitor-          ging in composition? and (iii) is it changing in struc-
   ing frequencies, vary according to the character        ture? Of these, regeneration is both the most
   of the changes to which the assemblages are             difficult question to answer and the question most
   responding.                                             often asked. Woods must regenerate to persist, but
4. Responses of fauna to 1–3; internal dynamics of         regeneration is often patchy and episodic. Woods
   animal species and interactions; direct influences      have survived even where regeneration has been
   on fauna, e.g. by pest management or deer hunt-         absent from much of the wood for much of the
   ing. Rapid changes from year to year are charac-        time. The more pertinent issue is whether a wood
   teristic, especially of small animals. Monitoring       is regenerating when it should, i.e. when it would do
   should be an annual process, but the problem is to      so naturally, or should do so under management.
   distinguish between long-term trends and short-             Recognising when adequate regeneration is tak-
   term fluctuations.                                      ing place is also problematical. The presence of
5. Changes in site condition as a consequence of           seedlings and saplings does not guarantee that
   (i) 1–4; (ii) physical processes acting directly, and   there will be recruitment to the canopy. The pre-
   thus factors in 1–4; (iii) long-term maturation,        sence of a thicket of saplings may be no more
   e.g. of soils; and (iv) pollution uptake.               effective in the long term than a sparse scatter:
   Recently proposed monitoring frameworks for             there is only so much room for canopy trees. On
woodland habitats (Kirby, 1994; Kupiec, 1997)              average, one large oak needs to generate only one
accept that anything more than a quick site inspec-        other large oak in 300–500 years to sustain the
tion by an experienced observer is time-consuming          species and the wood. Even in natural woods, the
and thus expensive. In practice, monitoring in             regeneration of a particular species is likely to be
woodlands is a choice of strategies:                       extremely irregular in space and time.
                                                               Against this background the critical question is:
*   Recording events (i.e. abrupt changes in the status    are enough individual saplings of the right mix of
    quo) when they occur, with their immediate con-        species developing beyond the reach of grazing and
    sequences. The record forms a basis for assessing      browsing when and where circumstances are right
    change at a later date.                                for regeneration? ‘Right’ for canopy trees would be
                                                                               5.1 Woodland and scrub     109



defined by availability of canopy space, plus an        the lateral spread of fertilisers from nearby fields
‘extension’ to cover advance regeneration of            through drift and leaching.
shade-bearing species. ‘Right’ for structure would         Measurement of natural changes would be
be defined by cover and stratification that normally    extremely laborious, and would certainly not be
allows enough light through to permit an under-         cost-effective. Measurement of most soil changes
storey to develop.                                      due to forestry operations is best done directly, e.g.
   Regeneration relates to the continued existence      flow in drainage ditches, extent of rutting and dis-
and future character of the woodland. Other attri-      tribution of surface disturbance during forestry
butes relate to its condition.                          operations. Measurement of change in nutrient
   A description of attributes indicating the condi-    status due to pollution inputs or forestry opera-
tion of woodland and scrub habitats is presented        tions can be achieved by:
below and summarised in Table 5.1.
                                                        1. repeated samples from the same points (i.e. the
                                                           equivalent of permanent plots); and
Shape and size
                                                        2. analysis of chemicals in run-off.
Shape and size is a particular issue for upland
woodland types, floodplain woodland and bog             Of these, the latter is expensive but offers the
woodland. It is rarely an issue in lowland woods,       possibility of constructing nutrient budgets. The
where most boundaries are sharply defined and           former should be far less expensive, especially if
static, but may be significant where the wood           the initial observations are repeated only when
abuts semi-natural non-woodland habitats.               there is a need to know.
Defining what counts as woodland is problematical          Observations of change due to forest opera-
in wood-pasture and regenerating woodland.              tions should be made before and after operations.
   Monitoring requires delimitation of the wood-        Baseline soil nutrient observations should be
land boundary. This is generally straightforward in     repeated on a need-to-know basis, but there is a
high-definition lowland farmland, but often diffi-      case for repeating them after 25 years, even if no
cult in upland pastoral environments. Delimiting        particular need has arisen in the meantime.
woodland requires a decision on the minimum size
of feature to be heeded (a fractal issue). Even on      Hydrology
rapidly changing boundaries, repeating observa-         This is particularly important for floodplain wood-
tions at intervals of 10 years would be sufficient      land and bog woodland, but it can also be an issue
from the point of view of the woodland.                 in any wood where streams, flushing zones and
                                                        water tables influence condition. Monitoring
Soil                                                    needs differ greatly between woodland types.
Forest soils change naturally, but profile develop-        In floodplain woodland, channel configuration
ment is usually very slow. Erosion is rarely severe,    can be tracked by a combination of aerial photo-
and soil creep on slopes is very slow. Rapid change     graphs and on-site observation of areas of erosion
occurs very locally when trees fall over. A change in   and deposition. Location of bank erosion and shoal
the dominant trees and shrubs may also change           deposition can be noted annually on site maps. The
humus characteristics. Forest operations can also       flooding regime is best followed from flow records
change soils. When drainage ditches are dug, pro-       of river managers, combined with on-site observa-
file character and microtopography are altered in       tion of main floods. The assessment of water qual-
the immediate vicinity. Felling and extraction can      ity would also normally depend on information
alter natural microtopography and mobilise soil         from river managers (see Section 5.10). In bog
nutrients.                                              woodland, the need is to assess water table fluctua-
   Pollution inputs are likely to be the major con-     tion and water quality.
cern at the present time. Throughout Europe there          In most woods there is a need to monitor any
is concern about acid rain, nitrate deposition and      streams and springs that may be present. In some
110    5 HABITAT REQUIREMENTS AND ISSUES



Table 5.1. A summary of the quality attributes providing an indication of the condition of woodland habitats,
and their recommended monitoring techniques

Attribute           Habitat properties                                 Monitoring technique

Size and shape      Area of individual wood and                        Aerial photographs (Section 6.1.3)
                    configuration of boundaries                        Phase I mapping (6.1.5)
                                                                       Comparison with historical maps (not
                                                                       described)
Soil                Structure                                          Soil cores (6.2.2)
                    Nutrient status                                    Chemical analysis (not covered) or run-off
                                                                       chemistry (not covered)
Hydrology           Watercourse configuration                          Mapping
                    Flooding regime                                    River flow data/visual inspection
                    Water chemistry                                    Chemical analysis (not covered)
                    Water table fluctuations                           Dipwells (6.2.1)
Composition         Stand pattern in managed forests:                  Stock mapping (6.5.1): covers all aspects
                    * extent of old, mid- and young growth             of stand pattern monitoring
                    * extent and configuration of

                      felling patches and canopy gaps
                    * rotation of managed stands

                    * stock of particular size classes

                    * thinning extent and degree

                    Communities                                        Quadrats (6.4.2) or transects (6.4.6), with
                                                                       NVC analysis (6.1.6) where NVC
                                                                       communities are Notified Features
                                                                       or important attributes
                    Species composition, richness and                  Species lists
                    diversity                                          Temporary plots (6.5.3)
                                                                       Plotless sampling (6.5.4)
Structure           Age class diversity                                Enumeration by permanent plots (6.5.2),
                    Horizontal and vertical structural                 temporary plots (6.5.3), plotless sampling
                    diversity                                          (6.5.4), mapping individual trees (Part III,
                    Retentions to natural death                        Section 15.2.4)
                    Thinning extent and degree
                    Deadwood: standing and fallen                      Enumeration (see above), measurement
                    * volume                                           of fallen wood and decay condition (6.5.5)
                    * size distribution

                    * spatial pattern

Dynamics            Open spaces                                        Stock maps (6.5.1)
                    * Extent and location                              Fixed-point photography (6.1.4)
                                                                       Aerial photography (6.1.3)
                                                                               5.1 Woodland and scrub     111



Table 5.1. (cont.)

Attribute            Habitat properties                          Monitoring technique

                     Seedling regeneration; composition,         Enumeration (see above)
                     number and distribution                     Ground vegetation condition (6.4)
                     Renewal of coppice stools
                     Amount and distribution of planting
                     Provenance of planted stock and natural
                     regeneration




forms of wet woodland (e.g. alder woods) the water      method that recent recruitment can be assessed
table is another important factor. One can also         and prospective long-term changes in stand com-
envisage the degree of flushing in slope woods          position picked up. Here, too, past irregularities in
becoming an issue, but it rarely if ever arises.        recruitment can be identified, subject to the limita-
    An annual check of springs and watercourses         tion that only the survivors provide information.
should be sufficient to alert managers to significant   Enumerations also provide an opportunity to
changes, supplemented by checks after extremely         record the condition of individual trees, e.g.
wet periods (or major floods) and after severe          crown vigour, degree of damage by squirrels, etc.
droughts. More detailed hydrological monitoring            The pattern of sampling determines what informa-
is, however, more difficult (see Section 6.2.1).        tion can be gathered on distributions of species
                                                        and size classes. A regular grid of plots can give
Stand extent and structure                              distribution information. Plotless sampling
This is the core of forest monitoring, and in forests   (Section 6.5.4) does not, but it can be more refined
managed for timber production this links standard       if samples are confined to units recognised on
forestry monitoring with environmental monitor-         stock maps. Transects (Section 6.4.6) give informa-
ing. In such cases, two classes of observations are     tion on small-scale patterns, such as groups and
normally required:                                      zonation. Plots and transects can be used to map
                                                        canopy gaps; aerial photographs can also be used.
1. stock maps (Section 6.5.1), which show the patch-
                                                           Rare features, e.g. pollard trees or large coppice
   work of stands; and
                                                        stools, are best mapped individually.
2. enumerations of trees and shrubs (Sections 6.4
                                                           Photographs record a great deal of detail in a
   and 6.5), which show composition and stand
                                                        non-quantitative and unedited fashion. They are
   structure.
                                                        best for giving a general impression of change,
Stock maps partition the stands according to age        demonstrating change to others and monitoring
of canopy and main species, show felling coupes,        features that were not initially thought to be
and give information on patch size and configura-       important. Fixed-point photographs (Section 6.1.4)
tion. They are simplifications that convey whole-       can be a cheap but effective method of revealing
site structure and work well in woods managed on        the main features of long-term change, provided
a moderate- to large-scale felling pattern and with     that enough care is taken to select a representative
coupe shapes. They show little or nothing of verti-     set of points and to record location and conditions
cal stand structure, or of fine detail.                 accurately.
   Enumerations of individual trees and shrubs             Changes can be identified by repeating an obser-
give information on size-class distributions, stand     vation. Permanent plots and transects have to be
composition and how these are related. It is by this    marked and re-found. In return for the extra effort
112   5 HABITAT REQUIREMENTS AND ISSUES



involved, they can provide detailed insights into       also intervene, and you cannot assume that the
stand dynamics, population changes and the fate         plan will be carried out to the letter. An annual
of individual trees. Repeated enumerations in tem-      record of management operations that affect the
porary plots give information on net change,            stock map (and major natural events) should be
but the fate of individuals is not known; this          made, i.e. the stock map should be updated
restricts the analysis of factors underlying change.    annually, and old stock maps should be retained
However, it is not essential to have permanent          as part of the record.
plots unless it is necessary to track the progress of      Change in stands is generally slow, but rapid
individual trees. A well-structured temporary sam-      change may occur unpredictably. In undisturbed
ple will generally be sufficient to monitor the pro-    stands general records need not be repeated at inter-
gress of a population.                                  vals of less than 10 years. When stand-changing
   In a managed semi-natural woodland it may be         events occur (e.g. windthrow, drought, disease), a
sufficient to generate a stock map. Rare and special    record should be made immediately after the event.
features can be mapped on to a stock map base.          If such an event takes place, it would be advanta-
Age-class distribution can be expressed in terms        geous to have observations shortly before the event,
of areas. If particular species are dominant in         so routine observations should not be too far apart.
parts of a wood, age class can also be expressed in     The optimal interval would probably be routine gen-
terms of species. Patch size and shape can be mea-      eral recording every 10 years, or thereabouts, com-
sured from maps. This approach is appropriate in        bined with small subsampling at 5 year intervals and
woods in which:                                         recording immediately after a major disturbance.
1. felling is done in well-defined patches; and
                                                        Regeneration
2. operations within a patch are relatively simple,
                                                        Seedlings and saplings are part of both the stand
   such as clear-cut felling.
                                                        and the ground vegetation. They are considered
In semi-natural woods managed on an irregular           separately because regeneration is a key indicator
basis, e.g. group felling or irregular thinning with-   of the state of a wood.
out clear felling, enumerations should form the            Estimates of seedling density and distribution
basis of monitoring. Likewise, in non-intervention      rarely provide useful information on the progress
woodland, it will be necessary to enumerate and/or      of regeneration. However, if regeneration of a par-
record permanent or temporary plots. Canopy gaps        ticular species is failing, observation of seedlings
should be defined and delimited.                        may indicate whether the failure is in seed produc-
   Monitoring dead wood is important, especially        tion or post-germination survival. In woods in
because of its biodiversity value to fungi, inverte-    which regeneration is not taking place, a seedling
brates, birds, etc. Dead wood can be partly recorded    survey will indicate the potential. However, obser-
in ordinary enumerations, i.e. snags and stumps.        vations over several years will be necessary to allow
Other dead wood elements (e.g. fallen wood, dead        for the ‘mast year’ phenomenon (years in which
branches on living trees, decay columns) have to be     trees produce an unusually large volume of seed).
separately recorded. The transect method (Section          Estimates of the distribution and density of
6.5.5) permits this for patches of 1 ha or more. A      small saplings (less than 1.3 m in height) are far
quicker alternative, based on indices for each dead     more meaningful. Saplings are established indivi-
wood element, is promising, but has not been ade-       duals that have the capacity to grow into trees, but
quately tested (Peterken, 1996). One can also map       almost all fail to become trees because they are
the categorical quantities of dead wood on a            killed by competition, browsing, breakage,
defined grid for the site.                              drought, etc. It is quite possible for a substantial
   Change in features recorded on stock maps            population of saplings to be permanently present,
(Section 6.5.1) is determined mainly by the plan of     yet for no regeneration to be occurring, i.e. there is
management, but catastrophic natural events may         no recruitment to the canopy.
                                                                                    5.1 Woodland and scrub      113



    Large saplings (taller than 1.3 m) should be moni-    grow, spread, decay and die. Variations in soil
tored as part of stand enumeration.                       moisture content due to periods of heavy rain or
    Permanent plots in which individual saplings          drought generate perpetual adjustments in the bal-
are mapped allow the population dynamics of               ance between species. Superimposed on these
regeneration to be understood. It is possible to          changes are responses to changes in the structure
follow the fate of individuals (merely counting the       and composition of the stand, both natural and due
number of seedlings present in a plot may conceal         to management.
the fact that none of them lives longer than three           The ground vegetation component of National
years). The ability to find the remains of saplings       Vegetation Classification (NVC) types may change
that were recorded on a previous occasion com-            (e.g. as a result of fire or grazing), but site character-
monly allows population turnover to be deter-             istics are the main control. If a different tree species
mined and the cause of death to be identified.            colonises the wood, there could be a change of NVC
This is only possible if monitoring occurs regularly.     type without any change in the species composition
If more than 5 years elapse between samples, the          of the ground flora. However, a map of NVC types
chances are that the remains of saplings will not         has value in monitoring as a basis for stratifying
be visible.                                               sample points. Exact delineation of boundaries
    Temporary plots and plotless sampling allow the       between types is not necessary to design the pattern
density, composition and distribution of seedlings        of sampling, and will rarely be sufficiently exact to
and saplings to be quantified, and to demonstrate         detect slight shifts in boundaries.
changes from previous observations, but the inter-           Change in ground vegetation diversity happens
pretation of these observations may be equivocal.         at various scales. Change is inevitable at the scale of
In general, if there is a need to observe small sap-      1 m2, and species turnover can be high in 10 m  10 m
lings (i.e. those missed by ordinary stand enumera-       plots (more than 40% of species over 15–20 years).
tions), it should be done by a method that allows         Change at larger scales is less, and change at the
the processes to be understood, i.e. by recording at      whole-site scale (i.e. colonisation and extinction) is
the level of individuals in permanent plots.              rare. Any assessment of ground vegetation change
    If regeneration takes the form of regrowth from       must take account of scale.
stumps, the simplest and most useful measure of              Whole-site monitoring requires a species list for
any individual would be the height above ground of        the site. This will include those localised species not
the tallest sprout. In many cases, such regrowth          detected in the aggregate of sample plots. The total
exceeds 1.3 m in the first or second season, and          number of species in all sample plots does, however,
would thus be recorded as part of ordinary stand          provide an alternative measure of whole-site species
enumeration. In woods in which regeneration               richness. Sample plots provide an opportunity to
takes the form of planting, one assumes that the          measure small-scale change and vegetation height,
manager will keep a record of what is planted and         to map distributions, and to take fixed-point photo-
where. If saplings are protected by tubes or sticks, it   graphs. The capacity to map changes may be import-
is simple to record their survival and growth.            ant if there is reason to suspect patchy change, e.g.
    If one is observing naturally regenerating indivi-    due to lateral fertiliser drift.
duals in permanent plots, observations should be             Sample plot recording of ground flora is nor-
at least annual. The optimum is probably two              mally worth repeating at 10 year intervals, so that
observations each year, at the start of the growing       a reasonably recent record is available for compari-
season and at its end.                                    sons when a substantial change is suspected, e.g.
                                                          after change in grazing pressure.
Ground vegetation
Ground vegetation changes continually. In addi-           Open spaces
tion to the seasonal cycle, the small-scale pattern       These are the ‘permanent’ open spaces, i.e. mainly
of species changes from year to year as individuals       glades in upland woods and rides in lowland
114     5 HABITAT REQUIREMENTS AND ISSUES



woods. For some species groups, the biodiversity of    there is a possibility that the most recent record of
open spaces may exceed that of tree-covered            condition before a major blowdown (which is
ground. Most take the form of grassland and tall       unpredictable) will be 19 years before the event.
herb communities (Section 5.4). A map showing the      Accordingly, the best tactic is to repeat records at
exact configuration of ‘permanent’ open spaces is      suitable intervals for each attribute, or after major
required as a baseline, which is part of – or a sup-   events. Subsamples can be recorded between
plement to – the stock map. The degree of shading      main recordings, or to determine whether a full
of rides and boundary characteristics should also      re-recording is worthwhile.
be recorded.                                              Recording of ground vegetation should be
   Newly created permanent open spaces should be       undertaken in the growing season. Strong seasonal
recorded as part of the annual updating of a stock     changes require that plots be recorded during a
map. Boundary conditions may be best recorded by       particular season. Lists should be compiled at two
fixed-point photography at 5 year intervals.           points in the season, to cover spring and summer
                                                       aspects. Enumeration of stand condition is best
Management requirements and external impacts           avoided during the peak of the growing season
Management and external impacts that require           (May to July). Winter recording is best for stand
monitoring include forestry practices and natural      structure. Thus, there can be a degree of comple-
disturbances, grazing and browsing intensity, fire,    mentarity in the annual monitoring programme.
pollution (particularly atmospheric inputs) and           Compartmentalisation of monitoring may be
public access and disturbance.                         desirable in larger and heterogeneous woods.
   Forestry practices are typically monitored by       Different monitoring programmes may be neces-
stock maps. Stock maps can also record large-scale     sary in areas undergoing different treatments (i.e.
natural disturbances; small-scale impacts can be       in what foresters used to call working circles).
detected by routine enumerations. See Chapter 7           Care should be taken to ensure that the monitor-
for more details on monitoring other impacts.          ing programme does not itself affect the condition
                                                       of the wood. Recording of permanent plots (includ-
                                                       ing transects) can damage vegetation and alter the
5.1.2      Specific issues affecting the
                                                       browsing patterns of deer. Marker posts can influ-
           monitoring of the habitat
                                                       ence both deer and forestry contractors.
Woodlands generally change slowly, but major
rapid change can happen as a result of natural
events, changes in grazing/browsing regimes, and       5.2     LOWLAND WOOD-PASTURES AND
forestry operations. Annual recording is very rarely           PARKLAND
justified, except at the level of an annual inspec-    5.2.1     Survey and monitoring requirements
tion. Rather, monitoring is most efficiently under-              and methods
taken by:
                                                       Attributes for assessing habitat condition
1. establishing baselines; and
                                                       Some wood-pastures require separate treatment
2. recording events.
                                                       from woodlands. Those that do are defined as con-
Baselines describe the condition at a particular       sisting of an open scatter of trees in a matrix of
date. The record can be repeated when there is a       pasture (i.e. parkland), and also include very open
need to assess change. Rigid adherence to a prede-     parts of mature woodlands (pine (Pinus spp.) woods,
termined recording interval is rarely necessary.       oak (Quercus spp.) woods and birch (Betula spp.)
However, if the interval between recordings is         woods). Although many woodlands sustain such a
large, there is an increased risk that no recent       high pressure of grazing and browsing that they
record will be available before a major change.        may eventually become wood-pastures, they should
For example, with a recording interval of 20 years,    be regarded as woodland for present purposes.
                                                                         5.2 Wood-pastures and parkland       115



   The trees in parklands are generally large and/or      saplings. Monitoring regeneration can thus be a
old. They are isolated from each other or distrib-        matter of recording the survival of marked samples
uted in small groups. Either way, the trees should be     of individuals.
regarded as individuals and recorded as such. The            In open woodlands, which have effectively been
ground vegetation can be regarded as grassland or         pastures with trees for at least decades, the aim
heathland (see Sections 5.4 and 5.6, respectively).       may be to regenerate woodland by removal of, or
A description of attributes indicating the condition      reduction in, grazing animals. The initial stages of
of wood-pasture and parkland habitats is presented        this process can be successfully monitored by fixed
below and summarised in Table 5.2.                        quadrats or transects (Section 6.5.2).
                                                             An annual inspection is desirable, preferably at
Shape and size                                            the end of the growing season, or before animals
The extent of parkland is extremely difficult to mea-     are admitted to pasture. Height and growth of
sure, principally because its boundaries are vague.       important individuals may be worth recording
Boundaries may be roughly delimited on a map, but         annually but, as growth during each of the past 5
a precise delimitation requires decisions on mini-        years or so can generally be assessed by inspection
mum density of trees and minimum mappable                 of individuals, any more general recording is only
patch size. Change is better assessed in terms of         worthwhile at this interval. Likewise, an annual
individual trees. Patches of closed woodland within       inspection is desirable during the early stages of
wood-pastures should be monitored as woodland.            restoration of grazed woodland.

Stands                                                    Management requirements and external impacts
The special feature of monitoring parkland is the         Management and external impacts that require
treatment of the stand as a population of individual      monitoring are similar to those for woodland and
trees. Change can be quantified in terms of num-          scrub and include forestry practices and natural
bers and distribution of trees of a particular class.     disturbances, grazing and browsing intensity, fire,
Each tree can be mapped and numbered, aided by a          pollution (particularly atmospheric inputs) and
recent vertical aerial photograph (Section 6.1.3).        public access and disturbance. Disturbance caused
    A baseline record of individual trees is required.    by public access and grazing are often of particular
This can be supplemented by photographs from              concern in wood-pastures.
recorded points of a sub-sample of trees (Section            Forestry practices are typically monitored by
6.1.4). The record will be a mixture of precise quan-     stock maps. Stock maps can also record large-
tities (e.g. girth) and classes (e.g. crown condition).   scale natural disturbances; small-scale impacts
    Dead wood is commonly an important compon-            can be detected by routine enumerations. See
ent. Most takes the form of decay columns and             Chapter 7 for more details on other impacts.
dead branches on living trees. A supplementary
record of fallen dead wood and stumps is desirable,
                                                          5.2.2      Specific issues affecting the
based on the transect method (Section 6.5.5).
                                                                     monitoring of the habitat
    Once a baseline is established there is rarely any
justification for repeating monitoring assessments        The critical features of parklands are generally:
at less than 10 year intervals. A sub-sample of trees
                                                          1. the mortality rate of existing trees, particularly
can be checked between main recordings.
                                                             the largest and oldest;
                                                          2. the amount and growth of recruitment;
Regeneration
                                                          3. the amount and condition of dead wood; and
Regeneration is commonly a critical issue in park-
                                                          4. the condition of ground vegetation.
lands. Unlike in woodland, however, regeneration
is generally achieved by planting with protection         The peculiar character of parklands emphasises
from grazing, or by protecting naturally set              the need for monitoring at the level of individual
116    5 HABITAT REQUIREMENTS AND ISSUES



Table 5.2. A summary of the quality attributes providing an indication of the condition of wood-pasture
habitats, and their recommended monitoring techniques

Attribute         Habitat properties                            Monitoring technique

Size and shape    Area of individual wood and                   Aerial photographs (Section 6.1.3)
                  configuration of boundaries
                                                                Phase I mapping (6.1.5)
                                                                Comparison with historical maps (not described)
Soil              Structure                                     Soil cores (6.2.2)
                  Nutrient status                               Chemical analysis (not covered) or run-off
                                                                chemistry (not covered)
Hydrology         Watercourse configuration                     Mapping
                  Flooding regime                               River flow data/visual inspection
                  Water chemistry                               Chemical analysis (not covered)
                  Water table fluctuations                      Dipwells (6.2.1)
Composition       Stand pattern in managed forests:             Stock mapping (6.5.1): covers all aspects of stand
                  * extent of old, mid- and young growth        pattern monitoring
                  * extent and configuration of

                    felling patches and canopy gaps
                  * rotation of managed stands

                  * stock of particular size classes

                  * thinning extent and degree

                  Communities                                   Quadrats (6.4.2) or transects (6.4.6), with NVC
                                                                analysis (6.1.6) where NVC communities are
                                                                Notified Features or important attributes
                  Species composition, richness and             Species lists
                  diversity                                     Temporary plots (6.5.3)
                                                                Plotless sampling (6.5.4)
Structure         Age class diversity                           Enumeration by permanent plots (6.5.2), tem-
                  Horizontal and vertical structural            porary plots (6.5.3), plotless sampling (6.5.4),
                  diversity                                     mapping individual trees (Part III, Section 15.2.4)
                  Retentions to natural death
                  Thinning extent and degree
                  Deadwood: standing and fallen                 Enumeration (see above); measurement of fallen
                  * volume                                      wood and decay condition (6.5.5)
                  * size distribution

                  * spatial pattern

Dynamics          Open spaces                                   Stock maps (6.5.1)
                  Extent and location                           Fixed-point photography (6.1.4)
                                                                Aerial photography (6.1.3)
                  Seedling regeneration; composition,           Enumeration (see above)
                  number and distribution                       Ground vegetation condition (6.4)
                  Renewal of coppice stools
                  Amount and distribution of planting
                  Provenance of planted stock and natural
                  regeneration
                                                                           5.3 Farmland boundary features    117



trees. It also offers excellent opportunities for retro-   overhanging vegetation and high banks. This will
spective monitoring, by using old maps, ground             inhibit aquatic vegetation but may provide good
photographs and aerial photographs.                        cover for animals. In contrast, wide drains with
                                                           sunlit water will favour the growth of vegetation.
                                                           Overall, therefore, it is generally advantageous
5.3      FARMLAND BOUNDARY FEATURES                        to aim for a variety of structures along stretches
5.3.1      Survey and monitoring requirements              of ditch. This can be achieved by rotational
           and methods                                     management.
                                                              In general, the condition of ditches is dependent
Attributes for assessing habitat condition                 on vegetation structure and composition, which is
Farmland and field boundaries commonly consist             in turn dependent on water quality and quantity
of fences, ditches, grassy banks, walls or hedges, or      (including seasonality), ditch structure and man-
various combinations of these. Fences by them-             agement. These attributes and influencing factors
selves hold little of interest for biodiversity.           should therefore be monitored. Vegetation can be
However, the other features may be of considerable         simply monitored by transects, with overall condi-
conservation importance and may be the prime               tion simply related to the number of submerged,
source of biodiversity within artificial farmland          floating and emergent wet bank species per 20 m
landscapes.                                                (see NCC, 1989). For further information on moni-
   Ditches may hold important aquatic, floating            toring ditch vegetation see Alcock & Palmer (1985).
and emergent plants, invertebrates and amphi-              Other attributes will normally be dependent on
bians, particularly where they occur alongside             specific site conditions.
semi-natural farmland habitats and are unpolluted             Grassland banks and strips adjoining fields are
by fertilisers, manure or pesticides. They may             also important sources of biodiversity in the farm-
also provide important feeding and breeding habi-          land landscape. Under favourable conditions they
tats for birds. Attributes used in assessing the con-      may have relatively rich plant assemblages and
dition of such habitats are water quality (in              hold scarce or rare farmland species. They may
particular, avoidance of eutrophication), water            also provide suitable habitats for invertebrates
quantity, ditch structure, vegetation structure and        (especially overwintering insects), small mammals
composition (i.e. avoidance of overgrazing or exces-       and nesting birds. However, their quality is highly
sive cutting and dredging). Water quality attributes       dependent on their plant species composition and
are covered in detail in Sections 5.9 and 5.10.            the soil nutrient status. Sown grasslands, which are
   Where possible, water should be maintained in           dominated by species-poor seed mixes and/or are
the ditch all year. If ditches regularly dry out, they     subject to fertiliser applications, are generally of
have little value for aquatic plants and animals,          low conservation interest. Further details on the
although the ditch bed may contain some wetland            attributes of grassland habitats and their monitor-
plants. Deep ditches can hold permanent pools of           ing methods are provided in Section 5.4.
water without interfering with field drainage. As a           At first sight walls may appear to be devoid
minimum, ditches or pools should hold 30 cm                of biological interest. However, old walls may har-
of water (ideally 1 m) in stretches at least 3 m long      bour rich lichen assemblages and potentially some
(Andrews & Rebane, 1994). The structure of the             rare or scarce species. They may also provide suit-
ditch is also important for other reasons. In parti-       able habitats for some plants (especially mosses,
cular, shallow margins provide favourable condi-           liverworts and ferns), invertebrates, amphibians,
tions for wetland plants, invertebrates and                reptiles, small mammals and birds. In general their
amphibians. Poached margins provide particularly           quality is primarily dependent upon their age (espe-
good conditions for many invertebrates (P. Kirby,          cially in the case of lichens, as these are extremely
1992), as well as feeding areas for birds. In narrow,      slow-growing) and the presence of micro-habitats
deep ditches, the bottom is often shaded by                (such as crevices). Particular species inhabiting
118    5 HABITAT REQUIREMENTS AND ISSUES



walls may also be dependent on specific condi-            species richness may be affected by structure
tions, such as a damp or shady environment with           (Pollard et al., 1974; P. Kirby,1992; Green et al.,
a particular aspect.                                      1994, Parish et al., 1994, 1995; MacDonald &
   Hedgerows are well known for their conserva-           Johnson, 1995; Lewis et al., 1999). Structure is pro-
tion importance in a wide range of farmland habi-         foundly affected by management, and probably
tats. Although they seldom hold rare species or           also by age. The complexity of a laid hedge provides
plant communities, they may hold relatively high          better habitats for invertebrates and birds than
diversities of plants and provide important sources       does a simple line of bushes managed by cutting
of food and cover for a wide variety of animals,          or coppicing. Hedge-laying also maintains more
especially insects and birds. Over 600 species of         dead wood, which is particularly important for
plant (including some endemic species such as             invertebrates.
the Whitebeam Sorbus devoniensis), 1500 insect spe-          The external form of the hedge is an important
cies, 65 bird species and 20 species of mammal            attribute of hedgerow quality: trimming reduces
have been recorded at some time living or breeding        structural diversity and flowering and fruit produc-
in hedgerows (Anon., 1995). They may also provide         tion. Severe trimming will also reduce a hedge’s
an important function in linking habitats in open         suitability for nesting birds (Lack, 1987). A hedge
farmland landscapes, thereby providing dispersal          that is trimmed into an ‘A’ shape with a wide base
routes for species that cannot cross large open           may also shade out ground vegetation and become
spaces (see, for example, Simberloff & Cox, 1987;         less suitable for associated animals. In contrast,
Bennett, 1990).                                           a hedge that has become leggy, perhaps as a result
   Attributes providing an indication of the over-        of browsing on lower growth by livestock, is poor
all quality of a hedgerow are species composi-            for scrub-nesting birds but is more likely to have a
tion, structure, whether it contains trees, and           well-developed and richer herbaceous ground
its relationship to other habitats (Andrews &             flora. However, the actual shape of a trimmed
Rebane,1994). Older hedges in England tend to             hedge may be less important for wildlife than is
have a greater plant species richness and associated      often claimed (Hill et al., 1995). The development of
structural diversity (Pollard et al., 1974), as well as   ground flora and associated animal communities is
more mature trees and dead wood. These attri-             much more likely to be influenced by accidental
butes, in turn, support rich communities of inver-        fertiliser applications and pesticide drift.
tebrates and other animals. In contrast, recently            Numerous methods have been used to describe
planted hedges tend to have low plant species             hedges, but they may not be the most applicable for
diversity, often being dominated by Hawthorn              monitoring. A summary of attributes giving an indi-
Crataegus monogyna or Blackthorn Prunus spinosa.          cation of habitat condition in hedgerows and recom-
Proximity to woodland has also been shown to be           mended methods for monitoring them is provided
important for species diversity (Wilmott, 1980).          below. This takes into account recommendations
Another important aspect of species composition           for a standard method for local Biodiversity Action
is the presence of native species. Low numbers of         Plan (BAP) surveys of hedges, which has been pro-
non-native plants will not be a problem (unless           posed by the UK Steering Group for the Species-rich
they are particularly invasive), but hedges that are      Hedgerow Biodiversity Action Plan.
dominated by native species will hold more insects           For monitoring hedgerow extent it is necessary
(Kennedy & Southwood, 1984). Old hedges are also          to adopt a consistent definition that can be readily
more likely to be linked to non-farmed habitats and       applied during field surveys. In this respect the
may be extremely valuable if they are, for example,       proposed UK Steering Group definition of a hedge-
connected to ancient woodlands or other habitats          row is ‘any boundary line of trees or shrubs less
of high conservation importance.                          than 5 metres wide, provided that at one time the
   In general, tall, dense and broad hedges are rich-     trees or shrubs were more or less continuous’. This
est in biodiversity. Plant, invertebrate and bird         broad definition is proposed because it avoids the
                                                                          5.3 Farmland boundary features     119



need to distinguish between lines of trees and rows      Hedgerows Regulations was found, but it was not
of bushes. Earth or stone banks or walls are not         predictive for middle-ranking hedgerows, and the
included, in accordance with the Species-rich            HEGS method cannot be used as a proxy for the
Hedgerow BAP. However, if these features occur           Hedgerows Regulations or vice versa. General sur-
in association with a line of trees or shrubs, they      veys can be carried out with these two methods
are considered to form part of the hedgerow.             together to maximise both ecological and context-
A length of hedge between connections with               ual information collected during surveys.
other hedges or other linear features (i.e. inter-
nodal length) is counted as a separate hedge.            Management requirements and external impacts
A hedgerow with a gap of more than 20 m is con-          Farmland boundary features are profoundly affected
sidered to be two separate hedges.                       by agricultural management practices (see Barr et al.
   Although the survey methodology proposed by           (1995) for hedges) and therefore these, as well as
the Steering Group is meant to be for species-rich       habitat quality, should be thoroughly monitored.
hedges, it could be applied to others. However,              Stocking density and grazing or browsing inten-
monitoring all hedges could be onerous and is            sity can have serious impacts on the vegetation of
unlikely to be a priority of a monitoring programme.     ditch margins, grassy banks and hedges. Stocking
It may be appropriate in many cases to target species-   density can be easily monitored by regular stock
rich hedges for monitoring beyond the simple             head counts or by inspection of farm management
assessment of extent. Species-rich hedges are            records obtained directly from the landowner.
defined in the Species-rich Hedgerow BAP as ‘any         However, stock will preferentially graze some
hedge that has 5 or more native woody species on         vegetation. Monitoring grazing intensity, e.g. by
average in a 30 metre length, or 4 or more in north-     marking hedgerow branches and recording
ern England, upland Wales and Scotland’. The BAP         damage or by excluding grazing from banks and
also recommends that hedges that contain fewer           ditch margins, may therefore be a more reliable
woody species but have a rich basal flora should be      indicator of impacts, but it is more difficult.
included if possible, although no practical criteria         The accidental or intentional application of fer-
have as yet been agreed for defining them on this        tilisers, pesticides and other agrochemicals (e.g.
basis. Table 5.3 summarises the attributes indicat-      plant growth regulators) to farmland boundary fea-
ing the condition of hedgerows, along with their         tures can have major direct and indirect effects on
recommended monitoring techniques.                       vegetation and animal communities. Farm records
   Rich et al. (2000) compared four different hedge      can again provide basic data on their use, but these
survey techniques on the same hedgerows: stand-          will be of little value in predicting impacts unless
ard 30 m lengths, 10 m plots, the Hedgerow               such agrochemicals are not used or are applied
Evaluation and Grading System (HEGS), and                infrequently. Such data may also be unreliable.
features of importance as defined in the UK                  Where fertilisers, pesticides, etc. are applied fre-
Government’s Hedgerows Regulations 1997. All             quently, impacts can only be assessed by more sophis-
methods identified variation between hedgerows           ticated procedures. Detecting fertiliser application
which could differentiate between hedgerow               may require chemical analysis of soils (Section 6.2.2)
types (e.g. parish/community boundaries, new             or water (Section 5.9.) Herbicide and other pesticide
hedgerows), or compare hedgerows in different            applications may be detected by chemical analysis of
areas (e.g. communities). The number of species          soil and/or plants or indirectly from quadrats
in 10 m lengths, 30 m lengths and the whole hedge-       (Sections 6.4.2–6.4.5) if there are marked changes in
row were highly correlated; surveys of sections can      botanical composition (e.g. a marked decrease in the
thus indicate overall species richness, although 30 m    broadleaved herb component). Liming can be
lengths gave better results than 10 m lengths. In        detected from soil pH analysis (Section 6.2.2.) These
general a good relationship between the HEGS             monitoring methods are, however, likely to be time-
value and the ‘importance’ as defined by the             consuming, complex and difficult to interpret, and
120    5 HABITAT REQUIREMENTS AND ISSUES



Table 5.3. A summary of the quality attributes providing an indication of the condition of hedgerows, and their
recommended monitoring techniques

Attribute         Habitat properties                                Monitoring technique

Physical          Extent                                            Phase I survey (Section 6.1.5) or aerial
properties                                                          photography (6.1.3)
Composition       Presence/frequency of indicator species           Species listings over defined hedgerow lengths,
                                                                    line transects (6.4.6), or quadrats (6.4.2–6.4.4)
                                                                    for hedge bottoms
                  Species richness (woody species and               Line transects (6.4.6) or quadrats (6.4.2–6.4.4)
                  ground flora)                                     for hedge bottoms
Structure         General structure and shape, including:           Subjective visual assessment for major
                  * length of gaps at top of hedge                  changes (e.g. trimming) or fixed-point
                  * length of gaps in woody plants                  photography (6.1.4)
                  * coverage at base of hedge

                  * shape of hedge (e.g. ‘A’, box, etc.)

                  Average height                                    Sample measurements with a graduated pole,
                                                                    or fixed-point photography (Section 6.1.4)
                  Width:                                            Sample measurements with a graduated pole
                  * average hedge width

                  * width of rough herbage composed

                    of native species at hedge base
                  Density                                           Subjective assessment or chequered board (see,
                                                                    for example, Fuller et al. (1989))
                  Dead wood                                         Transect and subjective assessment of part
                                                                    (6.5.5)
                  Presence of trees (live and dead)                 Include as target notes in Phase I surveys (6.1.5)



are therefore probably beyond the resources of most
monitoring programmes.
                                                            5.3.2        Specific issues affecting the survey
   As described above, the structure of hedgerows
                                                                         and monitoring of habitat
is primarily controlled by their management.                As there is the possibility of agricultural practices
Laying or coppicing is carried out to reinvigorate          having a major impact on farmland boundary fea-
the hedge and is best done on 8–12 year rotations.          tures, monitoring should be carried out frequently.
Trimming is routinely done to avoid excessive               In particular, it is prudent to carry out routine site
shading of crops, to maintain access and to retain          visits annually to make simple visual inspections of
the general shape of a hedge. To be most beneficial         features and to monitor management practices. More
for biodiversity, trimming is best carried out on           detailed quantitative assessments of features should
rotations of 2–3 years or longer (Andrews &                 be carried out at no more than 3 year intervals, or
Rebane, 1994; Hill et al.,1995). These management           immediately if annual inspections reveal apparent
practices can normally be easily monitored from             impacts or detrimental management activities.
farm records or simple visual inspections every                Some monitoring data can often be obtained
couple of years (as the effects of trimming are visi-       directly from landowners’ or managers’ farm records.
ble for several years).                                     However, as explained above, such general data may
                                                                 5.4 Grassland and herbaceous communities    121



Table 5.4. Native woody species that occur in hedgerows in the UK
These are typical species, not an exhaustive list.

