Macroinvertebrate-based index of biotic integrity for protection

Document Sample
Macroinvertebrate-based index of biotic integrity for protection Powered By Docstoc
					J. N. Am. Benthol. Soc., 2002, 21(4):686–700
   2002 by The North American Benthological Society

      Macroinvertebrate-based index of biotic integrity for protection of
                       streams in west-central Mexico

                                                      BRIAN M. WEIGEL1
 Wisconsin Department of Natural Resources, Bureau of Integrated Science Services, 1350 Femrite Drive,
                                  Monona, Wisconsin 53716 USA

                                                       LISA J. HENNE
           Los Alamos National Laboratory, Ecology Group (RRES-ECO), PO Box 1663, MS M887,
                                   Los Alamos, New Mexico 87545 USA

                                             LUIS M. MARTINEZ-RIVERA

                         ´                          ´
       Instituto Manantlan de Ecologia y Conservacion de la Biodiversidad, Universidad de Guadalajara,
                                                                 ´               .
               Independencia Nal. #151, Apartado Postal #64, Autlan, Jalisco C. P 48900 Mexico

          Abstract. The water and habitat quality of many streams in west-central Mexico are influenced by
       municipal and industrial effluent, as well as water diversion for irrigation, livestock, and deforestation.
       Restoration efforts have been hampered by a lack of clear standards against which to judge the degree
       and trend in environmental degradation. We describe the development and characteristics of a mac-
       roinvertebrate-based index of biotic integrity (IBI) designed to provide such standards. Eight metrics
       chosen a priori comprised the IBI: catch per unit effort, generic richness, % Ephemeroptera–Plecop-
       tera–Trichoptera genera, % Chironomidae individuals, Hilsenhoff Biotic Index, % depositional indi-
       viduals, % predator individuals, and % gatherer genera. Each metric distinguished sites with mini-
       mum or moderate human influence from sites with severe influence. The IBI was developed with
       data from 27 sites and validated with 6 others. Values from the development data set correlated well
       with the measures of human influence based on qualitative assessment of habitat and water quality
       (Pearson’s r                                      ´
                       0.86). IBI values for 7 sites on Rıo Ayuquila corresponded with a documented longi-
       tudinal pattern of human influence and the existing fish-based IBI (Pearson’s r          0.87). This macro-
       invertebrate IBI shows promise for developing biological standards, facilitating long-term monitoring,
       and improving ecological integrity of streams in west-central Mexico.
         Key words: IBI, macroinvertebrates, metrics, biomonitoring, bioassessment, water quality, streams,
       benthos, Mexico.

   The Sierra de Manantlan Biosphere Reserve is                    Point-source and nonpoint-source pollution
located in west-central Mexico, which is defined                are prevalent throughout Mexico. Only 7% of
here as the area having Pacific drainages be-                   the surface waters are considered in acceptable
tween the cites of Acapulco and Mazatlan (Fig.                 condition, whereas 86% are classified as con-
1). Humans are encouraged to live in the Re-                   taminated to excessively contaminated (SE-
serve in a manner compatible with preserving                   MARNAP 1999). Sugarcane agriculture neces-
its environmental integrity (Lyons and Navarro                 sitates use of stream water for irrigation and is
1990). Streams provide critical water resources                a major source of sediment to streams in west-
for society in this seasonally arid region, yet se-                                            ´
                                                               central Mexico. For example, Rıo Ayuquila and
rious declines in water and habitat quality                      ´
                                                               Rıo Juchipila have up to 99% of their volume
threaten both ecological integrity and the qual-               diverted for agricultural irrigation (Lyons et al.
ity of human life (Santana et al. 1993, Martınez                           ´
                                                               1998, Martınez et al. 2000). Agricultural fields
et al. 2000). To assist in improving stream eco-               are flooded and excess water is channelized
logical integrity, we tailored a popular assess-               back to the stream, but return flows carry a high
ment tool used in the United States to this re-
                                                               sediment load and have an altered water chem-
                                                               istry. Sugarcane processing produces highly en-
                                                               riched organic byproducts that are flushed di-
      E-mail address:             rectly to streams along with caustic cleaning so-

2002]                     MACROINVERTEBRATE IBI            FOR   MEXICAN     STREAMS                          687

   FIG. 1. West-central Mexican sites used to develop and validate the macroinvertebrate index of biotic integ-
rity (IBI). Development sites (1–27) not labeled on the figure include: 1-Arroyo San Jose del Tule; 2-Arroyo San
Jeronimo; 3-Arroyo Contla; 6-Arroyo Las Marias; 7-Rıo Cuzalapa at Cuatitlan; 8-Arroyo El Durazno; 22-Rıo         ´
                                ´                 ´                                ´
Attenguillo; 23-tributary to Rıo de la Pola; 24-Rıo de la Pola at Highway 90; 25-Rıo Potrero Grande. Validation
                                                            ´             ´                           ´
sites (a.–f.) not labeled include: a.-Arroyo Llanitos; b.-Rıo Tuxpan; c.-Rıo Cuzalapa at Patitas; f.-Rıo de la Pola
at Guachinanguillo. SMBR                           ´
                                 Sierra de Manantlan Biosphere Reserve.

lutions. Effluent from sugarcane processing con-            ecological indicators of stream integrity to help
tributes 40% of all the organic waste discharged           improve the environment and, ultimately, the
to streams (SEMARNAP 1999). Municipalities                 quality of human life.
of 80,000 people release raw sewage directly                  Aquatic macroinvertebrate assemblages pro-
to streams in the area. Common practices of ba-            vide an integrative measure of water chemistry
sin deforestation, construction, stream dredg-             and physical stream conditions (Cummins 1974,
ing, and cattle grazing in riparian areas are also         Vannote et al. 1980, Rosenberg and Resh 1993)
major sources of nonpoint pollution (Lyons et              to indicate the overall health of the system
al. 1995).                                                 (Meyer 1997, Karr 1999). The index of biotic in-
   Basin degradation results in conflicts among             tegrity (IBI) framework uses biota to provide
communities along streams, and between com-                scientifically defensible evidence of environmen-
munities and industry. Basin planning commis-              tal condition (Karr 1981, Plafkin et al. 1989, Ker-
sions were formed in 1997 but they have been               ans and Karr 1994, Barbour et al. 1996). In turn,
unsuccessful in bringing about change, largely             IBIs can be used to develop biological criteria
owing to their reliance upon circumstantial ev-            for stream protection (Yoder 1995, Barbour et al.
idence to support claims of stream degradation             2000). Several quantifiable attributes of the biotic
(LMM, personal observation). Scientists and ba-            assemblage (termed ‘‘metrics’’) that assess mac-
sin planning commissions need science-based                roinvertebrate assemblage structure, composi-
688                                     B. M. WEIGEL     ET AL.                            [Volume 21

