Docstoc

A1a

Document Sample
A1a Powered By Docstoc
					                                  DRAFT Dec 15, 2005



Evaluation of Design Templates for Monitoring Snake River Spring and Summer Chinook
                                       Salmon

                             Status and Trends Subgroup, CSMEP
                   Chris Jordan, Claire McGrath, Sam Sharr, Chris Beasley;
       Pete Hahn, Charlie Petrosky, Marc Porter, Jesse Schwartz, Eric Tinus, Paul Wilson


                                            Objectives

A primary goal of CSMEP is to improve the quality of data that are used to evaluate status and
trends of fish populations in the Columbia River basin. The EPA’s Data Quality Objectives
(DQO) Process is being used in a pilot effort in CSMEP to develop and evaluate monitoring
designs for spring and summer Chinook salmon in the Snake River basin (SRSS Chinook). The
last steps in the DQO Process to specify tolerable limits on decision errors (Step 6) and optimize
the monitoring design for obtaining data (Step 7; Appendix A). This document describes work
done to address Steps 6 and 7 for decisions regarding status and trends of SRSS Chinook salmon.

A simulation model is being developed as a tool for evaluating design templates for monitoring
fish at the scale of sub-basin in the Columbia River. A design template describes 1) the location
and temporal pattern of measurements (sampling design) and 2) the specific types of
measurements that are to be made (response design; Parnell et al. 2005). Design templates will
be compared in terms of associated cost ($/yr) and probability of error in decisions. The initial
objective, described in this document, is to evaluate alternative design templates for determining
the status of SRSS Chinook salmon. The ultimate objective is to develop a tool that can be
adapted for evaluating monitoring designs for other decisions, locations, and species.


                                           Background

Management decisions may be improved by quantifying sources of variation and reducing
uncertainty in parameter estimates. Monitoring strategies should be designed using knowledge
of spatial and temporal variability and residual variation that is associated with specific sampling
methods (Larsen et al. 2001). Some variation is real and reflects differences in magnitude of
factors that control the attribute of interest. Despite its name, natural variation may result from
environmental variability that is natural or anthropogenic in origin. Natural variation may be
manifest as temporal variation, spatial variation, or individual heterogeneity (e.g., variables
related to genetics). Estimates of the magnitude of spatial variation can be improved by
increasing the number of sampling locations, and estimates of the magnitude of temporal
variation (i.e., trend detection) can be improved by increasing the number of sampling events.
Estimats of individual heterogeneity can be improved by increasing the number of sampling
units.

Residual variation reflects sampling or measurement uncertainty, i.e., variation that cannot be
allocated to spatial or temporal components. Such uncertainty arises from imprecise or biased



                                                 1
                                   DRAFT Dec 15, 2005


methods of data collection or analysis, as well as from sampling. Residual variation can be
reduced by improving methods of data collection and increasing sample sizes.

The decision that is considered in this analysis is: ―Should Snake River spring and summer
Chinook salmon (SRSS Chinook) be delisted?‖ This decision is based on the abundance,
productivity, spatial structure, and diversity of SRSS Chinook over the prior 10 years (IC-TRT
2005). Decision errors stem from missing data and natural or residual variation. Improved
methods for monitoring fish will reduce residual variation (i.e., increase data quality), which, in
turn, reduces the probability of decision errors. However, even if residual variation were
eliminated, high spatial and temporal variation could still result in a high probability of decision
error. Quantifying the sources of variation, and subsequent implications for detection of status
and trends, is needed to develop efficient monitoring designs (Larsen et al. 2001).

The SRSS Chinook Evolutionarily Significant Unit (ESU) consists of 32 historic populations,
two of which have been extirpated. These populations occur within five major population
groups (MPGs) (Appendix B). The Interior Columbia–Technical Recovery Team (IC-TRT) has
developed procedures for evaluating status of an ESU, as well as rules for delisting the ESU (IC-
TRT 2005). The TRT process combines multiple performance measures that describe
abundance, productivity, spatial structure, and diversity, to evaluate status of the ESU. For
abundance and productivity, the covariance structure for a population is compared to viability
curves that depict threshold levels for risk of extinction (Figure 1). Threshold risk levels are
defined as 1%, 5%, and 25% probability of decreasing to below a critical number of spawners
per year for a generation in a 100-year period. The critical number is determined for a
population based on its assignment to one of three size/complexity categories. This assignment
is done using a spatially-explicit model of habitat variables and empirically derived fish/habitat
relationships. The TRT suggests a recovery goal of combined abundance and productivity above
the 5% risk threshold. For spatial structure and diversity, risk level for a population is assessed
using a hierarchical analysis of 12 metrics (Table 1). The TRT stipulates that uncertainty
associated with data be reported, so that probabilities of decision errors can be estimated. In the
past, levels of uncertainty that are associated with data rarely have been communicated.

In the present analysis, the decision whether to delist SRSS Chinook salmon will rely
performance measures commonly used to evaluate abundance, productivity, spatial structure, and
diversity (Table 2-). This project addresses the following questions:
        1) What data currently are available to inform this decision?
        2) What other data (i.e., proposed but not available currently) should be considered in this
        analysis?
        3) What levels of uncertainty, or error, are associated with these data (both available and
        proposed data)?
        4) How do different monitoring designs compare in the probability of decision error and
        cost associated with each design?
        5) What are the effects of different sources of variation on the probability of decision
        error?
The approach includes documentation of existing and proposed sampling schemes, analysis of
variation associated with particular datasets, assessment of annual cost to obtain each dataset,
and construction of a simulation model that describes the probability of decision error that is



                                                  2
                                   DRAFT Dec 15, 2005


associated with various sampling designs. It should be emphasized that the primary purpose of
this study is not to evaluate the status of SRSS Chinook salmon. Rather, decisions about model
development will be based primarily on utility for investigating the effect of parameter
uncertainty on decision errors.

[insert figure 1. A/P viability curves from IC-TRT, with indication of threshold population size
based on size/complexity category as described in Appendix B of TRT documents]

[insert table 1. SS/D criteria at the population level from IC-TRT]

[insert table 2. Performance measures and metrics to evaluate]


                                              Methods

1) What data currently are available to inform this decision?