                        Trees                                                       Shrubs

Common name                        Scientific name                  Common name              Scientific name

Alder                              Alnus glutinosa                  Blackthorn               Prunus spinosa
Crab Apple                         Malus sylvestris                 Broom                    Cytisus scoparius
Ash                                Fraxinus excelsior               Elder                    Sambucus nigra
Aspen                              Populus tremula                  Gorse                    Ulex europaeus
Downy Birch                        Betula pubescens                 Gooseberry               Ribes uva-crispa
Silver Birch                       Betula pendula                   Guelder Rose             Viburnum opulus
Bird Cherry                        Prunus padus                     Hawthorn                 Crataegus monogyna
Wych Elm                           Ulmus glabra                     Hazel                    Corylus avellana
Elm species                        Ulmus spp.
Gean (Wild Cherry)                 Prunus avium                     Juniper                  Juniperus communis
Wild Plum                          Prunus domestica                 Dog Rose                 Rosa canina
Holly                              Ilex aquifolium
Pedunculate Oak                    Quercus robur                    Wild Rose                Rosa spp.
Sessile Oak                        Quercus petraea                  Spindle                  Euonymus europaeus
Scots Pine                         Pinus sylvestris                 Bay Willow               Salix pentandra
Rowan (Mountain Ash)               Sorbus aucuparia                 Eared Willow             Salix aurita
Crack Willow                       Salix fragilis                   Grey Willow              Salix cinerea
Goat Willow                        Salix caprea                     Osier                    Salix viminalis
White Willow                       Salix alba                       Purple Willow            Salix purpurea
Whitebeam                          Sorbus aria

Sources: NCC (1988) and Stace (1997)


not be an accurate indicator of the actual impacts of      grasses and sedges is usually critical for monitoring
management practices. Furthermore, such data may           hedgerow ground floras, ditches and banks, and it
not be reliable, for example as a result of inconsistent   may be necessary to use a specialist botanist for
record-keeping. It is also possible that some activities   such detailed work.
(intentional or accidental) may be concealed if they
contravene management agreements or general
codes of good environmental practice.                      5.4        GRASSLAND AND HERBACEOUS
   Most general field monitoring of vegetation can                    COMMUNITIES
be carried out between May and September, but
                                                           5.4.1        Survey and monitoring requirements
should be carried out within the same 2 week per-
                                                                        and methods
iod of the year as the original survey when repeat-
ing monitoring of ditches, grassy banks or                 Attributes for assessing habitat condition
hedgerow ground flora.                                     Grassland and herbaceous NVC communities are
   Hedgerow shrubs and trees are normally fairly           designated features or attributes of broader habit-
easily identified and can therefore be monitored           ats of many SSSIs. A key requirement is therefore to
by most biologists with general field training (see        monitor their continued presence, extent and qual-
Table 5.4). However, vegetative identification of          ity. This can be carried out by repeat NVC surveys
122    5 HABITAT REQUIREMENTS AND ISSUES



incorporating properly replicated quadrat sam-            exotic or invasive species, or indicators of poor
pling (see Sections 6.1.6 and 6.4.2 for detailed dis-     condition may also need to be monitored. For
cussions) or by using appropriate mapping                 example, in species-rich Nardus grasslands, an
techniques. Simple repeat NVC mapping is not con-         absence or sparse cover of Crested Dog’s-tail
sidered to be appropriate for monitoring purposes.        Cynosurus cristatus and Perennial Ryegrass Lolium
   Many lowland sites are in enclosed areas on            perenne is considered to be necessary for acceptable
farms, and defining their extent is relatively simple.    condition (Davies & Yost, 1998).
On larger upland sites the grasslands may form part           If it is only necessary to establish the presence or
of a much larger complex, including blanket bog           minimum approximate population of a species
and heathland, and sampling is therefore required         that is likely to be reasonably common and detect-
to determine changes in extent and distribution. As       able, then simple look–see or count methods may
grassland composition is often strongly related to        be adequate (see Part III, Sections 15.2.1 and 15.2.2.)
soils and topography in such sites, careful stratifica-   However, if species are rare or difficult to detect, or
tion may be required to ensure that sampling              if accurate quantitative assessments (e.g. of cover)
designs are effective and efficient, and cover the        are needed, sampling procedures such as quadrat
full range of vegetation types present.                   or transect techniques will probably be required
   In addition, various aspects of the vegetation,        (Section 6.4.)
such as species richness, presence of particular              The dispersal of species within a grassland can
typical or indicator species, sward height and            be of use when determining quality. When many
cover, soil nutrient status, etc., may be needed to       rare species or typical indicators are dispersed
assess whether the quality of the vegetation is           throughout the grassland at medium to high fre-
being maintained. Species richness is an easily           quency rather than being in single isolated clumps
understood and measured variable, which can indi-         at low frequency, grasslands are often old and of
cate grassland quality. In general, the more species-     high quality with a long history of the same
rich a grassland, the more valuable it is for nature      management.
conservation. The diversity should be of species              Sward height, cover and litter are valuable indi-
characteristic of that community, and not extra-          cators that should be monitored as a matter of
neous species (e.g. non-native species or trees           course when collecting quadrat data or surveying
invading from adjacent woodland); some grass-             sites. When swards are ungrazed, height, cover and
lands are inherently more species-rich than others.       litter all increase, resulting in decreased reproduc-
One way to clarify that richness is an intrinsic          tion of many species and a decrease in the propor-
function of the community concerned would be              tion of small, short-lived species. The presence of
to assess the ratio of species listed in the appropri-    associated grassland features, such as anthills,
ate NVC table in Rodwell (1991 et seq.) to other          often adds diversity and they may also be worth
species, or to assess the ratio of constants, differ-     monitoring in their own right.
entials and preferentials to associates and other             Table 5.5 summarises the attributes that indi-
species (Rodwell, 1991 et seq.). Bear in mind that        cate the condition of grasslands and gives recom-
oversampling a local community will result in a           mended techniques for monitoring them. Detailed
number of constants higher than that indicated in         methods and reviews of monitoring grasslands are
the NVC tables.                                           outlined by Byrne (1991), Hodgson et al. (1995) and
   The presence and/or abundance of particular            Robertson (1999).
species may often be considered useful in helping
to define condition in many grassland and other           Management requirements and external impacts
vegetation communities. These most often relate           Monitoring of management practices (Sections
to desirable species that are of conservation impor-      7.1–7.3) may be as important as monitoring the
tance or species that are indicators of favourable        quality of the habitat. It should perhaps be carried
ecological condition. Undesirable species, such as        out as part of a site monitoring programme.
                                                                 5.4 Grassland and herbaceous communities          123



Table 5.5. A summary of the quality attributes providing an indication of the condition of grasslands,
and their recommended monitoring techniques

Attribute            Habitat properties                       Monitoring technique

Physical             Extent and distribution                  Phase I mapping (Section 6.1.5) with aerial
properties                                                    photography (6.1.3) for basic long-term monitoring
                                                              NVC surveys with quadrat sampling (6.1.6) for NVC
                                                              vegetation types
                     Soil nutrients                           Chemical analysis (not covered)
Composition          Characteristic communities               Quadrats (6.4.2) or transects (6.4.6), with NVC
                                                              analysis (6.1.6) where NVC communities are
                                                              Notified Features or important attributes
                     Functional components of                 FIBS (6.4.4)
                     vegetation
                     Species composition and richness         Mini-quadrats (6.4.3)
                     Presence/absence of typical/             Look–see or total counts (Part III, Sections 15.2.1
                     indicator species                        and 15.2.2,) quadrats (6.4.2, 6.4.3) or transects (6.4.6)
Structure            Sward height                             Drop-disc, ruler
                     Cover                                    Conventional quadrats (6.4.2) or point quadrats
                                                              (6.4.5) if precise measurements are required
                     Litter                                   Quadrats (6.4.2)




   Most grassland and herbaceous communities                 some vegetation types in preference to others,
need to be managed by mowing, burning or graz-               monitoring the stocking density alone for a site
ing to maintain their quality. There are often               may not reflect the true grazing pressure on valu-
severe practical difficulties in ensuring that these         able vegetation. Here, more detailed methods, such
operations are carried out at the right time each            as counting the proportion of grazed shoots or
year, and therefore monitoring of such activities            leaves, may be required.
will often be needed to establish a long-term view              Detecting fertiliser application may require chem-
of the stability of the community.                           ical analysis of soils (Section 6.2.2). Herbicide appli-
   Grasslands vary in their sensitivity to changes in        cation may be detected by chemical analysis of
management and subsequent recovery. For                      plants and/or soil or indirectly from quadrats
instance, application of fertiliser to nutrient-poor         (Sections 6.4.2–6.4.4) where there are marked
grassland may result in rapid and irreversible               changes in sward composition (e.g. a marked
changes, yet effects due to the absence of grazing           decrease in the broadleaved herb component).
can be reversed after even a decade. In setting the          Liming can be detected from soil pH analysis
timescale for monitoring, any threats to, and the            (Section 6.2.2.) Some management monitoring
sensitivity of, each site and community will there-          data can often be obtained from the landowners’
fore need to be considered.                                  or managers’ farm records. However, such data
   Monitoring stocking density may require regu-             may not be reliable, for example as a result of in-
lar stock head counts, although stock will often             consistent record-keeping. It is also possible that
graze some types of grassland in preference to               some actions (intentional or accidental) may be con-
others. When stock are absent, previous use can              cealed if they contravene management agreements
be assessed from dung. As stock selectively graze            or general codes of good environmental practice.
124     5 HABITAT REQUIREMENTS AND ISSUES



  On upland grasslands the impacts of grazing and       scattered vegetation to pastures, woodland clear-
burning can be monitored by using the method            ings and closed woodland canopies. The resulting
developed by MacDonald et al. (1998a,b) for SNH.        diversity and unusual combinations of plants of
                                                        woodland, rocky habitats and grassland growing
                                                        together on a pavement are key attributes, resulting
5.4.2      Specific issues affecting the survey         in a range of NVC types. Ward & Evans (1976)
           and monitoring of habitat                    regarded open pavements as the most important
Grasslands and herbaceous communities should be         floristically. Open pavements with a good ‘view’
monitored at 3 or 6 year intervals. Hay meadows         are also important from an earth science perspec-
should be surveyed before they are cut. Most gen-       tive, but the maintenance of existing woody cover
eral monitoring can be done between May and             is often necessary to maintain overall diversity.
September but should be carried out within the             Ward & Evans (1975, 1976) documented all the
same 2 week period of the year as the original          limestone pavements in Scotland and England and
survey when repeat monitoring.                          included species lists for all sites, with a crude
   Vegetative identification of grasses and sedges is   estimate of frequency. They limited floristic record-
usually critical for monitoring grasslands, so the      ing to grikes more than twice as deep as they were
work will need to be carried out by a specialist        wide, and this method should be followed for
botanist.                                               consistency. Species lists were used to create a flor-
   When carrying out assessments of grasslands it       istic index, which was used to rank pavements,
should be remembered that it can be difficult to        and deviations from this figure can be used to
record quadrats safely and effectively if inquisitive   monitor maintenance of diversity. Most surveys
stock are present. Trampling by stock can also          took about one hour by two botanists, and it is
affect tall herb vegetation, and lodging (vegetation    suggested that this should be repeated as closely
falling over as a result of excessive growth, wind or   as possible and at a similar time of year. Repeat
rain) has marked effects on the ease and accuracy       botanical surveys will vary depending on the bot-
of recording quadrats.                                  anist and the amount of effort, so measures should
                                                        be taken to standardise surveys (Rich & Smith,
                                                        1996). As site-based surveys return generally simi-
5.5      LIMESTONE PAVEMENT                             lar but rarely identical species lists, the floristic
                                                        index from repeat surveys should be within Æ20%
5.5.1     Survey and monitoring requirements
                                                        of the Ward & Evans (1975) baseline figure to
          and methods
                                                        account for sampling error and indicate mainte-
Attributes for assessing habitat condition              nance of the floristic index.
In Europe, limestone pavements are restricted              The earth science interest centres on mainte-
to Britain and Ireland, and internationally they        nance of the key physical features for which each
are a rare habitat. Most pavements are quite small      site is selected (such as the grikes), and their visibil-
and extend over only a few hectares. All examples       ity (i.e. they should not be covered by excessive
are important irrespective of size, and extent          vegetation). The variation in structure of grikes
should be monitored to check for damage and             and other typical karst geomorphological features,
encroachment. From a national study only 3% of          such as solution basins, erratics, runnels, etc., is
pavements in Britain were found to be undamaged         also important for the biological features of the
and only 13% were 95% or more intact (Ward &            pavements as it adds a diversity of habitats.
Evans, 1976).                                           Physical structure is unlikely to change except by
   The biological interest is provided by a variety     damage, although perched erratics tend to get
of microclimates, which results in a mosaic of          pushed off by vandals.
different plant communities. The development               During the botanical survey, notes can be kept of
of vegetation over pavements ranges from sparse,        damage by geologists (these are typically small
                                                                                     5.5 Limestone pavement       125



Table 5.6. A summary of the quality attributes providing an indication of the condition of limestone pavements,
and their recommended monitoring techniques

Attribute           Habitat properties                     Monitoring technique

Physical            Extent                                 Phase I or NVC survey (Sections 6.1.5 and 6.1.6),
properties                                                 aerial or fixed-point photography (6.1.3 and 6.1.4)
                    Removal of limestone, damage by        Field surveys
                    geologists
Composition         Floristic index                        See text and Ward & Evans (1975, 1976)
                    Characteristic communities             Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                           (6.1.6) where NVC communities are Notified Features
                                                           or important attributes
Structure           Cover of wood and scrub                Aerial or fixed-point photography (6.1.3 and 6.1.4)


chips taken off edges of grikes, often obvious              On pavements with deep grikes, low-intensity graz-
because of their lack of lichen or algae cover).            ing may be tolerated. Some improvements may be
Existing damage can be marked with a small spot             achieved locally if grazing is reduced or removed.
of enamel paint so that new damage can be                      Air pollution and acid deposition could well be
assessed, but this is usually a relatively minor prob-      damaging the geomorphological features as a
lem. Table 5.6 gives a summary of attributes useful         result of an increased rate of erosion, although
when assessing the condition of limestone pave-             this is a slow process and a long-term problem.
ments and the methods recommended for moni-                 Lichens may have some role in protecting the pave-
toring them.                                                ments from weathering.
                                                               The amount of public pressure, and hence the
                                                            amount of monitoring of damage required, will
Management requirements and external impacts                vary depending on catchment area. For example,
Variation in woody cover of pavements is import-            some sites in Scotland do not appear to be under
ant for the diversity of the habitat in its own right;      the same public pressure as those in England, and
in cases in which biological sites have been                there seems to be little point in monitoring erosion
selected to include this, maintenance of the exist-         or other damage from visitors unless it is identified
ing woody cover is regarded as desirable. If sites          as a potential problem.
have been selected for their earth science import-
ance, the requirement for open views should be
                                                            5.5.2      Specific issues affecting the survey
taken into account.
                                                                       and monitoring of habitat
   In the past limestone pavements have been
widely damaged by removal of limestone for rock-            It is recommended that these sites are monitored at
eries, walls and building materials, and minor              3–6 year intervals. This should be carried out
damage still occurs from geological sampling.               between June and September when the flora is
   Because of the difficulty of stock grazing on            fully developed.
limestone pavements, most sites are unlikely to be              The boundary of the limestone pavement should
under threat from an increase in cattle grazing             be taken as the edge of the exposed limestone. The
pressure but there may be a gradual impoverish-             small size and relative accessibility of most pave-
ment associated with sheep grazing. Some open               ments, and the existence of a standard method
pavements that support rarities should continue to          (Ward & Evans,1975), will not impose significant
be lightly grazed to prevent scrub encroachment.            logistical limitations on surveys.
126     5 HABITAT REQUIREMENTS AND ISSUES



   Fixed-point photography has worked well as a           having a cover of 25% or more of the main erica-
monitoring tool at a number of sites, and is useful for   ceous species (Calluna, Empetrum, Vaccinium or Erica
assessing changes in woody cover or limestone             spp.) (NCC, 1990a,b), and cover of these species
extraction. Aerial photographs are also useful for        is therefore an essential indicator of condition.
spotting limestone removal on larger pavements.           Gorse (Ulex spp.) may also be important on lowland
   NVC surveys with quadrat sampling of each              heaths. Much of the vegetation is naturally species-
grike are unlikely to be worth while, and may be          poor but no one plant species should cover more
difficult because of the patchy nature of the vegeta-     than 90% of the ground. This will allow the devel-
tion. A broad overview of the vegetation types            opment of diversity within the limits characteristic
should be taken. Quadrats (Section 6.4.2) or trans-       of the habitat. In the west, heathland may be very
ects (Section 6.4.6) may be useful for assessing          important for oceanic bryophytes and lichens.
changes in vegetation related to grazing.                 Heavy grazing and excessive burning (or poor burn-
   Particular care should be taken when undertak-         ing practices) may result in reduced cover of erica-
ing fieldwork because of the risks of falling on          ceous and other species.
slippery pavements in wet weather. Monitoring                In general, mixed-age stands of ericaceous spe-
should be carried out by two people, and other            cies on heathlands are more valuable than homo-
appropriate safety measures outlined in Box 2.11          geneous stands, as the former tend to have more
should be followed.                                       microhabitats for invertebrates, lichens, bryo-
                                                          phytes and higher plants, and they also indicate
                                                          that regeneration conditions are suitable. Bare
5.6      LOWLAND AND UPLAND HEATHLAND
                                                          ground may vary from large areas in recently
5.6.1      Survey and monitoring                          burnt stands to virtually none in closed mature
           requirements and methods                       stands; this is quite natural, but some bare ground
                                                          is usually desirable.
Attributes for assessing habitat condition                   Scrub (birch, pine) and Bracken Pteridium inva-
Heathlands are subject to significant reclamation
                                                          sion is often a problem on both lowland and upland
pressure, at least in the lowlands. Most heathland
                                                          heaths and may not be desirable, although it is an
types occur as part of habitat mosaics in which they
                                                          integral part of the habitat.
exhibit gradations into other communities (e.g.
                                                             A summary of attributes useful when assessing
grassland, blanket bogs). They show strong edge
                                                          the condition of heathlands is provided in Table 5.7
effects and are vulnerable to fragmentation. It is
                                                          together with recommended monitoring methods.
therefore important to monitor their extent and
                                                          An additional method for monitoring Heather
distribution.
                                                          Calluna vulgaris cover is given in MacDonald &
   Soils are a key feature determining the nature of
                                                          Armstrong (1989).
the heathland vegetation. They tend to be acidic,
nutrient-poor podzols and shallow peat. Soil pH
                                                          Management requirements and external impacts
may not always be low; in The Netherlands the
                                                          Heathlands are semi-natural habitats, and require
rarer heathland plant species are often associated
                                                          management techniques such as grazing, burning
with soils of pH greater than 5 (Roem & Berendse,
                                                          or cutting (with the possible exception of some
2000). Nutrient enrichment is usually very dam-
                                                          maritime heaths) to maintain structural and spe-
aging. Heathlands may be freely drained or more
                                                          cies diversity and prevent scrub encroachment.
or less permanently waterlogged. Wet heath types
                                                             Low-intensity grazing is often valuable in creat-
may depend on a high water table, which may
                                                          ing diverse microhabitats and is the preferred man-
require monitoring.
                                                          agement, although it is not always practicable
   In general the more diverse the heathland in
                                                          (Gimingham,1992). Grazing regimes need to be
terms of characteristic heathland species and struc-
                                                          adapted to local situations. Many upland heaths
ture the better. Heathlands are usually defined as
                                                          are managed as grouse moors by patchwork
                                                                           5.6 Lowland and upland heathland        127



Table 5.7. A summary of the attributes providing an indication of the condition of heathlands, and their
recommended monitoring techniques

Attribute           Habitat properties                         Monitoring technique

Physical            Extent and distribution                    Phase I mapping (Section 6.1.5) with aerial
properties                                                     photography (6.1.3) for basic long-term
                                                               monitoring
                                                               NVC surveys with quadrat sampling (6.1.6)
                                                               for NVC vegetation types
                    Soil pH and nutrients                      Soil analysis (6.2.2)
                    Bare ground                                Conventional quadrats (6.4.2)
                    Water table                                Dipwells or WALRAGS (6.2.1)
Composition         Characteristic communities                 Quadrats (6.4.2) or transects (6.4.6), with NVC
                                                               analysis (6.1.6) where NVC communities are
                                                               Notified Features or important attributes
                    Ericaceous and other keystone              Conventional quadrats (6.4.2), aerial
                    species cover                              photographs (6.1.3)
                    Species composition and richness           Conventional quadrats (6.4.2) or line/point
                                                               intercept transects (6.4.6)
                    Presence/abundance of                      Look–see or total counts (Part III, Sections 15.2.1 and
                    typical/indicator species                  15.2.2) quadrats (6.4.2 and 6.4.3) or transects (6.4.6).
Structure           Occurrence and scale of                    Transects (6.4.6) and fixed-point
                    horizontal and vertical                    photography (6.1.4)
                    structure (patchiness)                     Drop-disc or ruler for height
                    Age/physical structure of                  Plant size and demographic techniques (Part III,
                    ericaceous shrubs                          Sections 15.2.3 and 15.2.4)
                    Scrub invasion                             Fixed-point (6.1.4) or aerial photography (6.1.3)


burning, or as sheep walks. Overgrazing can occur,
normally near supplementary feeding locations or
                                                            5.6.2       Specific issues affecting the survey
around the lower margins of moors close to better-
                                                                        and monitoring of habitat
quality pastures. Grazing and burning are key fac-          It is recommended that heathlands are monitored
tors on many heaths, and much impact monitoring             at 6 year intervals. Late summer is an appropriate
will be directed towards this (Sections 7.1–7.2) A          time for monitoring floristic parameters, as the
method for monitoring the impacts of upland                 weather is likely to be better and the plants fully
land management practices has been developed                developed, but work can be carried out for most of
by MacDonald et al. (1998a,b) for SNH.                      the growing season (April–October) in these essen-
   Air pollution should be below the critical con-          tially evergreen communities. Early spring is the
centrations required to maintain the low nutrient           best time to monitor heather browsing, after win-
status of the heaths (SO2 10 mg mÀ3, NO2 30 mg mÀ3,         ter browsing has finished but before new growth
NH3 8 mg mÀ3) (English Nature, 1993). In recent             occurs. Access may be difficult late in the season
years concentrations of SO2 have declined as a              because of deer stalking or grouse shooting.
result of effective drives to remove such emissions             Sites containing heathlands are often very large
from power stations across Europe, whereas pollu-           and complex, with other related vegetation types
tion from NOx remains a problem.
128     5 HABITAT REQUIREMENTS AND ISSUES



such as blanket bogs intermixed. Monitoring may           examples are small and widely scattered, often
therefore have to be integrated with that of other        occurring as isolated, fragmented sites in the low-
habitats. The large scale can make access difficult if    lands, and this fragmentation imposes significant
vehicle tracks are absent, and walking though tall        limitations on their potential for recovery after
heather can be extremely tiring.                          damage. Fens are among the habitats that have
   The large size of many heathland sites also            undergone the most serious declines across
means that sampling will be essential, as it will         Europe. Swamps and reedbeds often occur around
not be possible to monitor the whole site. As heath-      the margins of lakes, lochs, pools and rivers
land composition is often strongly related to soils       (Sections 5.9 and 5.10).
and topography in such sites, careful stratification         These habitats are wetlands, and the rise and fall
may be required to ensure that all communities            of the water table and movement of water are
and subcommunities are adequately and efficiently         important factors in determining the plants and
sampled. Stratification according to ownership or         communities that occur. The height of the water
management may also be appropriate because this is        table, typically at or slightly above or below that of
likely to be the major factor determining the condi-      the substrate, appears to be especially important in
tion of the vegetation. In large areas of uniform moor-   controlling zonation and succession to other vege-
land it may also be efficient to carry out sampling       tation types. Hydrological regimes should there-
using a multi-level strategy (Part I, Section 2.3.3.)     fore be monitored, but this is a complex subject
   Accurately determining location can be difficult       that cannot be covered here (see Section 6.2.1.)
when mapping. Boundaries between communities              Similarly, water chemistry has a profound influ-
can be ecotonal in nature, and different surveyors        ence on wetland vegetation and should be carefully
may not be consistent in their interpretation of          monitored. Further information on this subject is
boundary locations.                                       provided in Section 5.9.
   Heather damage may be caused by inverte-                  Fen vegetation is variable but very distinctive
brates, especially certain moth species (e.g. Winter      and contains many species that are rare or scarce.
Moth Operophtera brumata caterpillars) and the            The type of vegetation and its richness are key
Heather Beetle Lochmaea suturalis. Damage can also        indicators of habitat quality. Wheeler (1989) pro-
be due to other factors, such as weather and fungal       posed that two botanical indices based on richness
diseases.                                                 indicators and rare species could be used for rapid
                                                          evaluation of sites; a similar approach could also be
                                                          used for monitoring.
5.7      FENS, CARR, MARSH, SWAMP                            There is often some variation in topography
         AND REEDBED                                      across a fen, which can be important for maintain-
5.7.1      Survey and monitoring requirements             ing diversity. The vegetation itself often forms
           and methods                                    small mounds with wetter areas between (and
                                                          sometimes shallow pools), allowing species of wet
Attributes for assessing habitat condition                and dry ground to grow adjacent to each other.
This group includes a range of habitats, each of          Variations in topography may also be associated
which presents its own problems for monitoring.           with old peat cuttings. Natural transitions to non-
Carr is essentially swampy woodland; monitoring           fen habitats are rare features and can be of high
techniques appropriate for woodlands will there-          value.
fore be important (Section 5.1.) Marsh monitoring            A summary of attributes that are useful in pro-
will include techniques appropriate for grasslands        viding an indication of the condition of wetlands
and herbaceous vegetation (Section 5.4).                  is provided in Table 5.8 together with recom-
   Fens may vary from small areas around a calcar-        mended monitoring methods. Rowell (1988) pro-
eous spring to large sites (e.g. the 300 ha Insh          vides practical advice on monitoring peatlands
Marshes near Kingussie); size is critical. Most           including fens.
                                                                    5.7 Fens, carr, marsh, swamp and reedbed       129



Table 5.8. A summary of the quality attributes providing an indication of the condition of wetlands, and their
recommended monitoring techniques

Attribute           Habitat properties                         Monitoring technique

Physical            Extent                                     Phase I mapping (Section 6.1.5) with aerial
properties                                                     photography (6.1.3) for basic long-term
                                                               monitoring
                                                               NVC surveys with quadrat sampling (6.1.6)
                                                               for NVC vegetation types
                    Soil pH and nutrients                      Soil analysis (6.2.2)
                    Hydrological regime                        Piezometer, dipwells or WALRAGS (6.2.1)
                    Water chemistry                            Macrophyte indicators for standing waters or
                                                               chemical analysis (not covered)
Composition         Characteristic communities                 Quadrats (6.4.2) or transects (6.4.6), with NVC
                                                               analysis (6.1.6) where NVC communities are
                                                               Notified Features or important attributes
                    Species composition and richness           Conventional quadrats (6.4.2) or line/point
                                                               intercept transects (6.4.6)
                    Presence/abundance of                      Look–see or total counts (Part III, Sections 15.2.1 and
                    typical/indicator species                  15.2.2) quadrats (6.4.2 and 6.4.3) or transects (6.4.6).
Structure           Vegetation height                          Drop-disc or ruler
                    Scrub invasion                             Fixed-point (6.1.4) or aerial photography (6.1.3)



                                                            species composition. High species richness is
Management requirements and external impacts
                                                            strongly related to low nutrient status. Nutrient
The main threats to fens are reclamation, drainage
                                                            enrichment by agricultural fertiliser run-off or sew-
and abstraction from aquifers, cessation of tradi-
                                                            age is therefore highly damaging. Rivers tend to
tional management practices such as grazing and
                                                            have high nutrients in their sediments, although
turf cutting, overgrazing, eutrophication, develop-
                                                            fens can occur in floodplain situations.
ment of scrub, and flood defences. Some of these
may require off-site monitoring, and large-scale
catchment protection may be required for fens               5.7.2        Specific issues affecting the survey
because of their dependence on the flow of ground                        and monitoring of the habitat
or surface water of an appropriate quality.
   Management of fen vegetation varies. Some                These habitats should be monitored every 3 years.
short fens are maintained by light grazing and its          The vegetation of wetlands is most developed late
associated trampling, the low nutrient concentra-           in the summer (July–September) and is best moni-
tions and scouring by water erosion. Reedbeds               tored in August when water levels are at their low-
should not be grazed. These and others, such as             est. The presence of breeding birds may also
Cladium (sedge) beds, may require regular cutting.          restrict access at other times of year.
Peat cutting and scrub clearance are also required             A high level of botanical skill is needed for NVC
in some sites.                                              surveys of fens and similar habitats because of the
   Minerotrophic or topogenous fens develop under           range of difficult groups, such as grasses, sedges
the influence of ground water, the nutrient content         and bryophytes, which form important parts of
of which is critically important in determining             the vegetation.
130     5 HABITAT REQUIREMENTS AND ISSUES



   It may be very difficult to place quadrats in tall      Topography is a second attribute. Lowland
swamp without damaging the vegetation; transects        raised bogs form deep peat deposits of variable
may be easier to record. If largish areas of uniform    depth (5–10 m) with a flat or gently sloping topo-
vegetation are picked to minimise edge effects,         graphy and sometimes a steeper edge. Most natural
quadrats can be crudely delimited by placing ran-       undisturbed bog surfaces usually show distinctive
ging poles sideways through the vegetation. Rowell      fine-scale variation with small drier hummocks
(1988) suggests the use of circular quadrats, which     and wetter hollows related to growth of Sphagnum
can be threaded through the vegetation. In either       and other plants.
case it can be difficult to see both sides of the          A third attribute is the water table. The water table
quadrats clearly without trampling vegetation all       may be maintained by both rainwater and ground
around.                                                 water (Lamers et al., 1999). It is higher than the sur-
   Fens are difficult habitats to survey. Tall swamp    rounding land and is therefore very susceptible to
vegetation is disorienting and difficult to walk        drainage. Invasion by birch or willow may indicate
through; there may be sudden changes to open            surface flushing or that the bog is drying out.
water and the surface may be unstable because of        Transects of dipwells may therefore be valuable to
floating vegetation. Aerial photographs may be          provide hydrological information but, as there are
invaluable for mapping inaccessible areas at a gross    long-term natural cycles of drying and wetting
scale. Some access by boat can help with surveying.     related to natural variations in climate, dipwell data
Chest waders are more useful than wellington boots.     may need to be correlated with rainfall.
Eye protection may be needed in reedbeds.                  A fourth attribute is the presence of (and
   Permanent markers may be difficult to relocate       preferably active formation of) the peat itself.
under water, in deep peat or in tall vegetation, but    Assessments of whether peat growth is active or
are unlikely to be interfered with because of their     not can be made by measuring peat depth and rates
location. Birds may perch on them, resulting in         of peat accumulation directly, although if decom-
localised nutrient enrichment from droppings.           position in the catotelm equals accumulation in
Vegetation can be quite heterogeneous, and is           the acrotelm, the net result is no peat accumula-
amenable to investigation through transects and         tion, despite the fact that peat is actively being laid
by stratified sampling.                                 down. Strictly speaking, active growth of peat is
                                                        therefore a feature of peat formation, not of peat
                                                        accumulation. Peat shrinkage is usually caused by
5.8      LOWLAND RAISED BOG                             drainage or other disturbance. The characteristic
5.8.1     Surveying and monitoring                      vegetation is dominated by Sphagnum spp. (espe-
          requirements and methods                      cially S. papillosum, and sometimes S. magellanicum),
                                                        and it is important that a healthy growth is main-
Attributes for assessing habitat condition              tained in wet conditions. To a large extent, if the
Raised bogs have been officially recognised as one      Sphagnum is healthy and growing, the remainder of
of Europe’s rarest and most threatened habitats.        the habitat should be in good condition.
Since 1840 the area of primary, active lowland             A summary of attributes that provide an indica-
raised bog in the UK has decreased from around          tion of the condition of lowland raised bogs is pro-
95 000 ha to 6000 ha, a decline of 95%. Only about      vided in Table 5.9 together with recommended
3800 ha of this remains intact, some 800 ha of          monitoring methods. Stoneman & Brooks (1997)
which are in Scotland. Extent is thus the first         and Rowell (1988) provide practical advice on moni-
important attribute to monitor. The most common         toring bogs.
causes of loss have been peat extraction or conver-
sion to agriculture or forestry. Mineral extraction,    Management requirements and external impacts
built developments and neglect probably account         The management of bogs can markedly affect the
for most of the recent losses.                          quality of the site. Grazing, burning, drainage,
                                                                                       5.8 Lowland raised bog    131



Table 5.9. A summary of the quality attributes providing an indication of the condition of lowland raised bog,
and their recommended monitoring techniques

Attribute        Habitat properties                     Monitoring technique

Physical         Extent                           Phase I mapping (Section 6.1.5) with aerial photography
properties                                        (6.1.3) for basic long-term monitoring
                                                  NVC surveys with quadrat sampling (6.1.6) for NVC
                                                  vegetation types
                 Water table                      Dipwells or WALRAGS (6.2.1)
                 Peat depth                       Soil cores (6.2.2)
Composition      Characteristic communities       Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                  (6.1.6) where NVC communities are Notified Features or
                                                  important attributes
                 Species composition and richness Quadrats (6.4.2–6.4.4) or transects (6.4.6)
                 Presence/abundance of typical/   Look–see or total counts (Part III, Sections 15.2.1 and
                 indicator species                15.2.2) quadrats (6.4.2 and 6.4.3) or transects (6.4.6)
                 Sphagnum cover                   Conventional quadrats (6.4.2) or transects (6.4.6)
Structure        Pattern (hummock/hollow, bog     Quadrats (6.4.2–6.4.4), transects (6.4.6), or fixed-point
                 pools, etc.)                     (6.1.4) or aerial (6.1.3) photography for large-scale surveys
                 Scrub invasion                   Fixed-point (6.1.4) or aerial photography (6.1.3)
Dynamics         Peat formation                   Growth of Sphagnum


forestry and scrub invasion can all damage the                 Maps are usually too small in scale to show the
vegetation. Under natural conditions raised bogs            detailed minor topographical variations, and find-
may have been lightly grazed or ungrazed (unli-             ing your location on a bog may be a significant
kely), and most will survive by themselves if the           problem. Furthermore, as there are fine gradations
water table and air pollution regimes are                   between many bog communities, there is likely to
satisfactory.                                               be significant variation in the boundaries drawn by
                                                            different surveyors; monitoring such boundaries is
                                                            unlikely to be reliable. Aerial photographs may be
5.8.2        Specific issues affecting the survey
                                                            really helpful for both location and boundary deli-
             and monitoring of habitat
                                                            mitation. Careful stratification may be required to
It is recommended that bogs are monitored at                ensure that the range of bog communities and sub-
intervals of not less than 3 years. They should             communities is adequately covered and sampling
ideally be surveyed in June–October when their              is carried out efficiently.
vegetation is fully developed, but as they support             Vehicular access is always undesirable and in
few annual or deciduous species it is possible to           any case is usually impossible. Trampling can
survey them in all seasons.                                 affect Sphagnum cover on some bogs, with foot-
   Air pollution (especially sulphur-based pollution)       prints remaining for 20–30 months. It is very easy
is known to damage Sphagnum communities, so                 to damage bog vegetation during a survey, and
favourable conditions in the long term will require         damage by trampling around permanent quadrats
pollution climates below the critical thresholds (SO2       is often excessive (Rowell,1988). Duckboards, lad-
10 mg mÀ3, NO2 30 mg mÀ3, NH3 8 mg mÀ3) (English            ders or inflatable mattresses can help to spread the
Nature, 1993). Ground water quality can also affect         weight of the surveyor. Permanent markers may be
bog communities (Lamers et al., 1999).                      difficult to relocate or may become overgrown by
132     5 HABITAT REQUIREMENTS AND ISSUES



Sphagnum; the sensitivity of the vegetation to tram-
                                                         Table 5.10. The trophic categories of waters
pling means that non-permanent techniques
                                                         in terms of nitrogen and phosphorus
should be used wherever possible. These effects,
coupled with small-scale variations from year to
                                                                                  Total P         Inorganic N
year, can make comparisons problematic.
                                                         Status                   (mgÀ1)          (mgÀ1)
   Ideally, the peat moss should be intact, but most
bogs have been damaged in one or more ways by            Ultra-oligotrophic       <0.005          <0.02
cutting, drainage, burning, grazing, agriculture,        Oligo-mesotrophic        0.005–0.01      0.2–0.4
forestry and other developments. If these activities     Meso-eutrophic           0.01–0.03       0.30–0.65
are still continuing, specific monitoring of their       Eu-polytrophic           0.03–0.1        0.5–1.5
effects may be required (see Chapter 7).                 Polytrophic              >0.10           >1.5
   Bogs can be dangerous and should be surveyed
with care. Safety guidelines outlined in Part I, Box     Source: From Vollenweider (1968).
2.11 should be followed.
                                                         are therefore fundamental to assessing condition
                                                         and must therefore be monitored and maintained.
5.9      STANDING OPEN WATER                             Other important plant-related attributes include NVC
                                                         community species richness, taking into account the
5.9.1      Surveying and monitoring
                                                         level of richness expected for the type of water body.
           requirements and methods
                                                         A rich assemblage of Potamogeton spp. in particular
Attributes for assessing habitat condition               is also a good indicator of high botanical quality. An
This habitat class includes both natural and artifi-     extensive fringe of emergent vegetation is also a
cial standing fresh waters, ranging in size from a       desirable attribute of an open water site, even if its
few square metres upwards, and therefore encom-          intrinsic value as fen habitat is not high.
passes a large variety of habitats in Britain, includ-      The abundance and availability of phosphorus
ing freshwater lochs, meres, reservoirs, gravel pits,    (Table 5.10) normally limits and therefore determines
ponds, canals and temporary pools. As a result of        the growth of phytoplankton and macrophytes
this variety and the special character of these habi-    (Mainstone et al., 1993). Under certain circum-
tats, their attributes and the monitoring methods        stances, nitrogen can be the limiting nutrient for
can only be outlined in this Handbook. Further infor-    aquatic plants, particularly if phosphorus concentra-
mation is available on these topics in the recom-        tions are very high as a result of enrichment from
mended reference sources listed at the end of the        sewage treatment works or internal loading.
book. Lagoons and other marine habitats are not             However, the availability of plant nutrients
covered by this volume. Ditches were discussed in        changes with the seasons as a result of a variety of
Section 5.3.                                             influences. Therefore, a more constant measure of
   Palmer et al. (1992) found that their plant com-      productivity is provided by alkalinity (Table 5.11).
munity classification closely followed classifica-       According to this scheme, water bodies are conven-
tions based on water chemistry (see, for example,        tionally classed as dystrophic, oligotrophic, meso-
Vollenweider, 1968; Ratcliffe, 1977). Submerged          trophic and eutrophic in increasing order of
and floating macrophytes form the primary basis          calcium carbonate concentration and productivity.
for the classification and selection of SSSIs for        A fifth class is marl lakes, which have the highest
freshwater habitats (NCC, 1989). For standing            levels of alkalinity but in which productivity is
waters, representative sites are selected for each       limited because phosphate is bound to the sedi-
of ten types of macrophyte community identified          ment and therefore unavailable for plant growth.
from a detailed study of water bodies throughout            Dystrophic waters include the small water-
Britain (Palmer, 1989; Palmer et al., 1992). The char-   bodies and pool systems commonly found on peat
acteristic macrophyte communities of these types         bogs. They are solely rain-fed and thus receive
                                                                                     5.9 Standing open water     133