tion, and function comprise the IBI (Fore et al.      passed the type and degree of environmental
1996, Karr and Chu 1999).                             pressures within the region (Lyons et al. 1998,
   Our goal was to develop a macroinvertebrate-                                 ´
                                                      SEMARNAP 1999, Martınez et al. 2000). Our
based IBI for the protection and restoration of       environmental condition assessment was based
streams in the Sierra de Manantlan Biosphere          upon methods used to develop the fish IBI for
Reserve. We sampled streams outside the Re-           west-central Mexico (Lyons et al. 1995). We mea-
serve to better capture natural environmental         sured stream water temperature, dissolved ox-
variation in west-central Mexico and determine        ygen, conductivity, turbidity, and estimated
how biota responds to a broader gradient of hu-       substrate and substrate embeddedness imme-
man influence (Hughes et al. 1986, Gerritsen et        diately prior to macroinvertebrate sampling.
al. 2000, Hawkins et al. 2000). Consequently, we         The goal was for the assessment tool to dis-
tailored the IBI to the larger region. We hypoth-     tinguish streams with human influence from
esized that 8 macroinvertebrate metrics would         relatively uninfluenced streams regardless of in-
distinguish minimally polluted from severely          herent environmental differences. We sampled
polluted streams. These metrics were selected a       outside the reserve in an attempt to understand
priori to increase objectivity (Norris and Haw-       a larger range of intensity for the different types
kins 2000). The IBI can be considered successful      of human activities. By increasing our sampling
if it corresponds with our assessment of human        area we also increased the likelihood of encoun-
influence, a documented spatial pattern of deg-        tering greater environmental variation. In at-
radation, and the existing fish-based IBI tailored     tempting to control for some inherent sources of
to the region (Lyons et al. 1995).                    variation, we limited our sites to permanent,
                                                      subtropical streams with a moderate gradient.
                    Methods                           We estimated basin area upstream of each sam-
                                                      pling point and classified it as either small
   We developed a macroinvertebrate IBI for           ( 400 km2) or large ( 400 km2) to test whether
west-central Mexico based on data collected           macroinvertebrates naturally responded to dif-
February through May 1999, and February 2000.         ferent stream sizes. The 400-km2 threshold co-
Thirty-three sites on 21 streams within 6 major       incides with the fish-based IBI tailored to the
basins represented natural environmental con-         region (Lyons et al. 1995).
ditions and human-influence types and inten-              We divided the sites into a 27-site IBI develop-
sities in the area (Fig. 1). These streams origi-     ment data set and a 6-site testing or validation
nate in mountains that can be 3000 m above            data set before doing any statistical analyses (Fig.
sea level and have subtropical to temperate cli-      1). We grouped our development sites to deter-
mates receiving 500 to 1600 mm of rainfall an-        mine effects of basin area on metrics, and test
nually (Martınez et al. 1993, Lyons et al. 1998).     whether metrics could distinguish among mini-
Streams experience subtropical conditions on          mum-, moderate-, and severe-influence sites.
high plateaus before flowing down to the trop-            We assigned each site to 1 of 3 possible
ical Pacific coast. The entire region has a distinct   stream condition categories, based on environ-
dry season from October to May during which           mental condition scores, physicochemical as-
very little or no rain occurs. Humans rely heavi-     sessment, and professional judgement: mini-
ly upon rivers for irrigation, subsistence fishing,    mum, moderate, or severe (Table 2, Fig. 1). For
sewage dispersion, and other services through-        the development data set, the minimum impair-
out the area.                                         ment category consisted of sites having the
                                                      highest environmental condition scores. Five
Assessment of environmental condition                 large and 4 small sites had an environmental
                                                      condition score 4. We included Arroyo Las
   Quantifying environmental conditions inde-         Marias, a small-basin site with an environmen-
pendent of the biota is imperative to IBI devel-      tal condition score    3.5, in the minimum cat-
opment (Karr and Chu 1999). Sites were scored         egory. Arroyo Las Marias appeared to be a min-
0, 0.5, or 1 in each of 5 stressor or habitat cate-   imal influence reference site, and physicochem-
gories: point-source pollution, nonpoint-source       ical analysis suggested it was similar to other
pollution, riparian quality, substrate, and water     minimum influence sites (Table 2). The stream
clarity (Table 1). These 5 categories encom-          had a lower environmental condition score than
2002]                    MACROINVERTEBRATE IBI          FOR   MEXICAN     STREAMS                        689

   TABLE 1. Assessment of environmental condition and scoring criteria. Sites received a score of 0, 0.5, or 1
(severe, moderate, and minimum ratings, respectively) for each stressor type. Summing values among the 5
stressor measures assigned sites an environmental condition score.

Stressor type
  Impairment rating                                 Habitat or water-quality attributes
Point-source pollution
  Severe                      Industrial wastewater or municipal effluent from moderate-size ( 100 people)
                                community within 30 river km (rk) upstream
 Moderate                     Some industrial wastewater or municipal effluent from small community with-
                                in 30 rk or major effluent 30 rk upstream
 Minimum                      No industrial wastewater or municipal sewage effluent within 30 rk upstream
Nonpoint-source pollution
 Severe                       Channel dredging, construction, 50% stream water diverted for irrigation, or
                                irrigation return water 30 rk upstream
 Moderate                     Basin is primarily in row crop agriculture, forests are mostly grazed or har-
                                vested for timber, 50% stream water diverted for irrigation, irrigation re-
                                turn water 30 rk upstream
 Minimum                      Basin is primarily forest, little or distant irrigation influence
Riparian quality
 Severe                       Dredge spoils, livestock grazing associated with 50% of bank length erod-
                                ing, stressors immediately adjacent to stream within 100 m upstream
 Moderate                     Severe stressors as listed but 100 m upstream or 50% of bank length erod-
                                ing, evidence of grazing but primarily 10 m perpendicular to stream
 Minimum                      Vegetation primarily intact, no detectable effects of grazing near stream within
                                100 m upstream
 Severe                       Sampling riffle 75% coarse substrate (diameter 1 cm) and embedded
                                 25% by fine sediment
 Moderate                     Sampling riffle either 75% coarse substrate, or 25% embedded by fines
 Minimum                      Sampling riffle 75% coarse substrate and 25% embedded by fines
Water clarity
 Severe                       Turbidity 15 ntu
 Moderate                     Turbidity 5–15 ntu
 Minimum                      Turbidity 5 ntu

