2003
SOUTHEASTERN NATURALIST
2(3):447–468
AN INDEX OF BIOTIC INTEGRITY FOR FISH ASSEMBLAGES IN OZARK HIGHLAND STREAMS OF ARKANSAS DANIEL C. DAUWALTER1,*, EDMUND J. PERT1,2, AND WILLIAM E. KEITH3 ABSTRACT - We developed an Index of Biotic Integrity (IBI) for wadeable Ozark Highland streams in Arkansas by using fish-assemblage data from 96 stream sites. All sites were classified as reference or non-reference based on both subjective and objective information, and we used uni- and bivariate statistics to examine 39 potential IBI metrics. Seven metrics were chosen for the IBI. They were percent (of individuals) as algivorous/herbivorous, invertivorous, and piscivorous; percent with black spot or an anomaly; percent green sunfish Lepomis cyanellus, bluegill L. macrochirus, yellow bullhead Ameiurus natalis, and channel catfish Ictalurus punctatus; percent invertivorous; percent top carnivores; number of darter Etheostoma and Percina, sculpin Cottus, and madtom Noturus species; and number of lithophilic spawning species. Trophic metrics contributed most to IBI scores, and metrics were most often correlated with chloride, nutrient, land use, road density, and sedimentation levels, which suggests that our IBI should be able to successfully differentiate stream conditions in wadeable Ozark Highland streams of Arkansas.
INTRODUCTION The Water Pollution Control Act of 1948 was implemented to protect U.S. waters (Alder et al. 1993). Now known as the Clean Water Act (CWA), due to 1972 amendments, this statute mandates state and federal agencies to assess and monitor U.S. surface water conditions. Water-quality monitoring programs implemented to meet CWA criteria have been conducted by using a variety of chemical and biological methods. Biological endpoints can be advantageous, especially regarding public understanding of water-quality goals (Barbour et al. 1999). Karr (1981) developed the Index of Biotic Integrity (IBI), a multimetric index using fish-community attributes, to assess stream-site quality. The IBI was developed to provide a quick, reliable, and easily understood method to assess local stream conditions (Karr et al. 1986), descriptors not prevalent in previous water-quality monitoring programs (Ward et al. 1986). Since its development, IBIs have been modified (see 1
University of Arkansas at Pine Bluff, Department of Aquaculture/Fisheries, 1200 North University Drive, Pine Bluff, Arkansas 71601. 2Current address: California Department of Fish and Game, 1416 9th Street, 12th floor, Sacramento, California 95814;
[email protected]. 3Arkansas Department of Environmental Quality, Water Division, Planning Section, 8001 National Drive, Little Rock, Arkansas 72219;
[email protected]. *Corresponding author, current address - Oklahoma Cooperative Fish and Wildlife Research Unit, 404 Life Sciences West, Oklahoma State University, Stillwater, OK 74078;
[email protected].