                                                             inherent and inseparable characteristic of such
Table 5.11. Alkalinity characteristics of different
                                                             features (see Part II, Section 2.1.2 and Glossary).
types of freshwater body
                                                             Consequently, a common requirement of monitor-
                                                             ing standing waters is to measure their character-
                            Alkalinity
                                                             istic nutrient and pH properties to ensure that
                   CaCO3                                     these are being maintained within natural fluctua-
Status             (mg lÀ1)       mequiv. lÀ1         pH     tions. Direct measurement of such chemical prop-
                                                             erties can be undertaken but interpretation may
Dystrophic         0–2            0.00–0.04           <6
                                                             not always be straightforward. However, as macro-
Oligotrophic       0–10           0.0–0.2             6–7
                                                             phytes are highly influenced by water chemistry,
Mesotrophic        10–30          0.2–0.6             c. 7
                                                             monitoring information on their distribution and
Eutrophic          >30            >0.6                >7
                                                             abundance can provide information on water
Marl               >100           >2.0                >7.4
                                                             chemistry conditions. Macrophytes have been
                                                             widely used for pollution monitoring of rivers in
Source: From Ratcliffe (1977).
                                                             Europe and the UK, but less so for standing waters.
                                                             Palmer et al. (1992) used the results of their analysis
water that contains no mineral salts dissolved from          of macrophytes (as described above) for the devel-
the underlying rocks (Andrews, 1995). Peat staining          opment of a ‘trophic ranking score’ system that
also reduces light penetration and macrophyte                allows assessment of changes in trophic status
growth. Consequently, productivity is low and                over time. See Palmer (1989) and Palmer et al.
they support a restricted range of flora and fauna.          (1992) for further details.
The water is also often too acidic to support fish.             Advantages of using macrophytes to monitor
However, the absence of fish and low numbers of              water chemistry include the fact that they are
other predators such as birds provides favourable            generally large and easy to identify with the
conditions for dragonflies, water bugs, midges and           naked eye, can be sampled rapidly, are present
other invertebrates.                                         throughout the summer months, and can act as
   Oligotrophic waters are typically upland lakes            accurate reflectors of overall conditions at a fixed
in areas with hard, nutrient-poor rock types. They           point within a water body (Bell, 1996). The disad-
have a low biodiversity and biomass of plants                vantages of using macrophytes are their seasonal-
and animals; fish are principally salmonids.                 ity, the lack of knowledge about their natural
Mesotrophic waters have the highest biodiversity             population fluctuations and difficulties with the
of standing fresh waters, often combining ele-               identification of some species.
ments of oligotrophic and mesotrophic systems,                  A standard method for surveying aquatic macro-
and also support rich and abundant macrophyte                phytes was developed by the Nature Conservancy
communities. Furthermore, relative to other types            Council (NCC) and has been used since 1975 to
of lake, they contain a high proportion of nation-           record aquatic information on standing water
ally rare and scarce species of aquatic plants               bodies throughout Britain. This has been used as
(Anon., 1995). Macro-invertebrates are also particu-         the basis for the botanical classification of standing
larly well represented. Eutrophic waters are more            waters described above (Palmer, 1989; Palmer et al.,
typical of lowland areas of Britain and support a            1992). The method entails walking the perimeter of
high biomass of vegetation (including plankton               the water body to record shoreline and shallow-
and macrophytes), and high numbers of fish                   water vegetation. Deeper water is sampled by
(usually coarse species such as cyprinids, Perch             means of a grapnel thrown from the bank at fre-
Perca fluviatilis and Pike Esox lucius) and birds, par-      quent intervals during the perimeter walk. Where
ticularly in winter.                                         possible, a boat is used, and grapnel samples are
   Nutrient status is therefore normally regarded            obtained from the bottom during transects of the
as a key attribute of water bodies, as it is an              lake and passages parallel to the shore. The cover of
134    5 HABITAT REQUIREMENTS AND ISSUES



all aquatic plants is recorded on a subjective            calibrated line until the disc is no longer visible,
DAFOR scale of abundance: dominant, abundant,             at which point the depth is recorded. The disc is
frequent, occasional, rare. Although this technique       then lowered further and raised until it reappears,
is suitable for the classification and conservation       at which point a second depth reading is taken. The
evaluation of standing water bodies it is too sub-        average of these depths is the final Secchi disc
jective and insensitive to be adequate for all but the    visibility reading. This reading provides a relative
most basic monitoring purposes. The DAFOR scale           measure of water clarity, but can also be used to
in particular is highly subjective and prone to con-      calculate the depth to which photosynthetic organ-
siderable interpretative variation between obser-         isms can occur. This is termed the euphotic zone
vers (see Section 6.4.2.)                                 (Zeu) and is between 1.2 and 2.7 times the Secchi
    A variety of methods have been used for moni-         disc depth (Moss, 1998). Secchi disc measurements
toring macrophytes, including satellite imagery           should be made under consistent light conditions
and aerial photography, grab and rake sampling,           and in calm water. Even then, measurements tend
subaqua diving and the use of sonar and remotely          to differ according to conditions and variation
operated vehicles (ROVs); see Bell (1996) for a           between observers. For more accurate measure-
review. However, none of these methods has been           ments, underwater light meters should be used,
developed to a stage of wide application and there        or turbidity (the concentration of suspended parti-
are no accepted standard protocols. Nevertheless, a       culate matter) can be measured by using a turbidity
suggested technique for the use of grapnel samples        meter or a suspended solids monitor.
to obtain semi-quantitative data on macrophyte               In addition to the primary influence of water
presence and frequency is provided in Section             quality, other relevant physical attributes of stand-
6.3.2. Other methods for monitoring attributes of         ing waters include the depth and profile of the
open water bodies are given in Table 5.12.                water body and its substrate type. Maintenance of
    Other important aspects of water chemistry that       these conditions and an overall diversity of the
influence the quality of standing water habitats          physical forms is important.
include the concentrations of dissolved oxygen,              The vegetation of water bodies can change rela-
ammonia, toxic substances such as heavy metals            tively rapidly as a result of changes in water qual-
(some of which may occur naturally, e.g. in acidic        ity, which in turn can be very rapid as a result of
waters), and pesticides. These cannot, however, be        pollution incidents. Vegetation monitoring should
regarded as direct attributes of standing water           therefore be carried out fairly frequently, probably
bodies, but are rather influencing factors that are,      at intervals of no more than 3 years. Sampling of
in turn, primarily influenced by external factors (e.g.   appropriate water-quality determinants should be
pollution). They should nevertheless be monitored         carried out at least annually, with replicate sam-
as they can have significant impacts on the condi-        ples collected on a number of occasions during the
tion of interest features. The monitoring of these        peak growing season and preferably at other times
water chemistry attributes is, however, a specialised     as well, especially for water bodies known to be
activity and the interpretation of results is complex     subject to pollution. Phosphorus concentration
(see Parr (1994) and Hellawell (1997) for reviews). It    measurements during the growing season should
is therefore recommended that specialist advice be        include total phosphorus as well as soluble reactive
obtained on such chemical analyses.                       phosphorus (SRP), as most SRP will be taken up by
    Water clarity is also an important factor, deter-     growing phytoplankton and macrophytes. Water-
mining underwater light intensities and hence the         quality monitoring is carried out by the regulatory
occurrence and vertical zonation of aquatic plants.       authorities on water bodies over 1 km2 in size (and
A simple relative measure of this can be obtained         a small number of others); these authorities should
by using a Secchi disc. Secchi discs are about 30 cm      therefore be contacted when such monitoring is
in diameter with alternating black and white or           required, to establish what data are routinely col-
yellow quarters. The disc is lowered slowly on a          lected for the water body in question.
                                                                                     5.9 Standing open water     135



Table 5.12. A summary of the quality attributes providing an indication of the condition of open water bodies,
and their recommended monitoring techniques

Attribute         Habitat properties                           Monitoring technique

Physical          Extent                                       Aerial photography (Section 6.1.3) or satellite-
properties                                                     based remote sensing (6.1.2)
Structure         Depth and profile                            Physical surveys with echosounders or depth
                                                               lines (not covered)
                  Substrate type                               Grab samples, subaqua or ROV inspections (not
                                                               covered)
Water             Nutrient status, pH, dissolved oxygen,       Trophic ranking score system (see text) or
chemistry         toxic substances, etc.                       chemical analysis (see text)
                  Turbidity/underwater light                   Secchi disc or light meter (see text)
Composition       Community type                               NCC method (see text) or NVC for detailed surveys
                                                               (6.1.6)
                  Macrophyte abundance or species              Quadrat or transect surveys by subaqua diving
                  richness                                     (Part III, Section 14.2.1)
                                                               Grapnel surveys (6.3.2 or 14.2.1)
                  Emergent vegetation                          Fixed-point photography (6.1.4), quadrat surveys
                                                               (6.4.2 and 6.4.3) or transects (6.4.6)


   Physical attributes are unlikely to change               circumstances aquatic plant productivity in fresh-
rapidly and therefore monitoring may only need              water systems tends to be limited by phosphorus
to be carried out at 5–10 year intervals, depending         availability. Phosphorus-rich pollutants, such as
on local circumstances. Additional and immediate            run-off from cereal fields, farmyard slurry, manure
monitoring may, however, be required if physical            and silage seepage, and effluent from sewage treat-
changes are known to occur at a site. See Table 5.12        ment works, are therefore the major causes of fresh-
for a summary of the attributes indicating the con-         water eutrophication (Klapper, 1991). However,
dition of open water bodies and recommended                 nitrogen can become limiting in waters in which
techniques for monitoring them.                             phosphorus concentrations are very high. An
                                                            increase in nitrate concentrations, resulting from
Management requirements and external impacts                agricultural run-off following fertiliser application
Many standing water bodies require no, or rela-             or the ploughing of old grasslands, may therefore
tively little, management to maintain their conser-         contribute to eutrophication in such circumstances.
vation interest. However, standing water bodies are            The effects of eutrophication may also be exacer-
increasingly subject to a number of detrimental             bated by excessive water abstraction upstream lead-
external impacts.                                           ing to a reduction in the quantity of water reaching a
   Pollution is probably the main impact on stand-          water body. This may increase nutrient concentra-
ing waters but tends to differ between lowland and          tions in the incoming water and increase residence
upland water bodies (Alexander et al., 1997).               time in the water body, thereby increasing the avail-
Nutrient enrichment (eutrophication) from pollu-            able time for nutrient uptake by plants.
tion is the main impact on standing waters in                  Eutrophication is less of a problem in the uplands
the lowlands because of the proximity of intensive          because of the absence of intensive farming and
agricultural activities and higher densities of human       the low human population density. Instead, oligo-
settlements. As described above, under most                 trophic lakes are prone to acidification from
136     5 HABITAT REQUIREMENTS AND ISSUES



pollutants in rain on account of their naturally low    the suitability of various methods in relation to
pH and poor buffering capacity.                         these and other considerations (such as the macro-
   Other threats include siltation, as a result of      phyte growth form) are provided in Bell (1996).
ploughing (for agriculture or forestry) or peat cut-       Safety is clearly a key consideration when carry-
ting on surrounding land, the introduction of alien     ing out fieldwork at large water bodies. Key safety
species of fish, and disturbance of waterbirds and      measures that should always be followed include
otters resulting from the use of water bodies for       proper training of personnel in safety aspects of
leisure activities.                                     aquatic monitoring (especially in the use of
   These external factors should therefore be moni-     boats); the correct use of appropriate safety equip-
tored where appropriate, but in many cases these        ment (e.g. life jackets when working over or along-
are likely to be the responsibility of the regulatory   side deep water); working in pairs or teams (never
authorities and may already be covered by ongoing       alone); and proper emergency planning (including
monitoring programmes.                                  notifying others of routes and expected return
                                                        times when working in remote locations). Other
                                                        safety precautions listed in Part I, Box 2.11, should
5.9.2      Specific issues affecting the
                                                        also be followed where appropriate.
           monitoring of the habitat
The monitoring of freshwater habitats is a specia-
lised subject and cannot be dealt with compre-          5.10      RIVERS AND STREAMS
hensively here. In particular, assessments of           5.10.1      Surveying and monitoring
water chemistry can be difficult and require specia-                requirements and methods
lised equipment. The interpretation of results in
relation to the condition of and impacts on features    Attributes for assessing habitat condition
of interest is also complex. It is therefore recom-     As with standing waters, river habitats exhibit a
mended that specialist advice be obtained on these      wide range of physical and biological variation,
subjects. Further information on these can also be      from headwater streams to mature reaches and
obtained from some of the recommended sources           estuaries. As a result of this variety and the special
listed at the end of the book.                          character of these habitats, their attributes and
   As described above, general assessments of           monitoring methods can only be outlined in this
water quality by using macrophyte indicators are        Handbook. Further information is available on these
easier than chemical analyses. A method for moni-       topics in the recommended reference sources
toring macrophyte presence and frequency is             listed at the end of the book.
described in Section 6.3.2.                                Ten major types of river have been identified
   There are a number of specific practical consid-     and used as a basis for SSSI selection (NCC, 1989).
erations to take into account when selecting            This classification was initially based on a Two-Way
appropriate methods for monitoring water                INdicator SPecies ANalysis (TWINSPAN) of macro-
bodies. In particular, the size of the water body       phyte data from 1055 sites on over 100 rivers
will considerably influence the efficiency of dif-      throughout Britain, which identified 54 subdivi-
ferent techniques and the resources required to         sions (Holmes, 1983). This has since been recently
sample it. Access is also an important consid-          updated following a re-analysis and the addition of
eration. Detailed and quantitative techniques           data from a further 459 sites to the original dataset
may be difficult or impossible to carry out along       (Holmes et al., 1998, 1999a). The overall structure of
deep tree-lined water bodies or on large shallow        the new classification is the same as that of the first
lakes. The distance from a road or navigable water-     version. The highest level consists of four broad
course may also restrict the use of some survey         groups (A–D) representing an environmental gradi-
methods; ROVs, boats, diving equipment, etc. are        ent from lowland eutrophic rivers to those that are
difficult to transport on foot. Tables summarising      essentially upland, torrential and oligotrophic.
                                                                                 5.10 Rivers and streams    137




Group        RCT         Description

A            I           Lowland, low-gradient rivers
             II          Lowland, clay-dominated rivers
             III         Chalk rivers and other base-rich rivers with stable flows
             IV          Impoverished lowland rivers
B            V           Sandstone, mudstone and hard limestone rivers of England and Wales
             VI          Sandstone, mudstone and hard limestone rivers of Scotland and northern England
C            VII         Mesotrophic rivers dominated by gravels, pebbles and cobbles
             VIII        Oligo-mesotrophic rivers
D            IX          Oligotrophic, low-altitude rivers
             X           Ultra-oligotrophic rivers


These four sub-groups are divided into 10 River         A more constant measure of productivity is provided
Community Types (RCTs) with subdivisions into           by alkalinity, as indicated by the amount of calcium
38 sub-types (see later).                               carbonate dissolved in the water. Definitions of the
   Macrophyte communities are highly influenced         various trophic categories according to nutrient sta-
by water-flow regimes, water nutrient status and        tus (Vollenweider, 1968) and alkalinity (Ratcliffe,
substrate type. These factors tend to vary across the   1977) are given in Section 5.9.
stages of a river as it flows from source to mouth.        Other important aspects of water chemistry that
Consequently, the classification, as described          influence the quality of river habitats include
below, reflects the different stages of a river as      dissolved oxygen concentrations, ammonia con-
well as its geology, water chemistry, substrate and     centration, turbidity, and concentrations of toxic
characteristic macrophyte communities.                  substances such as heavy metals (some of which
   Thus, although macrophyte communities may            may occur naturally, e.g. in acidic waters), and pes-
be the designated features of conservation interest     ticides. These cannot, however, be regarded as
within an SSSI or other site, other attributes of       direct attributes of river habitats, but they are influ-
nutrient status, pH and substrate should also be        encing factors, which are in turn influenced by
monitored (where feasible) as these are inherent        external factors (e.g. pollution). They should never-
and inseparable characteristics of each River           theless be monitored as they can have significant
Community Type. Similarly, underlying geology is        impacts on the condition of interest features.
also an inherent and inseparable characteristic, but       The monitoring of water chemistry is a specia-
this does not need monitoring as it is not expected     lised activity and the interpretation of results can
to change.                                              be difficult (see Parr (1994) and Hellawell (1997) for
   The principal factor in controlling the nutrient     reviews). It is therefore recommended that specia-
status of freshwater ecosystems is the abundance        list advice on such chemical analyses be obtained.
and availability of phosphorus, as this normally        Required data may also be collected by the envir-
limits the growth of phytoplankton (free-floating       onmental protection authorities as part of their
unicellular algae) and macrophytes (other aquatic       routine water quality monitoring programmes.
plants) (Mainstone et al., 1993). Under certain cir-       Alternatively, water quality can be assessed
cumstances, nitrogen can be the limiting nutrient       through macroinvertebrate indicators. This is,
for aquatic plants, particularly if phosphorus con-     however, a specialised technique and cannot be
centrations are very high as a result of enrichment     described here. A useful summary of the subject
from sewage treatment works. However, the avail-        can be found in RSPB/NRA/RSNC (1994); recent
ability of plant nutrients changes with the season.     reviews of the subject have been carried out by
138    5 HABITAT REQUIREMENTS AND ISSUES



Metcalfe-Smith (1994), Hellawell (1986, 1997) and       reliability of the results: I–III indicate whether
Wright et al. (1994). Information on techniques can     paired sites being compared were physically com-
be found in Part III, Chapter 20; see also Hellawell    parable; A–C indicate whether results may have
(1978) and HMSO (1978, 1980, 1983). Macrophyte          been affected by poor survey conditions or the
identification guides are available in Croft (1986)     effects of management; and a–c identify how
and various Institute of Freshwater Ecology publi-      many macrophyte species on the recording sheet
cations (www.ceh.nerc.ac.uk). Again, the environ-       were present. The greater the number, the better
mental protection authorities should be contacted,      the confidence in the results.
as such data may be available as part of their moni-       The MTR system has been used for monitoring
toring programmes.                                      water quality around effluent discharge points.
    As with standing waters, macrophytes may be         Reviews of survey results have shown that the
monitored to assess overall water quality (see          method is effective and efficient. It performed
Section 5.9 for advantages and disadvantages) as        best on river systems that were not already
well as their own condition as features of conserva-    enriched prior to the discharge being monitored
tion interest. Consequently, macrophytes have been      and worst on extremely enriched river systems in
widely used for monitoring the water quality of riv-    which one more discharge makes very little differ-
ers in Europe and the UK. The current technique for     ence. Scores are distorted if two sites being com-
this purpose being used in the UK is the mean trophic   pared are not similar in physical character and
rank (MTR) system (Holmes et al., 1999b) which devel-   when few species that are used to calculate scores
ops the earlier plant score system developed by the     are present. The method has given good results
Standing Committee of Analysts (1987).                  from clean oligotrophic rivers in south-west
    The MTR system is based on surveys of selected      England and the Lake District.
(usually common) aquatic macrophytes. These are            A brief review of general monitoring methods
assigned a number from 1 to 10 according to their       for macrophytes is provided in Bell (1996). At its
tolerance/preference for enriched or clean waters:      simplest, monitoring may focus on confirmation of
this is the species trophic rank (STR). Depending on    the presence of a particular River Community
the species cover value of listed taxa within a 100 m   Type. This can be easily carried out by using the
reach (recorded on a nine-point scale), a mean          standard method for river macrophyte surveys
trophic rank can be assigned. The method is             developed by the NCC for its national survey of
applied by surveying a 100 m length of river, pre-      river communities. Full details of this are given in
ferably by wading or a combination of wading and        Holmes (1983) and Boon et al. (1996a,b, 2002). In
walking along the banks in narrow rivers. Where         essence, the survey method involves recording
rivers are wadeable, or a boat can be used safely and   macrophytes at sites 1 km long (formed from two
effectively, the whole channel width should be sur-     contiguous 500 m lengths), situated 5–7 km apart.
veyed. For wide and deep rivers in which the cen-       Surveys include the entire channel and lower
tral channel is devoid of vegetation or cannot be       slopes of the banks, with separate records being
accurately surveyed because of its depth, turbidity,    made for macrophytes that are more or less perman-
etc., a strip 5 m wide down one side (ideally with      ently submerged and those that are typically sub-
little shading from trees) should be surveyed. If       ject to alternate inundation and exposure with the
water clarity is poor, a glass-bottomed bucket or       rise and fall of river levels. Terrestrial plants with
an underwater video camera should be used.              no special affinity for rivers are excluded from the
    Surveys should be carried out once or twice a       survey; although rare aquatic plants are recorded,
year between June and September. From the stand-        these are not used in the classification process. At
ard list of species, cover values are estimated on a    each site an estimate is made of the relative macro-
nine-point scale, with 1 being less than 0.1% cover     phyte abundance by using a simplified DAFOR-type
and 9 being more than 75% cover. MTR calculations       scale (1, rare; 2, occasional or frequent; 3, abundant
have three suffixes of confidence to assess the         or dominant) and a simple percentage cover scale
                                                                                 5.10 Rivers and streams   139



(1, <0.1%; 2, 0.1–5.0%; 3, >5%). Surveys are carried     scales, i.e. catchment or sub-catchment scale. It is
out by walking the banks and wading, or by boat          therefore primarily used as a broad-scale conserva-
for deeper rivers. However, the repeatability of this    tion evaluation tool, although evaluations are car-
technique and resulting consistency of classifica-       ried out by dividing rivers into a series of evaluated
tion is unknown, and it may be inadequate for all        catchment sections (ECSs). Monitoring of each ECS
but the most basic monitoring purposes. Certainly,       may be carried out on the basis of scores from the
the abundance and percentage assessments for each        model or individual SERCON attributes. However,
species are too subjective and crude for moni-           assessments of these attributes are to some extent
toring purposes. Vegetation composition monitor-         subjective and therefore variation between repeat
ing should instead be carried out by the                 surveys by different people may limit its value for
appropriate adaptation of quadrat and transect           monitoring compared with more objective and
techniques (see Section 6.3 for bankside vegetation      quantitative methods. A full SERCON assessment
and Part III, section 14.2.1, for aquatic vegetation).   is also time-consuming, but SERCON is currently
   A method for monitoring aquatic plant assem-          being revised; it is expected that version 2 will have
blages by rake or grapnel samples in deep, slow-         a slimmer variant as well as the full version.
flowing rivers is outlined in Section 6.3.2. Maps of     However, some SERCON attributes could form a
the cover of individual species may be appropriate       useful suite for monitoring.
for small stretches of shallow rivers. Mapping may          General monitoring of morphological attributes
be time-consuming, but it provides detailed and          of rivers can be carried out by using River Habitat
reproducible results that can justify the effort.        Survey methods (see Section 6.3.1.) More detailed
Further information on this technique is provided        and accurate monitoring of the extent of these
in Standing Committee of Analysts (1987) and             attributes will probably need to be carried out by
Wright et al. (1981).                                    the adaptation of other methods such as fixed-
   Other key features of river habitats include          point photography (Section 6.1.4), quadrats and
important morphological and hydrological attri-          transects (Section 6.4 and Part III, Section 14.2.1)
butes, such as channel width, depth, slope, capacity,    or by specialised monitoring techniques that can-
substrate type, flow velocity and flow rate (and its     not be described here.
seasonality) and the presence of various important          Monitoring the hydrological attributes of a river,
habitat features, such as riffles and pool sequences,    such as flow rates, is complex, time-consuming and
bars, meanders and waterfalls, etc. The effects of       expensive. However, sufficient information neces-
such attributes on river ecology are complex but         sary for basic conservation monitoring purposes is
these, together with biological data, have been          likely to be available from the environmental pro-
recently incorporated in a model that provides a         tection authorities.
comprehensive and integrated assessment of river            A summary of river attributes that require moni-
conservation value: SERCON (System for Evaluating        toring is provided in Table 5.13, together with
Rivers for CONservation). SERCON utilises existing       recommended methods for monitoring each.
habitat and species data for a range of river corridor   These are described more fully in Chapter 6.
attributes to apply classic conservation assessment
criteria, such as diversity, naturalness, representa-    Management requirements and external impacts
tiveness and rarity, but in a more rigorous manner       High human population density and the presence of
than has been done in the past (Boon et al., 1996a,b).   intensive agriculture directly affects river condition
   For the assessment of many river corridor attri-      and water quality. The frequency of rivers affected
butes SERCON depends on outputs from River               by, or at risk from, organic and chemical pollution
Habitat Surveys (Environment Agency, 2003) or,           from agriculture, domestic wastes and industry gen-
rarely, River Corridor Surveys (NRA, 1992) unless        erally increases in the lowlands (and in some areas
similar data are available from other sources.           towards the coast) as the land is more densely popu-
Generally, SERCON is intended to work at large           lated, agriculture is more intensive and industry is
140
      Table 5.13. A summary of the quality attributes providing an indication of the condition of rivers, and their recommended monitoring techniques

      Attribute                       Habitat properties                                        Monitoring technique

      Physical properties             Extent                                                    Aerial photography (Section 6.1.3) or satellite-based
                                                                                                remote sensing (6.1.2)
      River morphology                Channel width (bank, full and low flows), shape,          River habitat surveys (6.3.1) or fixed-point photography
                                      substrate type, presence of riffles, pools,               (6.1.4) for general surveys, quadrats (6.4.2 and 6.4.3) or
                                      meanders, water control structures, etc.                  transects (6.4.6) for detailed studies
      Hydrology                       e.g. water flow rates and depth                           Obtain from environmental protection authorities
      Water chemistry                 Nutrient status, pH, BOD, dissolved oxygen, toxic         Obtain from environmental protection authorities
                                      substances, etc.                                          or use mean trophic rank system (see text) or
                                                                                                macroinvertebrate indicators (see text)
      Vegetation extent and           Coverage of banks and watercourse                         Fixed-point photography (6.1.4), line transects or
      structure                       Vegetation height                                         quadrat methods (6.4 or 14.2.1) for detailed studies
      Vegetation composition          River community type                                      NCC method (see text) or NVC surveys (6.1.6) for detailed
                                                                                                studies
                                      Bankside and emergent vegetation species                  Line transects (6.4.6) or quadrat methods (6.4.2 and 6.4.3)
                                      abundance or richness
                                      Aquatic macrophyte species abundance or                   Grapnel surveys in slow moving deep water (6.3.2),
                                      richness                                                  quadrat or transect surveys (14.2.1) or mapping for
                                                                                                detailed studies in shallow water (see text)
                                                                                   5.11 Montane habitats    141



more common. This may also be exacerbated by             monitoring (especially in the use of boats), the cor-
water abstraction as this may lead to low flows,         rect use of safety equipment (e.g. life jackets when
which can contribute to high pollutant concentra-        working over or alongside deep water) and work-
tions and reduced dissolved oxygen concentrations.       ing in pairs or teams and never alone. Other safety
Overall pollution impacts may, however, be less in       aspects listed in Part I, Box 2.11, should also be
downstream sections on account of the dilution of        followed where appropriate.
pollutants by large volumes of water.                       Special care should be taken when working on
    In the upstream reaches and headwaters, pollu-       rivers that are liable to rapid changes in flow.
tion may occur from various types of discharge and
domestic waste. The ploughing and/or drainage of
moorland for forestry and agriculture can lead to        5.11     MONTANE HABITATS
high levels of silt and peat run-off, which increases    5.11.1      Surveying and monitoring
the turbidity levels of the water. Oligotrophic and                  requirements and methods
acid waters are also particularly susceptible to acid-
ification from acid deposition.                          Attributes for assessing habitat condition
    In addition to pollution and abstraction pro-        Montane sites include a range of habitats above the
blems, management actions may have detrimental           natural tree line. Ninety per cent of the UK resource
effects on the conservation interest of rivers. In       occurs in Scotland. Information on the monitoring
particular, flood defence measures such as bank          of many of the habitats present in the uplands is
strengthening and canalisation, removal of riparian      covered elsewhere (scrub in Section 5.1; grasslands
vegetation, dredging, and the installation of water      in Section 5.4; heathlands in Section 5.6; wetlands
control structures such as weirs, can have profound      in Section 5.7; streams and pools in Sections 5.9
impacts on river morphology and in turn on their         and 5.10; and blanket bogs in Section 5.12). Only a
biological interest. Fisheries management and            brief account is therefore provided here.
recreational activities can also have impacts               Key attributes of importance to montane habitats
through habitat modifications, disturbance and, in       are physical features such as bare cliffs, rocks, scree
the former case, the introduction of alien species or    and soil and the prolonged presence of snow
artificially increased populations of native fish.       patches. Montane vegetation types are rather dis-
                                                         tinct and often rare in the UK, such as alpine calcar-
                                                         eous grassland. Montane habitats commonly occur
5.10.2      Specific issues affecting the survey
                                                         in mosaics to form habitat complexes of particular
            and monitoring of habitat
                                                         collective importance for their flora. Species compo-
As with standing waters, the monitoring of rivers is a   sition can be very important, with many relict
specialised subject and therefore it is recommended      Arctic–alpine and endemic species occurring.
that specialist advice be obtained before planning       A summary of the key attributes that require mon-
and implementing a monitoring programme for              itoring is provided in Table 5.14, together with
these habitats. See Section 5.9 for a discussion of      recommended methods for monitoring each.
practical considerations regarding monitoring of
water bodies. As described above, routine monitor-       Management requirements and external impacts
ing of water quality and various hydrological factors    Montane vegetation is particularly vulnerable to
is carried out on many rivers by the environmental       heavy grazing (Section 7.1), accidental burning
protection authorities. Relevant data may therefore      (Section 7.2), erosion (Section 7.3) and air pollution.
be available for some sites.                             Montane habitats are known to be sensitive to air
   Safety is clearly a key consideration when carry-     pollution and acidification, so maintaining an
ing out fieldwork on rivers. Key safety measures         acceptable condition in the long term will require
that should always be followed include proper            pollution climates below the critical concentra-
training of personnel in safety aspects of aquatic       tions (SO2 10 mg mÀ3, NO2 30 mg mÀ3, NH3 8 mg mÀ3)
142      5 HABITAT REQUIREMENTS AND ISSUES



Table 5.14. A summary of the quality attributes providing an indication of the condition of montane habitats,
and their recommended monitoring techniques

Attribute          Habitat properties                      Monitoring technique

Physical           Extent                                  Phase I mapping (Section 6.1.5) with aerial
properties                                                 photography (6.1.3) for basic long-term monitoring
                                                           NVC surveys with quadrat sampling (6.1.6) for NVC
                                                           vegetation types
                   Exposed rock, scree and bare soil,      Aerial (6.1.3) or fixed-point (6.1.4) photography,
                   and snow lie                            quadrats (6.4.2) or transects (6.4.6)
                   Soil nutrients                          6.2.2
Composition        Characteristic communities              Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                           (6.1.6) where NVC communities are Notified Features
                                                           or important attributes
                   Species composition and                 Mini-quadrats (6.4.3)
                   richness
                   Presence/absence of                     Look–see or total counts (Part III, sections 15.2.1 and
                   typical/indicator species               15.2.2,) quadrats (6.4.2 and 6.4.3) or transects (6.4.6)


(English Nature, 1993). These aspects should there-            The complexity of some sites means that strati-
fore also be monitored though, as it may be diffi-          fied sampling will be required and will have to be
cult to monitor air pollution directly, data may            carefully designed to efficiently cover the range of
need to be drawn from wider-scale models.                   habitats and their localised variations. On account
                                                            of the large size of sites and the time required to
                                                            move between samples, multi-level sampling may
5.11.2       Specific issues affecting the
                                                            be appropriate (Part I, Section 2.3.)
             surveying and monitoring of habitat
                                                               If permanent quadrats are to be used, frost heave
It is recommended that uplands are monitored at 3           may result in loss of markers. Using good location
or 6 year intervals. They are almost always best            features, such as large boulders, and making
surveyed in July and August when the weather is             detailed measurements may be very helpful for
better and the vegetation fully developed. Access           relocating quadrats. Rock climbing bolts have been
may be difficult late in the season because of deer         used as markers but are now considered unsightly.
stalking or grouse shooting.                                Scree slopes are often mobile; permanent quadrats
    When planning surveys allow at least 1 day lost         are therefore inadvisable on this habitat.
to bad weather for each survey day. The logistics of           For assessing changes in many features of the
getting equipment into place can be very difficult;         vegetation, such as species richness and sward
often, cursory surveillance may be the best option.         height and cover, the techniques outlined in
Aerial photography (Section 6.1.3) may be cost-             Section 5.4 can be applied. The hanging quadrat
effective for large areas.                                  technique may be useful for recording quadrats
    Determining your location may be very difficult         on vertical surfaces (Rich & Matcham, 1995).
in wet weather, and especially on large uniform                Unfortunately, the NVC does not cover lichen
upland areas is virtually impossible from maps.             and bryophyte vegetation, and there is no work-
Global positioning systems (GPS) have distinct              able account of these vegetation types available.
advantages, despite their inaccuracies (although            James et al. (1977) give a preliminary conspectus
these are substantially improving over time).               of lichen communities, which may be of some
                                                                                          5.12 Blanket bog   143



use for some communities. The specialist upland              High water tables are essential for bogs to be
bryophytes and lichens also require expert                active (i.e. forming peat); most bogs have water
botanists.                                                tables near the surface except in drought condi-
   The persistence of snow patches varies from            tions. However, monitoring water tables alone is
year to year, but snow cover has usually gone by          not sufficient, even if it could be achieved in a
late July. The snow often acts to catch nutrients         meaningful way, and assessments of whether the
from wind-blown vegetation, and snow patches              bog is forming peat are also important. On account
are thus often relatively nutrient-rich compared          of the slow rate of peat accumulation (in the region
with surrounding ground.                                  of 10 cm over 100 years), it is not practicable to
   Alpine cliffs and rocks can support rich plant         assess this by measuring peat depth and rates of
communities. These habitats are dangerous to              peat accumulation. Instead, it is normally assumed
work on and often unstable. Roped-access work             from the active growth of Sphagnum. The presence
may be required for critical areas, but is very           of characteristic Sphagnum species can be used to
time-consuming. Particular care should be taken           infer the occurrence of active peat formation. In
when working in montane areas; all relevant safety        general, if the Sphagnum is healthy and growing,
recommendations outlined in Part I, Box 2.11,             the habitat should be in good condition. Other
should be strictly followed where appropriate.            typical species indicative of peat formation capabil-
                                                          ity are often locally important on blanket bogs, and
                                                          appropriate species may be selected on a site basis.
5.12      BLANKET BOG                                     For example, in the north and west Racomitrium
                                                          often replaces Sphagnum as the dominant bryo-
5.12.1      Surveying and monitoring
                                                          phyte, and Cotton Grass Eriophorum vaginatum is
            requirements and methods
                                                          an important peat-forming species on many high-
Attributes for assessing habitat condition                altitude bogs.
Under a number of conditions (such as suitable               In wet, humid climates the vegetation may be
rainfall–evapotranspiration regime and topog-             dominated by Sphagnum, but in drier conditions
raphy), blanket bogs often occur as the dominant          there is usually more Heather Calluna, Cross-
habitat type within extensive landscapes, and may         leaved Heath Erica tetralix, Eriophorum spp. and
form mosaics with other vegetation types; such            Deer Grass Trichophorum spp.. The significance of
extensive blanket bog landscapes are of particular        different species varies with altitude, longitude
importance. Habitat extent is therefore a key             and latitude. A significant proportion (perhaps
attribute.                                                more than 10%) of Sphagnum in the main bog com-
   The physical structure of bogs is also considered      munities is considered to indicate active peat for-
to define their condition and therefore there are         mation. Locally determined cover proportions
several structural attributes that should be moni-        should be derived from analysis of existing quadrat
tored. It has been suggested that blanket bogs            data and other historical observations. Similarly,
require at least 0.5 m of peat (NCC, 1990a) to separate   proportions of other species may be derived from
the vegetation communities from the underlying            existing data.
substrate and provide the appropriate hydrological           Certain NVC bog communities can also indicate
and chemical conditions for Sphagnum growth,              good, relatively undisturbed blanket bogs, depending
although many have on average 2–3 m of peat.              to some extent on geographical location (Rodwell,
Some blanket bogs may also have shallower peat            1991, vol. 2). The M17 Scirpus cespitosus–Eriophorum
areas because of their topography. Ideally, the           vaginatum blanket mire, often associated with the
peat mass should be intact, but most bogs have            M1 Sphagnum auriculatum bog pool community in
been damaged to some extent by cutting, drainage,         the pools or wettest areas, is the characteristic blan-
burning, grazing, erosion, afforestation or agricul-      ket bog type in oceanic parts of Britain, generally at
tural improvement (see below).                            low altitudes. The M18 Erica tetralix–Sphagnum
144    5 HABITAT REQUIREMENTS AND ISSUES



papillosum raised and blanket mire occurs over              A summary of attributes providing an indication
large areas in Caithness (A. Coupar, personal com-       of the condition of bog habitats and their recom-
munication) on cols and in depressions (Rodwell,         mended monitoring methods is given in Table 5.15.
1991) and has associated areas of the M2 Sphagnum           It is recommended that bogs be monitored at
cuspidatum/recurvum bog pool community. The              3 to 6 year intervals.
M15 Scirpus cespitosus–Erica tetralix wet heath may
occur in naturally better drained areas at the bog       Management requirements and external impacts
margin. The M19 Calluna vulgaris–Eriophorum vagi-        The management of bogs can markedly affect their
natum blanket mire is dominant on high-level             quality. Grazing, burning and drainage can all
blanket bog. NVC-based survey and monitoring             damage the vegetation, and even trampling can
is, however, by itself too general to differentiate      affect Sphagnum cover. Under natural conditions
between good and poor bog communities (such as           the blanket bogs are likely to have been lightly
M18), being particularly insensitive to changes in       grazed. Information on indicators of drying, burn-
structural attributes (e.g. as a result of grazing).     ing, grazing and trampling are given in MacDonald
Interpretation of the presence of so-called              et al. (1998a,b).
‘degraded’ communities should also be under-                Air pollution (especially sulphur-based pollu-
taken with care.                                         tion) is known to damage Sphagnum communities
   Woody species (except Salix repens, Betula nana,      so maintaining an acceptable condition in the long
Vaccinium spp. and Bog Myrtle Myrica gale) are           term will require pollution regimes below the criti-
generally not considered to be natural components        cal concentrations (SO2 10 mg mÀ3, NO2 30 mg mÀ3,
of blanket bogs. Therefore, invasion by other birch,     NH3 8 mg mÀ3) (English Nature, 1993).
willow or other woody species should be moni-               Stoneman & Brooks (1997) provide management
tored, although this may indicate surface flushing       advice for blanket bogs.
(unlikely) and the absence of grazing and burning,
rather than a drying out of the bog.
                                                         5.12.2     Specific issues affecting the survey
   Lastly, structure is very important in blanket bogs
                                                                    and monitoring of habitat
in determining hydrological functioning and the
presence of different features and niches. At a          Bogs should be surveyed from July to September
large scale, individual blanket bog units (mesotopes)    when their vegetation is fully developed, although
occur in a range of topographical positions such as      all attributes other than those pertaining to
watersheds, valley sides, spurs and saddle mires.        deciduous species can be monitored at other
These often form inter-related complexes (macro-         times of year.
topes) of greater interest than individual mesotopes.       Wet bog habitats can be very sensitive to tram-
   Within each mesotope it is also natural to have       pling and there is therefore a considerable risk of
variation in the communities present related to var-     damage from monitoring activities. Methods
iations in topography, hydrology and substrate fea-      should be chosen appropriately, according to local
tures, including transitions to vegetation on            conditions. In general, permanent quadrat or trans-
mineral soils. Pool and ridge patterning in some of      ect methods should be avoided and disturbance of
the northern bogs is of particular interest. All bogs    the bog surface kept to a minimum. In particular,
display some form of surface patterning, which           on vulnerable habitats, automated or remote moni-
represents an important source of biodiversity.          toring techniques should be used where appropri-
   Most relatively undisturbed bog surfaces usually      ate. The number of sampling locations should also
show distinctive fine-scale variation (microtopes)       be kept to the minimum necessary and the interval
with small drier hummocks and wetter hollows             between sampling occasions should be as long as
related to growth of Sphagnum and other plants,          possible.
although the large-scale features may be highly             As mentioned above, bog habitats can be very
restricted or locally frequent.                          extensive and this raises a number of problems. In
                                                                     5.13 Maritime boulders, cliffs and slopes   145



Table 5.15. A summary of the quality attributes providing an indication of the condition of blanket bog,
and their recommended monitoring techniques

Attribute       Habitat properties                       Monitoring technique

Physical        Extent                                   Phase I mapping (Section 6.1.5) with aerial photography
properties                                               (6.1.3) (or satellite-based remote sensing (6.1.2) for very
                                                         large sites). NVC surveys with quadrat sampling (6.1.6)
                                                         for detailed studies
                Water table                              Dipwells or WALRAGS (6.2.1)
                Bare peat                                Aerial photography (6.1.3), quadrats (6.4.2 and 6.4.3) or
                                                         transects (6.4.6)
                Peat depth                               Soil augers or levels from mineral ground
Composition     Characteristic communities               Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                         (6.1.6) where NVC communities are Notified Features
                                                         or important attributes
                Species composition and richness         Mini-quadrats (6.4.3) (quadrat size depending on scale
                                                         of vegetation)
                Presence–absence of typical or           Look–see or total counts (Part III, Sections 15.2.1 and
                indicator species                        15.2.2), quadrats (6.4.2 and 6.4.3) or transects (6.4.6).
Structure       Landscape and habitat mosaics            Aerial or fixed-point photography (6.1.3 and 6.1.4)
                Structural features (pool and            Conventional quadrats (6.4.2) or transects (6.4.6).
                hummock, and/or hollow and               Photographs from vantage points may be useful (6.1.3
                ridge as appropriate)                    and 6.1.4)
Dynamics        Peat formation                           Quadrat- (6.4.2 and 6.4.3) or transect- (6.4.6) based
                                                         assessment of indicator species


particular, vehicular access is likely to be impossi-       compartments for monitoring purposes (Part I,
ble, and is in any case undesirable because of the          Section 2.1.7.)
potential for lasting, if local, damage. A consider-           The size, remoteness and extreme weather con-
able amount of time is likely to be spent walking to        ditions of many bog sites also raise potentially
monitoring locations. Sampling strategies should            significant safety problems. Safety protocols (see
therefore be designed with this in mind, using              Part I, Box 2.11) should therefore be strictly fol-
techniques such as multi-stage sampling (Part I,            lowed. In particular, personnel should not carry
Section 2.3.4). Automated monitoring techniques             out monitoring alone at remote sites and should
(e.g. for water levels) may also be cost-effective          be properly equipped and trained.
but should only be used in very specific circum-
stances, e.g. the monitoring of a consented, but
potentially damaging activity. Similarly, expensive         5.13       MARITIME BOULDERS, ROCKS,
remote methods such as aerial photography (or for                      CLIFFS AND SLOPES
very large sites, satellite-based sensing) may be
                                                            5.13.1       Surveying and monitoring
financially viable.
                                                                         requirements and methods
   Variations in topography, hydrology and other
physical properties can lead to considerable hetero-        Attributes for assessing habitat condition
geneity in bog habitats. As a result, it may often be       The 4000 km of sea cliffs in the UK are a major
necessary to subdivide large and complex sites into         nature conservation resource of international
146    5 HABITAT REQUIREMENTS AND ISSUES



importance. They are important as extensive areas of        Management requirements and external impacts
natural habitat, which is often relatively little           An account of sea cliffs and their management
affected by human activity. Some cliffs have import-        has been presented by Mitchley & Malloch (1991).
ant geological exposures, but the geological, botani-       The main management tools are grazing, mowing
cal and zoological interests may not coincide. Cliff        and burning, although much of the lower parts
habitat, as defined here, includes all the vegetation       of cliffs are inaccessible and do not need to be
types in the NVC sea cliff vegetation chapter of            managed.
Rodwell (2000, vol. 5), from the vegetation of vertical         The main problem in maintaining the area of sea
sections at the base of a cliff to the flatter top parts.   cliff is that the upper edge is usually valuable farm-
   The extent of the sea cliff habitat is not easy to       land, which is expensive to use as a replacement for
measure because vertical projections are not repre-         areas lost to erosion. The most practical measure to
sented well on maps, although in practice often the         maintain area may be to ensure that no further
upper parts of cliffs are slopes and only the lower         truncation of the inland margin of the cliff vegeta-
parts are vertical. Natural erosion of cliffs results in    tion occurs, and accept loss on the seaward side.
regular loss of area. Rather than monitoring loss of            The main threats to sea cliffs are agricultural activ-
material to the seaward side (which could be moni-          ities, tourism, and coastal development and protec-
tored by photographs from the sea), it may be best          tion. Coastal protection works or uncontrolled
to concentrate on monitoring loss to agriculture,           dumping may prevent erosion and affect coastal pro-
etc. on the inland side.                                    cesses, leading to loss of interest. There may be some
   Coastal sea cliff vegetation and species composi-        risk from accidental fires, although this has probably
tion are important factors to monitor, as they are          decreased with the cessation of stubble burning. Oil
the basis of this unique habitat. The vegetation            pollution may be a serious risk in some places close
often shows marked zonation depending on geol-              to shipping lanes; the lower cliff communities may
ogy, erosion, geographical location and especially          be seriously affected by oil deposition.
the degree of exposure to wind and salt spray. The
lowest zones are primarily occupied by lichens and
                                                            5.13.2       Specific issues affecting the survey
some bryophytes, which grade into higher plant
                                                                         and monitoring of habitat
vegetation above. In some exposed sites the clifftop
vegetation grades into maritime heath, grassland            It is recommended that cliffs be monitored at 3 or
and scrub, which form an integral part of the cliff         6 year intervals. They can be surveyed between May
habitat, but in many cases they are truncated by            and October, although if annual species are import-
agriculture or development inland. Soft cliffs often        ant they should be surveyed in early summer. This
display a much wider range of vegetation than that          may conflict with the bird nesting season.
included in the NVC maritime cliffs section, and                Salt deposition during summer storms may have
much of it is of nature conservation interest.              a dominant influence on the zonation of the vege-
   The most important influence on the habitat is           tation and cause the death of some areas. It may be
the amount of salt spray, which is strongly influ-          worth monitoring salt deposition if damage is also
enced by situation and exposure. On the accessible          expected from herbicides etc. used on adjacent
upper parts of the cliff top, where salt deposition is      farmland.
weakest, structure and composition may be                       Unfortunately, the NVC does not cover lichen
strongly affected by management, especially graz-           and bryophyte vegetation, which is predominant
ing. The soil sodium : organic content ratio is a           at the lowest levels on cliffs; the best available
useful yardstick for assessing the influence of             account of these vegetation types available is
spray (Rodwell, 2000). Table 5.16 gives a summary           James et al. (1977). The specialist maritime bryo-
of the attributes providing an indication of the            phytes and lichens also require expert survey.
condition of maritime habitats and their recom-                 As vegetation composition is often strongly
mended monitoring methods.                                  related to soils and topography, careful stratification
                                                                                 5.14 Shingle above high tide    147