Performance measures were 1) abundance of natural-origin spawners (Figure 2), productivity,
defined as ratio of spawner to spawner (Figure 3), spatial structure, and diversity. A variety
metrics may be used to provide information on abundance, productivity, spatial structure, and
diversity of SRSS Chinook salmon. For example, productivity can be measured as ratios of
spawner to spawner, smolt to adult return, or smolts per female. Also, the specific data needed
to inform a performance measure can vary depending on characteristics of the population being
studied (Figures 2-3). To constrain the analysis, a subset of metrics was identified that was
considered adequate for evaluating status of the ESU.

A ―dataset‖ is defined herein as data that are collected using a particular sampling method. For
example, one-pass surveys of redds in a fixed area and multiple-pass surveys of redds that
comprise a spatial census are two distinct datasets. Datasets that were deemed relevant to
performance measures included counts of adult fish, counts of redds, age-structure of spawners,
percent of spawners of natural origin, counts of smolts, rates of smolt survival to Lower Granite
Dam, and sex ratio of spawners. Redd surveys were further divided into multiple datasets based
on intensity of surveys in space (fixed/non-representative and census) and time (one-pass,
multiple-pass, and census). Documentation of datasets available to inform the decision were
obtained from reports and direct communications with personnel at the Idaho Department of Fish
and Game, Oregon Department of Fish and Wildlife, Washington Department of Fish and
Wildlife, and the Nez Perce Tribe (Appendix B).


2) What other data (i.e., proposed but not available currently) should be considered in this
analysis?

In addition to the available data, datasets representing the statistical sampling of redds across
space will be considered.




                                                  3
                                                    DRAFT Dec 15, 2005


                                                                             PM
                                                                    Abundance of natural
                                                                      origin spawners
                                                                       Need spawner
                                                                   abundance information




                                Include jacks                                                               Exclude jacks
                                                                                                              Need age
                                                                                                            structure info




             Only natural origin                 Non-natural                               Only natural origin           Non-natural fish
              fish are present                   fish present                               fish are present                present
                                                Need origin info                                                         Need origin info




         Prerequisite PM:               Prerequisite PM:                               Prerequisite PM:                 Prerequisite PM:
         None                           None                                           None                             None

         Other required data:           Other required data:                           Other required data:             Other required data:
         1. Spawner                     1. Spawner                                     1. Spawner                       1. Spawner abundance
DATA     abundance                      abundance                                      abundance                        2. Spawner age structure
                                        2. % population                                2. Spawner age                   3. % population natural
NEEDS:                                  natural origin                                 structure                        origin




  Figure 2. Data needed to inform the performance measure: abundance of natural origin
  spawners.


                                                                                 PM
                                                                            Productivity
                                                                      (spawner to spawner)
                                                                    of natural origin spawners
                                                                   Prerequisite PM: Abundance
                                                                     of natl. origin spawners.
                                                                     Need age structure info




                                                                   Prerequisite PM:
                                                                   1. Abundance of natural
                                                                   origin spawners
                         DATA
                         NEEDS:                                    Other required data:
                                                                   1. Spawner age structure




  Figure 3. Data needed to inform the performance measure: productivity of natural origin
  spawners.


                                                                            4
                                   DRAFT Dec 15, 2005



3) What levels of uncertainty, or error, are associated with these data (both available and
proposed?

Uncertainty associated with datasets will be reported as variance (2) as a function of the mean
().

Measurement uncertainty: For existing datasets where the structure permits, measurement
uncertainty will be distinguished from natural spatial and temporal variability, to obtain
estimates for these three parameters. Measurement uncertainty can be estimated using
calibration studies.

Sampling uncertainty can be estimated by comparing variable estimates from samples to
parameter estimates from censuses (e.g., spawner abundance estimated from redd surveys
calibrated to spawner abundance measured at a video weir). Examples include calibration
studies for redd counts conducted for Lake Creek (Kucera and Beasley), and Catherine Creek
(Tinus).

Measurement uncertainty also can be estimated using quality control studies (e.g., rates of
hatchery fish that are mismarked, weir efficiency estimates from mark-recapture studies).

Parameter uncertainty.


4) How do different monitoring designs compare in the probability of decision error and
cost associated with each design?

Overview of model and cost estimates. Monitoring design templates have two components. The
sampling design stipulates the location and temporal pattern of measurements and the response
design describes the specific types of measurements (i.e., datasets) that will be collected. The
cost ($/yr) and benefit (probability of making a correct decision about status) will be investigated
for alternative monitoring designs using a simulation model. Assigning costs for monitoring
designs is fairly straightforward – the annual cost for obtaining a particular dataset will be
estimated for each of the 31 populations. Costs will vary solely with monitoring design applied
to the ESU, and will not depend on the particular input data that are used in analysis. Assigning
benefits associated with a particular monitoring design is more complex; this requires estimating
a number of model parameters, which are described in detail below.

Input data. A time series of distinct input data will be generated at the population level, to
represent ESU status under three scenarios: 1) Low risk of extinction, i.e., delisting is warranted,
2) high risk of extinction, i.e., delisting is not warranted; 3) moderate risk of extinction with
trend of risk increasing over time, i.e., delisting is not warranted due to uncertainty levels
overlapping with the viability curve. A ―sampling universe‖ of data that can be iteratively
sampled across a number of years will be generated.




                                                 5
                                  DRAFT Dec 15, 2005


Temporal variability. Realistic estimates of interannual variability will be incorporated for each
of the input datasets.

Spatial variability. In reality, data often are generated at the level of a reach, an individual
stream, or some other scale within a population (e.g., abundance and productivity metrics) but
may also be generated at higher levels of organization such as population and MPG (e.g, spatial
structure and diversity metrics). However, data in the model will be simulated at the spatial scale
of a population. For cases where data are generated at a finer level than decision rules are
applied and status is assessed, the spatial variance within the population-level dataset will be
included in the model, at levels deemed realistic using best-available information for individual
populations. Spatial variability that is associated with multiple sample sites within a population
will be estimated as the spatial component of variance for a dataset. For spatial variability at
scales larger than population, rules for aggregating population level data to MPGs and ESU have
been defined by the TRT (Appendix C).