usual for the minimum category because of non-          years; we used the 1999 sample for IBI valida-
point agricultural pollution and riparian stress-       tion and the 2000 sample for IBI development.
ors distant upstream. Severe influence condi-
tions were represented by 2 large and 2 small
                                                        Macroinvertebrate collection and metric selection
stream sites having chronic point-source pollu-
tion and an environmental condition score 1.5.             A D-frame net (0.3 m wide with 600- m
The remaining 13 sites characterized the mod-           mesh) was used to collect a semiquantitative
erate influence category (Table 2).                      sample from a riffle at each site. We kick-sam-
   The validation data set included a severe in-        pled an area of 0.5 m2 for 3 min and estimated
fluence site and 2 minimum influence sites. The           if we had at least 250 individuals. If our estimate
other 3 sites in the validation data set had mod-       was 250 individuals, we repeated the process
erate levels of nonpoint-source pollution and           up to 4 times. The catch per unit effort (CPUE)
each had additional sites within the basin used         metric represented the number of sampling at-
for IBI development (Fig. 1). Rıo Ayuquila at           tempts it took to acquire 250 individuals and
Zenzontla was the only site sampled in both             was, therefore, a surrogate measure of abun-
690                                      B. M. WEIGEL     ET AL.                            [Volume 21

 TABLE 2. Mean (and range) for selected stream water and habitat measurements at severe-, moderate-, and
minimum-influence sites.

 Human                     Temperature           oxygen             Conductivity           Turbidity
influence         n            ( C)               (mg/L)               ( S/cm)                (ntu)
Severe           4       23.3 (20.7–25.8)      4.6 (3.5–7.2)       804 (475–1400)        22.7 (4.0–41.4)
Moderate        13       21.9 (11.0–29.7)      7.5 (5.2–10.2)      307 (117–645)          5.2 (1.6–7.8)
Minimum         10       22.6 (18.0–28.0)      7.7 (6.9–8.4)       307 (88–966)           2.9 (1.2–5.5)

dance (Table 3). Because chironomids were             age effluent as well as nonpoint agricultural in-
abundant, CPUE at severe influence sites was           puts. Organic pollution tolerance values were
comparable to minimum influence sites, making          primarily based on Hilsenhoff (1987) where 0
it necessary to recalculate CPUE after excluding      least tolerant and 10       most tolerant. Wide-
chironomids.                                          spread dredging, construction, grazing, and
   Each sample was placed in a pan in the field        row crop agriculture suggested sedimentation
and 250 individuals were subsampled random-           could stress our streams (Weigel et al. 2000). We
ly. We scanned for rare taxa and then picked          expected % depositional individuals (Depo%I)
macroinvertebrates systematically starting at         to respond to high sediment inputs in this re-
one end of the pan to reduce the bias toward          gion having predominantly cobble–gravel
collecting large individuals (Vinson and Haw-         streams. Taxa listed as inhabiting depositional
kins 1996).                                           zones by Merritt and Cummins (1996) were con-
   After assessing ecological stressors among the     sidered ‘‘depositional’’. Studies relating macro-
streams and before identifying the macroinver-        invertebrate feeding ecology with environmen-
tebrate samples, we selected 8 metrics related to     tal condition suggested functional-feeding met-
composition, structure, and function to repre-        rics can detect stressors from multiple spatial
sent the macroinvertebrate assemblage (Table 3).      scales (Cummins 1974, Kerans and Karr 1994,
Kerans and Karr (1994) found that macroinver-         Wallace and Webster 1996). We chose % preda-
tebrate abundance reflected environmental con-         tor individuals (Pred%I) because they may be
dition. Experience also suggested that substan-       less likely to respond to stream-size differences
tially more effort was needed to collect a mini-      (Karr and Chu 1999), and % gatherer genera
mum number of individuals at severe influence          (Gath%G) because they are typically the most
sites. We used CPUE in lieu of more expensive         abundant functional-feeding group found in
efforts (e.g., quantitative sampling) required to     streams (Wallace and Webster 1996). These tro-
estimate abundance. CPUE is an important              phic-function metrics together allowed us to
component in fish IBIs (Lyons et al. 1995, Karr        characterize both herbivorous and carnivorous
and Chu 1999). We selected generic richness           forms of feeding. Trophic function, primarily
(GR) and % Ephemeroptera–Plecoptera–Tri-              based on feeding morphology, was assigned ac-
choptera genera (EPT%G) because they are uni-         cording to Merritt and Cummins (1996). We
versally used macroinvertebrate metrics. These        used % genera and % individuals for feeding-
metrics are easily calculated, stable among min-      ecology metrics in an attempt to buffer IBI val-
imum influence sites, and track water-quality          ues from numerical anomalies associated with
changes effectively (Wallace et al. 1996, Karr        extreme values in either GR or abundance.
and Chu 1999). Henne (1997) reported that ei-
ther no macroinvertebrates or exclusively chi-        Statistical analyses in IBI development
ronomids were found at sites stressed by point-
source pollution, so we included % chironomid            Metric responses to basin size and stress. Using
individuals (Midge%I) to indicate stress. The         the development data set, we examined varia-
Hilsenhoff Biotic Index (HBI), designed to de-        tion in metric values as a function of basin area,
tect organic pollution (Hilsenhoff 1987), was ex-     and determined if metrics responded along a
pected to reflect point-source pollution from          gradient of human influence. Metrics may re-
sugarcane processing and municipal raw sew-           spond similarly to human disturbance and nat-
2002]                   MACROINVERTEBRATE IBI          FOR   MEXICAN     STREAMS                        691

               TABLE 2. Extended.                      fluence sites. Summation of the metric scores
                                                       constituted the final IBI value.

 Embedded-                                             Statistical analyses in IBI validation
  ness (%)              Major stressors
                                                          Final IBI values and biological responses. A 2-
  25 (25)       Organic point-source discharge         way ANOVA indicated if the final IBI values
  16 (0–50)     Irrigation, grazing, construction      were different by environmental condition cat-
   5 (0–25)     Distant nonpoint pollution
                                                       egories and basin area. If the interaction term
                                                       was not significant, the model was rerun with-
                                                       out the term. In turn, if the basin area term was
ural environmental variation. We determined            not significant, we removed it and reran the
metric responses to basin area and stress using        model. Using the development data set, Pear-
a 2-way analysis of variance (ANOVA), with ba-         son’s correlation analysis determined how well
sin area and site condition category as main ef-       the macroinvertebrate IBI values could predict
fects and a basin area     site condition interac-     environmental condition values. Pearson’s prod-
tion term. For metrics that showed no significant       uct–moment correlation is the standard para-
effect of basin area, we ran an ANOVA with site        metric correlation analysis that measures the
condition category as the main effect. A Bonfer-       strength of the linear relationship between 2
roni multiple range test indicated differences         variables. We developed biocriteria classifica-
among condition categories for each metric.            tions and provided narrative descriptions of bi-
   Metric scoring criteria. Summary statistics,        otic and environmental conditions for categories
including the mean, median, and interquartile          of IBI values. The IBI categories were interpre-
and total ranges, were calculated for each site        tations of breaks among data in the IBI versus
condition category (i.e., minimum, moderate,           environmental condition plot. Component mac-
and severe) to determine scoring criteria. Nat-        roinvertebrate metric and environmental condi-
ural breaks and interquartile ranges among the         tion scores influenced IBI categories.
groups dictated scoring criteria for each metric.         Validation data set. We calculated IBI values
In general, we assigned 10 points to minimum,          for the 6-site validation data set. For each site,
5 points to moderate, and 0 points to severe in-       we compared the IBI value against the environ-

   TABLE 3. Macroinvertebrate metric definitions and their expected response to increasing perturbation.