Table 5.16. A summary of the quality attributes providing an indication of the condition of maritime habitats,
and their recommended monitoring techniques

Attribute       Habitat properties                      Monitoring technique

Physical        Extent                            Phase I mapping (Section 6.1.5) with aerial photography
properties                                        (6.1.3) for broad long-term changes
                                                  NVC surveys with quadrat sampling (6.1.6) for detailed
                                                  studies
                Soil salinity                     Soil analysis (6.2.2)
Composition     Characteristic communities        Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                  (6.1.6) where NVC communities are Notified Features or
                                                  important attributes
                Species composition and richness Mini-quadrats (6.4.3)
                Presence–absence of typical or    Look–see or total counts (Part III, Sections 15.2.1 and
                indicator species                 15.2.2), quadrats (6.4.2 and 6.4.3) or transects (6.4.6)
Structure       Zonation between vegetation types Transects (6.4.6), fixed-point (6.1.4) or aerial
                                                  photography (6.1.3)
                Pattern within vegetation types   Quadrat sampling (6.4.2 and 6.4.3), fixed-point
                                                  photography (6.1.4)


may be required to ensure that the range of quality
and extent is adequately covered.                           5.14       SHINGLE ABOVE HIGH TIDE
   Generally, the upper limit of cliff vegetation is
                                                            5.14.1       Surveying and monitoring
marked by a change to agriculture or develop-
                                                                         requirements and methods
ment, but in some localities there may be natural
transitions to moorland. The influence of the sea           Attributes for assessing habitat condition
and salt declines with increasing distance from             The major vegetated shingle structures of Britain
the sea and decreasing exposure, and it may be              have been reviewed in detail by Randall (1989) and
difficult to define the inland edges. If permanent          Sneddon & Ranwell (1993, 1994). There are esti-
markers are required, these are best established at         mated to be 6115 ha in the UK. These may be of
the landward edge on sites with significant                 considerable interest for their geomorphology in
erosion.                                                    addition to their distinctive plant and animal
   Safety considerations may prevent detailed map-          communities.
ping of vegetation for monitoring. There are severe            In general the extent of shingle should be rela-
practical difficulties in mapping vertical cliffs,          tively easy to monitor on the inland edge, though
especially on crumbly rocks. Rope work may be               the shore edge may be more dynamic. The develop-
required and this must be carried out by adequately         ment of a shingle beach is dependent on a supply of
trained personnel using appropriate equipment.              sediment and waves, winds and tidal currents.
The hanging quadrat technique may be useful for             Much material may be lost or supplied naturally
recording quadrats on vertical surfaces (Rich &             during storm episodes, but some loss may also
Matcham, 1995). For assessing changes in many               occur through shingle extraction or indirectly as a
features of the vegetation, such as species richness        result of coastal protection elsewhere. The supply
and sward height and cover, the techniques out-             of shingle to the site by natural processes is best
lined in Section 5.4 can be applied. For dwarf              monitored from continued measurements of ero-
shrub heath, see Section 5.6.                               sion or accretion at fixed points and may need to be
148    5 HABITAT REQUIREMENTS AND ISSUES



assessed over decades. Mass movement of shingle         Management requirements and external impacts
to or from a site is very difficult to quantify.        Most shingle sites do not require management.
   The vegetation types present are key attributes.     However, human pressures such as coastal defence
Shingle habitats include open pioneer stages close      works, development, shingle extraction and
to the sea, and grassland, heaths, scrub and moss-      recreational activities may need some monitoring.
and lichen-dominated vegetation on very old,               Shingle extraction has affected some sites and is
stable shingle further inland. Near the shore the       probably the single most damaging activity.
vegetation is typically open with many maritime         Military and tourism activity has also damaged
species. These decrease in abundance away from          some sites. Grazing may result in the loss of some
the shore as inland species increase. The NVC types     sensitive taxa, but most sites are ungrazed. Oil pol-
have been revised by Sneddon & Ranwell (1993),          lution may occur at some sites.
whose classification provides more detail than the         Shingle banks may be coastal defence features in
NVC but requires rationalisation. Some sites may        their own right, and they are often maintained by
be important for lichens; undisturbed shingles may      supplies of shingle from further up the coast.
have their own distinctive communities ( James          Coastal defence works elsewhere may therefore
et al., 1977), which may require monitoring in          starve some sites of their supplies.
their own right.
   Salinity, hydrology, and the stability, morph-
ology and composition of the shingle are principal      5.14.2      Specific issues affecting the survey
factors determining vegetation composition.                         and monitoring of habitat
Strong patterning within the vegetation may             It is recommended that shingle sites be monitored
occur related to shingle ridge structure, with dis-     every 3 years. Measurements of growth or loss of
tinct lines of Crambe or Glaucium on the shingle        shingle should be made every decade. Sites are best
ridges. These patterns may become blurred as            monitored between May and October, but nesting
humus builds up in the shingle and is colonised         birds may restrict access. Mapping can be difficult
by additional plants. For invertebrates this struc-     in uniform shingle structures, but is often rela-
turing of the vegetation and small-scale mosaics        tively traightforward provided details of topography
are more important than its composition.                are available. Aerial photographs can be invaluable.
   There may be strong zonation from the shore to
the inland edge that is of considerable interest. The
vegetation of the foreshore is strongly controlled      5.15     SAND DUNES AND STRANDLINE
by the environment and only physical damage will                 VEGETATION
markedly affect it. Chance determines which spe-
                                                        5.15.1     Surveying and monitoring
cies colonise the foreshore. Transitions to inland
                                                                   requirements and methods
communities are often truncated by anthropogenic
activities, or shingle communities may grade into       Attributes for assessing habitat condition
rocky or sandy habitats. Sites with a range of com-     Sand dune vegetation includes strandlines, dunes,
munities, including pioneer communities, are            dune slacks, dune heath and scrub. The different
especially valuable.                                    types form a complex, dynamic, sensitive ecosys-
   Shingle sites are often associated with other spe-   tem capable of rapid change, and each part creates
cial interest habitats, such as lagoons, sand dunes     its own problems for monitoring. Because of the
and saltmarshes, and the transitions between them       dynamic nature of sand dunes and their associated
can be of interest. The hydrology is often important    vegetation, long-term (10 year) views on the
for lagoons and saltmarshes. Table 5.17 sum-            amount of each habitat and natural succession
marises the attributes indicating the condition of      between them should be taken.
shingle, together with recommended techniques              There is a significant amount of information
for their monitoring.                                   available on the extent and composition of sand
                                                                   5.15 Sand dunes and strandline vegetation    149



Table 5.17. A summary of the quality attributes providing an indication of the condition of shingle,
and their recommended monitoring techniques

Attribute         Habitat properties                        Monitoring technique

Physical          Extent                                    Phase I (Section 6.1.5), NVC surveys with quadrat
properties                                                  sampling (6.1.6), fixed-point (6.1.4) or aerial
                                                            photography (6.1.3)
                  Hydrology                                 Piezometers, dipwells or WALRAGS (6.2.1)
                  Salinity                                  Soil analysis (6.2.2) and water chemistry analysis
                                                            (not covered)
                  Topography and land loss                  Level surveying or fixed-point height surveys
                                                            (not covered), aerial photography (6.1.3)
Composition       Characteristic communities                Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                            (6.1.6) where NVC communities are Notified Features
                                                            or important attributes (see also Sneddon & Ranwell
                                                            (1993) community types)
                  Species composition and richness          Mini-quadrats (6.4.3)
                  Presence/absence of                       Look–see or total counts (Sections 15.2.1 and 15.2.2),
                  typical/indicator species                 quadrats (6.4.2 and 6.4.3) or transects (6.4.6)
                  Zonation between vegetation types         Transects (6.4.6), fixed-point (6.1.4) or aerial photo-
                                                            graphy (6.1.3)
Structure         Pattern within vegetation types           Quadrat sampling (6.4.2 and 6.4.3), fixed-point
                                                            photography (6.1.4)




dunes (see, for example, Dargie, 1993). Although             monitoring, and different attributes will be required
the overall extent is relatively easy to monitor,            for different parts of the dune system.
the proportions of NVC types within each system                 The percentage cover of vegetation usually
can be very difficult to follow reliably as a result of      increases inland and the extent of bare sand
the intrinsic difficulties of surveying a complex            decreases. Bare sand can give a good indication of
habitat, the natural rate of change and the natural          the likely stability of dunes and indicate grazing or
continual gradation between the dune commu-                  public pressure, but occasional erosion of stabi-
nities. The typical standardised succession from             lised dunes may occur naturally and is central to
strandline to yellow dune to grey dune to fixed              long-term maintenance of slacks.
dune, and to dune heath in some sites, can be used              Sand dunes decrease in salinity and pH and
as a framework for understanding the dynamics of             increase in organic content with increasing dis-
the system, but the whole sequence is rarely seen in         tance from the sea. Nutrients are usually low
practice. Typically the vegetation patterns are              throughout except on the strandline. Soil analysis
strongly related to topography, soil pH, water               can therefore provide valuable information
table, nutrient availability and grazing. There are          directly relevant to the vegetation types. The
usually extensive mosaics forming complex pat-               water table is also very important for determining
terns across the dunes, and within each part there           the distribution of slacks and their associated vege-
may also be mosaics of vegetation. These complex-            tation, and may be worth monitoring if water
ities mean that a very clear set of requirements             abstraction is increasing. Water will vary in salinity
must be set out before attributes are selected for           depending on the local hydrology.
150      5 HABITAT REQUIREMENTS AND ISSUES



   Given the importance of grazing to dune sys-           recolonisation from a few permanent major
tems, it is worthwhile monitoring some aspects of         donor sites or as a patchwork mosaic of extinction
grazing, such as stocking densities, at the same          and recolonisation, and may require monitoring
time as other features (see Section 7.1). A summary       annually (a long-term view of the occurrence of
of attributes giving an indication of the condition       these must also be taken, such as over 25 years).
of sand dune and strandline habitats and their            Different parts of the dune systems may need to be
recommended monitoring methods is provided in             monitored at different times of year; for example,
Table 5.18.                                               slacks and strandlines are best monitored in July
                                                          and August, whereas yellow dunes are best moni-
Management requirements and external impacts              tored in May and June while annual plant species
The main management tools on dunes are grazing            are still present.
and scrub clearance. The structure of individual             As the composition of dune vegetation is often
stands is largely determined by grazing, the inten-       strongly related to soils and topography, careful
sity of which is site-specific and requires careful       sampling stratification may be required to ensure
adjustment. Undergrazing results in rank grass-           that the full range of communities is adequately
land and development of scrub, whereas heavy              and efficiently covered. For assessing changes in
grazing results in few flowers, poaching, erosion         many features of the vegetation, such as species
and uniform turf. Both Rabbit Oryctolagus cuniculus       richness and sward height and cover, the techni-
grazing and low nutrient concentrations are               ques outlined in Section 5.4 can be applied. For
important for the maintenance of diversity on             dwarf shrub vegetation see Section 5.6, and for
some dune systems.                                        slacks much of the fen section (5.7) will be relevant.
   Oil pollution episodes can have dramatic effects          Orientation on dune systems while monitoring
on strandlines on account of both the toxic effects of    can be extremely difficult, even with good topo-
the oil and the effects of the clean-up operation. Tidy   graphic maps. Permanent markers are notoriously
beach campaigns (for the prestigious ‘Blue Flag’          difficult to re-find and frequently become under-
awards) can also significantly damage strandline          mined or buried by sand.
habitats if cleaning is carried out mechanically, but
some cleaning of human rubbish by hand must be
made acceptable in some sensitive areas: unsightly        5.16      SALTMARSH
rubbish needs to be removed in order to improve           5.16.1      Surveying and monitoring
habitat condition (Llewellyn & Shackley, 1996).                       requirements and methods
   The major threats to dune systems are coastal
protection, tourism, golf courses, afforestation,         Attributes for assessing habitat condition
land claim for agriculture, sand extraction, military     There are estimated to be about 44 370 ha of salt-
use and access roads (Ranwell, 1972; Doody, 1985).        marsh in Britain, occupying about 10% of the coast-
The major causes of erosion on many dune systems          line (Burd, 1989).
are human feet and vehicles. Coastal defence works           Saltmarshes are dynamic habitats; natural
may also affect the supply of sediments and alter         change in extent is to be expected. They may be
coastlines, with knock-on effects for the occurrence      subject to periods of sediment erosion or accretion.
of these habitats. All these may require monitoring.      Sediment movement patterns may be quite com-
                                                          plex. Larger saltmarshes are intrinsically more
                                                          valuable than smaller ones because of the
5.15.2      Specific issues affecting the survey
                                                          increased range of habitats within them and the
            and monitoring of habitat
                                                          lesser disturbance occurring at the upper edges.
It is recommended that dune systems be moni-              The development of small steps or edges at the
tored at 3 or 6 year intervals. The strandline com-       outer margin of the marsh is usually an obvious
munities act as metapopulations, with repeated            sign of erosion. Accretion can be monitored by
                                                                                               5.16 Saltmarsh   151



Table 5.18. A summary of the quality attributes providing an indication of the condition of sand dunes
and strandline vegetation, and their recommended monitoring techniques

Attribute         Habitat properties                       Monitoring technique

Physical          Extent                                   Phase I mapping (Section 6.1.5) with aerial
properties                                                 photography (6.1.3)
                                                           NVC surveys (6.1.6) with quadrat (6.4.2) or belt
                                                           transect (6.4.6) sampling or fixed-point (6.1.4)
                                                           photography
                  Topography and sand                      Level surveying or fixed-point height surveys
                  accumulation/erosion                     (not covered), aerial or fixed-point photographs
                                                           (6.1.3, 6.1.4)
                  pH and nutrient content of sand          Soil analysis (6.2.2)
                  Water table                              Dipwells (6.2.1)
                  Ground-water salinity                    Conductivity meters (6.2.2)
Composition       Characteristic communities               Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                           (6.1.6) where NVC communities are Notified Features
                                                           or important attributes
                  Species composition and richness         Mini-quadrats (6.4.3)
                  Presence–absence of typical or           Look–see or total counts (15.2.1, 15.2.2), quadrats
                  indicator species                        (6.4.2, 6.4.3) or transects (6.4.6)
Structure         Percentage bare sand                     Quadrats (6.4.2–6.4.4), transects ( 6.4.6) or aerial
                                                           photography (6.1.3)
                  Tidal litter                             Quadrats (6.4.2–6.4.4) or transects (6.4.6)
Dynamics          Zonation                                 Transects (6.4.6), fixed-point (6.1.4) or aerial (6.1.3)
                                                           photography


measuring the increased height of a marsh relative          inland end of coastal-linked lakes, lochs and estu-
to fixed points, or the extension of its outer edge.        aries and natural communities inland are valuable,
Extent is thus a key factor in monitoring but is            but sites have often been truncated by sea walls,
subject to long-term natural changes.                       land reclamation or agriculture.
   Saltmarshes are quite complex habitats and pro-             Many saltmarshes are dissected by small creeks
vide a range of attributes that can be measured,            and channels, which provide microhabitats within
depending on the monitoring objectives. Natural             more uniform areas of marsh. The upper levels of
changes in species and vegetation may occur,                ungrazed or lightly grazed marshes are usually
coupled with changes in the sediments. It should            relatively rich in species, at least partly as a result
be possible to monitor the range of vegetation              of the range of microhabitats present, but lower
types present on saltmarshes fairly simply; most            marshes are intrinsically relatively species-poor.
sites have been mapped by using a simplified vege-          Some inland species may also occur near the top
tation survey (Burd, 1989). Species composition can         of weakly saline marshes. Salt pans and small pools
also be surveyed for selected species.                      within the marsh are an intrinsic part of many
   A key attribute is the vegetation zonation, deter-       marshes and also add diversity. Structural diversity
mined by tidal submergence. The zonation is                 within the vegetation may be very important for
usually simple to observe and map by using trans-           invertebrates. Variation in the salinity of the sedi-
ects. Transitions to freshwater swamps at the               ments adds to the floristic diversity.
152    5 HABITAT REQUIREMENTS AND ISSUES



Table 5.19. A summary of the quality attributes providing an indication of the condition of saltmarshes, and their
recommended monitoring techniques

Attribute         Habitat properties                        Monitoring technique

Physical          Extent                                    Phase I mapping (Section 6.1.5) with aerial
properties                                                  photography (6.1.3)
                                                            NVC surveys (6.1.6) with quadrat (6.4.2) or belt
                                                            transect (6.4.6) sampling or fixed-point (6.1.4)
                                                            photography
                  Saline sediments or water                 Conductivity meters, soil analysis (6.2.2)
                  Physiography (salt pans,                  Physical mapping, aerial or fixed-point photographs
                  creeks, etc.)                             (6.1.3, 6.1.4)
                  Organic litter                            Physical mapping, aerial or fixed-point photographs
                                                            (6.1.3, 6.1.4)
Composition       Characteristic communities                Quadrats (6.4.2) or transects (6.4.6), with NVC analysis
                                                            (6.1.6) where NVC communities are Notified Features
                                                            or important attributes
                  Species composition and richness          Mini-quadrats (6.4.3)
                  Presence–absence of typical or            Look–see or total counts (Part III, Sections 15.2.1 and
                  indicator species                         Section 15.2.2), quadrats (6.4.2 and 6.4.3) or transects
                                                            (6.4.6)
Structure         Zonation                                  Transects (6.4.6)
Dynamics          Accretion on existing marsh               Level surveying (not covered)
                  Tidal inundation                          Observation at peak high tides, water level monitoring


   Deposits of organic litter from vegetation are              The main threats are erosion and land reclama-
typical of the strandline on the upper shore, and           tion, which can be monitored from changes in
indeed contribute to the nutrient balance of some           extent, heavy grazing, which can be monitored
upper-shore communities (e.g. SM24 Elytrigia ather-         from the vegetation, and pollution. Turf cutting
ica saltmarsh). These deposits are of natural occur-        may cause damage locally but is generally sustain-
rence, but are now often supplemented with much             able. Oil pollution is generally damaging to salt-
human-generated rubbish and flotsam. A summary              marshes, at least in the short term, although some
of the attributes providing an indication of the            species are surprisingly tolerant. Natural degradation
condition of saltmarshes and their recommended              of oil is preferable to removal, as it may cause less
monitoring methods is given in Table 5.19.                  damage to the marshes. The implementation of man-
                                                            aged retreat for coast protection works may result in
Management requirements and external impacts                increased areas of upper marsh in the future.
In general, management of saltmarshes is restricted
to grazing, which can greatly modify the vegetation
                                                            5.16.2        Specific issues affecting the survey
structure and species richness. Saltmarshes import-
                                                                          and monitoring of habitat
ant for plants are probably best left ungrazed or
lightly grazed, but those that are of interest for          It is recommended that saltmarshes be monitored
birds may be grazed more heavily. Heavy grazing             at 6 year intervals. Longer-term studies may be
tends to result in poor vegetation zonation. A histor-      needed to assess loss and gain due to natural
ical view of the grazing regime should be taken.            changes in the coastline. They can be monitored
                                                                                           5.16 Saltmarsh    153



throughout the season, although if identification          Changes in the sea level, coupled with isobatic
of some species such as Atriplex and Salicornia is      rebound, may cause longer-term changes in salt-
required this is best done in August to September.      marshes. Local monitoring data may therefore
   As saltmarsh composition is variable, careful        need to be interpreted against these changes,
sampling stratification may be required to ensure       which will have to be extrapolated from the few
that the full range of communities is adequately and    sea-level monitoring sites.
efficiently covered. If permanent markers are to be        As a result of the tidal nature of saltmarshes and
used, it should be remembered that relocation can       the presence of creeks (which are often deep,
be difficult because of mud deposition. For assessing   muddy and complex), particular care must be
changes in many features of the vegetation, such as     taken to observe safety procedures when monitor-
species richness and sward height and cover, the        ing. Refer to Part I, Box 2.11, for details of the safety
techniques outlined in Section 5.4 can be applied.      precautions that should be followed.
6       *    Methods for surveying habitats




6.1         GENERAL HABITAT SURVEY AND                   available, which cover both imaging and non-
            MONITORING METHODS                           imaging systems. This section covers only imaging
                                                         systems. The principal differences between these
The methods described in Section 6.1 may be
                                                         systems relate to their:
applied to the surveying and monitoring of most
habitat types. Section 6.1.1 provides an overview of     *   modes of data collection (e.g cameras, scanners,
remote sensing technology, which includes both               radars etc.);
satellite-based remote sensing (Section 6.1.2) and       *   storage media (film or digital); and the
aerial photography (Section 6.1.3). Remote sensing,      *   platforms from which the instrument operates
Phase I habitat mapping (Section 6.1.5) and                  (aircraft or satellite).
National Vegetation Classification (NVC) surveys
                                                         The optimum data source for any project will
(Section 6.1.6) are principally survey techniques
                                                         depend upon the user’s requirements. To assess
for mapping and/or quantifying the extent of dif-
                                                         the suitability of different sources of imagery, the
ferent habitats at a variety of scales. This may be
                                                         general principles that govern their operation are
carried out for a number of different purposes:
                                                         outlined below.
*   audits of habitat resources;
*   the production of maps for management plans;
    and                                                  Data collection
*   general recording of changes in landscapes and       All remote sensing systems depend upon differ-
    habitats, e.g. to document the result of land-use    ences in the way in which ground objects interact
    changes or management practices.                     with solar radiation. We can tell the difference
                                                         between one object and another, and also infer
Such methods may also be used for basic monitor-         something about an object’s properties by the
ing of the presence, extent and distribution of habit-   way in which it reflects, transmits or radiates this
ats. Knowledge of the distribution and extent of         radiation across different parts of the electro-
habitats and vegetation types is useful for identify-    magnetic spectrum. Recording these variations in
ing site features and their approximate boundaries,      the visible parts of the spectrum can be achieved
defining monitoring units, defining homogeneous          photographically or electronically; however, varia-
strata for stratified random sampling and locating       tions at non-visible wavelengths require the use of
samples within defined habitats of strata.               electronic sensor technology.
                                                            Photography uses chemical reactions on the sur-
                                                         face of a light-sensitive film to record energy vari-
6.1.1        Remote sensing principles
                                                         ations within a scene. Most other remote sensing
The term ‘environmental remote sensing’ covers           instruments use sensors to detect these energy vari-
all means of detecting and measuring environmen-         ations, which are then converted to a digital read-
tal conditions from a distance. There is a huge          ing. Whereas a camera records an instantaneous
variety of remote sensing instruments currently          image across a whole field of view, most airborne

# RPS Group plc and Scottish Natural Heritage 2005.
                                                          6.1 General habitat survey and monitoring methods       155



or satellite-borne scanning systems depend upon             of radiation reflected at different spectral wave-
the forward motion of the platform to build up an           lengths varies for each vegetation type. If the ‘spec-
image from a series of scanned lines (these instru-         tral signature’ for a particular vegetation type is
ments are often called line scanners). As the instru-       known, then it is possible to identify all other
ment moves over the ground, the sensor readings             areas of the image with the same properties. This
vary according to the amount of light reflected back        is called digital image classification and there is a
from the ground and an image is built up line by            wide range of statistical pattern recognition techni-
line. Through geometrically controlling these read-         ques that support this type of digital image analysis.
ings it is possible to build up an image of light           However, the effectiveness of these techniques
variation over a geometric grid. Each grid cell, called     always depends upon the extent to which objects
a pixel (picture element), has a digital number that        can be differentiated on the basis of spectral and
corresponds to the amount of light reflected from           textural properties. Often it is not possible to distin-
the area on the ground covered by the pixel within a        guish between different species of plant, for exam-
specified part of the light spectrum. These digital         ple. Again the user must be aware that it may not be
numbers can either be displayed on a computer               possible to use remotely sensed imagery to inter-
screen or by using a specialised film recorder to           pret traditional ecological classification systems
produce a photographic image.                               (e.g. those based on plant types, cover and
   The size of the pixel determines the spatial             abundance).
resolution of digital imagery. This is in turn driven
by the height of the instrument above the ground            Resolution trade-offs
and the focal length of its lens system. Satellite-         One very important advantage of satellite-borne
borne instruments tend to provide digital imagery           sensing systems over airborne ones is that they
with a spatial resolution greater than 5 m (i.e.            provide the opportunity for regular coverage of an
objects smaller than 5 m can be distinguished).             area at relatively low cost. They are, therefore,
Airborne systems have a much higher potential               potentially attractive for monitoring purposes.
resolution but they tend to provide one-off cover-          The resolution and frequency of coverage is set by
age. With all remote sensing systems there is the           the satellite orbit and the viewing geometry of the
trade-off between spatial resolution and fre-               instrument (e.g. wide-angled or telephoto).
quency of coverage: no system provides high spa-                Geostationary satellites (e.g. Meteosat) are pos-
tial resolution with frequent coverage, although            itioned over the equator at altitudes of around
the most recent Earth observation systems such              36 000 km and therefore proceed at the same speed
as SPOT 5 do provide coverage of less than 10 m             as the Earth rotates. They can provide imagery on an
resolution over Europe with repeat coverage every           hourly basis over very large areas and are ideally
three days.                                                 suited for use by meteorological agencies for weather
                                                            forecasting and climate analysis. Typical pixel sizes
From data to information                                    are 4 km (i.e. a square of side length 4 km) at the
Remotely sensed images can be used to map vege-             equator; such imagery is therefore unlikely to be of
tation. Visual interpretation of aerial photography         any practical value for habitat analysis at a local level.
is a long-established procedure whereby the inter-          Higher-resolution imagery can be obtained by
preter is able to discriminate and outline different        bringing the satellite nearer to the ground.
vegetation types on the basis of size, shape, tone,         However, this requires the satellite to be in a polar
texture, context and shadows. It is possible to inter-      orbit and although the spatial resolution of the
pret all imagery in this way. However, most digital         imagery is improved (e.g. pixel size of 1 km)
imaging devices have the ability to record reflec-          the temporal resolution is reduced (e.g. 12 hours).
tance properties across many wavebands. Many of             The most common medium-resolution satellite-
these contain valuable information, which helps to          imaging systems are also meteorological ones,
discriminate between vegetation types. The amount           such as the NOAA AVHRR, which are commonly
156    6 METHODS FOR SURVEYING HABITATS



used in television weather broadcasts. These med-             As introduced above, a key issue with these
ium-resolution systems are still too coarse for most       different satellite-borne systems is the trade-off
land applications. As a consequence, there is a grow-      between temporal and spatial resolution (Table
ing family of high-resolution Earth-observing satel-       6.1). This has implications for their application in
lite systems. These achieve coverage at a resolution       habitat monitoring. Generally speaking, systems
that can be measured in metres by the use of long          such as AVHRR are used for monitoring global
focal length lenses, but in so doing reduce temporal       vegetation changes on a seasonal basis, whereas
coverage still further (i.e. from hours to days).          LANDSAT and SPOT imagery is used for national
    The Landsat series (seven since 1972) is perhaps the   land cover surveys such as the DETR Countryside
most well established of the ‘Earth-observing satel-       Survey 2000. It is likely that the new high-resolution
lites’. The existing instrumentation on Landsat 7, the     systems such as IKONOS will be applicable to site-
Enhanced Thematic Mapper (ETM), records in eight           level assessments. A major restriction to the practical
spectral bands across the visible, near middle and into    application of satellite imagery in the UK is the diffi-
the thermal infrared. The panchromatic (black and          culty of obtaining cloud-free imagery. The opportu-
white) band has a pixel size or a spatial resolution of    nities for acquiring cloud-free imagery are greater
15 m. All the remaining bands have a pixel size of         for satellites with shorter revisit periods. This high-
30 m, except the thermal channel, which has a poorer       lights one of the major advantages of aircraft sur-
ground resolution of 60 m. The satellite orbits at an      veys over satellite methods. Not only can aerial
altitude of 705 km and the time between repeat             surveys provide data with vastly improved spatial
coverages is 16–18 days, depending upon latitude.          resolution, but the operator is also able to restrict
A standard scene covers an area of 185 km  170 km         flying sorties to those days with clear conditions.
(approx. 31 000 km2). The French SPOT satellite series     As the technology develops, airborne remote sen-
(five since 1986) offers similar capabilities to the       sing may become more capable of supplementing
Landsat series but provides the added enhancement          field survey information on widespread habitats
of simultaneous acquisition of stereo pairs of images      and landscape features.
(600 km  120 km) from the SPOT 5 platform. Stereo            Aerial film camera systems are mature technolo-
coverage allows for the derivation of height informa-      gies that have been used successfully for many
tion directly from the imagery by using digital photo-     years. Digital camera technology is now also find-
grammetric techniques. The elevation accuracy from         ing its way into the airborne imaging marketplace.
the High Resolution Stereoscopic (HRS) instrument          Not only is processing time reduced (because film
on SPOT 5 is quoted as 10 m.                               development is no longer necessary) but also,
    The first of a new generation of commercial,           unlike scanning systems, each frame of imagery is
high-resolution satellite systems was launched in          captured in a single exposure, which facilitates
1999. Through reducing orbital altitude to 680 km          geometric restitution. One problem, however, is
and increasing the focal length of the camera,             that each frame must be downloaded between
Space Imaging’s IKONOS satellite is able to provide        each acquisition and so ultimately a limit is
imagery with ground resolutions of 1 m in pan-             reached at which it is not possible to download
chromatic mode and 4 m in multi-spectral mode.             quickly enough. Even with the fast pace of technol-
This spatial resolving power is comparable to that         ogy advancement, it is likely to be some time
of high-altitude aerial photography. IKONOS has            before airborne digital sensors will be able to com-
been followed by two other high-resolution                 pete with conventional film cameras in terms of
spaceborne systems: Quickbird (launched 2001) and          image quality and efficiency.
OrbView (launched 2003). Information on all of these          New airborne scanning instruments, such as the
systems is available from a number of websites but         Compact Airborne Spectrographic Imager (CASI),
the most useful UK site at the time of writing is that     also have considerable potential for habitat evalu-
hosted by infoterra, formerly the UK National              ation. One particular advantage of the CASI system
Remote Sensing Centre (www.infoterra-global.com).          is that its configuration is programmable and the
      Table 6.1. Specifications of some current Earth-observing satellite sensors

                                                                                                                           Cost of digital data
                                                         Nominal
                                                                                                                         (based on 2000 prices) b
      Sensor       Year of first   No. of spectral       spatial                Revisit      Image frame
      name         launch          bands                 resolution (m)         period a     (km  km)        £ per scene          £ per square km c

      IKONOS       1999                                                         2.9 days     13 Â 13          3042                 18
          Pan                      1                     1
          XS                       4                     4
      SPOT         1986                                                         26 days d    60 Â 60
          Pan                      1                     10                                                   800–1700             0.25–0.50
          XS                       3                     20                                                   800–1350             0.25–0.40
      LANDSAT                                                                   16–18 days   185 Â 170 e
          ETM      1999            7                     15–60                                                400                  0.01
          TM       1982            4                     30–60                                                160–2200             0.005–0.070
          MSS      1972            4                     80                                                   120–600              0.003–0.020
          AVHRR    1979            5                     1100                   12 h         2400 Â 2400      80                   negligible
      KFA          —               Film                                         —
          1000                                           5                                   80 Â 80          1800–2250            0.30–0.35
          3000                     Camera f              2–3                                 21 Â 21          2250–3000            5–7

      a
        For polar-orbiting satellites, the exact revisit period will depend on latitude.
      b
        Hard-copy imagery will be cheaper than digital data. The lower price ranges relate to older imagery.
      c
        These costs are given for comparative purposes only. It may not be possible to buy imagery on a per km2 basis.
      d
        The SPOT satellite is actually steerable and thereby can obtain, by request, more frequent coverage.
      e
        ‘Quarter scenes’ of LANDSAT imagery may also be purchased.
      f
        Costs here are based on scanned photography but it is also possible to purchase prints.




157
158    6 METHODS FOR SURVEYING HABITATS



precise number of spectral bands, their locations            Table 6.2 presents some common land cover and
and bandwidths can be selected in flight. A spatial       landscape features and poses three questions:
resolution of between 0.5 m and 10 m can be
                                                          1. What is the minimum ground resolution
achieved, depending on the flying altitude.
                                                             required to detect or measure them?
    LIDAR (a light detection and ranging instrument)
                                                          2. What scale of imagery will provide data at this
is often flown in combination with the CASI system.
                                                             resolution?
LIDAR is a laser range-finder that measures the time
                                                          3. What instrument platform is most commonly
of flight of a laser beam from the aircraft to the
                                                             used to provide this imagery?
ground and back. The information provides eleva-
tion data, which can be used for vegetation measure-      From Table 6.2, it is possible to identify the techni-
ment and mapping and is particularly useful for           cal solutions most appropriate to particular map-
providing textural definition. Other systems, such        ping tasks. Some key points are:
as synthetic aperture radar (SAR), offer similar spe-
                                                          *   No single type of imagery suits all purposes.
cialist facilities. For further information on such
                                                          *   The requirements for detection are less demand-
instrumentation refer to a standard remote-sensing
                                                              ing than those for measurement.
textbook (e.g. Lillesand & Kiefer, 1994).
                                                          *   Users wanting to identify or measure several dif-
                                                              ferent land cover features from the same imagery
Satellite specifications and cost                             will have to accept that some will be more prone
The major commercial sources of satellite data in             to errors than others.
the UK are the National Remote Sensing Centre             *   Remotely sensed imagery generally cannot pro-
(now Infoterra) and NPA Group (see Box 6.1 for                vide information that is directly comparable to
contact information). Table 6.1 provides an indica-           botanical classifications (e.g. Phase I habitat sur-
tion of the relative costs of imagery. Since satellite        vey and NVC).
data can be expensive (particularly as it is usually
                                                          One way of ensuring high mapping accuracies
necessary to purchase a whole scene even if the
                                                          would be to use large-scale imagery, but as photo
study site is only a fraction of the area), it is worth
                                                          scale increases, the number of images required to
contacting your local university’s geography and or
                                                          cover any given area increases geometrically. Scale
environmental sciences department before pur-
                                                          and cost are therefore inter-related.
chasing data, because they may know of existing
                                                             ‘Fitness for purpose’ can only be established by
images that may be bought at a much lower cost.
                                                          being very clear about the trade-offs and inter-
Again the major centres of remote sensing expert-
                                                          linkages between the following.
ise are readily found on the Internet.
   Aerial photographs can be obtained from a variety      *   Purpose: what do you need to know?
of sources. If it is necessary to commission a special    *   Method: what technical options do you have?
survey to meet the data requirements, then the costs      *   Economy: what can you afford?
of flying will have to be included. Monitoring will       *   Error: what types and level of error can you
necessitate several repeat surveys over time.                 tolerate?
Generally, there are no cost savings when multiple
                                                          Sections 6.1.2 and 6.1.3 describe the applications of
aerial surveys are done, but time series of satellite
                                                          satellite-based remote sensing and aerial photogra-
data can often be purchased with major discounts.
                                                          phy in habitat mapping and monitoring.

Applications of remote sensing
                                                          6.1.2        Satellite-based remote sensing
Satellite or aerial imagery can be used in a wide
range of applications, but selecting appropriate          Applications of satellite remote sensing
solutions from the growing range of technical             The early LANDSAT satellites were principally
options can be difficult.                                 employed for expansive crop inventory projects.
      Table 6.2. Data users’ requirements and recommendations

                                                                              Image recommendations

                                                                                             Smallest scale for:

      Data requirement                               Ground resolution (m)   Detection                 Measurement               Platform a

      Land cover classification and mapping
      Forestry
      Species                                                    0.1         1 : 5 000                 1 : 2 500                 LAA
      Isolated trees                                             0.3         1 : 12 000 to             1 : 6 400 to 1: 9 600     LAA
                                                                                1: 20 000
      Forest strips < 20 m wide                                 30           1 : 1 500 000             1 : 750 000               SAT
      Groups of trees(< 0.25 ha) (mainly                         1.0         1 : 64 000 to             1 : 32 000 to 1: 64 000   HAA
      broadleaved/conifers/mixed)                                               1: 125 000
      Mature forest stand                                       30           1 : 1 500 000             1 : 750 000               SAT
      (broadleaved/conifers/mixed)
      Felling                                                   10           1 : 500 000               1 : 250 000               SAT
      Non-forest
      Scrub (species)                                            0.3         1 : 12 000                1 : 6 400                 LAA
      Scrub (stand)                                              3           1 : 184 000               1 : 92 000                HAA
      Grasses (species)                                          0.1         1 : 3 200                 1 : 1 600                 LAA
      Grasses (stands)                                           3           1 : 184 000               1 : 92 000                HAA
      Peatland                                                   3           1 : 184 000               1 : 92 000                HAA
      Blanket bog                                                3           1 : 184 000               1 : 92 000                HAA
      Non-vegetated surfaces
      Rocks and cliffs                                           3           1 : 184 000               1 : 92 000                HAA
      Dunes                                                      3           1 : 184 000               1 : 92 000                HAA
      Built-up land                                              3           1 : 184 000               1 : 92 000                HAA
      Urban open space (e.g. cemeteries)                         3           1 : 184 000               1 : 92 000                HAA
      Derelict land                                              3           1 : 184 000               1 : 92 000                HAA
      Bare soil                                                  0.3         1 : 12 000                1 : 6 400                 LAA
      Extensive bare soil                                       10           1 : 500 000               1 : 250 000               SAT
      Transport routes                                           3           1 : 184 000               1 : 184 000               HAA
      Water
      Coastal water/estuaries                                   10           1 : 500 000               1 : 250 000               SAT




159
      Inland open water                                          3           1 : 184 000               1 : 92 000                HAA
      Table 6.2. (cont.)