Measurement uncertainty. Uncertainty that is associated with analytical methods (i.e.,
measurement error) will be estimated using calibration and quality control studies. This
parameter includes estimates for both precision and symmetry (bias) components of uncertainty.

Sampling uncertainty. Uncertainty that results from subsampling will be simulated using
monitoring efforts of various intensity (i.e., Varying proportions of the sampling universe will be
sampled, from low to high sampling effort). .

Evaluating alternate design templates. Alternate design templates to be evaluated include 1)
relatively high level of effort, 2) moderate level of effort (similar to ―best‖ current monitoring),
and 3) relatively low level of effort. For each of the three sets of input data, sampling under each
of the three monitoring designs will be simulated. The monitoring design acts as a ―filter,‖
which adds uncertainty to the input data. Annual estimates of the mean () and variance ()
will be generated. As an example, a dataset for ―abundance of natural origin spawners obtained
using single-pass fixed-area redd surveys‖ would consist of annual values for  and  for each
of the 31 populations within the ESU, over a number of years. Data will be analyzed and
decisions will be made using TRT-defined criteria (Figure 3).


(Notes on simulation of spatial components, subject to further development).
Abundance and productivity viability curves are compared to threshold curves that are based on
population size and spatial complexity (IC TRT document). Also, spatial structure and diversity
measures have an explicit spatial component. Thus, the spatial extent of the sampling universe
must be defined, in order to determine the area that a true spatial census would include. For
example, ―percent of historic spawning areas occupied‖ is a spatial structure metric that relies on
the definition of historic spawning areas. Historic spawning area might be defined as potential
spawning habitat, i.e., accessible habitat that has geomorphic and flow characteristics consistent
with spawning requirements of Chinook salmon. Using this definition, the sampling universe of
historic spawning habitat could be defined using a spatially-explicit habitat model (e.g., a GIS
database with data on geomorphology, flow, and migration barriers) or, alternatively, by
professional judgment.


                                                 6
                                   DRAFT Dec 15, 2005



Model procedure. Following is an outline for the modelling process:
  I. Generate data based on TRT rules – simulated time series by population
          a. Abundance by population and extinction risk
          b. Age structure by population/MPG/ESU
          c. Hatchery fraction by population/MPG/ESU
          d. Spatial-temporal distribution by pop/MPG/ESU
  II. Take input and generate ―monitoring data‖ using alternate monitoring programs
          a. Input and uncertainty
          b. All metrics necessary for decision
  III. Take monitoring data and put into decision rules
          a. Test 1. Can we do step III on entire raw dataset and get correct answer (Type I
              and II)
          b. Test 2. Can we do step III with monitoring data? Error rates on decision?
  IV. Sensitivity analysis, things to vary:
          a. Input data sets
          b. Monitoring design template: sampling design and response design
          c. Decision rules


 5) What are the effects of different sources of uncertainty on the probability of decision
error?

Item IV in the outline above describes sensitivity analyses that will reveal the effects of different
sources of uncertainty, including measurement uncertainty resulting sampling intensity and
method of measurement. Multiple model runs will discern measurement uncertainty from spatial
variability, because measurement uncertainty will vary depending on the analytic method being
simulated. Spatial and temporal variance will be simulated at realistic levels, but also may be
varied systematically to investigate sensitivity of model output to spatial and temporal variance.

Details on the mechanics of the simulation model
(note, work to be done here).
Generation of simulation dataset.
Define the parameters of the model
Mechanics: Coding/ programming: in VB, R, MatLab???


 [insert fig 3. CM’s drawing of TRT rules of data analysis and aggregation for A and P here]


                                     Results and Discussion


1 and 2) What data (currently available and proposed but not yet available) were
considered in this analysis?




                                                 7
                                  DRAFT Dec 15, 2005


Available data that are available to inform the decision include counts of adult fish, counts of
redds, age-structure of spawners, percent of spawners of natural origin, counts of smolts, rates of
smolt survival to Lower Granite Dam, and sex ratio of spawners. Redd counts were divided into
multiple datasets based on intensity of surveys in space and time. In addition to the available
data, datasets representing the statistical sampling of redds across space were considered (Table
2)


3) What levels of uncertainty, or error, are associated with these data (both available and
proposed?)

Following is an outline for the approach to estimate uncertainties associated with various
datasets. Uncertainty is reported as variance (σ2) as a function of the mean () for each
parameter.

Green = can estimate this (CJ working on presence/absence in space and time)
Yellow = points that need clarification
   1. Spawner Abundance: estimate abundance and σ2 where available (σ2 available for weir
       with MR). Need to add Grand Ronde-Imnaha and Lower Snake data.
          a. Count fish- Data available from IDFG projects.
                   i. DIDSON – no variance estimate
                          1. Upper Secesh in Secesh
                  ii. Video – no variance estimate
                          1. Lake Creek in Secesh
                 iii. Weir with MR – variance estimates from M-R for each weir
                          1. Upper Mainstem in SF Mainstem
                          2. Johnson Cr in EFSF
                          3. Pahsimeroi
                          4. Mainstem East Fork in EF Salmon River
                          5. Upper Salmon Upper Mainstem
                 iv. Weir without MR – no variance estimate, can apply estimates from weirs
                       with M-R
                          1. Rapid River in Little Salmon
          b. Redd counts. Uncertainty of Rindex proportional to run size. Uncertainty is
              associated with converting ―redds into fish.‖ Quantify the uncertainty observed
              for this conversion. Refer to Lake Creek analysis (P. Kucera and C. Beasley) and
              Catherine Creek analysis (E. Tinus draft).
                   i. Peak redd survey in subarea, Rindex
                          1. Rindex variance is proportional to run size.
                          2. Rindex → A uncertainty: for each site calculate variance using
                              regression for all years of data.
                  ii. Peak redd survey in total area, Rsingle census
                          1. Rsingle census variance.
                          2. Rsingle census → A uncertainty: for each site calculate variance using
                              regression for all years of data.