              Metric                                         Definition                            response
 Taxa richness and composition
 Catch per unit effort (CPUE)        No. of sampling episodes needed to collect 250 individu-    Increase
 Generic richness (GR)               Total number of taxa                                        Decrease
 % Ephemeroptera–Plecoptera–
  Trichoptera genera (EPT%G)         % of genera from mayfly, stonefly, and caddisfly orders        Decrease
 % Chironomidae individuals
   (Midge%I)                         Relative abundance of midges                                Increase
 Hilsenhoff Biotic Index (HBI)       Organic pollution tolerance; nxvx/N where nx       count    Increase
                                       of taxon x, vx   tolerance value of taxon x, and N
                                       count of individuals (0–10; 10    tolerant)
 % depositional individuals
  (Depo%I)                           Inhabitants of fine depositional substrate                   Increase
 Feeding morphology
 % predator individuals (Pred%I)     Carnivores; engulf or pierce prey                           Decrease
 % gatherer genera (Gath%G)          Collect deposited fine organic material                      Increase
692                                       B. M. WEIGEL   ET AL.                           [Volume 21

mental condition score to indicate how well the                                       ´
                                                      Two severe influence sites on Rıo Ayuquila with
IBI could predict human influence on a new set         large basins, downstream of sugarcane process-
of sites.                                             ing discharge (Fig. 1, site 12) and municipal
   Correspondence along Rıo Ayuquila. We deter-       sewage (Fig. 1, site 11), caused the interaction
mined whether macroinvertebrate IBI values            term to be significant in the EPT%G, Midge%I,
from 7 sites on Rıo Ayuquila corresponded to a        Gath%G, CPUE, and GR models. The sites also
documented longitudinal pattern of human in-          caused a significant basin area term (no signif-
fluence (Lyons et al. 1995). Six of these sites        icant interaction term) in the Pred%I model.
were in the development data set and 1 site was
in the validation data set. Lyons et al. (1995) and
Mercado et al. (2002) provided fish IBI values         Metric scoring criteria
for comparison. Pearson’s correlation analysis of
the macroinvertebrate and fish IBI values char-           Summary statistics for macroinvertebrate
acterized how well the macroinvertebrate IBI          metrics identified natural breaks and interquar-
corresponded to the gradient of environmental         tile ranges among sites grouped by environ-
conditions along Rıo Ayuquila.                        mental condition (Fig. 2). We established IBI
   All analyses were run using SAS statistical        scoring criteria for each metric based upon the
software (SAS Institute, version 8.0, Cary, North     summary statistics (Table 4). Scoring criteria de-
Carolina) and tests were considered significant        pended upon basin area for GR and EPT%G
at the p    0.05 level. Before running any anal-      metrics. Basin size scoring criteria were not war-
yses, we tested all macroinvertebrate metrics for     ranted for the other metrics because the metrics
normality and used biplots of the residuals and       primarily responded to differences in the level
predicted values to assess homogeneity of vari-       of stress within the severe category.
ance. Arcsin-square root transformations nor-
malized metrics calculated as proportions (e.g.,
EPT%G). Generic richness and HBI scores were          Responses of final IBI values
square-root transformed.
                                                         The 2-way ANOVA used to test for differenc-
                      Results                         es in final IBI values by environmental condition
                                                      category and basin area found no evidence of a
Macroinvertebrate assemblage characteristics          basin area site condition interaction (F 0.12,
  We collected and identified 79 macroinverte-         p     0.883). The final ANOVA model with site
brate taxa. Several taxa were ubiquitous among        condition as the main effect indicated the IBI
moderate- to minimum-influence sites, especial-        values were significantly different among con-
ly genera in the families Corydalidae, Elmidae,       dition categories (F      30.40, p   0.001). After
Baetidae, Leptophlebiidae, Tricorythidae, and         Bonferroni adjustments for multiple compari-
Hydropsychidae. The only plecopteran collected        sons, the moderate and minimum categories
was a perlid (Anacroneuria sp.), and typically        were not statistically different from one another
only 1 to 3 individuals were found at a site. Chi-    yet were both different from the severe category.
ronomids were the only taxon at sites with se-        Boxplots for the final IBI values of the develop-
vere point-source pollution.                          ment data set are shown by basin size and con-
                                                      dition category. Final values were not different
                                                      by basin area, and the trend suggested differ-
Effects of basin area and metric response to stress
                                                      ences among the minimum and moderate cate-
  The 2-way ANOVAs indicated significant ba-           gories (Fig. 3). A check of IBI performance
sin area effects on 6 of the 8 metrics, and that      showed that IBI values were correlated highly
each metric was able to distinguish among site        with environmental condition assessment scores
conditions (Fig. 2). Metrics distinguished mini-      from the development data set (Pearson’s r
mum and moderate sites from severe sites,             0.86, p     0.001) (Fig. 4A). The sites covered a
whereas no metrics distinguished between min-         broad range of environmental conditions but
imum and moderate sites after Bonferroni ad-          more sites represented the minimum influence
justments for multiple comparisons (Fig. 2).          end of the scale.
2002]                    MACROINVERTEBRATE IBI        FOR   MEXICAN    STREAMS                      693