160
                                                                                   Image recommendations

                                                                                                  Smallest scale for:

      Data requirement                             Ground resolution (m)          Detection                 Measurement              Platform a

      Marsh                                                  3                    1 : 184 000               1 : 92 000               HAA
      Saltmarsh                                              3                    1 : 184 000               1 : 92 000               HAA
      Canals                                                 5                    1 : 310 000               1 : 155 000              HAA
      Hydrological condition                                10                    1 : 500 000               1 : 250 000              SAT
      Interpretative information
      Habitat mapping
      Forest edge                                           30                    1 : 1 500 000             1 : 750 000              SAT
      Forest–agriculture edge                               30                    1 : 1 500 000             1 : 750 000              SAT
      Forest–water edge                                     10                    1 : 500 000               1 : 250 000              SAT
      Forest–abandoned land edge                            10                    1 : 500 000               1 : 92 000               SAT
      Fire breaks                                            3                    1 : 184 000               1 : 92 000               HAA
      Vegetative condition
      Insect effect                                          0.3                  1 : 12 000                1 : 6 400                LAA
      Disease                                               80.0                  1 : 3 900 000             1 : 2 100 000            SAT
      Pollution effect                                       0.3                  1 : 12 000 to             1 : 6 400 to             LAA
                                                                                     1 : 20 000                1 : 9 600
      Phenological stage                                     0.3                  1 : 12 000                1 : 6 400                LAA
      Resource parameter measurement
      Tree dimensions (height, basal area,                   0.3                  1 : 20 000                1 : 9 600                L–MAA
      crown diameter)
      Biomass statistics (annual production,                 0.3                  1 : 12 000                1 : 6 400                LAA
      plant density)
      Animal counts                                          0.3                  1 : 20 000                1 : 9 600                L–MAA
      Nesting trees                                          1.0                  1 : 40 000                1 : 20 600               MAA

      a
       LAA, low-altitude aircraft (150–3660 m); MAA, medium-altitude aircraft (3660–9150 m); HAA, high-altitude aircraft (9150–19 820 m);
      SAT, satellite (over 190 km).
      Source: Adapted from Aldrich (1979).
                                                          6.1 General habitat survey and monitoring methods   161




  Box 6.1 Contact information for obtaining                    Fax: 01793 414606
  remote sensing data                                          www.english-heritage.org.uk
                                                               Air Photographs Unit
  (correct at time of writing)                                 National Assembly for Wales
     Infoterra Ltd (formerly the National Remote               Cathays Park
     Sensing Centre)                                           Cardiff CF10 3NQ
     Atlas House, 41 Wembley Road, Leicester, LE3 1UT          Tel: 02920 823819
     Tel: 0116 2732300                                         Royal Commission on the Ancient and
     Fax: 0116 2732400                                         Historical Monuments of Scotland
     www.infoterra-global.com                                  John Sinclair House
     Nigel Press Associates Group                              16 Bernard Terrace
     Crockham Park, Edenbridge, Kent,                          Edinburgh EH8 9NX
     TN8 6SR                                                   Tel: 0131 662 1456
     Tel: 01732 865023                                         Fax: 0131 662 1499/1477
     Fax: 01732 866521                                         www.rcahms.gov.uk
     www.npagroup.co.uk                                        Public Records Office of Northern Ireland
     English Heritage                                          66 Balmoral Avenue
     National Monuments Record                                 Belfast BT9 6NY
     PO Box 569                                                Tel: 02890 255905
     Swindon SN2 2XP                                           Fax: 02890 255999
     Tel: 01793 414600                                         www.proni.nics.gov.uk




This agricultural monitoring role has continued              imagery. To improve classification accuracy, both
with the later satellites (e.g. the Monitoring               summer and winter imagery was used. Geometric
Agriculture by Remote Sensing (MARS) project                 errors were controlled by geometrically correcting
run by the Joint Research Centre on behalf                   the summer images to the Ordnance Survey
of EUROSTAT). The higher spatial resolutions                 National Grid and then resampling the winter
(c. 30 m) provided by these later satellites, particu-       images to fit. A land-cover classification system
larly the Thematic Mapper (TM), have provided the            was developed, which was appropriate both to
opportunity for pan-European and national land-              user requirements and to the ‘fitness for purpose’
cover mapping programmes.                                    of such medium-resolution imagery. However,
    At the European level, the principal initiative has      some of these classes (e.g. bracken) still had stan-
been in connection with the development of a co-             dard errors of around 33% (Pakeman et al., 2000).
ordinated information network on the environment                The 1990 Land Cover Map methodology has been
(CORINE). As part of this, each EU member state has          refined for a new land-cover map of Great Britain
been requested to provide a 1 : 100 000 scale land-          (LCM2000). The main reporting structure is based on
cover map for their national area. Although most             the Broad Habitats identified in the UK Biodiversity
member states have obtained this through manual              Action Plan. The relationships between these Broad
interpretation of appropriately scaled hard copy             Habitats and the LCM2000 Target Classes and Sub-
LANDSAT TM images, the UK was able to take advan-            Classes are given in Table 6.3.
tage of a national mapping initiative, the Land Cover           At an individual country level, similar satellite
Map of Great Britain (Fuller et al., 1994).                  image-based land-cover mapping initiatives have
    The production of the 1990 Land Cover Map of             been conducted by MLURI (the Macaulay Land
Great Britain was achieved using Landsat TM                  Use Research Institute) (Wright & Birnie, 1986;
162      6 METHODS FOR SURVEYING HABITATS




    Satellite-based remote sensing: summary of key              Precision   Satellite-based sensors commonly achieve
    points                                                      a resolution of 10–30 m. The most recent generation
                                                                of satellite sensors have an improved spatial resolution
    Recommended uses       Satellite-based remote sensing       of 1–4 m.
    may be used for measuring Broad Habitat extent, but is      Bias    The reflectance properties of areas of vegeta-
    not currently recommended for detailed site monitor-        tion can vary from image to image, owing to factors
    ing because the spatial resolutions are too poor.           such as cloud, moisture and season. These variations
    However, in conjunction with ground surveys and             must be considered for successful application of
    aerial photography, satellite imagery can be a useful       automated classification procedures.
    additional data source.
                                                                Expertise required    An understanding of the com-
      The method also has some potential for monitoring
                                                                puter hardware and software for image
    habitat quality through measurement of vegetation
                                                                processing and interpretation is essential.
    change (e.g. heather cover on a moorland) if a change
                                                                Experience in recognising ecological and land-cover
    from a ‘desirable’ to a ‘less desirable’ state can be
                                                                classes is also required.
    defined in terms of vegetation change that can be
    monitored with remote sensors.                              Equipment required     Contact information for
      The potential of the most recent generation of            the major UK satellite data re-sellers can be found in
    high-spatial-resolution satellite sensors has yet to be     Box 6.1. To view and interpret remotely sensed data, a
    fully explored but may be applicable for some site-level    computer capable of handling large quantities of
    assessments.                                                data and with a good quality printer is required. A GIS
                                                                (see Glossary) such as ArcInfo is also useful for
    Efficiency    Remotely sensed data are generally            manipulating and comparing images and adding extra
    cheaper to obtain than data obtained by field survey,       information.
    although the size of the study area will affect the cost    Key methodological points to consider     At present,
    and efficiency. Satellite imagery is available at differ-   satellite-based remote sensing does not provide a
    ent costs, according to the level of pre-processing that    reliable method for monitoring changes in
    the data have undergone. Some corrections will still be     semi-natural habitats.
    necessary in order to remove, for example, geometric            Satellite imagery can be used for site mapping but
    and atmospheric distortions.                                does not achieve the level of detail obtained from
      Field survey is still essential for ground truthing and   either Phase I surveys or aerial photography.
    calibration. This will probably be required for each            Airborne remote sensors are capable of
    separate image.                                             producing detail comparable with that of Phase I
                                                                Habitat Survey.
    Objectivity    Automated habitat classifications may be
    objective in the sense that classification is carried out   Data analysis     Calculation of habitat extent is
    by computer. However, analysis based on spectral            achieved with computer software. If data are input into
    signatures alone does not benefit from additional           a GIS, information derived from other sources, such as
    knowledge of size, shape, context, etc.                     ground survey, can be included.




Wright & Morris, 1997; G. G. Wright et al., 1997).              *   mapping the extent and condition of blanket bogs
SNH has also demonstrated the value of satellite                    (Box 6.2); and
imagery in several applications:                                *   woodland monitoring under the Earth
                                                                    Observation Network 2000 (Box 6.3).
*   primary stratification in relation to the choice of
    sample areas for the National Countryside                   In many of these applications, satellite imagery
    Monitoring Scheme (NCMS);                                   was used in combination with both aerial imagery
      Table 6.3. The relationship between terrestrial Broad Habitats and LCM2000 Classes

      Broad Habitats                     LCM2000 Target Class                LCM2000 Sub-classes and variants

      Broadleaved, mixed and yew         Broadleaved and mixed               Trees
      woodland                           woodland                              deciduous
                                                                               evergreen
                                                                             Scrub/shrub
                                                                               deciduous
                                                                               evergreen
      Coniferous woodland                Coniferous woodland                 Standing
      Boundary and linear features       (Not identified in LCM2000)         Felled
      Arable and horticulture            Arable and horticulture             Wheat
                                                                             Barley
                                                                             Oats
                                                                             Maize
                                                                             Oilseed rape
                                                                             Peas
                                                                             Field beans
                                                                             Linseed
                                                                             Sugarbeet
                                                                             Potatoes
                                                                             Unknown arable
                                                                             Horticulture
                                                                             Perennial crops
                                                                             Set-aside (bare)
      Improved grassland                 Improved grassland                  Agricultural/managed grass
                                                                               mown/grazed
                                                                               hay/silage
                                                                             Grass, semi-improved, reverting
                                                                             Grazing marsh
                                                                             Set-aside with grass or weeds
                                                                             Rough
      Neutral grassland                  Natural and semi-natural            Neutral – unimproved/neutral
                                         grasslands and bracken




163
164
      Table 6.3. (cont.)

      Broad Habitats               LCM2000 Target Class         LCM2000 Sub-classes and variants

      Calcareous grassland                                      Calcareous
      Acid grassland                                            Acid Nardus with rushes
      Bracken                                                   Bracken
      Dwarf shrub heath            Dwarf shrub heath            Closed heath
                                                                  ‘wet’ heath
                                                                  ‘dry’ heath
                                                                  gorse
                                                                Open heath
      Fen, marsh and swamp         Fen, marsh and swamp         Swamp
                                                                Fen/marsh
      Bog                          Bog                          Bog
                                                                  shrub
                                                                  grass (Molinia) or herbaceous
      Standing open water and                                   Water (inland)
      canals
      Rivers and streams
      Montane habitats             Montane (bare/heath)
      Inland rock                  Inland bare ground           Natural
                                                                Despoiled
      Built-up areas and gardens   Built-up areas and gardens   Suburban/rural developed
                                                                Continuous urban residential/
                                                                commercial industrial
      Supralittoral rock           Supralittoral rock and       Rock/shingle
      Supralittoral sediment       sediment                     Dune – with/without shrubs
      Littoral rock                Littoral rock and sediment   Rock, mud and sand
      Littoral sediment                                         Saltmarsh, grazed and ungrazed
      Oceanic seas                 Sea/estuary
      Sublittoral categories
                                                              6.1 General habitat survey and monitoring methods          165



and ground data. Such integrated approaches –                       Aerial photography has been widely used in
combining census data with sample data – are                     ecological applications. It has been used for both
increasingly used. So the question is not whether                inventory and monitoring at national to local
one source of data is better than another, but how               levels, especially as an adjunct to Phase I habitat
they can best be used together.                                  surveys. Whereas in the 1970s it could be said that
                                                                 the interpretation of aerial photos was a well-devel-
                                                                 oped area technically, recent developments in the
6.1.3       Aerial photography
                                                                 use of digital image acquisition and analysis tech-
Recommended uses                                                 niques have opened up a whole new range of
There is a wide range of potential uses for aerial               approaches including the advent of digital camera
photography. The relationships between resolving                 systems, on-screen interpretation and the linkages
power and photo scale (Table 6.2) hold true,                     with GIS technology for analysis and presentation.
and aerial photos are generally best employed to                    There are many examples of national surveys in
identify and classify structural components in the               the UK that employ aerial photography. Some are
landscape. They are not generally suited to species-             specifically related to baseline audit, others use
based classifications.                                           historical aerial photography to provide either a



  Box 6.2   Scottish Blanket Bog Inventory                       The method employs an unsupervised classification of
                                                                 Landsat 5 Thematic Mapper 30 m imagery, which is
  The Scottish Blanket Bog Inventory (SBBI) is an                validated by National Vegetation Classification (NVC)
  example of the use of remote sensing for characterising        ground survey. The classified image products will
  extensive biotopes more consistently, repeatably and           provide improved information on the vegetation
  cost-effectively than can be achieved by any other             communities and hydrological status of blanket bogs
  means. Undertaken by Scottish Natural Heritage, the            throughout Scotland.
  SBBI maps the extent, distribution and condition of               For further information, see Reid & Quarmby
  blanket bog vegetation throughout Scotland.                    (2000).




        Earth Observation for Natura 2000
  Box 6.3                                                        use of Earth Observation images and techniques for
  (EON2000)                                                      environmental monitoring applications.
                                                                    Although the spatial resolution was found to be
  Habitat inventory and land-cover change information is         insufficient to generate the more detailed inventories
  required across Europe to support the Natura 2000              required, the potential of basic classification for
  scheme. Developing a method for routine data collec-           baseline inventories at a national level was acknowl-
  tion over such a wide area presents obvious difficulties.      edged. Change detection approaches successfully
  Designated sites in Scotland, Austria and Finland were         flagged up ecological change and the method was
  selected to test the possibility of using imagery from         recognised by conservation organisations as being of
  space-borne sensors to derive forest habitat inventories.      practical application in targeting ground survey
     Landsat TM, IRS, LISS, SPOT Pan and IRS Pan images          resources. Lack of available satellite sensor data to
  (Table 6.1) were used for the inventories, and historic        evaluate ecological change was flagged up as a major
  Landsat TM images were used to validate and demon-             issue for an operational system.
  strate methods of change detection. An Internet-based             For further information see http://geospace.co.at/
  system was designed and implemented to facilitate the          EON2000.html.
166        6 METHODS FOR SURVEYING HABITATS




    Aerial photography: summary of key points                 when calculating areas with a planimeter or digitising
                                                              equipment, areas on slopes will be underestimated and
    Recommended uses
                                                              areas of high relief will be overestimated relative to
    *   Rapidly assessing the nature conservation resource    areas of low relief.
        in an area                                            Expertise required     Mappers should be trained in the
    *   Establishing a framework of data for a monitoring     recognition of different habitat types from photo-
        baseline                                              graphs, and in the use of stereoscopes. The ability to use
    *   Monitoring broad-scale changes in habitat extent      a planimeter or digitising equipment is necessary if
                                                              these items are used to process images. It is helpful if the
    Efficiency       Six days maximum AQE are required to     people analysing the photographs have on-the-ground
    evaluate each 5 km  5 km square (including one day       experience of the area involved and either have general
    field checking).                                          ecological and land-cover experience or have been
    Objectivity Reasonable as long as standard                trained in Phase I habitat surveying (see Section 6.1.5).
    methods are used for distinguishing between habitat       Equipment required      Overlapping aerial photographs,
    types and field checking is used to assess accuracy.      stereoscope, digitising equipment, pens, pencils, rulers,
    A problem arises in areas where few distinctive           light table, etc. for map-making.
    boundaries occur and a line has to be drawn               Key methodological points to consider       Good-quality
    between two habitats that grade into each other.          overlapping vertical aerial photographs are essential.
    This problem derives from a fundamental problem of        Colour photographs are preferred, if available. For
    using a map to represent natural variation and will       monitoring, two matching sets of photographs are
    lead to variations between interpretations by different   required, taken at the same time of year for the area
    people.                                                   being monitored. The time of year the photographs
    Precision      The errors involved in measuring habitat   were taken is important; some habitats are hard to
    areas on 1 : 10 000 scale maps are generally well below   identify at certain times of year. Some field checking
    5%, although some habitat types are more prone to         will be necessary, particularly for habitats that are hard
    errors than others.                                       to recognise accurately from photographs.
    Bias    Sources of bias arise from misidentification      Data analysis Areas of habitat can be measured via
    of habitat types from photographs and inaccurate          digitisation and analysis with a GIS package such as Arc
    mapping of boundaries. Unless slopes are included         View.



census of land cover change (e.g. Landscape Change            *   Slow changes in quantity, often associated with
in the National Parks; Countryside Commission,                    minor shifts in boundaries of the order of 1–5 m,
1991) or a sample of change (e.g. the NCMS;                       over 20–30 years are poorly identified (i.e. they lie
Mackey et al., 1998). There is a growing body of                  within the bounds of line positioning errors).
knowledge concerning the use of aerial photogra-              *   Changes in quality are poorly or seldom identified
phy for monitoring change. In general terms, the                  by using conventional point, line and area inter-
following hold true.                                              pretations or mapping conventions. (This type of
                                                                  change is most amenable to the use of digital
*   Changes can be divided into those relating to
                                                                  imagery.)
    quantity (i.e. changes in area or stock) and those
                                                              *   Changes in semi-natural communities that involve
    relating to quality (i.e. changes in composition or
                                                                  complex interdigitating boundaries, which are
    condition).
                                                                  often transitional rather than discrete, are extre-
*   Rapid changes in quantity or stock (e.g. major
                                                                  mely difficult to detect and quantify reliably.
    afforestation in the period from c. 1946 to c. 1988)
    can be reliably identified by using historical aerial     Time series of medium-scale photography (i.e.
    photography.                                              around 1 : 25 000 scale) can provide the most
                                                       6.1 General habitat survey and monitoring methods      167



reliable information on major transfers of stock,         (from the period 1940–60). The contact addresses
particularly in managed areas such as woodland            are given in Box 6.1. Other more recent sources
and arable land. They provide least reliable infor-       include the company ‘getmapping’ (www.getmap-
mation on changes in the quality or composition of        ping.co.uk). In addition, airborne scanning systems
land cover types or in complex and slow changes           (e.g. CASI) can be hired from independent organisa-
(e.g. in semi-natural cover types). To maximise the       tions. These images can be either digital or analo-
potential for detecting changes in these situations,      gue and can in some cases achieve a level of detail
it is recommended that longer time intervals              similar to that obtained with Phase I Habitat
between photographs be used.                              Surveys. A fuller review of this technique is given
                                                          in Pooley & Jones (1996).
                                                             A stereoscope provides a three-dimensional
Expertise required
                                                          image from overlapping pairs of photographs.
Interpretation of aerial photographs is a skill
                                                          This facilitates the identification of habitat types,
acquired largely through experience. Anyone
                                                          as the height and texture of vegetation becomes
undertaking analysis of aerial photographs should
                                                          more obvious. There are two sizes of stereoscope
be trained to a consistent standard in the recogni-
                                                          that are useful: a small portable set is easily
tion of features on photographs. Previous ground
                                                          employed for mapping but can only be used on a
knowledge of the area can make it easier to recog-
                                                          small part of a 22 cm  22 cm photograph at a time,
nise and interpret features on the photographs
                                                          whereas a large desktop mirror stereoscope allows
(although this might introduce some bias).
                                                          the viewing of the whole overlapping area.
   Interpreters will need to be familiar with the
                                                             For mapping, a set of coloured pens and pencils
use of stereoscopes and digitising equipment. If
                                                          plus drawing equipment is required. Crude area
on-screen interpretation and digitising is to be
                                                          estimates can be obtained by using a Romer dot
attempted, interpreters must be confident that
                                                          grid (a transparent overlay with a given density of
they can distinguish the same level of detail from
                                                          dots per square unit) or a planimeter. However, a
the screen as from traditional stereoscopic analy-
                                                          low-cost digitising tablet will provide more reliable
sis. Some additional training may be necessary;
                                                          measures and will enable maps to be digitised and
on-screen stereo-interpretation facilities represent
                                                          manipulated with a GIS package such as ArcView.
a new technology, and experience of best practice is
                                                             Some field checking is recommended when ana-
limited. A comprehensive background to aerial
                                                          lysing aerial photographs; the equipment required
photograph interpretation is provided by Lillesand
                                                          for this is similar to that required for Phase I habitat
& Kiefer (1994).
                                                          surveys (Section 6.1.5). The design of appropriate
                                                          field checks may require some statistical advice to
Equipment required                                        ensure appropriate methods that avoid biasing the
Good-quality aerial photographs are essential             results and ensure cost-effectiveness.
(Appendix 6). A 60% forward overlap is required to
provide stereo coverage for use with a stereoscope.
The choice between colour and black-and-white             Methods
photography will depend on the availability, sea-         Selection of photography
son of acquisition and quality of photographs, the        If the available sources of photographs are inade-
habitat type of interest and the preference and           quate, commissioning photography may be
experience of the individual interpreter. The             required. Some operators provide low-cost small-
Ordnance Survey (OS) hold and provide records             format aerial photographs (e.g. 35 or 70 mm). These
for the past 10 years, before passing these on to         can be useful, providing the equipment to analyse
national organisations. The English Heritage              them is available (for example, many older stereo-
National Monuments Record office in Swindon               plotters are set up to deal only with large-format
holds older OS photos, their own, and RAF records         photographs).
168    6 METHODS FOR SURVEYING HABITATS



Outline methods                                            estimation of the environmental factors responsible
Overlapping matched pairs of aerial photographs            for making up the landscape, supplemented by
are examined under a stereoscope, and blocks of            satellite data (Box 6.4). The classification is hierarch-
distinct habitat types are identified. It is possible      ical, split initially into linear features (e.g. hedge-
to mark outlines and calculate areas directly on           rows and tracks) and area features (eg. woodland,
the photographs. However, it is generally prefer-          scrub).
able to transfer habitat boundaries to Ordnance               As an example at a country level, the Land Cover
Survey base maps (usually 1 : 10 000 scale) by             Map of Scotland (carried out by the MLURI) was
using an optical device such as a Sketchmaster             more ambitious. It was a complete survey of
and standard colour codes. This mapping process            Scotland and identified many more habitat types
allows for the correction of relief distortion and         than did the previous surveys (Box 6.5).
camera effects. An easily interpretable map that              Although most studies have recommended
can be stored and copied as required is also pro-          that a standardised habitat classification is desir-
duced. Alternatively, a planimeter, a photogram-           able, they generally have created their own var-
metric plotting instrument or a digital                    iants. Comparisons between surveys are made
photogrammetric device can be used to map vege-            harder and an agreed standard classification has
tation directly from the photographs once the              not been formalised across the UK (but see Gilbert
boundaries have been identified.                           & Gibbons (1996) for a comparison of different
   Field checking is recommended to verify that            survey classifications). The introduction of the
habitat classifications are accurate.                      UK Biodiversity Action Plan and its recognition
                                                           of Broad Habitats (Table 6.3) may stimulate har-
Habitat classification and identification                  monisation of UK habitat classification systems.
The NCMS (Mackey et al., 1998) used the land classi-       Classifications should be decided upon with the
fication developed by the Institute for Terrestrial        reasons for the study in mind. The quality of
Ecology (ITE) for Cumbria (now CEH), based on an           photographs available will also affect the choice




  Box 6.4 National Countryside Monitoring                  sufficient detail. Instead, a stratified random sample
  Scheme (NCMS)                                            of 5 km  5 km squares (later 2.5 km  2.5 km) was
                                                           developed, covering 7.5% of Scotland’s land area.
  The NCMS is a Scotland-wide sample survey of land-       Land cover was interpreted in terms of 31 areal
  cover change. Utilising aerial photography from          features and five linear features. For each of the 467
  c. 1947, c. 1973 and c. 1988, the NCMS has quantified    sample squares, the interpreted features were
  the magnitude, rate and geographical variation in        mapped at a scale of 1 : 10,000. Land-cover maps for
  change over the second half of the twentieth century.    each sample square were digitised and processed on
  During this period, considerable changes took place in   a GIS, and overlay analyses allowed the computa-
  Scotland’s urban and rural environment, but prior to     tion of land-cover change between time
  the NCMS little could be said about the overall impact   periods. Statistical software allowed sample square
  of human activities on semi-natural habitats. The        data to be extrapolated to provide estimates of
  NCMS provides an objective account of the key            extent, change and interchange for a geographical
  changes and explains the changing relationships          region of interest. Standard errors and confidence
  between land-cover features.                             intervals provide a measure of the uncertainty in
      Aerial photography was used as the source for        the estimates due to sampling.
  land-cover interpretation, but it was impracticable         For further information see Mackey et al. (1998) or
  to map the whole of Scotland’s land cover in             www.snh org.uk/trends/landcover.
                                                              6.1 General habitat survey and monitoring methods           169




  Box 6.5 Land Cover of Scotland 1988 (LCS88)                    SURVEY METHOD
                                                                 Aerial photography specifically for the project
  This survey, carried out by the MLURI, was a complete          involved mainly panchromatic film, with some natural
  census of Scotland, covering 78 828 km2. The survey            colour coverage in the Central Valley. Collection took
  was intended to be a baseline for future monitoring,           place during 1987–9. Land-cover boundaries were
  for which a new land-cover classification system was           identified by interpreters with extensive local knowl-
  devised in consultation with SNH and Ecological                edge before digitisation. The digitisation of the dataset
  Advisors Unit within the then Scottish Office.                 and subsequent field checking took place between
                                                                 1989 and 1993.
  OBJECTIVES
  To provide a basic land-cover inventory for the whole          SUMMARY
  of Scotland, enabling studies of land-cover change, and        The LCS88 survey is widely regarded as an accurate
  to digitise the information for input into a GIS.              survey. Potential error sources included the referential
                                                                 error resulting from differences between people in
  CLASSIFICATION                                                 their interpretation of land-cover features. In an
  The habitat and land-use classification system was             assessment of this error it was found that interpreta-
  hierarchical and recognised summary, principal,                tion rates depend upon classification level: overall
  major and main land-cover features and                         error rates increased with increased detail. At the most
  sub-categories. From these different levels, 23 sum-           detailed level, the error was estimated at around 25%.
  mary features and 40 original land-cover features were         This reflects the issues relating to error and the use of
  identified. These were increased to 126 ‘single features’      one image source for all features, particularly at low
  (i.e. land-cover types) by sub-categories created from         levels in the classification hierarchy.
  the original 40. Area statistics were based on these 126          The use of mosaics, while allowing for the recogni-
  single features and mosaics of these features.                 tion of more habitat types, means that estimation of
     The use of mosaics of mixed land-cover features             single cover types is difficult. The survey area covered
  (defined as ‘visible mixtures of two land-cover                78 823 km2, whereas the area covered by single cover
  features in which the total area of each is below the          measurements was 54 817 km2. The remaining
  minimum mapping unit for separate identification’)             24 006 km2 is tied up in mosaics. For example, the
  stretched the number of categories to 1327 individual          single feature area of heather moorland in Scotland
  features. Although this increases the level of complexity      was estimated as 6882 km2. If the areas of all mosaics
  of the classification, it allows the description and mea-      containing heather moorland as the primary or sec-
  surement of land-cover combinations that would                 ondary feature are included, the total reaches
  otherwise be ignored, such as wooded bogs.                     16 922 km2. To estimate the amount of heather pre-
                                                                 sent in the mosaic areas requires an assumption about
  MONITORING                                                     the composition of the mosaics (e.g. 60 : 40). This may or
  A ten-yearly survey with LCS88 data as a baseline was          may not be valid.
  proposed. Two pilot studies have been conducted in                Sources: Gilbert & Gibbons (1996) and MLURI
  the Central Valley and Cairngorm areas.                        (1993).



of habitat types: colour photographs can allow                   monochrome photographs. Colour photographs
more types to be distinguished than monochrome                   may be easier to interpret.
ones.
   As an example, the guide to habitat identifica-               Mapping
tion given in Table 6.4 is based primarily on the                The mapping and calculation of areas from aerial
study by Langdale-Brown et al. (1980), which used                photographs requires the identification of features
170   6 METHODS FOR SURVEYING HABITATS



      Table 6.4. Characteristics of habitat types for aerial photograph analysis by means of a stereoscope

      Habitat type               Characteristics

      Deciduous woodland   Tallest vegetation with rounded crowns varying in height and
                           diameter with irregularly spaced occasional gaps
                           Light and medium tones
                           Coarse, irregular texture
                           Rounded shadows
                           Light, feathery appearance on winter photographs
      Coniferous woodland Tall, dense stands, small-crowned trees regularly spaced
                           Medium to dark tones
                           Medium regular texture
                           Pointed conical shadows
      Mixed woodland       Tones vary from light to dark
                           Irregular texture
                           Variously shaped shadows
      Scrub                Predominantly woody vegetation of medium height: shrubs,
                           bushes, occasional trees often interspersed with patches of grassland
                           Some distinct rounded crowns and areas of coalescing canopy
                           Mottled appearance due to mixture of woody vegetation
                           (medium and dark tones) and grasses (light tones)
      Dwarf shrub heath    Low vegetation
                           Very dark tones
                           Fine and regular texture giving a smooth, dark appearance
      Unimproved grassland No perceptible height
                           Irregular mix of light and medium tones
                           Fine texture, with occasional rougher or tussocky areas
                           Less regular than agricultural land
                           No signs of cultivation
      Improved grassland   No perceptible height
                           Mixture of light and medium tones
                           Finer and more regular texture than unimproved grassland
                           May be signs of improvement such as drainage lines,
                           walls and vehicle tracks
                           No signs of ploughing
      Wetland              Very varied appearance due to the variety of community
                           types occurring
                           Identification can be based upon local knowledge, the relief
                           of the area or the proximity of areas of open water
      Open water           Either very light or very dark tones
                           No texture
      Agricultural land    Low: no perceptible height
                           Textureless: very fine regular texture
                           Signs of cultivation such as plough lines or farm
                           machinery tracks
                                                          6.1 General habitat survey and monitoring methods     171



on the photographs that can be pinpointed on the            a different location and digitised photographs and
base map. Sometimes this is straightforward,                other files kept on computer should be backed up.
where there are a great number of anthropogenic
features, such as roads, buildings and field bound-         Sampling
aries, which appear on maps and can also be recog-          Given that aerial photography offers a synoptic
nised on photographs. However, there may be                 view of large areas, it is mostly used to provide
difficulties in areas of semi-natural habitat where         census information. However, it can be used to
there are few ground references. In such cases, it          provide estimates of habitat area in the same way
may be necessary to introduce local control, for            as ground surveying. One approach is to use strati-
example through use of a GPS. The identification            fied random sampling, which can improve preci-
of several points of known grid reference is also           sion and ensure adequate representation of less
essential if photographs are to be analysed or digi-        common features in the sample (this is sometimes
tised directly.                                             also known as ‘area-frame sampling’). The esti-
   Areas of uniform habitat type can be marked on           mates of habitat area obtained by this method
the base map with standard colour codes based on            have standard error terms that are directly related
those used for the Phase I habitat survey (Section          to the number of samples taken.
6.1.5). Colours of maximum contrast are used to                 Two issues arise from the above: the first relates
produce an easily interpretable map. If a mixture           to how to stratify and the second relates to how to
of two habitat types occurs with no uniform indi-           determine numbers of samples to obtain estimates
vidual blocks (or blocks are too small to map accu-         of a desired precision. A good worked example of
rately), use alternating lines of the colours for each      solutions to these issues is provided by the NCMS.
habitat. Make the thickness of the lines propor-            This shows how lower-resolution satellite imagery
tional to the relative predominance of the habitats.        can be used to provide adequate information for
Moodie (1991) used alphanumeric codes rather                stratification (in this case into upland, lowland,
than colours for improved grassland and arable              intermediate and urban classes). Alternatively,
land in order to draw more attention to areas of            stratifications could be provided by the ITE’s (now
unimproved semi-natural habitat on the map.                 CEH) Land Classification System, the Land Cover
                                                            Map of Great Britain or, indeed, more recent classi-
Field checking                                              fications such as SNH’s Natural Heritage Zones. The
Problems typically occur when attempting to dis-            choice of sampling intensity is often a simple trade-
tinguish between grasslands and agricultural land,          off between cost and precision. The NCMS sample,
and between improved and semi-improved grass-               for example, provided coverage of 7.5% of Scotland.
land. Areas identified as potentially unimproved            The ITE Countryside Survey provides a sample of
grassland in which no previous survey has                   less than 1% of Britain. The result is that area esti-
recorded this habitat should be checked from the            mates for NCMS are useful at a Scottish and local
ground by a surveyor trained in Phase I habitat             authority region level, whereas the Countryside
surveying. Local knowledge can be especially use-           Survey data are useful at a national level but
ful at this stage. Some limited field checking should       become increasingly uncertain at sub-national or
be carried out for all habitats to act as a calibration     regional levels. Fitness for purpose is a critical con-
for the aerial photograph analysis.                         sideration when designing sampling schemes and
                                                            it is always advisable to seek statistical advice in the
Data storage and analysis                                   design of such schemes.
Data storage
Good practice dictates that habitat maps created            Calculation of habitat areas
from aerial photographs should be stored in light-          The advent of digitising tablets has largely
proof cabinets to prevent colour fading. To                 removed the need to rely on manual methods of
ensure security, separate copies should be kept in          calculating areas from maps. The Romer dot grid,
172     6 METHODS FOR SURVEYING HABITATS



the planimeter and a host of other ingenious meth-       interpretation blocks on the basis of a checker-
ods have now been supplanted by standard func-           board design. This forces the need for correlation
tions for measuring map metrics within low-cost          around edges and acts as an internal check of
GIS or autocad facilities. These include not only        consistency.
area calculations but also a range of other metrics         Conventional methods of change detection
such as perimeter length (Wadsworth & Treweek,           essentially involve map overlay procedures and
1999). For those interested in fully exploiting the      hence the need for attention to detail in terms of
potential for deriving metrics from digital maps,        accurate geometry and labelling. An alternative
reference should be made to the spatial analysis         approach is borrowed from digital image proces-
program FRAGSTATS, which is specifically                 sing. In this, time series of aerial photographs are
designed for quantifying landscape structure (see,       scan-digitised and either co-registered (i.e. digitally
for example, Haines-Young & Chopping, 1996). The         superimposed) or registered to a common map
landscape indices provided by FRAGSTATS include          base and then superimposed. The advantage of
the following.                                           this method is that it makes no prior assumptions
                                                         about the structure of the landscape. Changes are
*   patch (e.g. mean patch size)
                                                         detected as differences between the images, and
*   edge (e.g. total edge)
                                                         the interpreter has to separate these according to
*   shape (e.g. average shape index: the patch para-
                                                         whether they are real changes or due to normal
    meter divided by the perimeter of a square of the
                                                         seasonal variations. This type of analysis is most
    same size)
                                                         extensively used in relation to the analysis of
*   core area (e.g. sum of core areas of each patch)
                                                         change from long time series of satellite imagery
*   landscape diversity (e.g. Shannon diversity index)
                                                         (e.g. to monitor tropical deforestation processes).
*   contagion and interspersion indices
                                                         It has only recently been applied to aerial photo-
                                                         graphy, but it offers major advantages in that it
Comparisons between photographs                          enables changes in quality to be assessed and can
The standard methods for change detection aim to         deal with changes across gradational boundaries.
control errors by close attention to geometric and       Research is currently being conducted at the
mapping accuracy. So, for example, the NCMS              MLURI to develop a knowledge-based change detec-
methodology used high-precision stereoplotting           tion system (SYMOLAC) that uses this approach
equipment to ensure accurate location of bound-          combined with advanced artificial intelligence
aries at each image date so that when the maps           methods. For an in-depth review of methods for
were overlaid any changes observed were real             change detection, refer to Lunetta & Elvidge (1999).
rather than artefacts of the method. For ‘look-             Change detection also poses problems in valida-
back’ studies conducted by the MLURI (alongside          tion. In general, changes are measured from a
the development of the LCS88), interpretation            present that is known to a past that is not (i.e.
errors were controlled by providing analysts with        historical interpretation has no equivalent ground
the 1988 interpretation and asking them to note          truth information). In the absence of historical
only actual changes. This approach both reduced          ground information with which to validate
the opportunity for changes due to line misregis-        the interpretation of the historical photography,
tration and improved the amount of intelligence          the only assumption that can be made is that the
being given to the interpreters. In both the NCMS        known error rates calculated for the present day
and the study of changes in the National Parks           apply equally to the historical dataset. It is essential
(Countryside Commission, 1991), the repeatability        to understand that, even with this assumption, the
of the photographic interpretation was tested by         error rates attached to the change dataset (i.e. when
having some areas examined by two different inter-       one interpretation is subtracted from the other)
preters. Where many interpreters are involved,           may be poorer as the errors may not coincide
another device to aid consistency is to allocate         spatially and will thus be inherited by the change
                                                          6.1 General habitat survey and monitoring methods    173



dataset. An analysis of sampling systems for                 Photographic records can also provide important
change detection accuracy assessment is provided             information on management operations and visual
in Chapter 15 of Lunetta & Elvidge ( 1999). Where a          evidence of changes in some environmental vari-
considerable investment is planned in developing a           ables, such as degree of flooding. Although not
habitat monitoring system based upon interpre-               directly related to monitoring, photographs can
tation of historical sequences of imagery, careful           also be useful for site familiarisation during desk
attention should be paid to the topic of accuracy            studies and provision of illustrative materials for
and the method of validation.                                reports, etc. They can also be very useful for con-
                                                             vincing people that change has actually occurred
Summary of advantages and disadvantages                      on a site if the timescale of change has been slow
Advantages                                                   and hence not very noticeable. The main value of
                                                             repeated photographs is that they provide a quick
*   Aerial photography provides a relatively quick
                                                             visual impression of change through time.
    assessment of extent of Broad Habitat types.
                                                                Fixed-point photography from ground stations
*   It can be used as a monitoring baseline.
                                                             is applicable to a wide range of habitats, but it
*   Historical trends can be examined by using past
                                                             must be recognised that in many more open habi-
    photographs.
                                                             tats major changes in the pattern of vegetation
*   The time taken for analysis compares favourably
                                                             communities will be more readily assessed by
    with a ground survey.
                                                             stereoscopic examination of aerial photographs
*   Boundaries can generally be more quickly and
                                                             (Section 6.1.3), provided the ground is not steeply
    accurately mapped than by a ground survey.
                                                             sloped. On steep slopes, fixed-point photography
                                                             can be more accurate than aerial photography.
Disadvantages                                                Photographs can also be used to record individual
                                                             quadrats if taken from directly above, and these
*   Fewer habitat details can be distinguished than
                                                             can be analysed objectively at a later date if neces-
    with a Phase I habitat survey.
                                                             sary. Other alternative/additional methods to con-
*   If photographs are of poor quality, accurate analy-
                                                             sider include Phase I mapping (Section 6.1.5) and
    sis is not possible.
                                                             NVC surveys (Section 6.1.6).
*   Some habitats can be hard to distinguish on
                                                                Fixed-point photography of permanent quadrats
    photographs, necessitating field checking of
                                                             is often used for monitoring individual species,
    results. Interpretation errors increase with the
                                                             particularly fungi, lichens and bryophytes (see
    level of detail that is attempted.
                                                             Part III).
*   Measured habitat area will be underestimated for
    slopes unless three co-ordinates are used to digi-
    tise maps. Likewise, high-altitude areas will be
                                                             Time efficiency
    overestimated relative to low-altitude ones.
                                                             Photography is a cost-effective method for record-
*   Unknown bias is introduced if habitat extent
                                                             ing and monitoring change, being relatively cheap
    is estimated from incomplete photographic
                                                             and straightforward to carry out. If the use of ran-
    coverage.
                                                             ging poles is necessary for lining up repeat photo-
                                                             graphs or indicating vegetation height (see below),
                                                             it will be more efficient to employ two people.
6.1.4       Fixed-point photography
                                                                The time required to sort, document and store
Recommended uses                                             photographs should not be forgotten or underesti-
Fixed-point photography is considered to be an               mated as this is critical to the success of photo-
essential part of many monitoring programmes,                graphic monitoring programmes. Interpretation
as it provides a relatively simple method of record-         of changes is normally by subjective visual assess-
ing broad changes in vegetation and habitat.                 ment and therefore relatively quick and easy.
174      6 METHODS FOR SURVEYING HABITATS




  Fixed-point photography: summary of key points                  *   Medium-speed black and white film can be used.
                                                                      Colour will allow greater precision when distin-
  Recommended uses                                                    guishing features/boundaries. High-quality
                                                                      (i.e. sharp and correctly exposed) photographs
  *   Monitoring broad changes in habitat extent and
                                                                      are required. Avoid very bright days, when
      type, e.g. through succession and patch dynamics
                                                                      shadows create contrasting images that lack
  *   Photographic records are useful for many purposes
                                                                      detail
      other than monitoring, e.g. recording management
                                                                  *   Standard methods should be used that allow
      operations and site familiarisation
                                                                      photographs to be repeated with the same camera
  Efficiency Relatively quick to carry out in the field by            configuration at the same position at the same time
  one person; some additional time required to manage                 of year
  photographic records; subjective analysis is quick and          *   It is useful to include graduated ranging poles in
  easy                                                                photographs as markers against which vegetation
  Objectivity     Normally subjective, although semi-                 height or water levels can be assessed
  objective counts or measurements can be obtained                *   Linkage of photographic recording to other moni-
  with appropriate set-ups                                            toring projects (especially aerial photography) is
  Precision     Not measurable, as data are usually                   useful
  qualitative                                                     *   It is essential that photographs and details of
  Bias Not measurable, as data are usually qualitative                their location, direction, timing and camera con-
  Expertise required     Only a basic knowledge of photo-             figuration, etc. are properly stored so that succes-
  graphy; other techniques are easily learnt                          sive photographs can be easily retrieved and
  Equipment required        Basic photographic equipment              compared
  and tripod (cost c. £500)
  Key methodological points to consider
                                                                  Data analysis    Photographs are normally analysed
  *   Little need to anticipate changes that are likely to take   subjectively. Analysis is therefore sensitive to indivi-
      place, especially if comprehensive coverage is attained     dual interpretation