                                                 8
                                          DRAFT Dec 15, 2005


                      iii. Multiple redd surveys in total area (true spatio-temporal census), Rmultiple
                           census
                           1. Rmultiple census variance.
                           2. Rmultiple census → A uncertainty: for each site calculate variance
                                using regression for all years of data.
                   iv. Statistical sample redd surveys, Rstatistical
                           1. Rstatistical variance is a function of sample design and effort.
                           2. Rstatistical → A uncertainty: for each site calculate variance using
                                regression for all years of data.
                    v. Estimate redds from habitat/intrinsic potential, H
                           1. H → A uncertainty for relationships established within MPGS
                           2. H → A uncertainty for relationships established between MPGS
    2.   Age structure: (estimate separately for hard parts vs tags?)
            a. For ESU, construct length at age regression for each year by aggregating all data
                available upstream of LGR. Calculate variance from age-length key, Rmix, or
                other.
            b. For individual populations and/or hatcheries, construct length at age regression
                for each year using data from weirs and carcasses. Calculate variance from age-
                length key, Rmix, or other.
    3.   Hatchery Fraction: variances associated with hatchery mismark rate and stray rates.
            a. Population mismark rate and associated variance, get estimates from hatcheries
                and apply the rate for closest upstream hatchery to individual populations. Any
                estimate for correctly marked but mis-classified at weir?
            b. Population stray rate = % spawners in population that are non-natural origin, from
                tagged fish at weirs and in carcasses.
    4.   Presence/absence in space: CJ
          Spatial sampling (redd surveys) uncertainty proportional to spatial balance.
          Estimated using point samples (weir) – large uncertainty based on habitat covariates
            (and for example dependent on number of spawners – reasoning that with more
            spawners they will spread out in space more, so, e.g.: ―If and only if A > 2 QET‖)
    5.   Presence/absence in time: CJ
          Spatial sample uncertainty proportional to temporal coverage
          Point sample in time (weir) with ―no‖ error

Notes on estimating σ2(A) :
                   Where measurement approach               Paramaterize (data available to     Who?
                   affects data quality                     estimate uncertainty?)
    Spawner        Spawners with weir → know A with         MR data from some, apply to all     CM
    abundance      no error for population
                   Redd index → know Rindex with            NPT in SFSR and IDFG Pahsimeroi,    CM
                   uncertainty proportional to run size,    MFSR
                   Rindex → A
                   Statistical sample → know Rhat with      Fxn of sample design and effort :   CJ
                   uncertainty- Function of sample design   MFSR
                   and effort and expansion of Rhat → A
                   with uncertainty                         expansion of Rhat → A with          CM
                                                            uncertainty : ODFW Grande Ronde,
                                                            Paul Kucera
                   Single pass census → know R census,      Construct for Pahs/SFSR multiple    CM




                                                            9
                                          DRAFT Dec 15, 2005

                   expand Rcensus → A with uncertainty in     pass dta
                   spawning duration
                   True spatio-temporal census; R→ A          Use ODFW Grande Ronde                    CM
                   expansion
                   Habitat/intrinsic potential based, H →     Talk with Michelle McClure               CJ
                   A within or between MPGs
    Age            Age structure for ESU with and without     Uncertainty in ESU length at age         CM
    structure      error on annual basis on populations       relationship (IDFG LGR annual
                   without weirs but known with ESU           estimate)
                   error on populations with weir             Uncertainty in distribution of age
                                                              structures between populations (IDFG
                                                              explore weir – carcass data)
    Hatchery       Population with weir → known with          Mis-assignment rate and stray rate for   CM
    fraction       no error                                   hatchery by population
                                                              Stray rate by closest upstream
                                                              hatchery
                   Redd based A → carcass surveys →            (just use above method)                 CM
                   error proportional to sampling effort
                   and design
                   Habitat based A → Hatchery fraction        (just use above method)                  CM
                   from closest population and closest
                   hatchery

    Presence       Spatial sampling (redd surveys)            Make up rules that are consistent with   CJ
    absence in     uncertainty proportional to spatial        TRT documents
    space          balance
                   Point samples (weir) – large uncertainty   Same                                     CJ
                   based on habitat covariates (If and only
                   if A > 2QET)
    Presence       Spatial sample uncertainty proportional    Same                                     CJ
    absence in     to temporal coverage
    time           Point sample in time (weir) know with      Same                                     CJ
                   ―no‖ error




Redd counts are a key source of data for all performance measures. Redd counts are used to
estimate number of spawners for abundance and productivity metrics, spatial distribution for
spatial structure, and (diversity?). Redd counts are conducted using a variety of methods, which
have varying levels of associated error (Table 2).


Table 2. Estimates of precision associated with various methods of redd surveys. Precision
is assessed in terms of magnitude and symmetry (i.e., the likelihood to overestimate or
underestimate). Unshaded cells represent datasets that are currently available. Light
shading represents datasets that are not currently available but will be considered in this
study. Dark shading represents datasets that are not currently available and will not be
assessed in this study.
                                                                Temporal extent
Spatial extent              One pass                        Multiple pass                Census
Fixed, non-                 ―Fixed-once‖                    ―Fixed-multi‖                ―Fixed-cen‖
representative              Data currently available;       Data currently available;    Data currently available;
                            done commonly                   done commonly                done rarely
(non-sym error is assoc     Large error associated          Non-symmetric error          Large error associated
with this)                  Non-symmetric error             associated                   Non-symmetric error


                                                              10
                                      DRAFT Dec 15, 2005

                         associated                  ∆-0.75                      associated
                         Too comfortable                                         Too comfortable
                         ∆-1                                                     ∆-0.5

Statistical              ―Stat-once‖                 ―Stat-multi‖                ―Stat-cen‖
                         Data not available;         Data not available;
                         implement                   implement
                         Error
                         (noisy, i.e., time has
                         error, but probably
                         symmetrical)
Census                   ―Cen-once‖                  ―Cen-multi‖                 ―Cen-cen‖
                         Data currently available;   Data currently available;
                         done rarely (can            done commonly
                         estimate by subsampling     Less error
                         from cen-multi datasets)    Symmetric
                         Less error                  N(mean ∆=0, variance
                         Symmetric (but time         ∆=0.01)
                         stil. has error)
Notes: Any time a fixed point in time is used there will be a symmetry issue, if the point that is fixed in
time non-representative. Because redds integrate (add) over time, it is probably better to survey later than
earlier.