Biological responses to environmental stressors       sites with similarly high IBI values. An IBI value
                                                         65 was higher than expected for Rıo Ayuquila
   We interpreted biological responses to human       below the irrigation return, a site with an en-
influence, establishing narrative descriptions for     vironmental condition 2.0.
multiple ranges of IBI values using physico-
chemical data and final IBI and environmental
condition values for each site (Table 5). The most                          ´
                                                      Correspondence along Rıo Ayuquila
obvious biological response was the association         The macroinvertebrate IBI accurately corre-
of point-source pollution with either no macro-       sponded with a documented spatial pattern of
invertebrates or high abundance and dominance                                            ´
                                                      environmental condition in the Rıo Ayuquila
by chironomids at ‘‘very poor’’ sites. Blood-red      basin. The macroinvertebrate and fish IBI values
chironomids from the subfamily Chironominae           were correlated strongly among the 7 sites
were the only organisms found at Rıo Ayuquila         (Pearson’s r  0.87, p   0.001) (Fig. 4B).
sites 11 and 12 (Fig. 1), below sugarcane pro-
cessing and municipal sewage discharges. Or-
ganic pollution from a large feedlot was asso-                           Discussion
ciated with Arroyo Tecolote from which only 41
                                                      Determining human influence from inherent
individuals were collected in 4 CPUEs. HBI and
                                                      environmental condition
metrics calculated as proportions are sensitive
to number of individuals collected, so there             An IBI must distinguish influences of human
were not enough individuals to calculate an IBI       activities from natural environmental variation
at Arroyo Tecolote.                                   (Hughes et al. 1998, Karr and Chu 1999). Stream
   Rıo Potrero Grande was the only ‘‘poor’’ site.     size was a potential nonhuman influence on
It was associated with severe cattle grazing im-      metric scores. We accounted for this variation by
mediately upstream, and it had a high Midge%I         creating separate scoring criteria for small and
(33%) and Depo%I (85%). The HBI indicated or-         large streams. Our macroinvertebrate IBI ver-
ganic pollution was a problem.                        sion may not work for different stream types or
   Nonpoint-source pollution associated with          streams outside of the sampling region because
dredging or construction and diversion for ir-        of other natural environmental variation not in-
rigation were typical influences at ‘‘fair’’ sites.    vestigated. Different stream types outside of the
The assemblage had high Depo%I and Gath%G.            region should be validated.
Moderate HBI scores indicated organic pollution          Significantly different metric scores between
was problematic but not as severe as at the very      large and small streams indicated that the met-
poor to poor sites.                                   ric score was a function of basin area. Scoring
   Slight to moderate levels of nonpoint-source       criteria can be developed to account for natural
pollution appeared to influence ‘‘good’’ sites.        environmental effects if the relationship with
Pred%I was a strong, positive indicator of en-        the metric is monotonic (Hughes et al. 1998).
vironmental condition. However, dredging and          EPT%G and GR differed between large and
bridge construction at a Rıo Ayuquila site (Fig.      small streams. At most of the large, minimum
1, site 13) was associated with predatory odo-        influence sites, EPT abundance was high and
nates. The site had a high Pred%I and otherwise       only 1 CPUE was needed to collect 250 individ-
fair conditions. Other common characteristics of      uals. Thus, high abundance and dominance
good sites were moderate GR and Depo%I.               meant that a smaller area was sampled, which
                                                      in turn may have resulted in lower GR scores
IBI values for the validation data set                (Vinson and Hawkins 1996). The size-specific
                                                      scoring criteria appears to account for basin
   The independent validation data set indicated      area effects on metric values because no signif-
that the macroinvertebrate IBI could predict hu-      icant differences in final IBI values as a result of
man influence well at 5 of the 6 sites. IBI values     basin area were observed (Fig. 3). Lyons et al.
were 0, 65, 65, 70, 70, and 70, whereas environ-      (1995) also found stream size influenced metric
mental condition scores were 0.5, 2.0, 3.5, 4.0,      scores of the fish-based IBI tailored to this re-
4.0, and 4.5, respectively. The validation data set   gion. Evidence of stream size affecting macro-
consisted of 1 site with a low IBI value and 5        invertebrate richness and composition was
694                                      B. M. WEIGEL      ET AL.                             [Volume 21

  FIG. 2. Boxplots of component macroinvertebrate index of biotic integrity (IBI) metrics by human influence
category (n   27). Analysis of variance indicated if basin area affected the metric score, and if differences
among minimum, moderate, and severe categories were detected. Metrics distinguished severe from moderate
2002]                    MACROINVERTEBRATE IBI            FOR   MEXICAN        STREAMS                       695

  TABLE 4. Scoring criteria and classification for component metrics of the macroinvertebrate index of biotic
integrity (IBI) tailored to west-central Mexican streams. Sites were scored 0, 5, or 10 points for most metrics.
A dash indicates that the fair (5) score was not represented. Component metric scores are summed to yield a
multimetric IBI value. Abbreviations for metrics as in Table 3.

        Metric                              Poor (0)                  Fair (5)                 Good (10)
CPUE                                          50                         –                        50
  Basin area   400 km2                        13                       14–22                      23
  Basin area   400 km2                        11                         –                        12
  Basin area   400 km2                        32                       32–38                      38
  Basin area   400 km2                        35                       35–55                      55
Midge%I                                       25                        5–25                       5
HBI                                            5.0                   4.25–5.0                      4.25
Depo%I                                        75                       55–75                      55
Pred%I                                         4                        4–14                      14
Gath%G                                        48                       44–48                      44

found in the development process of some IBIs             eling is not used to determine metric inclusion.
(e.g., Ohio EPA 1987, DeShon 1995) but not in             In contrast, Norris and Hawkins (2000) recom-
others (e.g., Maxted et al. 2000). Despite having         mended selecting metrics a priori based on con-
significant interaction or basin area terms,               ceptual understanding of the biota to avoid cir-
Midge%I, Gath%G, Pred%I, and CPUE did not                 cularity in IBI development and other issues
warrant scoring criteria based upon basin area            with interpreting IBI values. We believe that in
because the effects were a result of differences          our study there was a small tradeoff of insight
in stress within the severe influence category.            gained from empirical modeling for increased
                                                          objectivity by selecting metrics a priori.
IBI development                                              Sites may have too few macroinvertebrate in-
                                                          dividuals to calculate an IBI value. We recom-
   We hypothesized which biological attributes            mend a site be assigned a ‘‘very poor’’ rating
would be affected by the stressors within our             and tentatively given an IBI value         0 if 4
region, and selected metrics a priori based on            CPUEs produce 100 individuals. Other IBIs
conceptual understanding of stream macroin-               also require a standardized minimum number
vertebrate biology and empirical modeling of              of individuals to calculate an IBI value (e.g.,
the relationship between macroinvertebrates               Ohio EPA 1987), which reduces problems with
and human influence over the past 15 y (e.g.,              having a small number of a particular taxon
Vannote et al. 1980, Resh and Rosenberg 1984,             radically changing the overall IBI value. A min-
Ohio EPA 1987, Plafkin et al. 1989, Karr and              imum of 100 individuals is further justified be-
Chu 1999). Karr and Chu (1999) recommended                cause Hilsenhoff (1987) required 100 individ-
selecting metrics that respond systematically to          uals to calculate his index. Larsen and Herlihy
stressors along a gradient of increasing intensi-         (1998) discussed the importance of standardi-
ty. IBIs tailored to regions in which the biota has       zation and how taxa richness estimates changed
unique or unusual responses to disturbances               as the number of individuals sampled using a
may miss critical information if empirical mod-           fixed-count protocol changed. Extremely low


and minimum categories, but moderate and minimum were not statistically different after Bonferroni adjust-
ments for multiple comparisons. For metrics with a significant basin area category interaction, we included
a small basin and large basin box respectively per category. Box represents the 25th and 75th percentiles, thick
line within a box represents the median, whiskers represent the 5th and 95th percentiles, and solid dots represent
sites 1.5 times the interquartile range from the end of the box. Abbreviations for metrics as in Table 3.
696                                        B. M. WEIGEL      ET AL.                              [Volume 21