Expertise required                                                   Using the full procedure detailed below will
A basic knowledge of 35 mm single-lens reflex                     increase the time and hence costs of monitoring,
photography is essential. Although photographic                   and is therefore best restricted to cases where:
monitoring is relatively simple, and can probably
be carried out effectively by most people if they                 *   quantitative measurements are going to be made
follow procedures carefully (as described below or                    from the photographs for analysis; or
in other methodological descriptions), training in                *   you are working in a habitat or landscape with
photographic monitoring is desirable.                                 little in the way of distinctive permanent features
   If the fixed-angle method described below is to                    with which to orientate yourself by using previous
be used, it is recommended that each step be learnt                   photographs in the field.
and rehearsed thoroughly before embarking on the
actual photography at the site. The procedure is                  Equipment required
simple once mastered, but care should be taken to                 The equipment required (see Appendix 6) for
avoid mistakes, especially when taking and regis-                 fixed-point photographic monitoring requires a
tering the first photograph in a programme.                       capital expenditure of about £500. However, this
                                                             6.1 General habitat survey and monitoring methods    175



equipment can be used to monitor many sites and                    Digital cameras are now available and allow easy
most of it is likely to be of use for other purposes.           computer storage and manipulation of images.
    The basic requirement of fixed-point photogra-              Many digital cameras can now match the resolu-
phy is that the system must enable the precise                  tion of film. If the technology remains in long-term
relocation of the camera position in both the hor-              use, digital images are preferable to film.
izontal and the vertical plane. There must there-
fore be provision of a permanent marker, whether
                                                                Field methods
already existing or installed especially for the pur-
                                                                Number and layout of photographs
pose (see Appendix 5 for a discussion of the use of
                                                                Perhaps the most difficult decisions to make in the
permanent markers). The use of a firm tripod that
                                                                planning of photographic monitoring is the num-
prevents camera shake and allows the camera to be
                                                                ber and location of positions from which photo-
fixed at the right angle above the right spot is also
                                                                graphs are to be taken. Where possible, locations
recommended.
                                                                should be integrated with other monitoring pro-
    The type of tripod used is particularly important
                                                                jects and also aim to include representative views
if the fixed-angle method (described below) is to be
                                                                of all the interest features on the site and all
used. For this method NCC (1987) recommend the
                                                                major forms of management of each. The Site
use of a Linhof Propan pan and tilt head (Model II).
                                                                Management Statement or Management Plan
This can be used with many tripods, but care must
                                                                should therefore be consulted, together with any
be taken when purchasing to ensure that it is com-
                                                                available survey information (particularly habitat
patible and that any existing pan and tilt head on
                                                                maps) in order to identify appropriate locations.
the tripod is removable. The use of a ball and socket
                                                                It is often useful to take photographs that show a
head between the tripod and the pan head is also
                                                                range of scales including the subject in relation to
recommended to avoid the necessity for levelling
                                                                its surrounding landscape, the subject itself and
the tripod.
                                                                representative detailed shots of the subject.
    For long-term monitoring purposes monochrome
                                                                    The number of locations chosen, and the range
photographic prints are normally recommended as
                                                                of photographs taken at each, will clearly depend
the best material (NCC, 1987) as the negatives and
                                                                on the size and complexity of the site, but should
prints generally last longer in long-term storage than
                                                                not be allowed to become unworkable. In some
do colour equivalents, particularly if properly pro-
                                                                cases it may be appropriate to devise a programme
cessed for archival storage. If monitoring is also
                                                                whereby different locations are recorded in differ-
used for informing management decisions in the
                                                                ent years and at different return intervals, depend-
short term, colour film can provide more informa-
                                                                ing on the speed of change anticipated in the
tion and better discrimination between habitat
                                                                feature or attribute (see below). However, it is pre-
types. You can always use two film backs or camera
                                                                ferable to take more photographs than necessary
bodies with both monochrome and colour films. If
                                                                than not to take enough.
colour slide film is used, the slides can last for several
decades if infrequently projected and stored in cool,
dark and dry conditions.                                        Frequency and timing of photographs
     The speed of the film used is also an important            In most habitats and situations vegetation changes
factor to consider. Fast films will yield less fine             detectable from fixed-point photography are unli-
detail than will slow films because of their inherent           kely to be noticeable at intervals of less than 5–10
grain size, but allow the use of faster shutter speeds          years. However, where sudden changes are
and/or smaller apertures to give greater depths of              believed to have occurred, e.g. as a result of man-
field. However, as the photographs are taken from               agement or natural events, such as storms, photo-
tripods, slow shutter speeds can often be used satis-           graphs should be taken to record changes
factorily in suitable conditions. On balance, there-            irrespective of the time elapsed since the previous
fore, a medium-speed film is the best compromise.               monitoring. Some habitats (e.g. sand dunes) are
176    6 METHODS FOR SURVEYING HABITATS



notoriously dynamic and therefore prone to mostly         Fixed-angle system
gradual but potentially rapid changes. Photographic       All the equipment necessary for this system is
monitoring of features and their attributes in such       described above (see Appendix 6) and should be
habitats should therefore be more frequent. The           used in strict accordance with the following NCC
recommended frequencies for monitoring habitat            (1987) procedure.
features are summarised in Chapter 5.
   It is important that the time of year at which         A  Initial photographs
photographs are taken within any particular               A1 Locate or install permanent marker (see
sequence is consistent. If possible, a date should be        Appendix 5), making sure enough information
chosen when seasonal vegetation change is relatively         is recorded for its relocation.
stable so that the effects of annual variations in        A2 Set up tripod over the marker; centre with aid
weather or vegetation are minimised. Alternatively,          of a plumb bob slung from base of centre col-
some phenological cue may be used, such as the               umn, and adjust its height so that the camera
flowering of a key species. Where regular manage-            lens will be at a standard height above ground
ment is practised (e.g. mowing), photography should          level.
be consistently before or after such activities. In       A3 Using ball and socket head: set pan head to level
woodlands, features of tree- and shrub-layer struc-          position (red line) on tilt scale; fit pan head to
ture are best recorded during the winter. In some            tripod, level pan head, clamp ball and socket.
habitats, winter and summer records may be useful,           Without ball and socket head: set pan head to
for example for comparison of high and low water             level position (red line) on tilt scale; fit pan head
levels.                                                      to tripod, level pan head by manipulating tripod.
                                                          A4 Fix camera to pan head, in horizontal config-
Taking initial and repeat photographs                        uration, positioned so that the back of the base-
The simplest method of taking repeat photo-                  plate aligns with mark scribed across the head.
graphs is to relocate the fixed points and use            A5 Supporting camera, release the two locking
previous photographs as guidance for lining up               knobs on the pan head, and move camera until a
the shot. This may be adequate for many gen-                 suitable view is framed. Lock knobs tight.
eral purposes, but for systematic monitoring              A6 Adjust camera settings and take photographs.
two systems for fixed-point photography are                  NCC (1987) recommend two photographs are
recommended by NCC (1987) and are described                  taken. However, given the relatively low cost of
below. The first is the ‘fixed-angle system’                 film compared with that of staff time, we
(Bignal, 1978) used to monitor Loch Lomond.                  recommend that bracketed exposures are
This is recommended for National Nature                      taken, i.e. one photograph at the predicted
Reserves (NNRs) and other situations in which                ideal exposure, one underexposed by one f-stop,
intensive observations over a long period of                 and one overexposed by one f-stop. Digital
time are envisaged. The alternative ‘centre pole             photography is less expensive than film.
system’, devised by M. J. D’Oyly, is less time-           A7 Select a feature that appears at the centre of the
consuming in the field and is recommended                    horizontal field of the photograph. Step back at
where there are problems with access, topogra-               least two paces, and take a compass bearing
phy or a shortage of resources.                              over the lens centre of the camera to the
   The method chosen to relocate photographs will            selected feature. (If there is no suitable feature it
depend on the actual use to which they will be put;          may be necessary to locate a ranging pole, or
there is little point in precisely relocating points by      similar marker, in the middle view, but this is
using the fixed-angle system if one is using the             laborious unless an assistant is available.)
photographs to make a simple subjective assess-              Record the bearing.
ment of broad changes.                                    A8 Record tilt angle from the pan head scale.
                                                         6.1 General habitat survey and monitoring methods    177



A9 Make sure that the following have been                        height, lens used and focal length, any filters
    recorded: date, time of day, site, location refer-           used, film type, exposure details, film roll and
    ence, compass bearing, tilt angle, camera                    exposure numbers. Notes on weather condi-
    height, lens used and focal length, any filters              tions and features shown in the view should
    used, film type, exposure details, film roll and             also be recorded.
    exposure numbers. Notes on weather condi-               B9   Before dismantling equipment and moving on,
    tions and features shown in the view should                  make sure that all views from the point have
    also be recorded.                                            been repeated.
A10 Before dismantling equipment and moving on,
    consider whether one or more additional
                                                            Centre pole system
    photographs could usefully be taken from the
                                                            The centre pole system requires the equipment
    same point (e.g. perhaps as part of a panoramic
                                                            detailed in Appendix 6 except the pan head and
    view). Also (although not mentioned in the
                                                            compass, and was devised to monitor woodland
    NCC instructions), it is recommended that a
                                                            cover in an area already covered with permanent
    photograph of the tripod is taken before it is
                                                            transects marked by metal stakes at 100 foot
    removed from the fixed marker. This will help
                                                            (approximately 30 m) intervals (NCC, 1987). The
    relocation of the marker on future occasions.
                                                            method involves photographing forwards and
B   Repeat photographs
                                                            backwards along the transect at each stake. This
B1 Locate permanent marker (see Appendix 5).
                                                            generally requires taking a large number of photo-
B2 Follow steps A2, A3 and A4 as described above.
                                                            graphs in a limited time, and the system was
B3 Holding compass, step back two paces and,
                                                            designed with this in mind. This method is perhaps
    sighting over the lens centre of camera, select
                                                            best used in areas in which transects or gridlines
    distant feature that corresponds to the bearing
                                                            are already marked, unless it is decided to set up
    recorded for the first photograph. In the
                                                            permanent markers before photographic monitor-
    absence of a suitable feature, align assistant
                                                            ing takes place, which will involve additional time
    with ranging pole in middle of view or, if single-
                                                            and expense (see Appendix 5 for advice on setting
    handed, align ranging pole by taking a back
                                                            up permanent markers).
    bearing on the camera.
                                                               The main feature is a standard ranging pole
B4 Align camera so that selected distant feature or
                                                            placed at a set distance from the camera; the camera
    ranging pole is central in the horizontal field of
                                                            is centred on the mid-point of the pole. The proce-
    view. Clamp horizontal scale of pan head.
                                                            dures for first and repeat photographs are the same,
    Remove ranging pole (if used), or move to most
                                                            and follow guidelines detailed by NCC (1987).
    appropriate location if required in shot for
    scaling or estimation of vegetation height.             1. Set up camera at marker, normally in horizontal
B5 Set tilt scale to angle recorded for first photo-           configuration (although vertical configuration
    graph. Clamp tilt scale.                                   can be used on steep slopes if necessary), on tri-
B6 Check against print of first photograph that the            pod. Centre over the marker with the lens a stand-
    same view is again framed. If there is clearly a           ard height from the ground.
    discrepancy, not explained by the passage of            2. Measure a standard distance along transect
    time, check, and, if necessary adjust to corres-           towards next marker, and plant the ranging pole
    pond with the original photograph, and record              (upright and carefully aligned) at this point.
    new bearing and tilt angle.                             3. Free the ball and socket head, and centre the aim
B7 Adjust camera settings and photographs (see A6).            of the camera on the mid-point of the ranging
B8 As under A9, make sure that the following have              pole.
    been recorded: date, time of day, site, location        4. Clamp the tripod head, adjust camera settings
    reference, compass bearing, tilt angle, camera             and take two photographs.
178    6 METHODS FOR SURVEYING HABITATS



5. From all but the first marker, repeat steps 2–4             Ideally this should be done at a simple level as soon
   with the camera directed back to the previous               as each new set of photographs is taken.
   marker.                                                        In some instances, simple mapping of habitat
                                                               changes, such as area covered by scrub, may be use-
The type of pole and its distance from the camera
                                                               ful. In turn this can provide quantitative estimates of
should be standardised for any particular set of
                                                               habitat area and change. Mapping is normally done
photographs. The pole should also be buried to a
                                                               by eye, although in some circumstances images can
standard depth on each occasion. Alternatively, a
                                                               be scanned or digitised and areas, etc, calculated by
ranging pole support can be used on all occasions
                                                               computer. Aerial photographs are, however, more
so that the tip of the pole rests on the ground and
                                                               appropriate for such analysis techniques.
the pole is held vertically.
                                                                  Box 6.6 outlines the problems that may be
                                                               encountered with fixed-point photography, and
Data storage and analysis
                                                               some possible solutions.
Films should be processed and record sheets and
photographs indexed and filed as soon as possible
after shots are taken. A good indexing and filing              Summary of advantages and disadvantages
system is essential. Where possible, the slide index           The main advantage of photography for moni-
should be computerised and stored in a database                toring is that there is little or no scope for
or spreadsheet for easy search and retrieval.                  subjectivity in anticipating the changes that
Duplicate sets of photographs should be stored in              are likely to take place (NCC, 1987). In any
appropriate separate locations. Storage of digital             other form of recording there is always the pos-
images is simpler.                                             sibility that some critically important observa-
   Analysis is normally by subjective comparison               tion will be omitted from the record simply
(by eye) of a series of photographs taken over                 because of an assumption that change would
a period of time, recording any obvious changes.               affect particular attributes of the habitat or be



  Box 6.6      Likely problems and solutions                      Difficulties may also arise on mobile habitats. On
                                                               active dunes, for instance, the natural movement of
  The most likely problems to be encountered are due to        dune ridges can lead to permanent markers being
  inconsistent camera configurations and inaccurate            undercut or buried. Even if the precise horizontal
  relocations. These can be avoided by carefully               location can be found, it may differ considerably in
  following the prescribed methods and thorough and            height. Similar changes can occur on saltmarshes,
  accurate recording of data.                                  cliffs, and habitats prone to landslides. In such
      Other problems can be expected to arise from the         situations a regular system of photographic recording
  very changes that are being monitored. Tall herbage or       points may be inappropriate and it may be better to use
  woody growth may develop on or just in front of the          carefully selected camera locations that can be
  camera position. This may lead to a temporary                reasonably expected to remain stable. In some
  interruption in photographic records. If this is likely to   circumstances the construction of purpose-built plat-
  be for a long time (e.g. through tree growth) or is          forms for photographic recording may be necessary.
  unacceptable for other reasons, then it may be appro-           Weather conditions can also cause problems. Wet
  priate to relocate the observation point. This should be     and misty conditions should be avoided, and very
  precisely recorded. Sometimes a degree of ‘gardening’        bright and sunny weather can also be problematic.
  in the foreground may be justifiable, although in other      Shadowing obscures useful detail and can confuse
  cases it may be unacceptable disturbance to natural          interpretation and identification of features and
  processes.                                                   attributes.
                                                              6.1 General habitat survey and monitoring methods             179




  Phase 1 habitat mapping: summary of key points                 Expertise required      Surveyors must be able to recog-
                                                                 nise the dominant and other characteristic plant spe-
  Recommended uses                                               cies necessary for the identification of Phase I survey
                                                                 habitat types. Further botanical expertise is desirable to
  *   Providing relatively rapid records of vegetation and
                                                                 enable extra information to be recorded in the form of
      wildlife habitat over large areas of countryside
                                                                 target notes
  *   Providing an objective basis for identifying sites
                                                                     Mapping skills are also necessary for surveying in
      warranting more detailed surveys (e.g. Phase II,
                                                                 the field and for the production of master maps
      also known as NVC) or deserving consideration for
                                                                 Equipment required       No specialist equipment
      protection
                                                                 required for surveying; basic requirements include
  *   Preparation for planning a monitoring programme,
                                                                 Phase I manual, botanical field guides (see Appendix 3),
      including identification of features and sampling
                                                                 binoculars and maps. Equipment for producing master
      area boundaries
                                                                 maps and calculating habitat areas includes basic
  *   Monitoring large-scale changes in the extent and
                                                                 office equipment and Romer dot grids
      distribution of distinct Broad Habitat types
                                                                 Key methodological points to consider
  Efficiency   Relatively rapid, ranging between c. 1 and        *   Surveyors must work to a consistent standard to
  6 km2 per surveyor per day
                                                                     ensure accuracy and compatibility between surveys
  Objectivity Reasonably objective as long as surveyors          *   Planning of fieldwork is necessary to ensure that the
  are adequately trained in surveying and mapping
                                                                     survey area is covered in the field season and habi-
  techniques
                                                                     tats are visited when key species are readily
  Precision    The errors involved in measuring habitat
                                                                     identifiable
  areas on 1: 10 ,000 scale maps are generally likely to be      *   Aerial photographs can be useful to increase the
  well below 5%, although the original boundaries may
                                                                     speed and efficiency of mapping, particularly in
  be more variable
                                                                     areas of difficult or restricted access
  Bias    Estimates of habitat area can be biased by             *   If habitat areas are estimated by using sampling
  misidentification of vegetation or inaccurate
                                                                     methods, care must be taken to avoid bias arising
  mapping, especially in fragmented and mosaic
                                                                     from the non-random distribution of habitats
  habitats. The use of aerial photographs as an adjunct
  to boundary mapping can help to address this. Small            Data analysis      Areas of different habitat types are
  rare habitat types can be over or underestimated if            best calculated manually by using a Romer dot grid.
  areas are calculated from maps by using sampling               Maps can also be digitised, analysed and stored in a GIS.
  techniques. Habitat areas on hillsides will be                 Areas can be expressed as percentage of a given area
  underestimated if dot grids are used on                        covered by each habitat type, or as total area covered by
  two-dimensional maps. If areas are measured with a             each habitat type. Data can be used as a baseline for
  planimeter or with digitising equipment, altitude can          future monitoring, but considerable care must be
  be included from spot heights to allow some                    taken in the interpretation of changes because of
  consideration of slope to be made                              potential inconsistencies between surveys



in a particular direction. Photographic recording in             change that cannot be quantified or tested by
the field is also relatively quick and simple in com-            objective statistical methods.
parison with other monitoring methods and pro-
vides an easily interpretable visual picture of
                                                                 6.1.5        Phase I habitat mapping
change with time.
   The main disadvantage of photographic moni-                      Anyone carrying out a Phase I habitat survey will
toring is that it only gives broad indications of                require the Nature Conservancy Council (NCC)
180    6 METHODS FOR SURVEYING HABITATS



publication Handbook for Phase I Habitat Survey: A        periods for illustrative and interpretative purposes,
Technique for Environmental Audit (NCC, 1990a). It is     but are likely to be too insensitive and unreliable
beyond the scope of this section to reproduce the         for detecting small changes. Phase I mapping may
specific habitat classifications and codes necessary      be appropriate for distinct habitat types with
for Phase I survey, which are contained in the NCC        sharply delimited boundaries in which mapping
handbook. This section therefore presents a synopsis      can be carried out with the aid of aerial photogra-
of the manual for general information and reference       phy (Section 6.1.3). Under such circumstances,
purposes. The reader is referred to the NCC hand-         apparent changes in Broad Habitat extent and dis-
book itself for specific definitions and procedures.      tribution can be treated with reasonable
                                                          confidence.
Recommended uses
Phase I habitat surveying is a standardised system        Time efficiency
developed by the NCC for classifying and mapping          Surveyor fieldwork rates will depend on many
wildlife habitats in all parts of Britain. Phase I sur-   factors, including the relative competence of sur-
veys can provide, relatively rapidly, a record of semi-   veyors and the topography, complexity, number of
natural vegetation and wildlife habitat over large        target notes recorded, and accessibility of the area
areas of countryside. Habitat classification is based     to be surveyed. The scale of mapping also affects
principally on vegetation, augmented by reference         survey rates.
to topographic and substrate features, particularly          Phase I survey rates per surveyor have ranged
where vegetation is not the dominant habitat              from 0.8 km2 to 6.4 km2 per day. Assuming a total
component.                                                of 90 field survey days per year, a total of 81–580 km2
   The information provided by a Phase I survey           can be covered by an individual surveyor in one field
has many uses: it can provide an objective basis          season (NCC, 1990a). In practice, the upper end of
for determining whether a site merits more                this scale is extremely ambitious and should not be
detailed Phase II surveys (Section 6.1.6) or whether      used for calculating effort required to survey a site.
it deserves consideration for protection as an SSSI,         As a further approximate guide to the break-
Local Nature Reserve, etc.                                down of mapping stages, the time taken to produce
   In a monitoring context, an initial Phase I survey     and analyse a 5 km  5 km 1 : 10 000 scale habitat
is a useful precursor to the design of a new mon-         map is:
itoring programme for a site. Information from a
                                                          *   Field survey and production of fair copy 8–10 days
Phase I map can be used to establish feature and
                                                          *   Production of final copy from fair copy 1.5–2.5
sampling area boundaries and on occasions to iden-
                                                              days
tify strata for stratified sampling, although differ-
                                                          *   Analysis of final copy by using dot grid 1.0–1.5
entiation of vegetation types may be better carried
                                                              days
out by using Phase II (NVC) surveys. A Phase I map
can also be used as a clearly defined baseline for        Based on Phase I surveys of Cumbria and
monitoring changes. However, variations by sur-           Lancashire 1983–88 from NCC (1990a).
veyors in the identification of habitat types and
boundaries are sufficiently high to significantly         Expertise required
limit the reliability of changes deduced from repeat      Surveyors should be competent botanists with an
Phase I mapping (see Box 6.10; see also Cherrill &        aptitude for accurate field recording and mapping.
McClean ( 1999a, b), who found only c. 26% corres-        It is essential that surveyors be adequately trained
pondence between maps). Therefore, data from              to ensure accuracy and consistency both within
Phase I mapping are only likely to be suitable for        and between surveys. Discrepancies between sur-
detecting large-scale changes in Broad Habitat            veyors can be reduced if surveyors are trained to a
types over relatively long time periods. Such             uniform standard. Detailed descriptions of Phase I
broad data can be useful when collected over long         habitat types, colour codes and alphanumeric
                                                         6.1 General habitat survey and monitoring methods      181



symbols are given in the Handbook for Phase I Habitat          Each distinct habitat unit is recorded in the field
Survey Field Manual (NCC, 1990b). A thorough knowl-         by using standard coloured pencils or alternative
edge of the major vegetation types and habitats is          lettered/alphanumeric codes. Colours and codes
necessary, including their dominant and character-          should be entered directly on to copies of the
istic species. Further botanical skills are desirable:      large-scale Ordnance Survey maps.
these allow extra information concerning species               There are advantages in mapping directly in col-
composition to be recorded by using target notes,           our in the field, but some surveys have chosen to
which draw attention to particular features of              use pen or pencil only, mapping habitat bound-
interest.                                                   aries and using codes to identify habitat types.
   Surveyors should also be trained in other field-         This method is quicker and more convenient, par-
work skills, including the use of binoculars in             ticularly in wet conditions and when recording
vegetation survey, mapping techniques, naviga-              uncomplicated areas. The use of colour is prefer-
tion and route finding, habitat identification,             able in complex areas and where there are large
and indications of trophic status, soils and                amounts of semi-natural vegetation. Pencil marks
land management. Cherrill & McClean (1999a,b)               can also be altered at a later date, something that
recommend aerial photographs to help improve                should be avoided.
mapping.                                                       It is important to standardise the minimum size
   The work is physically demanding, so surveyors           of habitat unit to be mapped. It is suggested that at
should be fit and healthy.                                  1 : 10 000 scale all habitat units larger than 0.1 ha
   Writing and numerical skills are required for the        should be mapped, and at 1: 25 000 all units larger
production of target notes and reports, and the             than 0.5 ha should be mapped (NCC, 1990a,b). It is
ability to produce neat final maps is essential if          possible to map smaller units such as ponds, and
cartographers are not employed.                             target notes can also be used to draw attention
                                                            to small areas of noteworthy habitat. It is also
Equipment required                                          important to agree protocols for mapping fragmen-
Appendix 6 summarises the equipment required                ted or mosaic habitats.
for Phase I habitat surveying.                                 The overall aim of a target note is to give a concise
                                                            picture of the nature conservation interest of a site
Field methods                                               in the context of its land use and management. They
Outline method                                              must be clear, succinct and informative; even the
Ideally, a trained surveyor will visit every parcel of      briefest description can enhance the usefulness of
land in the area to be surveyed. The vegetation is          the habitat map. Target notes are used to provide
mapped on to Ordnance Survey maps, usually at a             extra detail in particular habitat types, or to point
scale of 1 : 10 000. An area of vegetation is assigned      out areas of interest that would otherwise not be
to one of some 90 specified habitat types, identified       recorded by using the standard habitat codes. They
on the map by standard colour codes or symbols. In          are very important as they can provide an indication
addition, further information is recorded by the            of areas that might require further study, as well as
use of dominant species codes within many habitat           providing useful additional information on any
types, and by the use of descriptive ‘target notes’,        other features of note identified by the surveyor,
which give a brief account of particular areas of           such as uncommon or rare plant species, mammal
interest. Habitat types and codes are described in          signs, bird species, etc. A target note is recorded as a
detail by NCC, (1990a,b). Only the standard colours         red circle with an individual number on the map;
in the Berol Verithin series should be used. These          explanations of the reasons for the target notes are
are available from stationers or from Berol Ltd,            included with the final survey report.
Oldmeadow Road, King’s Lynn, Norfolk PE30 4JR.                 Dominant species in each habitat unit should be
Marks made with these pencils, however, do not              recorded wherever possible by using standard spe-
photocopy or scan into computers well.                      cies codes given by NCC (1990b).
182    6 METHODS FOR SURVEYING HABITATS



   In practice much of the mapping can often be              time so that no gaps are left, has much to recom-
carried out from public rights of way, with the use          mend it. However, some habitats are best surveyed
of binoculars at relatively short ranges to identify the     at different times of year from others; woodlands
vegetation. This avoids the time-consuming process           in spring, grasslands in midsummer, heathlands in
of seeking access permission, although this will be          late summer and autumn and open waters between
necessary in areas where no rights of way occur.             mid-June and September (see Box 6.9).
However, the quality of grazed grassland can be                 To survey an area one habitat at a time at the
very difficult to assess without visiting the actual site.   ideal time for each habitat is likely to be costly and
   Aerial photographs may also be useful as an               time-consuming, involving repeated visits to each
adjunct to ground surveying. Although aerial                 area. A suitable compromise would be to survey
photography is no substitute for fieldwork when              areas most rich in woodland in spring and early
carrying out Phase I surveys, the availability of            summer, areas most rich in grasslands in midsum-
contemporary aerial photographs at a suitable                mer, and areas most rich in moorland later in the
scale can increase the speed and efficiency with             season. Within these areas all habitats should be
which field surveys are carried out and the accu-            surveyed at the same time.
racy of mapping boundaries. Photographs can also                The field season should be considered as starting
be used to map areas with difficult or restricted            in late March – early April in southern and central
access or for mapping the interiors of large woods.          Britain and late April – early May in the north of
   Informal or fixed-point photographs may also be           Britain. The season generally ends about mid-
useful, especially for subsequent interpretation of          October, although it may be possible to undertake
differences that might simply be the result of sur-          some surveying in November if the weather is
veyor variation. For large sites a large number of           mild. End-of-season surveys should generally be
photographs might be required, but this may not              restricted to checking previously surveyed areas;
be practicable.                                              data from such surveys should be treated with cau-
                                                             tion because many plant species will no longer be
Choice of scale                                              apparent.
Phase I surveys are mapped onto either 1 : 10 000               Each day’s fieldwork should be carefully
or 1 : 25 000 scale Ordnance Survey maps.                    planned to ensure that the maximum amount of
Generally, countrywide Phase I surveys have                  ground is covered, and to minimise back-tracking
been carried out at either scale, but there has              and overlap. Care should be taken to ensure that
been an increasing tendency to standardise on a              the whole area is covered; a gap may mean that
scale of 1 : 10 000 despite some of the advantages           another visit will be necessary.
of the smaller scale.
    There is no doubt that, for some uses, a 1 : 10 000      Data storage and analysis
scale is desirable as it allows greater detail to            The field maps made by surveyors are transferred
be recorded, but it is recognised that for very              to ‘fair’ maps either by the surveyors themselves or
large sites, such as in the Scottish Highlands, a            by cartographers. Surveyors are likely to be more
1 : 25 000 scale survey may be the only economic-            precise, because they are familiar with the areas
ally feasible option. If surveys are carried out at this     being surveyed, whereas cartographers will gener-
scale it is recommended that full use be made of             ally produce more consistent and neater maps. Fair
target notes to provide greater detail.                      maps can be monochrome or colour, but the final
                                                             objective is to produce an accurate, full-colour
Survey preparation                                           master habitat map, which has a high visual impact
A work programme should be planned carefully at              and is easy to interpret.
the beginning of the survey to ensure that the area             The procedure for the preparation of master
to be surveyed is covered in one field season. A             maps has varied from survey to survey; for a sum-
systematic approach, completing one map at a                 mary of different methodologies see NCC, (1990a,b).
                                                        6.1 General habitat survey and monitoring methods      183



   Master maps should be stored in lightproof cab-         Wadsworth & Treweek, 1999). The time required
inets to minimise colour fading. Colour copies             to digitise both the base map and the habitat
should also be kept for security. Habitat maps are         overlay will be considerable but, once completed,
most easily reproduced by photocopying, either in          data processing and extraction and area calcula-
colour or in monochrome. A monochrome copy of              tion is very quick and accurate. This also allows
the master map allows black and white copies to            high-quality colour plots to be made whenever
be made as required. Habitat area measurements,            needed.
completed target notes and general description                For monitoring purposes, it is recommended
sheets should be stored on paper in grid square            that areas of all habitats be measured as described
order. Fair maps produced in drawing packages              above, preferably with the more accurate and
can be stored electronically.                              sophisticated GIS technology. If only a rough idea
                                                           of habitat extent is required, estimates of habitat
Data analysis                                              abundance can be obtained by using sampling pro-
For monitoring purposes, either to establish a base-       cedures. Refer to NCC (1990a,b) for further details
line or for comparison with a previous baseline            of sampling strategies for area estimation.
survey, the area covered by each habitat type can             Calculated areas should be entered into a spread-
be measured from Phase I habitat maps. In order to         sheet for graphical and statistical analysis, and for
simplify this task, NCC (1990a, b) suggest that the        ease of data storage and retrieval.
90 or so Phase I habitat classifications be combined
to give 34 categories for measurement. Consistent
                                                           Summary of advantages and disadvantages
use of these groups will allow quick comparisons to
                                                           Advantages
be made between different surveys and will facili-
tate the compilation of regional and national stati-       *   Provides relatively rapid record of vegetation and
stics on habitat extent.                                       habitat type over large areas of countryside.
   Measurements of the extent of each of the               *   The use of standard methods and recording pro-
different types of habitat in the area covered by              cedures allows easy general comparisons between
the survey can be made with a Romer dot grid.                  different surveys to be made.
Planimeters are not sufficiently accurate for the          *   Habitat maps can provide valuable information
measurement of small areas. A dot grid is a trans-             for planning site monitoring programmes.
parent plastic sheet covered in regularly spaced           *   The level of detail obtained in a Phase I survey is
dots at a given density (e.g. 10 per cm2). The grid            higher than that obtained by using aerial photo-
is placed over the map and the numbers of dots                 graphy or remote sensing.
falling in each habitat type are counted. The area of      *   The use of descriptive target notes can draw atten-
each habitat is calculated from the map scale                  tion to areas that merit further study and can
and dot density. For example, at 1 : 10 000 scale              record additional ecological information that
with a grid dot density of 10 per cm2, 1 cm = 100 m;           might be of interest.
1 cm2 = 0.01 km2; 1 dot = 0.01/10 = 0.001 km2. So if
one habitat type is covered by 125 dots, its area =
                                                           Disadvantages
125 Â 0.001 = 0.125 km2.
   The advent of digitising tablets has largely            *   Requires substantial amount of fieldwork before
removed the need to rely on manual methods of                  maps can be compiled and the data analysed.
calculating areas from maps. Dot grids and plani-          *   Habitat classes are relatively broad, so finer-scale
meters have now been supplanted by standard                    variation will be missed.
functions for measuring map metrics within low-            *   Discrepancies between individual surveyors can
cost GIS or autocad facilities. These not only                 lead to biased area estimates.
include area calculations but also provide a range         *   A problem can arise when trying to accurately
of other metrics such as perimeter length (see                 map boundaries between habitat types where
184     6 METHODS FOR SURVEYING HABITATS



    there is a gradual transition between the two. This    a feature may be defined as a rich diversity, distinct-
    can lead to biased area estimates.                     ive mosaic or zonation of NVC types. Where vegeta-
*   Errors made on original survey maps cannot be          tion features have not been clearly defined in NVC
    checked without returning to the field.                terms or other ways, a subsequent NVC survey may
*   Areas where access is denied, or where access is       be used as a post hoc means of providing a precise
    problematic, such as large, dense woodlands or         definition of features on the basis of a standardised
    wetland, can be difficult to survey from the           technique and classification system. Such surveys
    ground.                                                may also reveal or highlight previously unrecorded
*   Overall, the method is too insensitive and unreli-     features of interest.
    able for most site monitoring requirements, but it        Where NVC types are not named as whole fea-
    is particularly useful in EIA studies.                 tures, they may be important attributes of broader
                                                           habitat features (see Davies & Yost, 1998).
A case study of the use of repeat Phase I habitat
                                                           Furthermore, the presence of locally determined
surveys for monitoring vegetation at three sites is
                                                           constant species, preferential species (i.e. those
presented in Box 6.7.
                                                           that predominantly occur in one community type)
                                                           and associated species (i.e. other species that occur
                                                           in the NVC type) is also a monitoring requirement
6.1.6        National Vegetation Classification
                                                           for some habitat types. Thus, to some extent the
            (NVC) surveys
                                                           NVC is often used as a basis for defining the condi-
Recommended uses                                           tion of vegetation. However, great care should be
The National Vegetation Classification (NVC), pub-         taken in the use of NVC datasets from British Plant
lished in British Plant Communities (Rodwell et al.,       Communities as standards for defining ‘poor’ and
1991 et seq.), is the standard phytosociological clas-     ‘good’ examples of vegetation stands. Such data
sification method in Britain. It offers a reliable         should not be used to normalise management of
framework for identifying vegetation types, inter-         stands to become ‘perfect matches’ to the NVC com-
preting the ecological factors that control them,          munity tables, because the maintenance of the local
and assessing their importance in a national and           variation invariably present is central to the conser-
local context. The NVC is also often used to describe      vation of biodiversity. The NVC types are idealised
a field technique of vegetation survey (also known         summary classes, which provide reference points
as Phase II survey) derived from the protocol used in      for classification of vegetation occurring in the field.
the original classification study. Thus although the          Where NVC types are site features or attributes
NVC was not developed as a monitoring tool, it can         that require monitoring, a logical starting point is
provide a conceptual framework and practical tools         to carry out an NVC survey if this has not been
for monitoring vegetation (Rodwell, 1997). As this is      recently done. Indeed, the establishment of the
a rather complex subject with some potential pit-          NVC types present will often be a prerequisite for
falls, the uses and misuses of NVC for monitoring          setting the correct objectives and limits for a site.
are discussed in some depth below.                         Guidelines for identifying attributes that define
   Perhaps the most valuable use of the NVC is as a        condition (see, for example, SNH, 2000) will differ
precursor to the establishment of a monitoring             according to NVC type. Where NVC surveys are to
programme. NVC surveys can provide inventories             be carried out, communities and sub-communities
and maps of NVC communities and subcommu-                  should be identified by using properly replicated
nities at a site. First, these may be used as a basis      quantitative quadrat-based methods (see below)
for the identification and characterisation of site        rather than subjective visual assessments, even if
features (Rowell, 1993); see Part I, Section 2.1. At its   these are done by experts.
simplest level, a Notified Feature (i.e. one that is          In theory, repeat NVC mapping could be used for
listed in the designation citation) may be defined as      monitoring both the extent and the composition of
an NVC community or subcommunity. Alternatively,           habitats (in terms of NVC types present, not species
                                                             6.1 General habitat survey and monitoring methods             185




Box 6.7 A case study of the use of Phase I survey              therefore relate to the date of aerial photography rather
for monitoring vegetation                                      than of the field survey; this difference is important if
                                                               accurate rates of habitat change are required. The repeat
INTRODUCTION                                                   survey was based on vertical colour aerial photographs.
This study (Dargie, 1992) examined the effectiveness              Errors due to tilt were noticed on both maps but
of repeat Phase I survey as a monitoring data source           comparison of fixed-point and line distances suggested
for use at sites requiring information on habitat              that these were small. There were discrepancies in habi-
change. Three sites previously surveyed using Phase I          tat definition between the two surveys. The first survey
were selected: Benacre Broad, Suffolk; Skipwith                did not separate Salix cinerea scrub and omitted a small
Common, North Yorkshire; and Hannah’s Hill,                    area of unimproved neutral grassland. More seriously,
Northumberland.                                                there was confusion between marsh or marshy grassland
                                                               and inundation vegetation habitat types in both surveys,
METHOD                                                         plus differences in the separation of wet and dry heath.
The sites were resurveyed by, as far as possible, the             However, the maps and interpretation, suggesting
same methods that were used in the original site               an expansion of woodland and a corresponding decline
survey. The two completed habitat maps for each site           in heathland, bracken and wetland, were well received
were produced at the same scale, and the new map was           by local experts, and indicated habitat losses not fully
overlaid on top of the old one, with a clear acetate grid      appreciated by local managers.
of 1 mm squares placed on top of both.
   For each repeat survey map polygon, the types of            (3) Hannah’s Hill, Northumberland
first survey habitat beneath were noted, with the              Both surveys were carried out without the use of aerial
number of 1 mm squares that the habitat occupied.              photographs, and major differences were found
Each repeat survey habitat type had, at the end, a             between them. The repeat survey had a much higher
count of 1 mm squares for each of its underlying first         density of habitat boundaries, and few of these
survey habitat types. These data were converted into           coincided with boundaries from the first survey.
hectares and entered into a transition matrix, with            Overall it was felt that the boundary displacement
repeat survey habitat codes entered along the top and          represented considerable planimetric error rather
the first survey codes in the left-hand column. This           than actual changes in habitat extent.
matrix presented a concise record of measured
change, with row (first survey) and column (second             CONCLUSIONS
survey) totals giving the quantities of each habitat at        Perhaps the most important finding of this study was
the two dates of survey.                                       that the use of aerial photographs in conjunction
                                                               with Phase I surveys considerably increases the
RESULTS                                                        accuracy of the surveys and the confidence placed in
(1) Benacre Broad, Suffolk                                     the results. It is essential that monitoring surveys
In addition to some relatively small faults in the repeat      should be accurate and perceived to be reliable if
mapping, two major errors were identified. First,              management decisions are to be based upon their
mudflats created by the loss of reed swamp were                findings. The accurate mapping of vegetation
recorded as brackish lagoons because they had been             boundaries is therefore vital.
inundated by salt water at the time of the survey. Second,        The correct identification of habitat types is also
local experts considered that both surveys had mistaken        very important, but mistakes in identifying habitat
areas of reed swamp for inundation vegetation.                 types can be more easily corrected than can errors in
                                                               determining boundaries.
(2) Skipwith Common, North Yorkshire                              If accurate measures of habitat change are required
The first survey was made by using field checking to           it is essential that precise field methods are used in
place Phase I categories on to a boundary map derived          conjunction with aerial photography and maps
from aerial photographs. The boundaries might                  derived from good photographic rectification.
186      6 METHODS FOR SURVEYING HABITATS