4. How do different sampling designs compare in the probability of decision error and cost
associated with each design?

Results of modelling exercise will be described in this section.


 5) What are the effects of different sources of uncertainty on the probability of decision
error?

Results of sensitivity analyses will be described in this section.

Methodological issues to consider (not sure how to incorporate yet)…
For performance measures that have a spatial component, the spatial extent of the sampling
universe influences potential sources of bias in sampling designs. For example, if the universe is
consistently defined as larger than it is in reality, spawning will be observed in a relatively small
area of potential habitat. In turn, risk will be defined as greater than it is in reality, Alternatively
if we always define it too small will tend to underestimate risk.




The models ability to simulate relative changes (rather than absolute value) can be addressed by
running different scenarios reflecting changes in abundance, productivity, spatial structure and
diversity.



                                                      11
                                 DRAFT Dec 15, 2005


Contribution of two sources of uncertainty (




                                               12
                                 DRAFT Dec 15, 2005


References

IC-TRT (Interior Columbia basin Technical Recovery Team). 2005. Viability Criteria for
Application to Interior Columbia Basin Salmonid ESUs, DRAFT June 2005.

Larsen, D.P., T.M. Kincaid, S.E. Jacobs, and N.S. Urquhart. 2001. Designs for evaluating local
and regional scale trends. BioScience 51:1069—1078.

Parnell, I.J., D.R. Marmorek, and M. Porter. 2005. Collaborative Systemwide Monitoring and
Evaluation Project, Definition and Evaluation of Design Templates. DRAFT. Prepared for
Columbia Basin Fish and Wildlife Authority, Portland, OR. Prepared by ESSA Technologies
Ltd., British Columbia, Canada. 94 pp.




                                              13
                                 DRAFT Dec 15, 2005


                      Appendix A – CSMEP DQO Process Steps 6 – 7

Step 6. Specify tolerable limits on decision errors

   a. What are the primary management questions of interest that are the focus of the design
      work?
   b. What are the acceptable limits on decision errors (―reasonable‖ range of values for  and
      )?
   c. How is uncertainty in data around these questions currently considered in the decision
      making process and how is it communicated?
   d. What are the key performance measures (PMs) relevant to these management questions?
   e. What are the effect sizes of interest for these key PMs?
   f. What are the primary analytical methods by which changes in these PMs will be
      evaluated?
   g. What are the sources of total error in such evaluations (process error, sampling design
      error, measurement error?). How will designs manage/minimize any of these sources of
      error?

Step 7. Design for Obtaining Data

   a. Outline potential L, M, H options for sampling designs (i.e. where and when), and
      response designs (how and what).
   b. How will tradeoffs among potential designs be evaluated?
   c. What additional work (field/analytical) will be required to move this process along and
      improve designs?




                                              14
                                    DRAFT Dec 15, 2005

Appendix B. Data summary for SRSS Chinook salmon. Note that other data are available in the basin – this table summarizes only those
data relevant to the decision to “Delist SRSS Chinook salmon,” using select performance measures. Methods are described for obtaining
the following datasets: adult fish counts, redd counts, age-analysis, % natural origin, smolt counts, smolt survival estimates, and sex ratio
of spawners.



MPG          TRT           Sampling area       Count         Count redds         Cohort /     % natural       Count          Estimate      Sex ratio   Other
             Population    (“subpopulation”)   adult fish    (single or          age-         origin          smolts         smolt         of
                                               (if weir is   multiple pass,      analysis                                    survival      spawners
                                               m-r           a subarea of
                                               estimate      spawning area
                                               done as       or total census)
                                               well?)
             SF            Upper mainstem      Weir count    Multiple pass       Length       %natural        Smolt trap     Survival      Sex ratio   Genetics in
SF Salmon    Mainstem                          with MR       ground and          frequency    origin          And            estimated     spawners    hand to do
                                               est.          single aerial in    using LGR    hatchery        (snorkel       with PIT      from        parental
                                                             same subarea        age-length   marks and       and            tags          carcasses   exclusion
                                                             (24.5 km)           key          tags directly   electrofish                              here like in
                                                                                              handling fish   for parr)                                Pahsimeroi.
                                                                                              at weir
                           Weir to EFSF                      Single pass                                      Snorkel
                           confluence                        subarea                                          surveys
                           EFSF confluence
                           to mouth (nothing
                           done)
             Secesh        Upper Secesh        DIDSON        Multiple pass       Length       Carcass         Trap           PIT tag       Carcass
                                                             ground subarea      frequency    surveys         (electrofish
                                                             (40.1 km);                                       for parr)
                                                             aerial from
                                                             Loon Cr to
                                                             Warm Springs
                                                             Cr.
                           Lake Creek* (is     Video weir    Multiple pass       Length       Carcass         Smolt trap,    PIT tag       Carcass
                           part of Upper                     ground subarea      frequency    surveys,        snorkel
                           Secesh)                           (20.8 km,                        remotely        surveys
                                                             mouth to                         sensed clips    (electrofish
                                                             Willow Cr)                       through         for parr)
                                                                                              video weir
                           Below DIDSON                      Single pass                                      Snorkel        PIT tag few
                           to mouth                          aerial                                           surveys,       Chinook at
                                                                                                              smolt trap     screw trap
                                                                                                              below Lick