  FIG. 3. Boxplot of macroinvertebrate index of biotic integrity (IBI) values by human influence categories and
basin area (small and large respectively) (n   27). Analysis of variance indicated if basin area affected the IBI
value, and if differences among minimum, moderate, and severe categories were detected. IBI distinguished
severe from moderate and minimum categories, but moderate and minimum were not statistically different
after Bonferroni adjustments for multiple comparisons. There was no significant basin area or basin area
category interaction effect on IBI values. Boxplot interpretation as in Fig. 2. The solid dot represents Arroyo
Tecolote at 1.5 times the interquartile range from the end of the box.

macroinvertebrate abundance in west-central               HBI scores, especially at very poor sites with se-
Mexican streams is an indication of severe en-            vere organic pollution where they were the pri-
vironmental impairment.                                   mary, if not only, taxon present. Hemoglobin en-
   Response to human activity type and intensity.         ables Chironominae to thrive despite low dis-
The component metric scoring criteria repre-              solved oxygen levels associated with organic en-
sented our best professional interpretation of            richment. Riparian alterations, sedimentation, and
summary statistics (Table 4). We took a conser-           turbidity typically influenced the poor to fair sites,
vative approach in developing criteria for GR             which corresponded to the moderate influence
and CPUE. Some minimum influence sites on                  category in Table 2. The increase in Depo%I at
large streams naturally had a low GR (12 taxa),           poor to fair sites was likely a response to an in-
whereas severely influenced sites on small                 crease in sand and silt. Increased turbidity at these
streams tended to have up to 13 taxa (Fig. 2).            sites was associated with increased runoff and
Streams with severe influence having a basin               sediment disruption (e.g., cattle grazing, construc-
area slightly larger than the 400-km2 threshold           tion, irrigation water returns). Gatherers are gen-
may have a GR that approaches the range for               eralists that thrive in depositional zones having an
minimum influence sites. Therefore, we only                abundance of fine particulate organic matter (Wal-
have scoring criteria that distinguish minimum            lace and Webster 1996). It is reasonable to expect
and severe sites for the GR metric for large              changes in riparian vegetation, and an increase in
streams. Our data and experience indicated that           sedimentation and suspended inorganic solids to
CPUE was extremely low at severe influence                 hamper the filter feeders, scrapers, and shredders
sites but it was not an issue between moderate            more than gatherers, in turn increasing Gath%G.
and minimum sites. As a result, we only des-                 We designed the study to include a range of
ignated scoring criteria for severe and minimum           human influence types and intensities but the
categories of catch effort (Table 4).                     results revealed many good to excellent and
   Certain macroinvertebrate metric ranges or bi-         very poor sites in contrast to few poor to fair
ological attributes corresponded with final IBI cat-       sites (Fig. 4A). Similarly, some metrics displayed
egories (Table 5). Chironomids strongly influenced         strong overlap between moderate and mini-
2002]                   MACROINVERTEBRATE IBI           FOR   MEXICAN    STREAMS                        697

  FIG. 4. Relationship between macroinvertebrate index of biotic integrity (IBI) and (A) environmental con-
dition in the development data set (n                               ´
                                        27), and (B) fish IBI along Rıo Ayuquila (n    7). Triangles represent
3 sites and diamonds represent 2 sites with the same values.

mum influence sites (Fig. 2). It is not surprising       ment, and the complementary nature of multi-
that the range for moderate influence sites was          metric indexes, a primary strength of the IBI ap-
large. Slightly stressed sites tend to have some-       proach (Karr and Chu 1999).
what lower scores in 1 or 2 metrics, whereas
minimum influence sites consistently have op-            IBI validation
timal scores for each metric. This result exem-
plifies the importance of relying upon several             The IBI may be more likely to indicate human
attributes of the assemblage to indicate impair-        influence than the environmental condition as-
698                                       B. M. WEIGEL       ET AL.                              [Volume 21

 TABLE 5. Guidelines for interpreting macroinvertebrate index of biotic integrity (IBI) values for west-central
Mexican streams. Abbreviations for metrics as in Table 3.

Value and
  rating                               Biological response to environmental condition
75–80             Comparable to the minimum influence systems in the region. Macroinvertebrates are abun-
Very good           dant. GR and EPT%G are near the maximum for the size of stream. Chironomids are
                    absent or Midge%I is very low. Low HBI scores and Depo%I indicate organic pollution
                    and sedimentation are undetectable. High Pred%I and low Gath%G indicate a predomi-
                    nance of specialized feeders. Nonpoint-source pollution is minimum or distant from the
60–70             Generally low levels of nonpoint-source pollution have influenced some aspects of the ma-
Good                croinvertebrate assemblage. GR an EPT%G are usually high but may not be at the maxi-
                    mum for the size of the stream. Often Midge%I, Depo%I, and Gath%G increase. Some
                    organic pollution may be reflected in the HBI scores.
50–55             Point-source pollution may be present distant from the site or in low quantities. Moderate
Fair                to severe nonpoint-source pollution or diversion for irrigation are typically the major
                    stressors. GR and EPT%G are moderate to low for the size of stream. Odonates and oth-
                    er depositional taxa become more prevalent. HBI scores and Midge%I typically run mod-
                    erate to high.
25–45             Point-source pollution is generally present but is intermittent or not immediately at the site.
Poor                Nonpoint-source pollution can be severe. Abundance is typically low but, if high,
                    Midge%I is high. HBI scores and Depo%I suggest very tolerant assemblages.
0–20              System is severely degraded by point-source pollution. Zero values can result from all wa-
Very poor           ter being diverted for irrigation or collecting 100 individuals in 4 CPUEs. Chironomids
                    are typically the only macroinvertebrate fauna present and are usually abundant. If other
                    taxa are present, their abundance is low.