  NVC surveys: summary of key points                         *   Time of year for survey; some communities are best
                                                                 surveyed in appropriate seasons (e.g. woodland in
  Recommended uses                                               early summer) (see Box 6.9)
                                                             *   Where mapping is to be carried out, follow standard
  *   Baseline descriptions and mapping of vegetation as
                                                                 mapping methods, such as Phase I overlain with
      a precursor to designing a monitoring programme
                                                                 standard NVC codes
  *   Confirmation of NVC types present when these are
                                                             *   Scale of mapping: for most purposes 1 : 10 000 is
      features or attributes for which monitoring is
                                                                 probably adequate
      required
                                                             *   Boundaries of vegetation types and mosaics can be
  *   Interpretation of changes in vegetation
                                                                 difficult to define
  Efficiency     Fairly slow and labour-intensive            *   Ensure representative      quadrats    are   recorded;
  Objectivity     Reasonably objective provided trained          quadrat sizes should be selected by following the
  surveyors are used, but some subjectivity over bound-          NVC guidelines
  aries of communities and especially subcommunities         *   Surveyors can cover 5–200 ha per day depending on
  Precision      No data available                               topography and complexity of vegetation
  Bias    Identification of communities and vegetation       *   Comparisons of repeated simple NVC mapping
  may be subject to significant bias from individual             cannot provide sufficiently reliable estimates of
  surveyors                                                      changes in extent or distribution of NVC types to be
  Expertise required     Competent botanist with NVC             appropriate for most monitoring purposes
  experience                                                 *   There are a few vegetation types not covered in the
  Equipment required       Quadrats/tape measures, maps,         original classification, though these are relatively
  floras, NVC guide, ruler, compass, spirit level with           rare
  angle measuring device, coloured pencils
                                                             Data analysis     A high level of statistical analysis on
  Key methodological points to consider
                                                             properly replicated and sampled data (e.g. G-tests) is
  *   Is an NVC survey necessary: are there more efficient   required to assess change with any degree of
      means to gather the data you require?                  confidence at the community level




within a community). Given that some interest fea-           detailed techniques such as point quadrats; see
tures are defined in terms of the NVC, this will be          Section 6.4.5) and will not detect gradual change
required in some instances. The extent of a selected         before large changes have occurred. Measurement
habitat could be measured simply from the NVC                of structural changes may show the consequences
maps and any changes in area assessed directly.              of alterations in management regimes long before
However, monitoring data from NVC mapping are                the NVC classes change.
only likely to be able to detect change at a fairly              NVC mapping is also time-consuming (see
crude scale (e.g. change of community, large-scale           below) and therefore costly. NVC repeat mapping
damage) and over a long time period. Furthermore,            is therefore not recommended as a general method
assessing changes in composition within a commu-             for monitoring. If NVC mapping is to be employed
nity from NVC mapping data alone is fraught with             for other long-term monitoring purposes, it is advi-
difficulties. It is at best very crude, and confidence       sable to have repeated all or part of the survey at
can only be placed in obvious changes at a commu-            the time the baseline is set up, to assess repeatabil-
nity level after inspection of historical data. The          ity and obtain error estimates to see whether the
mapping data will not be sufficient for assessing            method will give the information required.
management impacts such as those arising from                    Instead of conventional repeat mapping it is
fertiliser application, etc. (these require more             recommended that NVC communities and
                                                        6.1 General habitat survey and monitoring methods     187



subcommunities be monitored by more objective                 Although such analyses may be based on data
and quantitative means, such as those employing            from NVC surveys, the classification framework
quadrat- or belt-transect-based sampling methods           can be used in conjunction with survey techniques
(see Sections 6.4.2 and 6.4.6). An initial NVC survey      that are not themselves of the NVC type.
can play a vital role in designing the optimum
sampling strategy and methods for such pro-                Time efficiency
grammes. In particular, NVC maps can identify              The time taken to survey an area depends on the
the approximate boundaries of vegetation types.            complexity of the site, its accessibility, the scale of
Irrespective of whether NVC vegetation types are           the survey and the skill of the surveyor. A simple
features in their own right, or attributes of fea-         ‘walkover’ (better defined as a ‘site appraisal’) sur-
tures, NVC boundaries can be used to define sam-           vey will be quicker than one supported by quadrat
pling areas and sampling strata for stratified             data (the latter is always preferable). Quadrats in
sampling. The maps can also be used to locate              species-poor heathland may take 5 minutes to
transects across vegetation boundaries where               record, in species-rich hay meadows over 30 min-
changes in the extent of vegetation stands need to         utes, and large 50 m  50 m woodland quadrats
be monitored. Although, as mentioned above, NVC            over 60 minutes on top of the time taken to do
mapping can be time-consuming, the cost of a one-          the site appraisal and lay out the grid.
off initial survey may often be offset by savings             As a crude guide in average lowland areas, a pair
obtained through the reduced sampling require-             of experienced surveyors (one mapping and one
ments of an optimally designed monitoring                  quadratting) can map about 25–50 ha in lowland
programme.                                                 grassland or mire (of low habitat diversity) per day,
    A potential role for the NVC with respect to           excluding travel and access permission time. In the
monitoring is its use as an analytical and predictive      uplands where there are large tracts of similar low
tool (Rodwell, 1997). Each NVC vegetation type             diversity vegetation, experienced surveyors may be
characterised can be related to particular envir-          able to cover 100–150 or even 200–300 ha per day.
onmental conditions, including climatic, edaphic           Sites that are difficult to see over or are very com-
and biotic influences, the combination of which            plex (e.g. sand dunes, reedbed–fen complexes, mire
favours the development of the vegetation type,            mosaics) may take considerably longer, with less
and the continuance of which is essential to sustain       than 5 ha per day being covered.
it. Although available information on NVC commu-              Allow about double the survey time for typing
nities and subcommunities varies, this provides a          up quadrats, drawing neat maps, analysing data
valuable predictive capacity. This can illuminate          and writing the report if the survey takes up to a
spatial contrast between vegetation types and tem-         week. The time taken to write up is proportionately
poral relationships in successions, regressions or         smaller for larger surveys but could still take up to
lines of deflected development, where serial pro-          half as long again as the actual data collection.
cesses are still ongoing or where management or
other environmental changes have stimulated                Expertise required
vegetation change. Despite being incomplete,               Surveyors should be competent field botanists,
such a predictive framework of spatial and tem-            able to identify the majority of species in the
poral relations enables the fabric of a site and its       field. Vegetative identification of grasses, rushes
features to be related to potential conditions. The        and sedges may be essential for some commu-
NVC can thus help to define objectives, identify           nities, such as grasslands and mires, as will know-
possible management options and aims and pro-              ledge of bryophytes, such as Sphagnum spp. in bogs
vide some basis for calculating costs and benefits.        and Drepanocladus spp. in flushes. A list of botanical
Cooper & Rodwell (1995) outline a preliminary              identification guides is given in Appendix 3.
attempt at such a process for the English Nature              Training in survey and recognition of NVC types
Yorkshire Dales Natural Area.                              is important to ensure consistency, especially if
188    6 METHODS FOR SURVEYING HABITATS



more than one surveyor is being used to survey a          to Phase I (Section 6.1.5) can be applied to NVC map-
large site. Surveyors do not need to know all the         ping, although the detail required means that more
NVC communities initially, but they should be             time is required as well as a higher level of expertise.
familiar with those likely to be encountered in              The general procedure for NVC mapping is given
the course of the survey. There is no doubt that          in Box 6.8. If the site already has Phase I maps, take
recognition of vegetation types improves with             copies into the field and work from them if no
experience and knowledge, but it is also well             aerial photographs are available, overlaying NVC
known (if unquantified) that different surveyors          codes as appropriate and checking vegetation
may put boundaries between communities and                boundaries as you go.
especially subcommunities in different places;               In addition to these procedures, it is useful to
this limits the accuracy of the method. Although          take general photographs of the site and quadrats.
general ecologists may be able to map boundaries          These can convey an enormous amount of informa-
of communities and note some changes (e.g.                tion not gleaned from the species listed in a quad-
encroachment of scrub into heathland), they may           rat. Depending on resources and time available,
not have the expertise and experience to assess           these may be taken by using fixed-point photog-
whether community types are changing at the               raphy (see Section 6.1.4).
same time.
                                                          Best times for survey
Equipment required                                        The best times to survey particular vegetation types
The equipment required is listed in Appendix 6            are shown in Box 6.9. These are usually determined
Although photographs are not essential, they can          by the time at which the vegetation is best devel-
be highly informative and will provide invaluable         oped. It is inevitable that some surveys will be
supplementary information about typical and aty-          carried out outside these times. There is little
pical examples of vegetation. It may also be worth        point in surveying woodland ground flora commu-
taking a trowel and a pH meter to take a brief look       nities after July, and most wetland communities
at soils (Section 6.2.2), which can also provide infor-   are best surveyed in late summer.
mation about the vegetation type.                            For monitoring, try to repeat the survey within
                                                          two weeks of the original day of recording.
Field methods                                             Differences between seasons are most significant
There is no formal definition of what ‘an NVC sur-        in spring and early summer.
vey’ actually comprises or what its standards or
methods of quality assurance should be. However,          Scale of base maps
the following field techniques, derived from the          In general, 1 : 10 000 maps will provide the basic
protocol used in the research carried out to develop      background information. A 1 : 5 000 scale is usually
the NVC, have been widely adopted. A series of NVC        necessary to monitor extent of habitats on most
field guides is currently being prepared by the Joint     SSSIs reasonably well. Larger-scale maps (e.g.
Nature Conservation Committee (JNCC).                     1 : 10 000, 1 : 25 000, 1 : 50 000) are only likely to be
   Many details of the field survey, such as the scale    useful for summarising data at a larger scale on a GIS.
of mapping, will have to be decided before going into         Smaller-scale maps at 1 : 2500 or even 1 : 1250
the field. As the composition of vegetation is a func-    may be more use for detailed work. The main limi-
tion of the management, soils and climate, it is useful   tations with such maps are that recent digital cov-
to be aware of the influence of these factors before      erage may not be available, they may lack features
going into the field; very useful syntheses of each       visible on the ground (making location difficult to
major group are given by Rodwell et al. (1991 et seq.)    pinpoint), and large pieces of paper are hard to
   NVC mapping should be seen as a more detailed          handle in the field, especially in wet weather.
form of survey than the basic NCC Phase I habitat             The problems of deciding how small an area to
survey (NCC, 1990a, b). Much of the advice relating       map are discussed in Box 6.10.
                                                           6.1 General habitat survey and monitoring methods             189




  Box 6.8 General procedure for NVC mapping                      fundamental to the quality and accuracy of the sur-
                                                                 vey methodology. They should be identified on site,
  1. Start with boundaries and edges of sites, and work          or within a day or two before they deteriorate. If
    inwards if possible, working from known locations            samples need to be kept for longer then they should
    to unknown locations.                                        be pressed and dried or, in the case of bryophytes,
  2. By eye, select areas of uniform vegetation for              kept in dry paper packets to ensure the specimens
    mapping and quadrats. Selecting ‘uniform’ areas              do not go mouldy.
    does depend to some extent on experience and              4. At each new unknown vegetation type or distinct
    knowledge of the site, and is best done by walking           variant, record enough quadrats to identify the NVC
    through the areas rather than by selecting from a            type from tables in the field, or to allow identifica-
    distance. Knowledge of key preferential species is           tion in the office. Again, record at least five, and
    essential for identifying homogeneous stands.                preferably ten, quadrats. Record vegetation details,
    Communities that differ often appear as such on the          again noting the most characteristic species in each
    aerial photographs, so these can be used as a guide          vegetation type by using the DAFOR scale.
    to identifying stands. However, not all communities       5. Cross-reference the location of each quadrat on the
    or sub-communities will stand out on the aerial              map. This will allow the quadrat data to be referred
    photographs; care should therefore be taken when             back to particular stands and/or communities.
     identifying different stand types.                       6. Colour in the areas covered by each vegetation type
  3. At each new known vegetation type record a                  on the map to ensure full coverage, and mark on the
    quadrat of appropriate scale (see Box 6.11 below)            appropriate NVC code (it will usually be more
    and continue to record until at least five, and pre-         efficient to ensure coverage at the time than to
    ferably ten, quadrats have been recorded from the            return later to fill in gaps). The standard Phase I
    site as a whole, deliberately selecting them to show         habitat colours (JNCC, 1993) are often useful for
    the variation you are recording in that community            varied sites, although this is less useful for sites with
    or sub-community (this will be invaluable for any-           many different communities and subcommunities
    body repeating your work in the future). It may be           of the same general type (e.g. blanket bogs). If spe-
    worth recording more quadrats of any particularly            cific colours have been used for particular commu-
    interesting vegetation types. Ensure that the quad-          nities in previous surveys it may help to use the
    rat is sampled entirely within a stand and that it           same ones again. If time is limited, or if mapping
    does not include elements of the surrounding com-            onto aerial photographs, the areas of each vegeta-
    munities; avoid edges and transitions. Do not forget         tion type can be marked off with a fine liner pen and
    to record details of vegetation height, cover                marked with the appropriate NVC code.
    (including dwarf shrub, bryophyte, lichen and bare        7. Draw firm lines for clear boundaries and dotted
    ground), aspect and slope with the species list.             lines for less certain ones; mark mosaics of
    Quadrats do not need to be marked or relocatable             vegetation as mosaics (see also below).
    unless so desired. Noting the most characteristic         8. As soon as practicable after the survey (it may be
    species in each vegetation type by using the DAFOR           worth doing this every two days while it is fresh in
    scale can provide additional information. Correct            the mind), draw up neat maps and work out the
    identification of the plants encountered is                  identity of unknown vegetation types.




Boundaries of vegetation                                      upland heathland coinciding with the transition
In the field there are often sharp changes in NVC             of peat to mineral soils). In other cases, there is a
type directly related to sharp environmental                  more gradual change from one NVC community
boundaries (e.g. change from blanket bog to                   to another (for example. species-rich Nardus
190    6 METHODS FOR SURVEYING HABITATS




  Box 6.9   The best times for NVC surveying

  Woodlands: April–June (some types can be surveyed             Aquatics: July–September
    into October)                                               Swamps and tall herb fens: July–September
  Scrub: April–October                                          Saltmarshes: August–September
  Hedges/boundaries: April–October                              Sand dunes: May–August
  Heaths and mires: April–October                               Sea cliffs: May–October
  Hay meadows: May–June or before they are cut                  Shingle: May–October
  Upland grasslands: late May/early June – early/               Ruderals: June-October
    mid -September




grasslands merging into upland heathland). This         compare these with the constants in the tables (a
causes problems for mapping in the field and for        possible short-list may be obtained by checking
monitoring the extent of habitats where transi-         the index of species in the back of each volume)
tions can be related to management (for example,        and take into account the less frequent species.
a relaxation of grazing may result in reversion         Always check the description of the vegetation
of Nardus grassland to upland heathland). Other         against the text and full table. Full details of how
communities may form intricate mosaics (for             to interpret the tables are given in each British
example, springs of the alpine pioneer forma-           Plant Communities volume. Maps of the known dis-
tions of Caricion bicoloris-atrofuscae occur in scat-   tributions of communities in the NVC volumes are
tered clusters in upland heaths). Some problems         very incomplete and should not be used to judge
of deciding on boundaries are discussed in              whether the NVC type identified is likely to be
Box 6.10.                                               correct or not.
                                                           There are two computer programs available
Selection of quadrat size                               that can be used to help allocate quadrat data to
Quadrat sizes should be selected as appropriate for     NVC communities: MATCH and TABLEFIT (Box
the scale of the vegetation in the field (see           6.12). Both use the basic NVC table data but differ
Appendix 4). For linear or irregular features, quad-    slightly in other measures used to give a diagnosis
rats of different shape but equivalent area may be      (e.g. species richness). They give a measure of
required;     some     small    stands    such    as    ‘goodness of fit’ against the defined NVC types,
bryophyte flushes may need to be sampled in             and must only be used as an aid to identification
their entirety. The appropriate quadrat sizes as        and not for definitive results. They are also no
described in the NVC are shown in Box 6.11. The         substitute for expert interpretation. There is little
advantage of using these sizes is that the data can     to choose between them (Palmer, 1992) and both
be directly related to the NVC.                         cost about the same. A slight advantage of MATCH
                                                        is that if data are being handled in VESPAN (see
Identifying communities                                 Box 6.12) the quadrats can be exported to MATCH
In practice in the field, vegetation types are best     singly or in groups. The results given must be
named from the tables in British Plant Communities      taken as guidance only and must only be used as
(Rodwell, 1991 et seq.) rather than from the keys,      an aid to identification because odd community
which do not always work well and contain               placements can be given. For example, vegetation
relatively little information. From the quadrats        with constant Carex nigra in inland Wales has
recorded, establish which species are the constants     been assigned to a dune slack community by
(those occurring in over 60% of the quadrats),          MATCH.
                                                            6.1 General habitat survey and monitoring methods              191




Box 6.10      Common problems and solutions                    Mosaics are often very hard to map in the field; if
                                                               detailed notes of the vegetation are taken it may be
SCALE                                                          possible to map them later from a distance (e.g. the
It is often difficult to decide what is the smallest area      opposite hillside) or from recent aerial photographs.
to map; this also partly depends on the objectives of          Mosaics may also occur within communities related to
the survey and is best judged in the field from the scale      the growth forms of different species, e.g. shorter grass
of the mapping, the community involved and the                 between large clumps of Deschampsia cespitosa (Tufted
context. Three birch trees in the middle of a bog are          Hair-grass). In these cases, if it is difficult to decide
probably not worth mapping as an NVC type, but are             whether this is really one vegetation type or two
worth noting as present. Despite their small size it is        forming a mosaic, look for qualitative differences in
worth mapping all petrifying springs with active tufa          the composition of the vegetation related to the two
formation as they are listed in the EC Habitats                components. If the Deschampsia simply occurs as
Directive and are a habitat of European importance.            Deschampsia among the same species as in the shorter
One small 1 m  1 m clump of Typha sp. (Bulrush) in            grass it is probably one vegetation type; if it is
a Phragmites sp. reedbed is not a different vegetation         consistently associated with other species it may be
type as it occurs at low frequency in such vegetation,         best to treat it as two communities.
but a 10 m  5 m patch might be worth mapping
separately.                                                    COMMUNITIES THAT DO NOT FIT THE NVC
                                                               It is very difficult to describe when a vegetation type
BOUNDARIES BETWEEN VEGETATION TYPES                            is a variant of an NVC community and when it does
Deciding on the exact boundary between communities             not fit properly. Dealing with communities that do
is often difficult. The most practical approaches are to       not clearly fit the NVC, which may form up to 5–10%
pick out ‘good’ areas, which equate to known or                of the vegetation in some surveys, can be approached
recognisable vegetation types, and either mark an              in two ways. First, allocate them to the nearest-fit NVC
approximate boundary between them as a dotted                  type and say in the report that this has been done.
line or mark a transition zone including the problem           However, for monitoring purposes, this could lead to
vegetation. This may require quite a bit of walking to         the exaggeration of change. Second, present the
and fro, recording quadrats if needed, until the               quadrat data and describe them as communities in
differences can be clarified. The variability and errors       their own right (e.g. ‘hazel hedges’) or give a
associated with different surveyors placing boundaries         description relating to two or three NVC types between
in different places have not been quantified, but are          which the community may lie. Given such a complex
likely to be particularly significant in sites with many       range of communities, it is possible that some may
transitions between similar vegetation types. Such             actually be included under different chapters of the
variability is a major factor limiting the usefulness of       NVC (for example, woodland Cardamine amara (Large
repeat NVC mapping for monitoring purposes. If                 Bittercress) flushes are included under mires rather
monitoring of the extent and boundaries of vegetation          than woodlands). Very detailed localised sampling may
is required, appropriately designed quadrat- or                also result in many species becoming ‘constants’
transect-based sampling should be used.                        locally when they are not constants in the wider
                                                               national context. Contact the local survey teams such
MOSAICS                                                        as those from English Nature or Countryside Council
The traditional treatment of complex mosaics of                for Wales as they could have knowledge of other
vegetation is to mark an area as a mosaic and estimate         vegetation types that have been recorded in the area
what proportion is occupied by each vegetation type.           of survey.
192    6 METHODS FOR SURVEYING HABITATS




  Box 6.11 Sizes of quadrat for NVC surveys

  Woodlands:                        50 m  50 m for canopy and shrub layers, 4 m 4 m or 10 m 10 m
                                    for ground layer
  Scrub:                            10 m 10 m, exceptionally 50 m 50 m in open patchy scrub
  Hedges/boundaries:                30 m lengths for canopy, 10 m lengths for ground flora
  Heaths and mires:                 2 m 2 m or 4  4 m, exceptionally 10 m 10 m in very impoverished
                                    or grossly structured vegetation
  Grasslands:                       2 m 2 m or 4 m 4 m where large tussocky vegetation is present
  Aquatics:                         2 m 2 m or 4 m 4 m
  Swamps and tall herb fens:        4 m 4 m or 10 m 10 m
  Saltmarshes:                      2 m 2 m or 4 m 4 m
  Sand dunes:                       2 m 2 m or 4 m 4 m
  Sea cliffs:                       2 m 2 m or 4 m 4 m
  Shingle:                          2 m 2 m or 4 m 4 m
  Ruderals:                         2 m 2 m or 4 m 4 m where large tussocky vegetation is present




  Box 6.12 Computer programs for vegetation                Huntingdon PE17 2LS (or under ‘software’ on http://
  analysis                                                 mwnta.ac.uk/ite), price £80 + VAT.
                                                              TABLECORN, DECORANA and TWINSPAN, 1996,
  Details are correct at time of writing.                  M. O. Hill
      VESPAN 3 (VEgetation and SPecies ANalysis), 1997,       A DOS-based program for analysing vegetation data.
  A. J. C. Malloch                                         TABLECORN converts data from TABLEFIT into
    A Windows-based program for handling, presenting       DECORANA and TWINSPAN formats. Available from
  and analysing vegetation data including TWINSPAN         Centre for Ecology and Hydrology(CEH), Monks Wood,
  and DECORANA. Available from Biological Sciences,        Abbots Ripton, Huntingdon PE17 2LS (or under ‘soft-
  University of Lancaster, Lancaster LA1 4YQ, price £100   ware’ on http://mwnta.ac.uk/ite), price £75+ VAT.
  + VAT; upgrades from previous versions £50 + VAT.           FIBS
      MATCH 2, 1997, A. J. C. Malloch                         FIBS      (Functional   Interpretation   of   Botanical
      A Windows-based program for comparing vegeta-        Surveys) was developed by the Unit of Comparative
  tion data against the NVC. Available from Biological     Plant Ecology for English Nature (see Section 6.4.4).
  Sciences, University of Lancaster, Lancaster LA1 4YQ,      CANOCO (CANOnical Community Ordination),
  price £100 + VAT; upgrades from previous versions £25    1988, C. J. F. Ter Braak
  + VAT.                                                      A Fortran program for canonical community ordi-
      TABLEFIT 1.0, 1996, M. O. Hill                       nation by correlation analysis, principal components
      A DOS-based program for comparing vegetation         analysis (PCA) and redundancy analysis. Available from
  data against the NVC and the EC CORINE types.            Microcomputer Power, 111 Clover Lane, Ithaca,
  Available from Centre for Ecology and                    NY 14850, USA. Tel: (607) 272 2188.
  Hydrology(CEH), Monks Wood, Abbots Ripton,
                                                         6.1 General habitat survey and monitoring methods         193



Data storage and analysis                                   can then be analysed with the assistance of MATCH
Storage                                                     and TABLEFIT programs to allocate an NVC type and
Quadrat data can simply be filed for later use (cross-      similarity coefficient (see below) to each systematic
referenced in report, stating storage location), writ-      sampling point on the grid. The overall extent of
ten up with the report or held on a computer. The           each NVC type on a site may then be estimated as
best computer program for handling, presenting              a frequency of occurrence from the combined sam-
and analysing data in a format compatible with              ples, and compared between surveys by appropriate
the NVC is VESPAN (Box 6.12) but others are avail-          tests (e.g. G-tests or 2 tests). If the grid is permanent,
able. Maps can be stored on GIS or in hard-copy             the NVC type at each location may also be used for
format.                                                     spatial analysis of change.
                                                               Standard multivariate ecological analytical
                                                            techniques such as TWINSPAN, DECORANA and
Changes in NVC types present                                CANOCO (see Box 6.12) can also be used to analyse
A crude analysis of changes in NVC types present at         larger quadrat or vegetation datasets if data are
particular points can be made by comparing repeat           collected appropriately. These are extremely
NVC survey maps. However, it is often difficult             powerful tools. However, quadrat data derived
to be certain whether a change is real or simply            from NVC surveys for monitoring purposes are
due to a different surveyor. Change from a heath-           unlikely to be suitable for such detailed analysis.
land to a grassland community is likely to be real,         Be aware that TWINSPAN is sensitive to the order
but a change from one heathland sub-community               in which species are input (Tausch et al., 1995;
to the most closely related sub-community is more           Dirkse, 1998).
likely to be an artefact. The original data may need
to be examined with care.                                   Analysis and prediction of change
   The monitoring of changes in the extent and              As described above, the NVC provides a powerful
distribution of NVC types should be carried out             tool for analysing and predicting change. For
by using properly replicated quadrat- or belt-              example, the use of MATCH or TABLEFIT similarity
transect-based sampling programmes. Replicated              coefficients to look at changes in the species com-
permanent quadrats may be used, but there are               position of communities over time is shown in
problems with this approach (see Byrne, 1991;               Figure 6.1 (see also Rich et al., 1991, 1992).
Robertson, 1999). Individual quadrats can rarely                As many of the communities are related,
be precisely relocated to allow them to be re-              changes from one to another may occur with
recorded meaningfully (a 10 cm error in relocating          changes in management, soils, natural succession,
a 2 m  2 m quadrat can result in 5% error), and            etc. By understanding how the communities relate
usually too few are taken to allow for appropriate          to one another the direction of change can be
replication. Repeated estimates of Domin cover              understood and therefore interpreted against the
values (see Box 2.3) by different surveyors have            field situation. For example, the expected direction
been found to show differences of up to five                of changes in the MG12a Festuca arundinacea (Tall
Domin scale points for the same quadrats (Leach,            Fescue) grassland, Lolium–Holcus sub-community
1988). In addition, grazing can markedly alter              following changes in three different environmen-
vegetation structure without significantly affect-          tal factors are shown in Figure 6.2. Such changes
ing its composition.                                        may occur over periods ranging from a year to
   A possible approach may be to obtain Domin               decades depending on the strength of the driving
cover values from at least five quadrats (see               force. Details of many of these relationships are
Section 6.4.2) placed in representative areas of            mentioned in the NVC volumes (Rodwell, 1991
vegetation within an approximate distance from              et seq.), but these should only be regarded as indica-
each of a number of sampling points on a system-            tive and there is no substitute for appropriately
atically placed grid. The five sets of quadrat data         replicated experiments.
194    6 METHODS FOR SURVEYING HABITATS



 0.7                                                     *   The NVC classification and survey method can be
 0.6                                                         used to describe a site and is often used to identify
 0.5                                                         and define features and/or attributes that require
 0.4                                                         monitoring.
 0.3
                                                         *   The NVC classification and associated datasets can
                                                             assist with the definition of objectives and the
 0.2
                                                             setting of limits for a feature.
 0.1
                                                         *   A baseline NVC mapping survey can be used to
   0
   1970    1975     1980    1985     1990     1995           identify approximate boundaries of vegetation
                                                             types as an aid to optimising the design of detailed
                    MG6             MG5                      objective     and      quantitative     monitoring
                    MG7             MG1                      programmes.
Figure 6.1. Changes in MATCH similarity coefficients
                                                         *   The NVC can be used as an analytical tool for
for a hay meadow following introduction of grazing
                                                             investigating changes in vegetation and predict-
and fertiliser application in 1975 (hypothetical
                                                             ing the outcome of management, etc.
example). The changes in coefficients show a                However, the main disadvantages are that
change from a hay meadow community MG5 to a              NVC mapping can be time-consuming, it requires
grazed pasture community MG6. Coefficients for           competent botanists, the boundaries between
MG7 ryegrass leys (which can be derived from MG6         communities or even the identity of a commu-
by applying more fertiliser and heavy grazing)           nity may vary with the surveyor, and it may be
and MG1 False Oatgrass (Arrhenatherum elatius)           too crude to reveal gradual or subtle vegetation
grassland (a rank unmanaged grassland type) are          changes (e.g. as a result of management changes).
also shown.                                                 Therefore, for monitoring purposes:
                                                         *   The NVC should not be used for normalising vege-
                                                             tation types when setting feature objectives and
Summary of advantages and disadvantages                      limits.
The NVC can provide a conceptual framework and           *   Repeat NVC mapping surveys alone should not
practical method for monitoring sites that have              be used for monitoring the extent or distribution
been selected on the basis of their vegetation com-          of vegetation communities; other supporting
munities. In particular:                                     information is required.


                    MG6 Lolium–Cynosurus grassland
                    MG11 Festuca–Agrostis–Potentilla grassland
                    MG12a Festuca arundinacea grassland, Lolium–Holcus sub-community
                    MG12b Festuca arundinacea grassland, Oenanthe sub-community


                                                                     Nutrients +         MG6
                                                                      grazing


                                                     Impeded      MG12a
                                                     drainage                      Salinity

                                               MG11                                     MG12b

           Figure 6.2. Expected directional changes in an MG12a community following changes in
           three environmental factors.
                                                                                     6.2 Physical attributes   195



                                                            only be found on alkaline soils). For further informa-
6.2      PHYSICAL ATTRIBUTES
                                                            tion on these and other soil characteristics that may
The following sections give a brief overview of the         be of interest, such as salinity and redox potential,
physical attributes of a site that are most likely to       consult Jones & Reynolds (1996).
require monitoring, namely hydrology and soils.                It should be remembered that soils are biological
Both of these attributes can have considerable              entities, which include vital communities of soil
effects upon plant and animal communities.                  organisms. The health of these fungus, animal
   For the purposes of habitat monitoring, it is unli-      and plant populations is also an important attri-
kely that detailed analyses of hydrology and soils          bute of soil condition. Soils are also influenced by
will need to be carried out as a regular monitoring         the vegetation growing in them as well as by the
activity. However, some consideration will usually          chemical composition of the underlying substrate.
need to be given to these aspects at some point, and        For example, the decomposition of heather litter
so the key issues involved are summarised.                  releases acids, which tend to lower the pH of the
                                                            soil underneath the plants. These inter-relations
                                                            between physical and biological properties of the
6.2.1       Hydrology                                       soil and other ecosystem components are often
                                                            complex but should be borne in mind when inter-
Although probably the most fundamental aspect
                                                            preting the results of simple monitoring of phys-
of wetlands and particularly bog ecosystems,
                                                            ical attributes.
hydrology is all too often ignored in monitoring
programmes or added as an afterthought (Lindsay
                                                            Fertility
& Ross, 1994). One of the main reasons for this
                                                            A broad indication of fertility can be obtained by
is that it is a complex subject. Although dipwells
                                                            measuring total base cations, which indicates the
or piezometers, etc., can be easily installed, the
                                                            supply of plant macronutrients (e.g. nitrogen, potas-
interpretation of results in terms of impacts on
                                                            sium and phosphorus) available in soil solution.
features of conservation interest is often difficult.
                                                            Fresh soil samples and laboratory analysis are
Therefore, although it is strongly recommended
                                                            required.
that appropriate hydrological monitoring be carried
out where necessary, a detailed treatment of the
                                                            Soil pH
subject is beyond the scope of this Handbook. It is
                                                            The pH is a measure of the acidity or alkalinity of
therefore recommended that further specialist
                                                            a solution. It is measured on a logarithmic scale
advice on the subject is obtained from appropriate
                                                            from 1 to 14: a pH below 7 indicates acidity, a pH
experts and organisations such as SEPA, the
                                                            of 7 indicates a neutral solution (e.g. pure dis-
Environment Agency, and specialists within the
                                                            tilled water) and a pH above 7 indicates alkalinity.
UK statutory conservation agencies.
                                                            A pH reading is a measurement of the activity of
                                                            hydrogen or hydroxyl ions. Soil pH can be meas-
6.2.2       Soil characteristics                            ured in two ways (both of which can also be used
                                                            for measuring the pH of water).
The study and classification of soils and soil charac-
teristics is a large subject with a considerable            Indicator strips
amount of literature devoted to it. However, for            Strips of indicator paper that change to a particular
the purposes of habitat monitoring, assessments of          colour when immersed in a solution of a particular
many soil characteristics are unlikely to be required.      pH are commercially available. Wide-range paper is
It will generally be sufficient to restrict soil analysis   available, which covers the whole range pH 1–14, but
to nutrient status, pH, structure, texture and moist-       this is less accurate than using narrow-range paper.
ure content, because these have the greatest effect            A small amount of soil is mixed with distilled
on vegetation (for example, some plant species will         water, and the paper is immersed in the solution
196     6 METHODS FOR SURVEYING HABITATS



for several minutes before the pH is measured.               measure the depths of the soil horizons. Soil sam-
Indicator paper is cheap and quick, but less precise         ples from the different horizons can be taken for
than a pH meter.                                             chemical analysis. A less destructive and faster
                                                             method is to take core samples with an auger, a
pH meters                                                    hollow cylinder that is twisted into the ground
If available, a portable pH meter is the best way            and removes a cylindrical soil core.
to measure pH in the field. A standard electronic               Measuring the depth of peat layers may require
pH meter has an electrode, which is inserted into            several interlocking rods with the soil corer fixed
a solution of one part soil mixed with two parts             to the end, because peat layers can be very deep. In
distilled water. Some pH meters also have a tem-             situations in which the underlying sediments are
perature probe, which, when inserted into the                very different from the peat, the depth of peat can
solution, compensates for differences in the tem-            be measured by using radar and seismology.
perature of the solution from room temperature               However, these techniques are still partly experi-
(pH readings are affected by temperature).                   mental, and are expensive (Lindsay & Ross, 1994).
   Meters need to be calibrated against solutions of            Soil texture is a measure of the size of the
known pH; buffer tablets to make up solutions of             particles that make up the soil. Soil type is a con-
pH 4, 7 and 9 are generally supplied with pH meters.         tinuum from clay (particle size < 2 mm) through
At least two of these should be used to calibrate the        silt (2–50 mm) to sand (50–2000 mm) and gravel
meter before any readings are taken. It is also a            (> 2000 mm) and is based on the relative propor-
good idea to check the calibration at regular inter-         tions of different particle sizes in the soil. For
vals, because some meters have a tendency to drift           most biological purposes, the coarse distinctions
after a while. Portable pH meters are accurate to            of sand, silt and clay soils will be sufficient, but
within Æ0.01 pH units. Laboratory-based instru-              more accurate classifications are available should
ments can be accurate to Æ0.001 pH units.                    they be required (e.g. the texture scale used by the
   If soil is being measured in the field, it is possible    United States Department of Agriculture; see Jones
to take samples directly by hollowing out a small            & Reynolds (1996)).
hole for the electrode, filling this with                       Field assessments of soil texture from soil pro-
distilled water, and inserting the electrode (with           files are likely to be sufficient for most monitoring
the temperature electrode inserted close by).                purposes. However, if necessary particle size can be
Otherwise, samples can be mixed in beakers in the            measured by passing dried soil samples through
field, or collected for later analysis in a laboratory. If   sieves of successively smaller mesh sizes, although
the latter course is chosen, soil samples should be          this is less accurate for small particles than for
sealed in airtight bags or containers to prevent the         larger ones. A better method is to separate soil
soil from drying out before the pH is measured.              particles by sedimentation. A sample of soil is
                                                             mixed in a clear cylinder filled with water.
Soil profile and texture                                     Organic material will float to the top, and the soil
A soil profile is a vertical cross-section down through      particle types will settle according to size (larger,
the soil, which exposes layers or horizons. An exam-         heavier particles will settle out to the bottom faster
ination of the profile and the type of soil horizons         than small, light ones). The percentages of each
present can provide a valuable general indication of         particle type can be measured (Jones & Reynolds,
the chemical and physical condition of the soil. For         1996) and can be used to classify the soil texture
example, the smell of hydrogen sulphide (rotten              according to the FAO classification (FAO, 1977).
eggs) or presence of a gleyed horizon (dark grey                Soil texture and physical properties obtained
mottled with orange-red iron oxide patches) indi-            from soil profiles are important attributes of soils
cates alternating wet and dry soil conditions.               but change slowly. Therefore, although measure-
   Soil profiles can be examined by digging a verti-         ments need to be carried out over long time peri-
cal, smooth-faced pit and using a measuring tape to          ods, observations are only needed infrequently.
                                                 6.3 River morphology and aquatic vegetation composition      197



Soil moisture
                                                          6.3       RIVER MORPHOLOGY AND AQUATIC
The moisture retention capacity of soil is correl-
                                                                    VEGETATION COMPOSITION
ated with its organic matter content and differs
between soil types. For ecological purposes, soil         As discussed in Sections 5.9 and 5.10, survey and
moisture is divided into three types according to         monitoring of aquatic habitats is a complex and
its availability to plants, as follows.                   specialised subject. Therefore a detailed treatment
                                                          of the subject is beyond the scope of this Handbook
1. Free-draining water drains away or evaporates
                                                          and only brief accounts of survey and monitoring
   soon after rain but may be available to plants for
                                                          methods are provided below. It is recommended
   several days, depending upon the soil type (sandy
                                                          that further specialist advice on the subject be
   soils drain much quicker than clays) and the
                                                          obtained from appropriate experts and organisa-
   drainage characteristics of the site (an imperme-
                                                          tions such as the Environment Agency, SEPA,
   able layer, such as an iron pan, can impede drai-
                                                          Centre for Ecology and Hydrology and specialists
   nage and lead to soils becoming waterlogged).
                                                          within the UK statutory conservation agencies or
2. Plant-available (capillary) water occupies the gaps
                                                          key staff within consultancies.
   between soil particles and is of most importance
   for plants as it remains in the soil for longer than
   does free-draining water.                              6.3.1       River Habitat Survey
3. Plant-unavailable (hygroscopic) water forms a thin
   film on the surface of soil particles, which cannot    Recommended uses
   be taken up by the root systems of plants. (Most of    The River Habitat Survey (RHS) has been developed
   the water in clay soils is held in plant-unavailable   in response to the need for a nationally applicable
   form, and this is why they are more prone to dry-      classification of rivers based on habitat quality. It is
   ing out during droughts than are silty or loamy        a logical development of the River Corridor Survey
   soils, despite containing higher amounts of water.)    (RCS), which is principally a map-based system for
                                                          surveying 500 m lengths of river (see NRA (1992)
The plant-available water content of a soil can be        for detailed methods). The RCS provides informa-
estimated from the loss of weight of the soil at          tion on the location of habitats and plant assem-
field capacity (i.e. the state attained after the soil    blages within river channels, margins, banks and
has been saturated and allowed to drain for about         the river corridor. Most information is subjective
two days) after being air-dried. Keep soil samples in     (except for some basic river channel measurements)
airtight     plastic     bags    to    prevent    any     and therefore too prone to inter-surveyor variation
evaporation before analysis in the laboratory.            to be used for most monitoring purposes. However,
Measure the mass of each soil sample, then allow          RHS provides a more detailed and objective means
it to dry naturally (e.g. spread it on paper for a        of recording river habitat characteristics that can be
week) before weighing it again. The difference            used for broad monitoring purposes.
between the two masses is approximately the                  RHS provides four distinct but related outputs:
amount of plant-available water held in the soil.
    Electronic soil moisture meters are also avail-       *   a standard field survey method (which can be used,
able for taking measurements directly in the field.           where appropriate, as a consistent framework for
                                                              collecting data for general monitoring purposes);
                                                          *   a computer database, containing information
Soil organic matter                                           from a national reference network of UK sites;
A determination of soil organic matter content can        *   predictions of which physical features are
indicate overall biological activity in a soil and the        expected to be present within an RHS site at a
speed of turnover of nutrients. It is readily meas-           particular location (taking into account altitude,
ured by igniting dried samples in a laboratory and            height of source, etc.);
measuring mass loss after burning.                        *   a scheme for assessing habitat quality.
198      6 METHODS FOR SURVEYING HABITATS




  River Habitat Survey: summary of key points                    2–3 days and no expertise is required beyond a
                                                                 biological background and familiarity with river
  Recommended uses                                               habitats
                                                                 Equipment required     Standard recording forms, maps
  *   To survey the general physical and vegetation struc-
                                                                 and a ranging pole
      ture of a river
                                                                 Key methodological points to consider
  *   To monitor broad changes in river structure and
      vegetation, e.g. as a result of pollution or river engi-   *   Fieldwork is the basis of assessments, supplemented
      neering works or other impacts                                 with information from maps
                                                                 *   The procedure is carried out on 500 m reaches, with
  Efficiency Relatively rapid: average time for surveying
                                                                     10 spot checks carried out within each
  standard 500 m lengths of river is c. 60 min
                                                                 *   The survey methods have been developed for
  Objectivity    Semi-objective through standardised
                                                                     rivers in England and Wales but have now been
  forms
                                                                     expanded for Scotland and Northern Ireland as
  Precision     Tests have shown that inter-surveyor
                                                                     well
  variation is low and consistent results are achieved irre-
                                                                 *   More detailed quantitative analysis of channels and
  spective of position of spot checks within 500 m reaches
                                                                     vegetation may be required for specific monitoring
  Bias    Subjective assessments are prone to individual
                                                                     requirements
  interpretation, but this is minimised by training and a
  detailed methods manual                                        Data analysis    The data are amenable for rapid transfer
  Expertise required     Must be carried out by an               to the RHS database and analysis by RHS programs or
  accredited and trained surveyor. Training takes only           SERCON (see Section 5.10.1).