                                                                                15
                                         DRAFT Dec 15, 2005

MPG           TRT           Sampling area       Count         Count redds         Cohort /    % natural       Count         Estimate     Sex ratio   Other
              Population    (“subpopulation”)   adult fish    (single or          age-        origin          smolts        smolt        of
                                                (if weir is   multiple pass,      analysis                                  survival     spawners
                                                m-r           a subarea of
                                                estimate      spawning area
                                                done as       or total census)
                                                well?)
                                                                                                              Cr
                                                                                                              confluence
                                                                                                              (Alan) can
                                                                                                              be operated
                                                                                                              for
                                                                                                              Chinook
              EFSF          Johnson Creek       Weir with     Multiple pass       Length      Examination     Trap          PIT tags     Carcass
                                                MR est        ground, subarea     frequency   at weir
                                                              (41.4 km)
                            Mainstem EFSF                     Multiple pass                   Carcass         Snorkel                    Carcass
                                                              ground subarea                  survey
              Little        Rapid River         Weir count                        Length      Hatchery        Snorkel       Alan plans
              Salmon                            considered                        frequency   marks           Alan plans    to put a
                                                a census                                      directly        to put a      screw trap
                                                                                              handling fish   screw trap    in here
                                                                                                              in here
                            Slate Creek                       Multiple pass                                   Snorkel?
                                                              gournd
                                                              subarea(?) (6.6-
                                                              15.4 km) –are
                                                              these still done?
                                                              ISS report 91-
                                                              96
                            Whitebird Cr                                                                      Snorkel
Middle Fork   Chamberlain   Chamberlain Cr                    Single pass         Length      Carcass         Snorkel                    Carcass
              Cr                                              ground subarea      frequency   survey
                            West Fork                         Single pass         Length      Carcass         Snorkel                    Carcass
                            Chamberlain Cr                    ground subarea      frequency   survey
              Big Creek     (Only one)                        Single pass         Length      Carcass         Snorkel                    Carcass
                                                              aerial subarea      frequency
                                                              (hurow site)
                                                              and multipass
                                                              ground (NPT
                                                              And IDFG)
              Lower MF      (Only one)                        Single pass                                     snorkel
              Mainstem                                        aerial subarea
              Camas Cr      (only one)                        Single pass                                     snorkel



                                                                                16
                                 DRAFT Dec 15, 2005

MPG   TRT           Sampling area       Count         Count redds        Cohort /      % natural   Count     Estimate   Sex ratio   Other
      Population    (“subpopulation”)   adult fish    (single or         age-          origin      smolts    smolt      of
                                        (if weir is   multiple pass,     analysis                            survival   spawners
                                        m-r           a subarea of
                                        estimate      spawning area
                                        done as       or total census)
                                        well?)
                                                      aerial subarea
      Loon Cr       (only one)                        Single pass        Length        Carcass     Snorkel              Carcass
                                                      aerial (Thurow)    frequency
                                                      subarea,
                                                      multiple pass
                                                      ground
                                                      (Thurow)
      Sulphur Cr    (only one)                        Single pass        Length                    Snorkel              Carcass
                                                      ground subarea     frequency
                                                      and single pass
                                                      aerial (Thurow)
      Bear Valley   Elk Creek                         Single pass        Length        Carcass     Snorkel              Carcass
      Cr                                              ground subarea,    frequency
                                                      single pass
                                                      aerial (Thurow)
                    Mainstem                          Single pass        Length        Carcass     Snorkel              Carcass
                                                      ground subarea,    frequency
                                                      and single pass
                                                      aerial subarea
                                                      (Thurow)
      Marsh Cr      Mainstem                          Multiple pass      Length        Carcass     Trap      PIT tags   Carcass
                                                      ground census      frequency
                                                      (11 km), and       (Carcass –
                                                      single pass        very few**)
                                                      aerial (Thurow)
                    Cape Horn Cr                      Multiple pass      Length        Carcass     Snorkel              Carcass
                                                      ground subarea,    frequency
                                                      and single pass
                                                      aerial (Thurow)
                    Beaver Creek                      Multiple pass      Length        Carcass     Snorkel              Carcass
                                                      ground subarea,    frequency
                                                      and single pass
                                                      aerial (Thurow)
                    Knapp Cr                          Multiple pass      Length        Carcass     Snorkel              Carcass
                                                      ground subarea,    frequency
                                                      and single pass
                                                      aerial (Thurow)



                                                                        17
                                     DRAFT Dec 15, 2005

MPG        TRT          Sampling area       Count         Count redds         Cohort /    % natural    Count         Estimate   Sex ratio      Other
           Population   (“subpopulation”)   adult fish    (single or          age-        origin       smolts        smolt      of
                                            (if weir is   multiple pass,      analysis                               survival   spawners
                                            m-r           a subarea of
                                            estimate      spawning area
                                            done as       or total census)
                                            well?)
           Upper MF     (only one)                        Single pass                                  Snorkel
           Mainstem                                       aerial subarea
           (and upper                                     (Thurow)
           tribs)       Pistol Cr                         Single pass                                  Snorkel
                                                          aerial subarea
                                                          (Thurow)
Upper      North Fork   (only one)                        Single pass         Length      Carcass      Snorkel                  Carcass
Mainstem                                                  aerial subarea;     frequency
Salmon                                                    multiple pass
                                                          ground (36.8
                                                          km)
           Lemhi        (only one)                        Single pass         Length      Carcass      2 screw       PIT tags   Carcass
                                                          aerial subarea;     frequency   (very few)   traps (one               (very few)
                                                          multiple pass                                above
                                                          ground subarea                               Hayden Cr
                                                          (51.7 km)                                    confluence,
                                                                                                       one above
                                                                                                       mouth);
                                                                                                       snorkel
           Pahsimeroi   (only one)          Weir with     Multiple pass       Length-     Weir         Trap;         PIT tags   carcass        Also do
           R                                MR            ground census;      frequency                snorkel                  (and           parent
                                            estimate      single pass                                                           estimated at   analysis to
                                                          aerial census                                                         weir)          estimate
                                                                                                                                               contribution
                                                                                                                                               from
                                                                                                                                               different
                                                                                                                                               parent types
           Upper        (only one)                        Multi- pass         Length      Carcass      Snorkel                  Carcass
           Salmon                                         aerial (census?)    frequency
           Lower                                          (59.0 km)
           Mainstem
           East Fork    Mainstem East       Weir with     Single pass         Length      Weir         Trap;         PIT tags   Carcass        Genetics in
           SR           Fork                MR est        aerial subarea;     frequency                snorkel                                 hand to do
                                                          multiple pass                                                                        parental
                                                          ground subarea                                                                       exclusion
                                                          (15.3 km)                                                                            here like in
                                                                                                                                               Pahsimeroi.