sessment. Macroinvertebrate monitoring inte-             strate. We may have introduced sampling bias
grates disturbances throughout the macroinver-           by only collecting macroinvertebrates at the
tebrates’ life cycles (Rosenberg and Resh 1993),         rocky margin. Additional sites affected by irri-
whereas environmental condition assessment is            gation return water should be assessed and, if
often based on point-in-time estimates. Our en-          necessary, the IBI modified to detect this influ-
vironmental assessment was inherently more               ence. Validation with additional sites in the fair
subjective and qualitative than the bioassess-           category should assist in assessing IBI perfor-
ment. Nonpoint-source pollution, for example, is         mance.
extremely difficult to quantify and we may have              Macroinvertebrates tend to be sensitive to cer-
underestimated episodic inputs. Despite the              tain types of human activities, whereas fish are
problem of independently assessing environ-              sensitive to others, suggesting application of
mental condition, the macroinvertebrate IBI was          both macroinvertebrate- and fish-based indexes
correlated with human influence (Fig. 4A).                improves ecosystem assessment (Plafkin et al.
  The validation data set indicated that the IBI         1989, Lyons et al. 1995). The different IBIs
could predict human influence at 5 of 6 sites.            showed a relative response consistent with the
The Rıo Ayuquila site downstream of the irri-            pattern of known human influence (Fig. 4B).
gation water return (Fig. 1, site e.) had a good         Differences in fish and macroinvertebrate IBI
IBI value of 65 despite having poor substrate,           values may reflect different sampling dates that
high turbidity (13.6 ntu), and an environmental          corresponded with episodic stressors.
condition score      2.0. We sampled a turbulent            In conclusion, the macroinvertebrate- and
area along the stream margin because the re-             fish-based IBIs used together provided an as-
mainder of the channel was exclusively sand              sessment framework for protection of stream in-
and our sampling protocol required larger sub-           tegrity in west-central Mexico. Natural-resource
2002]                    MACROINVERTEBRATE IBI            FOR   MEXICAN      STREAMS                          699

scientists in this area have long understood the          DESHON, J. E. 1995. Development and application of
influence of point-source inputs on stream water               the invertebrate community index (ICI). Pages
quality but have had difficulty curbing the pol-               217–244 in W. S. Davis and T. P. Simon (editors).
lution because of a lack of scientific evidence of             Biological assessment and criteria. Tools for water
                                                              resource planning and decision making. Lewis
the influence of perturbation on biota (LMM,
                                                              Publishers, Boca Raton, Florida.
personal observation). The bioassessment tools
                                                          FORE, L. S., J. R. KARR, AND R. W. WISSEMAN. 1996.
developed in our study clearly indicate sugar-                Assessing invertebrate responses to human activ-
cane processing plants and municipalities were                ities: evaluating alternative approaches. Journal of
key contributors to serious declines of environ-              the North American Benthological Society 15:
mental condition. The assessment tools devel-                 212–231.
oped also give managers scientifically defensi-            GERRITSEN, J., M. T. BARBOUR, AND K. KING. 2000. Ap-
ble rationales for reducing stream inputs from                ples, oranges, and ecoregions: on determining
nonpoint sources.                                             pattern in aquatic assemblages. Journal of the
                                                              North American Benthological Society 19:487–
               Acknowledgements                           HAWKINS, C. P., R. H. NORRIS, J. GERRITSEN, R. M.
                                                              HUGHES, S. K. JACKSON, R. K. JOHNSON, AND R. J.
   This research was funded in part by Instituto
                                                              STEVENSON. 2000. Evaluation of the use of land-
         ´             ´              ´
Manantlan de Ecologıa y Conservacion de la
                                                              scape classifications for the prediction of fresh-
Biodiversidad, Universidad de Guadalajara. Ad-                water biota: synthesis and recommendations.
ditional support was from University of Wis-                  Journal of the North American Benthological So-
consin through Latin American, Caribbean, and                 ciety 19:541–556.
Iberian Studies program’s NAVE research grant,            HENNE, L. J. 1997. Development of a community-
Department of Zoology by a J. J. Davis travel                 based biological monitoring program for the
grant, and Departments of Rural Sociology and                 Ayuquila River, Jalisco, Mexico: a preliminary
Zoology by the Integrated Graduate Education                  study. Masters Thesis, Department of Urban and
and Research Training program. We thank Ar-                   Regional Planning, University of Illinois, Urbana,
turro Carranza-Montana, Norman Mercado-Sil-                   Illinois.
va, and Luis Ignacio Inıquez-Davalos for field
                         ´       ´                        HILSENHOFF, W. L. 1987. An improved biotic index of
                                                              organic stream pollution. Great Lakes Entomolo-
assistance, Jeff Dimick for taxonomy assurance,
                                                              gist 20:31–39.
and Paul Rasmussen for editorial assistance.
                                                          HUGHES, R. M., P. R. KAUFMANN, A. T. HERLIHY, T. M.
John Lyons and Stanley Dodson provided tech-                  KINCAID, L. REYNOLDS, AND D. P. LARSEN. 1998.
nical advice throughout the study. Emily Stan-                A process for developing and evaluating indices
ley, Monica Turner, John Magnuson, Ken Potter,                of fish assemblage integrity. Canadian Journal of
and Martin Griffith improved early drafts. We                  Fisheries and Aquatic Sciences 55:1618–1631.
sincerely appreciate the advice from Will Cle-            HUGHES, R. M., D. P. LARSEN, AND J. M. OMERNIK.
ments, Leska Fore, David Rosenberg, and an                    1986. Regional reference sites: a method for as-
anonymous reviewer, which greatly improved                    sessing stream potentials. Environmental Man-
the manuscript.                                               agement 10:629–635.
                                                          KARR, J. R. 1981. Assessment of biotic integrity using
                                                              fish communities. Fisheries 6(6):21–27.
                  Literature Cited                        KARR, J. R. 1999. Defining and measuring river health.
BARBOUR, M. T., J. GERRITSEN, G. E. GRIFFITH, R. FRY-         Freshwater Biology 41:221–234.
                                                          KARR, J. R., AND E. W. CHU. 1999. Restoring life in
   BASTIAN. 1996. A framework for biological criteria         running waters: better biological monitoring. Is-
   for Florida streams using benthic macroinverte-            land Press, Washington, DC.
   brates. Journal of the North American Benthol-         KERANS, B. L., AND J. R. KARR. 1994. A benthic index
   ogical Society 15:185–211.                                 of biotic integrity (B-IBI) for rivers of the Tennes-
BARBOUR, M. T., W. F. SWIETLIK, S. K. JACKSON, D. L.          see Valley. Ecological Applications 4:768–785.
   COURTEMANCH, S. P. DAVIES, AND C. O. YODER.            LARSEN, D. P., AND A. T. HERLIHY. 1998. The dilemma
   2000. Measuring the attainment of biological in-           of sampling streams for macroinvertebrate rich-
   tegrity in the USA: a critical element of ecological       ness. Journal of the North American Benthologi-
   integrity. Hydrobiologia 422/423:453–464.                  cal Society 17:359–366.
CUMMINS, K. W. 1974. Structure and function of            LYONS, J., G. GONZALEZ-H., E. SOTO-G., AND M.
   stream ecosystems. BioScience 24:631–641.                  GUZMAN-A. 1998. Decline of freshwater fishes
700                                          B. M. WEIGEL      ET AL.                               [Volume 21