   RHS aims to predict, with statistical probability,            based surveys for bankside vegetation or shallow
those features that ought to occur in unmodified                 rivers and grapnel samples (Section 6.3.2) for
examples for the full range of river types in                    deeper, slow-moving water.
England and Wales (although the system has been
extended to Scotland). The scheme for assessing                  Time efficiency
habitat quality comprises a simple five-band classi-             The technique is relatively simple and quick.
fication (excellent, good, fair, poor, bad) based on a           According to Raven et al. (1997) the time taken to
comparison of the observed features with those                   survey a 500 m length of river (i.e. the standard
expected for the particular river type. Crude mon-               survey unit) varies according to the complexity of
itoring can therefore be carried out by reference to             the site and ease of access, i.e. inherent factors
these broad scales, or in relation to specific targets           that affect any river surveying. Simple sites may
set for attributes that are measured as part of                  take only 35–40 min to complete, whereas those of
the RHS.                                                         complex character may take an hour or more. The
   Other alternative and additional methods to                   average time for a site in their study was just
consider for morphological features include aerial               under 1 h.
photography (Section 6.1.3), fixed-point photogra-
phy (Section 6.1.4) or adaptations of quadrat                    Expertise required
(Sections 6.4.2 and 6.4.3) or transect (Section 6.4.6)           The RHS survey must be carried out by an
techniques. If quantitative assessments of plant                 accredited surveyor. However, the training for
abundance are required, detailed objective meth-                 this is relatively simple (2–3 days) and does not
ods should be used, such as transect- or quadrat-                require specialist knowledge beyond a biological
                                               6.3 River morphology and aquatic vegetation composition      199



background and familiarity with rivers. Although
specialist geomorphological or botanical exper-
                                                        6.3.2      Grapnel surveys of aquatic
tise is not needed, consistent recognition of fea-
                                                                   macrophytes
tures included on the field survey form is              Recommended uses
essential.                                              The technique has been developed by George Hinton
                                                        at English Nature for the semi-quantitative assess-
Equipment required                                      ment of species presence–absence. It is a simple but
No specialist equipment is required other than          fairly time-consuming technique for establishing the
standard field equipment, maps and the RHS              frequency of occurrence of individual species, which
manual and standard recording forms.                    can probably be used on lakes of up to 100 ha or
   Adaptation of standard recording forms on to         slow-flowing deep rivers.
portable data logging devices is possible and may          The data collected can provide an accurate semi-
be cost-effective for large monitoring projects.        quantitative assessment of abundance and distri-
                                                        bution for most species, although it tends to miss
                                                        some low-growing plants such as Fontinalis and
Field methods
                                                        rosette-forming species. The dominant species (in
An illustrated manual (Environment Agency, 2003)
                                                        biomass terms) is consistently identified and the
has been produced and should be consulted for full
                                                        method picks up species contributing as little as
details on field procedures.
                                                        5% of the total biomass.
                                                           Other alternative or additional methods to
Data storage and analysis
                                                        consider include the use of double-headed rakes
Data are recorded on standard recording forms
                                                        or grab samples (see Bell, 1996). Rakes may be
and can easily be transferred to the RHS computer
                                                        used from the shore by throwing out on a line.
database, either manually or by optical reading.
                                                        Grabs are also less prone to bias against small
Data analysis programmes are included as part of
                                                        species, which are not detectable by grapnels.
the system, for example for assessment of overall
                                                        However, these methods collect too much mate-
habitat quality. Data may also be used for SERCON
                                                        rial to be used effectively for anything other than
evaluations (see Section 5.10.1). Alternatively, spe-
                                                        establishing presence. Rake methods may there-
cific attributes can be analysed separately.
                                                        fore be best if boat access is difficult, if the species
                                                        to be sampled is very rare or localised, or if it is
Summary of advantages and disadvantages                 only necessary to establish that a species is
The method provides an up-to-date, tried and            present.
tested standard means of recording the general             Subaqua equipment or ROVs may be used for
physical and vegetation character of river reaches,     carrying out transects or surveying quadrats etc.,
which is part of a national surveying scheme. It is     but these methods are very slow, expensive and
also relatively quick and simple. Although not          require specialist equipment and expertise.
designed specifically for monitoring, it can be
used for monitoring the overall quality of river
habitats by a simple five-point scale. The survey       Time efficiency
methods can also be used as a framework for col-        Grapnel surveys are fairly slow to carry out,
lecting data for general monitoring of specific         but can produce detailed assessments of the fre-
attributes.                                             quency of occurrence and distribution of species.
   However, the technique is largely qualitative        Approximately 25 stations across four or five trans-
and only semi-objective. It may therefore also be       ects can be sampled in a day. A small water body
necessary to develop site-specific quantitative and     requires at least ten transects of 25 stations and
objective monitoring methods for some key attri-        therefore such a water body would take at least
butes (see Section 5.10.1).                             two days to survey.
200      6 METHODS FOR SURVEYING HABITATS




  Grapnel surveys: summary of key points                     Equipment required     Grapnel and graduated cord and
                                                             a boat and compass or preferably a GPS for large water
  Recommended uses                                           bodies; safety equipment; standard recording
                                                             equipment
  *   Semi-quantitative assessments of the abundance
                                                             Key methodological points to consider
      (frequency) of aquatic macrophytes in lakes or
      large slow-moving rivers
                                                             *   The technique uses a grapnel on a graduated cord to
  *   Mapping the distribution of species and aquatic
                                                                 take small samples of vegetation from the bottom of
      macrophyte communities
                                                                 the water body; this is repeated 10 times at each
  Efficiency     Relatively slow, but can provide detailed       sample location to obtain a frequency estimate of
  results                                                        presence
  Objectivity    Objective                                   *   By sampling at many locations a semi-quantitative
  Precision     Moderate                                         map of distribution and abundance can be
  Bias    Biased against small-leaved species                    developed
  Expertise required    Competence in identifying
  aquatic vegetation from fragments of vegetation.           Data analysis    Standard data techniques for the
  Boat handling and GPS experience necessary in many         analysis of presence–absence frequency data can
  situations                                                 be used



   However, in larger water bodies, most aquatic             A presence of 1 is recorded for each species
vegetation will occur in a relatively narrow zone,           detected, regardless of its apparent abundance.
with little or no vegetation growing where the               The method thus provides a frequency of 0–10 for
water is deepest. Samples can therefore be targeted          each species at each sampling point.
in this zone, and this will increase the efficiency of           Repeat transects at Bosherston Lakes (Hinton,
the method.                                                  1989) showed that the mean frequency values for
                                                             each species from 250 grapnel drops was within
Expertise required                                           10% of the grand mean (based on 2000 drops).
It is essential that surveyors be able to identify aqua-         At each sample point the depth is recorded by
tic plants from fragments of vegetation. However, as         using graduations on the grapnel cord. It is recom-
the range of aquatic macrophytes is fairly limited,          mended that basic chemical parameters should
especially within geographical regions and certain           also be recorded at the mid-point of each transect.
types of water body, such techniques can be fairly           Turbidity should be measured by using Secchi discs
rapidly learnt.                                              (Section 5.9.1) and supplemented if possible by in
    For all but the smallest lakes, samples will need to     situ pH and conductivity readings (see Jones &
be obtained by boat. Boat-handling experience is             Reynolds, 1996).
therefore an essential requirement for the surveyor              For small lakes the position of each sample point
or an assistant. Experience with the use of GPS is also      can be ascertained by compass bearings on fixed
necessary for surveying large water bodies.                  points such as church towers, jetties, etc. However,
                                                             this can be time-consuming and inaccurate on
Equipment required                                           large lakes. In these circumstances the use of a
The equipment required is listed in Appendix 6.              GPS may be appropriate. Recent work reported by
                                                             Hinton (1997) found that a hand-held GPS is able to
Field methods                                                provide position fixing accurate to within 25 m.
Ten samples are obtained at each sampling station,           During fieldwork the GPS can be used to navigate
with the use of a small 7 cm grapnel on a cord.              to the selected stations (identified by ten-figure
                                                                      6.4 Ground and shrub vegetation      201



grid references) at which the boat is anchored          distribution targets. It is, however, time-consuming
while grab samples are carried out.                     and tends to underestimate the presence of certain
   Grapnel surveys should be carried out during         species. It can only be used in still or slow-flowing
the period from June to September, but the precise      water.
timing will depend on the species; some species
may require surveying earlier in the year. Repeat
surveys should be at similar times of the year,         6.4       GROUND AND SHRUB VEGETATION
unless seasonal variation is being investigated,
                                                        6.4.1       Total counts of individuals
and repeated at least every three years. Some
rarer species may require monitoring more fre-          Recommended uses
quently. Initial surveys may have to be more fre-       This method is appropriate for monitoring the
quent to establish how much variation occurs, both      presence or abundance of rare and/or distinct
seasonally and annually, and to determine the opti-     species that are attributes of habitat quality. The
mum monitoring time.                                    target species are normally chosen because of
   Working from boats can be hazardous; suitable        their rarity value in order to maintain a close
safety clothing should be worn. Surveyors should        check on population levels and to avoid situations
also be aware of the risk of catching Weil’s disease.   in which a species may disappear from a site
Toxic blue-green algae may also be a hazard.            through inappropriate management. These can
Consult a relevant health and safety policy (e.g.       include species that are typical of a particular
Part I, Box 2.11) for further information.              community or habitat, or other ‘indicator’ species
   Care should be taken to clean fragments of vege-     expected to respond quickly to changes in the habi-
tation from the grapnel when moving from site to        tat. It may also be an appropriate method to be
site; some aggressive introductions can grow from       applied to invasive weedy species such as creep-
fragments.                                              ing thistle, in situations in which information is
                                                        needed on the degree of the potential conservation
Data storage and analysis                               problem, again in response to management of the
The data can be used to calculate overall mean          area.
frequency tables for each species over the lake.           For some situations, counts of the total number
In addition, geographically referenced species fre-     of individuals may be more appropriate than the
quency data can be stored on a spreadsheet or           total area covered. An additional application of this
database and displayed over a scanned 1 : 10 000        method is in the assessment of condition, which
map by using a standard GIS such as ArcView             may be indicated by a minimum population of a
(Hinton,1997).                                          particular species, or even a semi-quantitative
   It should be remembered, when using such             threshold. For example, good condition may be
analyses to identify trends in abundance, that          defined as a particular species being dominant or
some aquatic plants are annuals, and populations        merely present.
will fluctuate from year to year as sites are colo-        Where precise estimation of population num-
nised and recolonised. Many species fluctuate con-      bers is difficult, especially if distribution is highly
siderably, both seasonally and annually. Five-year      dispersed, then quadrat- or transect-based
means of frequency may be used to remove some of        sampling methods should be considered (see
this natural variation from the data.                   Sections 6.4.2–6.4.5). However, total counts of
                                                        individual plants are most often associated with
Summary of advantages and disadvantages                 the monitoring of particular species of conserva-
The technique provides a simple method for              tion importance. This method is therefore treated
obtaining semi-quantitative estimates of species        in more detail in Part III under the look–see
presence– absence. Data can therefore be readily        method (Section 15.2.1) and total counts
used for monitoring overall abundance and               (Section 15.2.2).
202     6 METHODS FOR SURVEYING HABITATS




  Total counts of individuals: summary of key                   Precision    Not relevant to a single measurement
  points                                                        Bias Bias occurs at high abundance
                                                                Expertise required Dependent on the difficulty of
  Recommended uses        Accurate and absolute population      species identification, but characteristic features
  counts of the changes in number of rare and/or                can be quickly learned if only a few species are being
  distinct:                                                     monitored, even if all growth stages are to be recorded
                                                                Equipment required       Basic recording equipment i.e.
  *   ‘typical’ species of a habitat
                                                                field notebooks, standardised recording forms,
  *   ‘indicator’ species of the effect of management
                                                                identification notes/guides and hand-held computers
      changes on the habitat,
                                                                on occasions
  *   ‘indicator’ species of the health of the habitat,
                                                                Key methodological points to consider
      including invasive species e.g. Ragwort (Senecio
      jacobaea) and Creeping Thistle (Cirsium arvense)          *   Select species that are suited to this method; it is not
  Efficiency    Relatively quick if the area is small and the       appropriate for large and complicated sites
  species are large and/or obvious; extremely laborious         *   Search site or area systematically without damage to
  and time-consuming if the site is large, the species              vegetation
  small, or the vegetation diverse                              *   Consider other methods for estimating populations
  Objectivity Objective, but subjective variations do               for large sites and for populations that may be large
  exist (see below)                                                 or widely distributed




6.4.2          Conventional frame quadrats: cover               fewer quadrats have to be recorded to detect the
               and density estimates                            same level of change. Thus, changes in values
                                                                (e.g. of species abundance) are more likely to be
Recommended uses                                                real than if a new set of quadrats is selected each
Visual estimation of plant cover and abundance (or              time. Although measurement error can still occur
the presence–absence of species) is a common                    (for example, through the incorrect identification
method of describing vegetation, which can be                   of species), permanent quadrats reduce sampling
applied to the whole study area or to sample plots.             variation from data and provide a clear illustration
There are several different approaches to the sub-              of change at a particular point.
ject, but most involve the placing of convent-                     There are, however, a number of disadvantages
ional quadrats (i.e. not mini-quadrats as described             of permanent quadrats. Repeated recording con-
in Section 6.4.3, nested quadrats as described in               centrated in the same small area may itself cause
Section 6.4.4, or point quadrats as described in                changes in the vegetation through trampling or
Section 6.4.5) and/or transects (see Section 6.4.6).            continued removal of material for identification.
   Quadrats may be permanent or temporary.                      It is also possible that quadrats may ‘naturally’
Permanent quadrats have fixed positions, which                  become unrepresentative of the area as a whole
are relocated at each recording period. Temporary               during the course of a monitoring scheme, and
quadrats are placed in the area to be monitored but             this may significantly bias results if only a few
removed after recording is done; a new set of tem-              are placed. (This can be overcome by recording
porary quadrats is placed in the same area at the               sets A and B in year 1, B and C in year 2, C and D
next recording period.                                          in year 3, etc.; see Greig-Smith, 1983.) Finally, mark-
   The major advantages of permanent quadrats                   ing and relocating quadrats can be very time-
are that there is usually less variation due to                 consuming, although recent developments with
the effects of sampling in the change data, and                 laser range-finders (Total Stations) are making
                                                                             6.4 Ground and shrub vegetation          203




  Frame quadrats: summary of key points                      recording forms, identification notes/guides and
                                                             hand-held computer on occasions
  Recommended uses      Monitoring of vegetation when        Key methodological points to consider
  estimates of cover or density are required (e.g. for NVC
                                                             *   Surveyors should familiarise themselves with site
  surveys)
                                                                 and vegetation prior to survey
  Efficiency   Relatively time-consuming; larger
                                                             *   Optimum quadrat size should be estimated and the
  quadrats are difficult to search thoroughly; efficiency
                                                                 number of quadrats required calculated from preli-
  of cover estimates can be particularly poor when
                                                                 minary survey data
  physical fatigue sets in
                                                             *   Care must be taken when choosing whether to
  Objectivity Cover values can be very subjective
                                                                 estimate cover by percentages or cover bands (e.g.
  Precision Deriving cover values can be very
                                                                 the Domin scale)
  imprecise, especially with large quadrats and
                                                             *   Precision and accuracy of estimates can be increased
  inexperienced surveyors
                                                                 by subdividing quadrats
  Bias   More conspicuous species, such as those in
                                                             *   Trampling of vegetation may affect the ease with
  flower or those forming clumps, may be given
                                                                 which species can be seen or cover values
  overestimated cover values
                                                                 estimated
  Expertise required   Considerable expertise is required
  to enable the accurate identification of vegetative
  vascular and lower plants; some (easily learnt) skills     Data analysis    Standard methods can be applied
  required for improving estimates of cover values           providing that the sample is representative of the site
  Equipment required     Quadrats; basic recording           Only non-parametric tests can be applied to the
  equipment, i.e. field notebooks, standardised              analysis of non-linear cover scores



quadrat relocation much easier in some habitats              vegetation is usually layered, percentage cover
(see Appendix 5).                                            values can add up to more than 100%.
   The advantages and disadvantages of permanent                This method of describing the vegetation
versus temporary plots are discussed in more detail          involves listing the species present within the sam-
in Part I, Section 2.3.2. However, recent thoughts           ple area, and attaching to each species an assess-
are that monitoring schemes based on randomly                ment of abundance. A number of recording
located permanent quadrats should be avoided                 formats and techniques have been in place for
unless minimising sampling variation is of prime             many years; the most frequently encountered and
importance.                                                  applied include the DAFOR, Braun–Blanquet and
   Cover estimates provide a good description of             Domin scales, which are alternatives to estimating
the contribution that each species makes to the              the actual percentage cover. The Daubenmire scale
vegetation community, and they are sensitive to              is used for estimating cover alone but is not widely
short-term fluctuations in season or management.             used. The most widely used method in recent years
One major advantage of cover as a quantitative               has been the Domin scale, which is most frequently
measure is that different plant groupings (mosses,           encountered by ecologists in its use by the NVC (see
herbs, grasses and shrubs) can all be evaluated in           Section 6.1.6).
comparable terms. If the vegetation has a distinct              A once widespread approach that is now rarely
layered structure (i.e. shrubs and undergrowth),             used for quadrats is the DAFOR scale, which com-
the cover of species in each layer is measured sepa-         bines subjective assessments of frequency and
rately (Bonham, 1989). Cover is normally expressed           cover into five classes: dominant, abundant, fre-
as a fraction, percentage, or amount of cover on a           quent, occasional and rare. There are no defini-
scale or index basis. It must be remembered that, as         tions for these classes. Prefixes are also used to
204    6 METHODS FOR SURVEYING HABITATS



refine and qualify assessments by using terms           identified and scored. Time taken will be further
such as ‘very’ or ‘locally’. These prefixes have been   dependent on quadrat size, vegetation diversity
widely applied but frequently misused. Some sur-        and surveyor experience. For example, a 2 m  2 m
veyors have modified the DAFOR scale with quan-         quadrat on a species-poor heathland may only
titative cover estimates (e.g. the ROFAD scale: < 5%    require 5–10 min to record, whereas an equivalent
rare, 6–20% occasional, 21–50% frequent, 51–90%         quadrat on a species-rich grassland may require up
abundant, 91–100% dominant; see Williams et al.         to 30 min. These times exclude that required to
(1998)).                                                mark permanently or relocate the site of a quadrat
   DAFOR values are especially subjective and           if this is necessary. Time efficiency can also be
imprecise and therefore of limited use for monitor-     increased with additional surveyors.
ing. Different observers tend to have different defi-      Recording can also be slower in wet vegetation
nitions for the terms. Given the variability of         and in very bright sunlight. It is also easier to miss
DAFOR ratings they will only indicate change in         species under these conditions.
species abundance (and therefore be useful as sig-         Time savings can be made if certain species or
nals for action) if the change is very large, or if     groups of species are omitted (especially species
smaller changes are consistent over a number of         that are difficult to identify quickly in the field,
years. This may be useful if monitoring only needs      such as certain bryophytes or vegetative grasses)
to detect an obvious disastrous change at a site        but the resultant data may be of only limited use.
(Byrne, 1991).                                          Savings in time can be re-allocated to recording
   The Braun–Blanquet and Domin scales give             more sites or quadrats.
more information than DAFOR, but are still subjec-         Fewer quadrats are needed than for presence–
tive and are affected by the same over- and under-      absence methods (Section 6.4.3) to detect a given
estimating problems for certain species as is           level of change.
DAFOR. Domin values vary greatly with observer
and species condition.                                  Expertise required
   More recent monitoring methods have looked at        Considerable expertise is required to be able to
the value of recording presence–absence rather          identify correctly the full range of plant species
than cover as a means of vegetation monitoring,         (including bryophytes and macrolichens) typically
with frequency as the favoured approach (Byrne,         recorded in cover value assessments (see Appendix 3
1991; Rich et al. 1991; Hodgson et al., 1995). These    for identification references). This skill can be
developments have aimed to produce more consis-         especially demanding for vegetative material.
tent data, and have been favoured because moni-         Experience is further required to estimate cover
toring is comparatively quick, they reduce the          values to a consistent level, although even then
amount of observer error through removing the           they are generally inaccurate.
need to record cover values, and they are less tiring
to carry out. These frequency methods are               Equipment required
described in Section 6.4.3 and are probably more        Temporary quadrats require some form of delimi-
appropriate for most vegetation monitoring pur-         tation to ensure accurate recording. There are
poses other than where estimates of cover and           many alternatives, some constructed beforehand
density are specifically required, e.g. for NVC sur-    and others constructed on site. The latter mostly
veys (Section 6.1.6).                                   consist of pegs (e.g. metal skewers or tent pegs)
                                                        and wire, string or cord to link between the
Time efficiency                                         pegs. Hammers or mallets are sometimes required
Considerably more time is required to estimate          to push the pegs into hard ground. There are
scores or cover values than is needed to record         advantages to using thick cord or line: it is less
simple presence–absence data. This is further           prone to tangling and stretching and it often
increased if bryophytes and lichens must be             copes better in wet conditions. Plastic-coated cord
                                                                       6.4 Ground and shrub vegetation     205



may be preferable to string. Length should be reg-       chances of detecting species, and are therefore also
ularly checked as it can change according to envir-      more likely to improve observer consistency, even
onmental conditions and stretching. It is important      though they have larger edge effects (Leach et al.,
to maintain constancy in the shape of the quadrat;       1992). If the data are to be used for NVC analysis,
a set-square or other means of triangulation is          quadrat size must be consistent with those used by
needed for measuring right angles.                       NVC surveys (Box 6.11).
   An alternative method, which is easier to use in         Second, a sampling strategy needs to be
the field (but harder to carry around between sam-       designed. This is complex and itself involves a num-
ple points), is a pre-made square (of wood, plastic or   ber of key decisions (outlined in Part I, Figure 2.6)
metal). These can be further subdivided into grids       including selection of the appropriate means of
(by using wire or string at regular intervals along      locating samples (i.e. randomly or systematically),
the quadrat margins), which can considerably             optimisation of the sampling (e.g. by stratification
improve the accuracy of cover assessments.               of the site and/or by multi-level sampling) and the
   Field notebooks, clipboards or weatherwriters         calculation of the number of samples needed to
and specially designed recording cards are tradi-        provide sufficient precision to meet the monitor-
tionally used to collect floristic data. These meth-     ing objectives. These issues are described in detail
ods of data capture require the further stage of         in Section 2.3.
transferring the information to a computer-based            Once each quadrat has been placed, a list of spe-
and/or paper filing system back in the office. Much      cies is recorded and each species assigned a value on
time can be spent on this, and errors in data tran-      whichever measurement scale system you are
scription can occur during this process. To allow        employing. To increase accuracy this normally
more time for data interpretation, records can be        involves placing the species in a broad category to
captured in the field by using a small hand-held         begin with, and then refining the record if necessary
computer, although it is possible that recording         later in the recording process (Bonham, 1989).
errors may be more difficult to check with this             Cover estimates can be made more accurately by
method of data capture. Trials in the Unit of            division of the quadrat area into smaller units; this
Comparative Plant Ecology at Sheffield University        makes estimates more objective by concentrating
showed that computers were less efficient than           the area of search, but takes more time.
paper when used for direct field recording.                 The range of cover values for each ‘layer’ is given
Optical data scanners do not cope well with              from 0 to 100%, and possibly divided into a number
muddy or creased field cards. However, consider-         of categories, with each category assigned a rating
able developments are obviously occurring in this        or scale. Use of a scale is optional but it recognises
field and this form of data entry may become more        the low accuracy of cover measurements. Most
common in future.                                        scales have unequal class intervals, which allow
   The marking and relocation of permanent quad-         easier estimation of cover-to-area relationship but
rats is covered in Appendix 5.                           are harder to analyse statistically. A finer break-
                                                         down of scale toward the lower scale values allows
Field methods                                            better estimation of less abundant species.
Before embarking on a quadrat survey, a number              A more elaborate approach, again little used in
of key decisions must be made. First, a decision         recent years, is the Braun–Blanquet scale, a semi-
has to be taken on the optimum quadrat size; this        quantitative method, which gives a combined esti-
is discussed in Appendix 4 (and see also Box 6.11).      mate of abundance and cover. The sampling unit is
Quadrat sizes normally vary from 1 m  1 m to            called a ‘releve’ and its size is based on the minimal
4 m  4 m, although in certain species-poor moor-        area concept. Cover is measured on a scale based on
land habitats 10 m  10 m may be more appropriate.       a range of percentage cover values (see Table 6.5).
The use of smaller quadrats is normally favoured as      A second scale that measures the grouping of a
they are quicker to set up and record, increase the      species is less widely used.
206     6 METHODS FOR SURVEYING HABITATS




  Box 6.13 Frame quadrats: likely problems and                   than with scales that have broad categories (Bonham,
  solutions                                                      1989).
                                                                   DAFOR values are especially subjective and impre-
  The more complex the habitat or vegetation commu-              cise and therefore of limited use for monitoring.
  nity (particularly grasslands, which can often consist of      Different observers tend to have different definitions
  a diverse and dense mixture of plants), the more likely        for the terms. Given the variability of DAFOR ratings,
  it is that differences will occur between recorders.           they will only indicate changes in species abundance
  Some plants can be especially challenging to identify,         rather than quantifying any change.
  particularly small non-flowering individuals of grass             Small species tend to score lower than conspicuous
  and sedge species. Even the same observer re-recording         ones. Similarly, a species in full flower is likely to
  the same quadrat can produce different lists                   receive a higher rating than one that is in a vegetative
  (Robertson, 1999).                                             condition at the time of sampling. Species that form
      Visual estimates of cover by means of the Domin or         clumps are also more conspicuous than species that
  a related scale will not necessarily pick up small             are more evenly scattered and would be preferentially
  changes in abundance, particularly if the recording is         classed in an abundance scale (Kershaw & Looney,
  not carried out at both times by the same observer             1983). This applies to all methods of estimating cover.
  (Greig-Smith, 1983). It is certainly not easy to detect, for   Cover can be particularly difficult to estimate accu-
  example, a 10% change in cover from 50% to 40% for a           rately in areas of diverse vegetation and tall vegetation.
  species; both would be given the same value if the                The Braun–Blanquet and Domin methods give more
  Domin scale was used. In any event, there would                information than does DAFOR, but are still subjective
  always be uncertainty over whether a detected change           and are affected by the same overestimating problems
  in cover was a reflection of the changed abundance of          for certain species as DAFOR.
  the species or was due to observer error (Hodgson et al.,         Differences between surveyors can be reduced, at
  1995).                                                         least to some degree, by training. More consistent
      Even an experienced observer may assign a species          results can also be achieved by defining the values or
  to a scale value lower or higher than that actually            scoring terms more precisely. The use of pairs of
  occupied by the species. Because the mid-points of             observers will produce more consistent results
  each class interval are widely spaced, there can be a          (Hooper, 1992) but means that more expense will be
  large variation in data between investigators. In addi-        incurred. Cover estimates made by using mini-quad-
  tion, in a species-rich herbaceous community errors of         rats (Section 6.4.3) or nested quadrats (Section 6.4.4)
  estimates are more likely with finer scale intervals           will also be less prone to error.



   The Domin scale is a modified Braun–Blanquet                     It is convenient to record species lists and rat-
approach, and is now the most widely used scale.                 ings onto standard survey cards; the area covered
The increased number of divisions (which often                   by the species list should be clearly recorded
relate to fractions, e.g. 1/2, 3/4) (see Table 6.5)              (Byrne, 1991).
enables a more detailed assessment to be made of                    Many communities contain species that are able
plant coverage; a relatively high degree of consis-              to coexist because they occupy a different temporal
tency can be obtained with experienced recorders.                niche. Therefore, the vegetation should be
However, if you have to decide on the exact percen-              recorded at a time of year when the majority of
tage cover to obtain a Domin value, more informa-                species are likely to be visible above ground, large
tion is maintained by using percentage cover than                enough for easy identification and before any
by use of the Domin value. The use of the Domin                  major perturbation of the site occurs (e.g. burning,
scale within the context of the NVC is described in              introduction of grazing or haymaking). However,
Section 6.1.6.                                                   this may not always be possible. Sampling in future
                                                                      6.4 Ground and shrub vegetation       207



               Table 6.5. The Domin and Braun–Blanquet scales

                                                                           Braun–Blanquet
               Value                Domin scale                            scale

               10                   91–100%                                —
                9                   76–90%                                 —
                8                   51–75%                                 —
                7                   34–50%                                 —
                6                   26–33%                                 —
                5                   11–25%                                 76–100%
                4                   4–10%                                  51–75%
                3                   <4%–frequent                           26–50%
                2                   <4%–occasional                         6–25%
                1                   <4%–rare                               1–5%
                +                   Insignificant: normally                <1%
                                      1–2 individuals with
                                      no measurable cover


years should always be carried out at the same time     care should be taken to check that the transformed
of year as the initial survey, although flexibility     data meet the assumptions of parametric tests.
in some years may be required to accommodate            Cover scales are non-linear, so values (even from
climatic variations (i.e. early or late summers).       randomly located quadrats) cannot be analysed by
Suggested times for surveying the floristic compo-      parametric statistics; non-parametric statistics
sition of vegetation are given in Box 6.9).             based on ranks (such as the Mann–Whitney test)
   Some common problems and solutions that may          (see Table 2.3) can be used. There has been some
be encountered with frame quadrats are given in         work on transforming Domin values so that para-
Box 6.13.                                               metric statistical analysis is possible (Currall, 1987).
                                                        However, if it is intended to collect data suitable for
Data storage and analysis                               parametric statistical analysis, it would be preferable
This method will produce percentage cover or            to collect accurate percentage cover values and use a
Domin scores for a large number of species in           suitable transformation, or use a linear scale. This
each quadrat. Some way of summarising the impor-        need not be any more time-consuming and the data
tant changes is required. One way to gain a visual      produced will be more detailed than Domin scores
impression of changes in scores for each species        transformed to percentages (Byrne, 1991).
would be simply to plot Domin scores on the y axis         Another means of identifying the more impor-
and time on the x axis. This will demonstrate any       tant changes between years is to list for each quad-
clear increases or decreases in scores, and statisti-   rat those species that have increased or decreased
cal tests designed for ranked data (Section 2.6) can    by a large amount (e.g. two or more Domin points)
be used to test for significant changes over time.      in a consistent manner over the sampling period.
   Percentage cover data are not normally distribu-     A signal for action could be registered if a large
ted, and either should be analysed with non-            consistent decrease in a species was recorded in
parametric tests or, more commonly, the data can        a majority of quadrats, but several years’ data
be transformed (Part I, Section 2.6.3) into a closer    would probably be needed before a signal could
approximation of a normal distribution and ana-         be registered, and only large changes could be
lysed with parametric statistics (e.g. t-tests), but    detected by this method (Byrne, 1991).
208      6 METHODS FOR SURVEYING HABITATS



   Where Domin data are collected from quadrat            *   The inaccurate relocation of quadrats can ser-
sizes comparable to those used for NVC surveys,               iously bias the monitoring.
the data can be used to establish and monitor
the presence of NVC communities (see Section              Other measures that may be taken in frame quad-
6.1.6).                                                   rats, such as vegetation height, presence–absence
                                                          and sub-plot frequency, are discussed in the section
                                                          on mini-quadrats below (Section 6.4.3).
Summary of advantages and disadvantages of
cover scores
Advantages
                                                          6.4.3       Mini-quadrats
*   Cover values provide good descriptions of the con-
                                                          Recommended uses
    tribution that each species makes to vegetation
                                                          Frame quadrats can provide a simple, efficient and
    communities.
                                                          reliable way of determining the frequency of spe-
*   Recording by using Domin values is widely used
                                                          cies, assessed by recording the presence or absence
    (e.g. by NVC surveys) and understood.
                                                          of species within each of a set of randomly located
                                                          samples. They can also be used to monitor species
Disadvantages                                             composition if the total number of species in ran-
                                                          dom mini-quadrats is recorded.
*   Estimation of cover can vary significantly from          The most common application of the technique
    recorder to recorder.                                 has been through the use of random mini-quadrats,
*   Recording all species present, including bryo-        which was developed in the late 1980s by the
    phytes and macrolichens, can be time-consuming.       England Field Unit of the NCC. Their method was
                                                          designed primarily to monitor a series of grassland
*   Inaccuracies in identification occur for difficult    transplants.
    species.                                                 Temporary quadrats are located either comple-
*   Domin and other cover scales are non-linear and       tely randomly across the survey area or according
    therefore values, even from randomly located          to a stratified random sampling approach (see
    quadrats, can only be analysed by using less          Part I, Section 2.3.3 for further details).
    powerful non-parametric tests.

                                                          Time efficiency
Summary of advantages and disadvantages of                Experienced recorders are usually able to record
permanent quadrats                                        between 50–100 quadrats in a full day with the ran-
Advantages                                                dom mini-quadrat method. For planning purposes,
*   Reduced variation in data due to the effects of       assume a rate of ten (10 cm  10 cm) quadrats per
    sampling means that fewer quadrats are likely to      hour based on recording all species. The time taken
    be needed to detect a given level of change.          to record quadrats is obviously slower in long vegeta-
                                                          tion and during periods of bad weather (Byrne, 1991).

Disadvantages
                                                          Expertise required
*   Repeated recording may cause changes in the           Considerable expertise is required to be accurate in
    vegetation.                                           plant identification, especially for lower plants
*   Low numbers of plots, which were originally located   and certain vegetative phases of grasses, sedges
    within representative areas of habitat, may become    and rushes (see Appendix 1 for identification
    unrepresentative of the whole site over time.         references). Cover values are not estimated, so
*   Establishment and relocation of quadrats can be       experience of their estimation is not required for
    difficult and time-consuming.                         this method.
                                                                                6.4 Ground and shrub vegetation        209




  Mini-quadrats: summary of key points                        states of grasses, bryophytes and macrolichens are to
                                                              be recorded
  Recommended uses       Monitoring floristic change in       Equipment required Quadrats, identification notes,
  relation to land management changes. Mini-quadrats          field guides and basic recording equipment (i.e. field
  are relatively quick to record. More traditional quadrat    notebooks, standardised recording forms); hand-held
  recording using cover values is less so                     computer on occasions
  Efficiency   Individual quadrats are quick to record;       Key methodological points to consider
  more quadrats are needed to detect change than stan-
                                                              *   Measuring presence–absence is very quick to record,
  dard frame quadrats
                                                                  but the estimate of frequency will be influenced by
  Objectivity Objective where only presence–absence
                                                                  quadrat size
  data collected
                                                              *   Preliminary trials should be carried out if necessary
  Precision    Poorer than other methods but this is offset
                                                                  to establish optimum quadrat size and the number
  by its efficiency, and it can be quite precise if lots of
                                                                  required to achieve the desired level of precision
  quadrats are used
                                                              *   Patchiness in species distribution will reduce the
  Bias   Frequency can be biased against species with a
                                                                  likelihood of a randomly placed quadrat finding
  more clumped distribution. Shoot frequency will be
                                                                  the species
  biased against smaller plants, but rooted frequency
  does not have this problem, although it is slower to        Data analysis      Dependent on material collected If
  assess accurately                                           presence–absence recorded, random mini-quadrats are
  Expertise required    Competence required in plant          commonly analysed by a 2 test (see Part I, Section 2.6.4);
  identification; special expertise required if vegetative    for other measures see Data storage and analysis below




                                                              purpose of having a reasonably large number of
Equipment required
                                                              quadrats is to increase precision and the likelihood
Field notebook, clipboards or weatherwriters and
                                                              of detecting change. Abundance of each species is
specially designed recording cards are traditionally
                                                              objectively measured by its frequency (the total
used to collect floristic data. These methods of
                                                              of its occurrences in the quadrats: for example,
data capture require the further stage of transfer-
                                                              presence in 80 out of 100 quadrats equals 80%
ring the information to a computer-based or paper
                                                              frequency). These frequencies can then be analysed
filing system back in the office. Much time can be
                                                              to show relatively small changes in abundance. The
spent on this, and errors in data transcription can
                                                              point of using small quadrats is that only a small
occur during this process; see also Section 6.4.2.
                                                              area has to be searched, and this increases the
                                                              chance that different observers will record the
Field methods
                                                              same list of species.
There are no simple rules for determining quadrat
                                                                 Shoot frequency (which includes all species
size for frequency surveys. The size of quadrat
                                                              rooted within the quadrats plus any aerial shoots
needed to monitor a species depends on the fre-
                                                              within or overhanging the quadrat) is generally
quency of the species and its distribution (see
                                                              used for ease of recording. Measurements of root
Appendix 4 and Box 6.11). However, for the widely
                                                              frequency provide a less biased estimate of the
used ‘mini-quadrat’ system for estimating species
                                                              frequency of occurrence of smaller plants and have
frequency in lowland grasslands, temporary quadrats
                                                              been recommended for monitoring wherever-
of size 10 cm  10 cm are used. In tussocky heath-
                                                              possible (Robertson, 1999). However, assessing root
land, 1 m  1 m quadrats may be more appropriate.
                                                              frequency can be much more time-consuming and
   Generally, for presence–absence, at least 100
                                                              can cause damage to vegetation. Which definition
mini-quadrats are required per survey area. The
210    6 METHODS FOR SURVEYING HABITATS



of occurrence is adopted is largely dependent on                     Use the second co-ordinate of the pair to deter-
monitoring objectives. Where vegetation composi-                     mine the number of paces taken away from the
tion in terms of sward or canopy is the attribute of                 marker (at 908) and place the quadrat at this
concern, defining occurrence according to shoot                      point. Either return to the bamboo cane and con-
frequency is appropriate. If the recovery of species                 tinue pacing to the next co-ordinate on the long
or species populations is of concern, root frequency                 axis or else proceed directly to the next point. If
should be monitored.                                                 the latter, it is important to return to a fixed
   The random location of samples should be car-                     landmark at intervals, to avoid cumulative loca-
ried out according to the principles outlined in                     tional errors building up. An alternative is to use a
Section 2.3.3. Once their co-ordinates have been                     GPS. Recorders should avoid looking at the vege-
obtained, quadrats should be located and recorded                    tation when pacing out quadrat locations, so as to
as follows.                                                          avoid any possibility of ‘choosing’ the final
                                                                     location.
1. First, establish and measure axes (i.e. the bound-             3. Place the quadrat on the ground once the correct
   aries of the monitoring plot) for determining ran-                position has been reached. It is important to use a
   dom co-ordinates. Measurements may be paced                       standard method for this. This might be to face
   out or defined more accurately with tape mea-                     straight ahead and place the quadrat at arm’s
   sures. Use random number tables or a spreadsheet                  length, or to place the quadrat immediately in
   package such as Excel to generate random co-                      front of the right foot. The quadrat should be
   ordinates for the required number of quadrats.                    carefully lowered into the vegetation, avoiding
   Pairs of random co-ordinates can also be gener-                   pushing vegetation from outside into the quadrat
   ated from the VegAn computer package (Blake,                      area. This can be difficult with small quadrats in
   1988). Enter random co-ordinates on the record-                   tall vegetation. If it proves very difficult a larger
   ing sheet.                                                        quadrat should be considered.
2. Work along the longer axis of the plot. Find the               4. Record litter, bare ground and vegetation height
   random number pair with the lowest co-ordinate                    (see Box 6.14), if necessary before the vegetation is
   for this axis. Pace the required distance along this              disturbed by searching for species. Litter and bare
   axis and mark the spot (e.g. with a bamboo cane).                 ground should be assessed as amounts visible



  Box 6.14 Measuring vegetation height and                        undisturbed individual at every placement of the rule.
  structure                                                       Exact protocols for this type of measurement should be
                                                                  decided and written down so that the method can be
  The most usual method for measuring height is to                repeated exactly in the future.
  assess average height crudely by eye, with a ruler                 The principle that the rule is placed randomly or
  placed upright in the vegetation.                               systematically in the centre of the quadrat is
     Another widely used technique is to employ a long            important. The quadrats in which the vegetation is
  ruler with a light disc that is able to slide along the rule.   measured should either be randomly located or on a
  The disc is dropped from a certain standard height and          stratified (restricted) random basis.
  stops when it comes into contact with the vegetation.              For random samples, the results for mean height
  The height measured is not necessarily true vegetation          from each quadrat are compared against the results of
  height, as it is affected by the weight of the disc and the     other surveys by using t-tests or equivalent.
  density and flexibility of the vegetation, but it is quick         For   stratified   random      samples,   an   overall
  to record.                                                      estimatefor the site is calculated (see Part I) and
     Other methods are to pull the vegetation straight            changes tested by using t-tests or non-parametric
  before measuring the length or to measure the tallest           equivalents.
                                                                               6.4 Ground and shrub vegetation         211



   from directly overhead. If litter or bare ground            Data storage and analysis
   is present but at less than 50% cover, a tick is            Simple analysis of whether changes have occurred
   recorded. If the cover of litter or bare ground             in the proportions of species present are normally
   is 50% or more, the figure ‘50’ is recorded.                made through the preparation of contingency
5. Record all vascular plant species present within            tables and analysed with 2 tests (see Part I) or
   the quadrat area (according to roots or shoots),            G-tests (see Sokal & Rohlf, 1996). Long-term changes
   and bryophytes if required, using a tick for                in species frequency (i.e. percentage of quadrats
   mature plants, a letter ‘S’ for a seedling, and a           containing each species) may also be analysed
   tick and an ‘S’ for mature plants and seedlings.            with, for example, Cochran’s test of linear trend.
   A seedling is defined as a plant having six or fewer           If subdivided quadrats have been used to obtain
   true leaves or any plant with cotyledon leaves.             a frequency measure for each quadrat, the propor-
   This distinction is not made for annuals such as            tion (p) of sub-quadrats containing each species can
                                                                                                     p
   Fairy Flax Linum catharticum. The species contribut-        be transformed by using a sin–1 ( p) transforma-
   ing most to cover (as viewed from directly overhead)        tion (see Part I, Section 2.6.3) and usually analysed
   is noted by a circled tick.                                 with parametric statistics.
                                                                  For further information see the statistical refer-
Other measurements that can be made are                        ence texts listed in Part I, Box 2.5.
abundance (number of plants of each species in
each quadrat) and sub-plot frequency. To collect
                                                               Summary of advantages and disadvantages
sub-plot frequency data, the quadrat is divided
                                                               Advantages
into sub-quadrats (usually 25 or 100 divisions),
and presence–absence of each species is recorded               *   The use of small quadrats increases species detec-
in each to give a frequency for each quadrat. This                 tion and recording rates.
measurement is more sensitive than simple                      *   Presence–absence data are quick to collect.
presence–absence in the whole quadrat and thus                 *   Different field workers are likely to achieve very
fewer quadrats will be needed to detect a given                    similar results. There is no need to mark and relo-
level of change.                                                   cate permanent quadrats.
   Some likely problems and solutions that may                 *   There are no problems with subjectivity and there
be encountered with this method are outlined in                    is no need to allocate cover values, only presence
Box 6.15.                                                          or absence.



  Box 6.15    Mini-quadrats: likely problems and               monitoring scheme. The England Field Unit found
  solutions                                                    10 cm  10 cm quadrats to be suitable for monitoring
                                                               the commoner species on grassland sites except where
  The larger the quadrat, the higher the frequency values      the vegetation became very coarse or tall during the
  will be and the harder and slower it will be to search for   monitoring period. However, if only selected species
  species. You should therefore use a quadrat size             had been recorded (rather than all species), larger
  appropriate for the vegetation being sampled (see            quadrats might also have been practical.
  Appendix 4). The appropriate quadrat size for fre-               Although recording rooted frequency is considered
  quency measurements varies between species; a set of         to be preferable, shoot frequency was used (in the
  nested quadrats of differing sizes may be required (see      England Field Unit’s study) to save time in recording, as
  Section 6.4.4).                                              it was felt that difficult and time-consuming decisions
     Some vegetation will be too coarse or too tall for        (particularly with grasses) would have to made in
  very small quadrats to be used. However, once estab-         determining whether or not a species was rooted
  lished, quadrat size should not be changed during a          within a quadrat.
212       6 METHODS FOR SURVEYING HABITATS



Disadvantages                                                a simple method for surveying and monitoring
                                                             grassland vegetation by us