                                                                             18
                                        DRAFT Dec 15, 2005

MPG          TRT           Sampling area        Count         Count redds        Cohort /      % natural       Count      Estimate   Sex ratio   Other
             Population    (“subpopulation”)    adult fish    (single or         age-          origin          smolts     smolt      of
                                                (if weir is   multiple pass,     analysis                                 survival   spawners
                                                m-r           a subarea of
                                                estimate      spawning area
                                                done as       or total census)
                                                well?)
                           Herd Cr                            Single pass        Length        Carcass         Snorkel               Carcass
                                                              aerial subarea;    frequency
                                                              multiple pass
                                                              ground subarea
                                                              (Shoban) (16.4
                                                              km)
             Yankee Fork   West Fork                          Single pass        Length        Carcass         Trap;      PIT tags   Carcass
                                                              aerial subarea;    frequency                     snorkel
                                                              multiple pass
                                                              ground subarea
                                                              (11.6 km)
                           Mainstem                           Single pass                                      Snorkel                           Shobans
                                                              aerial subarea                                                                     proposing
                                                                                                                                                 to put weir
                                                                                                                                                 in here
             Valley Cr     (only one)                         Single pass        Length        Carcass         Snorkel,   PIT tags   Carcass
                                                              aerial subarea;    frequency                     seining
                                                              multiple pass                                    parr by
                                                              ground subarea                                   NOAAF
                                                              (33.2 km)
             Upper         Mainstem and         Weir with     Single pass        Length        Visual at       Trap;      PIT tags   Carcass     Genetics in
             Salmon        tribs (includes      M-R           aerial subarea     frequency     weir, cwt       snorkel                           hand to do
             upper         Alturas Lake                       (in multiple                     scan at weir;                                     parental
             mainstem      Creek sampling                     locations);                      carcass                                           exclusion
                           program)                           multiple pass                                                                      here like in
                                                              ground subarea                                                                     Pahsimeroi.
                                                              (14.0 km)
             Panther       No sampling - extirpated
             Creek -
             extirpated
Grande       Wenaha        Most of spawning     N/A           Multiple pass      Carcass       External        N/A        N/A        Carcass     Egg
Ronde –                    area                               ground, census     scale         marks/cwt                                         retention
Imnaha                                                                           analysis or   from                                              rate in
                                                                                 length-freq   carcasses                                         carcasses
Check data                                                                       applied to
                                                                                 length at
                                                                                 age key



                                                                                19
                              DRAFT Dec 15, 2005

MPG   TRT            Sampling area       Count         Count redds        Cohort /      % natural   Count        Estimate      Sex ratio   Other
      Population     (“subpopulation”)   adult fish    (single or         age-          origin      smolts       smolt         of
                                         (if weir is   multiple pass,     analysis                               survival      spawners
                                         m-r           a subarea of
                                         estimate      spawning area
                                         done as       or total census)
                                         well?)
      Lostine        Lostine             Weir with     Multiple pass      Carcass       External    Trap rkm 3   PIT tags      Carcass     Egg
                     mainstem for        MR est.       ground, census     scale         marks/cwt                                          retention
                     weir and SGS;                                        analysis or   from                                               rate in
                     SGS only for                                         length-freq   carcasses                                          carcasses
                     Wallowa and                                          applied to
                     tribs upstream of                                    length at
                     Lostine R mouth                                      age key
      Minam          Most of spawning    N/A           Multiple pass      Carcass       External    Trap rkm 1   Trap          Carcass     Egg
                     area                              ground, census     scale         marks/cwt                catches/PIT               retention
                                                                          analysis or   from                     tags                      rate in
                                                                          length-freq   carcasses                                          carcasses
                                                                          applied to
                                                                          length at
                                                                          age key
      Catherine      Most of spawning    Weir with     Multiple pass      Carcass       External    Trap rkm     PIT tags      Carcass     Egg
      Cr             area                MR est.       ground, census     scale         marks/cwt   32                                     retention
                                                       (with some         analysis or   from                                               rate in
                                                       property access    length-freq   carcasses                                          carcasses
                                                       issues)            applied to
                                                                          length at
                                                                          age key
      Upper          Most of spawning    Weir with     Multiple pass      Carcass       External    Trap rkm     PIT tags      Carcass     Egg
      Mainstem       area                MR est.       ground, subarea    scale         marks/cwt   299                                    retention
                                                       (access to         analysis or   from                                               rate in
                                                       private land       length-freq   carcasses                                          carcasses
                                                       issues)            applied to
                                                                          length at
                                                                          age key
      Imnaha         Most of spawning    Weir with     Multiple pass      Carcass       External    Trap rkm 7   PIT tags      Carcass     Egg
      River          area                MR est.       ground, census     scale         marks/cwt                                          retention
      Mainstem                                         (with minor        analysis or   from                                               rate in
                                                       property access    length-freq   carcasses                                          carcasses
                                                       isues)             applied to
                                                                          length at
                                                                          age key
      Big Sheep Cr   Most of spawning    N/A           Multiple pass      Carcass       External                 N/A           Carcass     Egg
                     area                              ground, census     scale         marks/cwt                                          retention



                                                                         20
                                          DRAFT Dec 15, 2005

MPG             TRT             Sampling area        Count         Count redds         Cohort /      % natural    Count         Estimate      Sex ratio     Other
                Population      (“subpopulation”)    adult fish    (single or          age-          origin       smolts        smolt         of
                                                     (if weir is   multiple pass,      analysis                                 survival      spawners
                                                     m-r           a subarea of
                                                     estimate      spawning area
                                                     done as       or total census)
                                                     well?)
                                                                                       analysis or   from                                                   rate in
                                                                                       length-freq   carcasses                                              carcasses
                                                                                       applied to
                                                                                       length at
                                                                                       age key
                Lookingglass                                       Multiple pass
                Creek -                                            ground surveys
                extirpated                                         ? (yes-J. Hesse)
Lower Snake     Tucannon                             Weir          Yes, mainstem       Yes           Weir, cwt    5 ft rotary   Yes           Carcasses     Fecundity
Check data                                                         rkm 34 - 84                       recoveries   screw trap                  and           and adult
                                                                                                     from         at rkm 3                    broodstock    pre-spawn
                                                                                                     carcasses                                collection    mortality
                                                                                                     and
                                                                                                     broodstock
                                                                                                     collected
                Asotin –                                           Index
                extirpated?