    and fisheries in selected drainages of west-central          GROSS, AND R. M. HUGHES. 1989. Rapid bioas-
    Mexico. Fisheries 23(4):10–18.                              sessment protocols for use in streams and rivers:
LYONS, J., AND S. NAVARRO-P. 1990. Fishes of the Si-            benthic macroinvertebrates and fish. EPA/440/4-
    erra de Manantlan, west-central Mexico. South-              89-001. Assessment and Water Protection Divi-
    western Naturalist 35:32–46.                                sion, US Environmental Protection Agency, Wash-
LYONS, J., S. NAVARRO-P., P. A. COCHRAN, E. SANTA-              ington, DC.
    NA-C., AND M. GUZMAN-A. 1995. Index of biotic
                             ´                              RESH, V. H., AND D. M. ROSENBERG (EDITORS). 1984.
    integrity based on fish assemblages for the con-             The ecology of aquatic insects. Praeger Scientific
    servation of streams and rivers in west-central             Publishers, New York.
    Mexico. Conservation Biology 9:569–584.                 ROSENBERG, D. M., AND V. H. RESH (EDITORS). 1993.
      ´                                                         Freshwater biomonitoring and benthic macroin-
    GARCIA. 2000. Aquatic ecosystem pollution of the
           ´                                                    vertebrates. Chapman and Hall, New York.
    Ayuquila River, Biosphere Reserve Sierra de             SANTANA-C., E., S. NAVARRO-P., L. M. MARTINEZ-R.,´
    Manantlan. Pages 165–181 in M. Munawar, S.                  A. AGUIRE-G., P. FIGUEROA-B., AND C. C. AGUI-
    Lawrence, I. F. Munawar, and D. Malley (editors).                                        ´
                                                                LAR-G. 1993. Contaminacion, aprovechamiento y
    Aquatic ecosystems of Mexico: status and scope.                         ´                         ´
                                                                conservacion de los recursos acuaticos del Rıo    ´
    Ecovision World Monograph Series, Backhuys                  Ayuquila, Reserva de la Biosfera Sierra de Man-
    Publishers, Leiden, The Netherlands.                             ´
                                                                antlan, Jalisco-Colima. Tiempos de Ciencias (Gua-
      ´                                                         dalajara) 30:29–38.
    VARA-G. 1993. Climas de la Reserva de Biosfera          SEMARNAP (SECRETARIA DEL MEDIO AMBIENTE RE-
    Sierra de Manantlan y su region de influencia.               CURSOS NATURALES Y PESCA). 1999. Estadisticas
    Agrociencia 2(4):1–10.                                      del Medio Ambiente. Instituto Nacional de Estad-
MAXTED, J. R., M. T. BARBOUR, J. GERRITSEN, V. PORET-           ı               ´
                                                                ´stica, Geografıa e Informatica y la Secretaria de
    TI, N. PRIMROSE, A. SILVIA, D. PENROSE, AND R.              Medio Ambiente, Recursos Naturales y Pesca,
    RENFROW. 2000. Assessment framework for mid-                                ´
                                                                Mexico DF., Mexico.
    Atlantic coastal plain streams using benthic mac-       VANNOTE, R. L., G. W. MINSHALL, K. W. CUMMINS, J.
    roinvertebrates. Journal of the North American              R. SEDELL, AND C. E. CUSHING. 1980. The river
    Benthological Society 19:128–144.                           continuum concept. Canadian Journal of Fisheries
MERCADO-S., N., J. LYONS, G. SALGADO-M., AND M.                 and Aquatic Sciences 37:130–137.
    MEDINA-N. 2002. Validation of a fish-based index         VINSON, M. R., AND C. P. HAWKINS. 1996. Effects of
    of biotic integrity for streams and rivers of central       sampling area and subsampling procedure on
    Mexico. Reviews in Fish Biology and Fisheries (in           comparisons of taxa richness among streams.
    press).                                                     Journal of the North American Benthological So-
MERRITT, R. W., AND K. W. CUMMINS (EDITORS). 1996.              ciety 15:392–399.
    An introduction to the aquatic insects of North         WALLACE, J. B., J. W. GRUBAUGH, AND M. R. WHILES.
    America. 3rd edition. Kendall/Hunt Publishing,              1996. Biotic indices and stream ecosystem pro-
    Dubuque, Iowa.                                              cesses: results from an experimental study. Eco-
MEYER, J. L. 1997. Stream health: incorporating the hu-         logical Applications 6:140–151.
    man dimension to advance stream ecology. Jour-          WALLACE, J. B., AND J. R. WEBSTER. 1996. The role of
    nal of the North American Benthological Society             macroinvertebrates in stream ecosystem function.
    16:439–447.                                                 Annual Review of Entomology 41:115–139.
NORRIS, R. H., AND C. P. HAWKINS. 2000. Monitoring          WEIGEL, B. M., J. LYONS, L. K. PAINE, S. I. DODSON,
    river health. Hydrobiologia 435:5–17.                       AND D. J. UNDERSANDER. 2000. Using stream mac-
OHIO EPA (ENVIRONMENTAL PROTECTION AGENCY).                     roinvertebrates to compare riparian land use
    1987. Biological criteria for the protection of             practices on cattle farms in southwestern Wiscon-
    aquatic life: Volumes I—II. Volume I: The role of           sin. Journal of Freshwater Ecology 15:93–106.
    biological data in water quality assessment. Vol-       YODER, C. O. 1995. Policy issues and management ap-
    ume II: Users manual for biological field assess-            plications for biological criteria. Pages 327–343 in
    ment of Ohio surface waters. Surface Waters Sec-            W. S. Davis and T. P. Simon (editors). Biological
    tion, Division of Water Quality Monitoring and              assessment and criteria. Tools for water resource
    Assessment, Ohio Environmental Protection                   planning and decision making. Lewis Publishers,
    Agency, Columbus, Ohio. (Available from: Ohio               Boca Raton, Florida.
    EPA, P.O. Box 1049, Columbus, Ohio 43216–1049).                                     Received: 28 September 2001
PLAFKIN, J. L., M. T. BARBOUR, K. D. PORTER, S. K.                                           Accepted: 12 June 2002