Sources of Information for Idaho:
ISS – Annual cooperative reports, 5 yr summary reports for 97-01, 92-96.
GPM - Database
Statewide Redd Survey – Database
Table B2 for SFSR Sp/Su Chinook in text format (on CSMEP site) P. Kucera and C. Beasley

Sources of information for Oregon (Grande Ronde-Imnaha and L. Snake):
LSRCP annual reports
Personal communications Brian Jonnason, ODFW NE Fisheries Research, La Grande (and Eric Tinus, ODFW Columbia River Coordination Program, Clackamas)

Tranquilly, J.V., B.C. Jonasson, M.L. Keefe, and R.W. Carmichael. 2004. A compendium of Grande Ronde River and Imnaha River basins Spring Chinook salmon spawning
ground surveys conducted from 1948 through 2003. Fish Research and Development, Northeast Region, Oregon Department of Fish and Wildlife, La Grande, Oregon.


Sources of information for Washington (Tucannon):
Gallinat, M.P., J. Bumgarner, L. Ross, and M. Varney. 2001. Tucannon River spring Chinook salmon hatchery evaluation program 2000 annual report. Washington Department
of Fish and Wildlife, Fish Program, Science Division, Olympia, WA.




                                                                                      21
                                  DRAFT Dec 15, 2005



Appendix C – Rules for aggregating population-level data to evaluate status of an ESU, and
                                related issues to consider


Aggregating data from populations to MPGs:
Step 5 Decision rules states that ―if half of the populations in an MPG are viable, and other
extant populations in the MPG are maintained, then the MPG is viable.‖ If there is only one
population within the MPG, then data must be collected from that population. When multiple
populations are within an MPG, data from one or more sampled populations may be used to
estimate unsampled populations. Rules need to be established about whether data need to be
gathered from within every population (which populations, frequency, etc), and, if not, how to
make estimates for unsampled populations. For example, rules may establish habitat criteria for
populations to be considered as similar populations.

Evaluating status using viability curves:
Need rules for which A and P metrics to use (or how to combine individual metrics) to assess
viability, and how many years of data to use in the assessment. Need numbers on axes of the
viablity curve. Need method for assessing confidence levels of A and P, and rules for classifying
as viable or not if confidence intervals cross the viability curve.

Also need rules for populations that are not sampled at all. Do we estimate values for A and P
from similar populations, and if so, what are the rules for doing this? Or do we simply assume
the population remains viable if similar populations are viable?

Decision rules:
Step 5 – Decision Rules give the strategy for combining estimates at the population level to MPG
level. Need to clarify these rules in the case where not all populations are directly sampled, and
where populations are sampled with different intensity.


----/




                                               22
                                       DRAFT Dec 15, 2005

Table . Method details. NOTE DATA USE COLUMNS NEED TO BE CORRECTED.                          <insert method descriptions.doc.>
                                                                                                                                              Data use
 Data need                  Method                                                    Description
                                                                                                                                          A    P    SS   D
Abundance       A1      Census weir               Weir with direct count that can be considered a census (e.g., DIDSON, video)               
of fish         A2      Weir with MR              Weir with direct count (trap) and mark-recapture to estimate weir efficiency (may use
                                                  mark rate on carcasses as recapture)
                                                                                                                                             
                A3   Weir without MR              Weir with direct count (trap) without MR estimate.                                         
                A4   MR survey, no weir           Closed population (no weir) with mark-recapture (traps or electrofishing)                  
Abundance /     B1   Cen-cen                      Spatial census, temporal census                                                                
spatial         B2   Cen-multi                    Spatial census, multiple passes in time                                                        
distribution
of redds
                B3   Cen-once                     Spatial census, one pass in time                                                               
                B4   Stat-cen                     Statistical sampling in space, temporal census                                                 
                B5   Stat-multi                   Statistical sampling in space, multiple passes in time                                         
                B6   Stat-once                    Statistical sampling in space, one pass in time                                                
                B7   Fixed-cen                    Fixed sub-area (index area), temporal census                                                   
                B8   Fixed-multi                  Fixed sub-area (index area), multiple passes in time                                           
                B9   Fixed-once                   Fixed sub-area (index area), one pass in time                                                   
Age structure   C1   Tags                         Record known age of tagged and remotely-sensed spawners in weirs or fish-ladders                
of spawners     C2   Hard parts                   Measure annuli on scales, fin-rays, or otoliths                                                 
                C3   Length at age                Rely on already established length-age relationships                                           
Origin of       D1   Hatchery marks, handle       Record % fish with hatchery marks on marked and trapped spawners
spawners             fish
                                                                                                                                             
               D2    Hatchery marks,               Record % fish with hatchery marks on remotely-sensed spawners in weirs or fish-
                     remotely sensed               ladders
                                                                                                                                             
               D3    Hatchery marks on             Record % fish with hatchery marks during carcass surveys
                     carcasses
                                                                                                                                             
Sex ratio of   E1    Carcass survey                Record sex of carcasses during ground survey                                                          
spawners       E2    Female per redd               Must already know # redds
                     expansion
                                                                                                                                                         
Abundance      F1    Juvenile trap                 Trap with direct count and mark-recapture to estimate trap efficiency                             
of smolts      F2    Electrofishing                Count fish using multiple pass electrofishing depletion                                          
               F3    Snorkel survey                Count fish in survey reaches and estimate density                                                
Survival of    G1    Mark recapture                Trap and mark, recapture and detect                                                               
smolts         G2    Mark resight                  Trap and mark, resight and detect (ex of resight - remote PIT tag detections)                     
H:\work\rme-csmep\reports\ ST-Steps6-7Dec19-05ccm.doc



                                                                             23
                                       DRAFT Dec 15, 2005

Notes:
Smolt abundance and survival not needed for spawner:spawner productivity; would be needed for SARs or recruit (smolt) per spawner.
Cost of not having certain information is higher at some sites than for others.




                                                                            24
DRAFT Dec 15, 2005




           25

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:10/20/2011
language:English
pages:25