448
Southeastern Naturalist
Vol. 2, No. 3
Simon and Lyons (1995) for a summary) for regional (Fausch et al. 1984, Miller et al. 1988), Mexican and European (Didier and Kestemont 1996, Lyons et al. 1995), and specific (Lyons et al. 1996, Minns et al. 1994) applications. The IBI has also been adopted by many states (e.g., OEPA 1987) to assess the status of their waterbodies in compliance with CWA section 305(b). Distinct fish assemblages occur in the state of Arkansas (Matthews and Robison 1988) and they have been shown to coincide with ecoregions (Keith 1987, Rohm et al. 1987). An IBI has been developed by using fish assemblages in Arkansas’ Ouachita Mountains ecoregion (Hlass et al. 1998). The Ouachita Mountains IBI reflected stream conditions at sites impacted by different timber harvest techniques. Although a multimetric index using macroinvertebrates has been developed for Missouri Ozark streams (Jones et al. 1981), no such index exists to monitor and assess Ozark streams in Arkansas. We chose the Ozark Highlands ecoregion in Arkansas as a regional framework to continue IBI development in the state. We used fish assemblages as our biological measure to develop the IBI because fish have several advantages as biological indicators, and fish-collection data were available through the Arkansas Department of Environmental Quality (ADEQ; formerly the Arkansas Department of Pollution Control and Ecology [ADPCE]) fish-collection database. Our goal was to use fish assemblages to develop an IBI for wadeable Ozark Highland streams in Arkansas that can successfully differentiate stream conditions. Thus, we: 1) identified non-redundant metrics that differentiated between reference and non-reference stream sites; 2) determined the percentage agreement between original reference-site classifications and IBI classifications; 3) determined relative metric contributions to IBI scores; and 4) assessed relations between IBI metrics and physicochemical and land-use variables. FIELD-SITE DESCRIPTION The Ozark Highlands ecoregion is located in north-central and north-western Arkansas, north-east Oklahoma, and south-central Missouri, U.S.A. (Omernik 1987) (Fig. 1). Mountainous terrain, steep gradients, and fractured limestone geology characterize and form the fastflowing, spring-fed streams of the region (Robison and Buchanan 1988). Recently, many Arkansas counties in the Ozark Highlands have experienced rapid human population increases. From 1990 to 1999, Benton, Carroll, Marion, Sharp, and Washington County populations increased 41.9, 20.9, 24.2, 21.1, and 29.3%, respectively (U.S. Census Bureau 2001). Ozark Highland land-use in Arkansas is 36.2% agriculture, 60.8% forested, 0.8% urban, and 2.2% water (Smith et al. 1998). Norfork, Bull Shoals, and Beaver reservoirs were formed in 1943, 1951,
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
449
and 1964 respectively (Robison and Buchanan 1988), and constitute much of the regional surface water. Agricultural practices are a major cause of regional water-quality problems. Many of these problems result from high production of animal wastes (e.g., poultry) that have the potential to contaminate surface and ground waters because of local geology (ADPCE 1996). The ichthyofauna in this ecoregion is highly diverse, consisting of at least 89 species, with cyprinids, percids, and centrarchids contributing most to relative abundance of fishes in Ozark Highland streams (Giese et al. 1987). METHODS Data Fish-assemblage data collected during base flow conditions at 96 stream sites from 1963 to 2000 were used to develop the Ozark Highlands IBI (Fig. 1); all but four collections were made after 1982. Fishes were sampled by using a one-pass backpack (or rarely a barge) electrofishing sample until field crews observed no new fish species being collected and no new habitat types sampled (Keith 1987). Although the distance sampled varied at some sites, it was estimated to be approximately 75 mean stream widths (MSWs) in length on average (Dauwalter 2002). Dauwalter (2002) used 15 of the 96 sites to determine that 51 MSWs could be sampled without jeopardizing fish species richness and relative abundance estimates. To simulate electrofishing samples equal to 51 MSWs in length, all fish collections were rarified by randomly removing 32% (i.e., (1 – [51 MSW / 75 MSW]) x 100) of the individuals from each collection. Site Classifications We classified all 96 fish-collection sites as reference or non-reference. Reference sites are generally undisturbed sites, within a relatively homogenous region, that are selected on the basis of human disturbances, stream size, stream channels, refuges, and zoogeography, and Figure 1. Location of 96 fish collection sites in the Ozark H i g h l a n d s ecoregion (black) in or near Arkansas (grey). All sites along ecoregion boundaries were considered representative of Ozark Highland streams.
450
Southeastern Naturalist
Vol. 2, No. 3
are used as benchmarks by which other regional streams are compared (Hughes et al. 1986). Classification methods incorporated: 1) reviewing locations of all point source dischargers in the ecoregion; 2) identifying sites located in “typical” ecoregion areas (Omernik 1987); 3) reviewing delineated watersheds of candidate sites for areas containing known non-point source pollutants; 4) utilizing professional judgment to identify sites known to be impaired; and 5) by conducting field reconnaissance to confirm the suitability of stream sites as representative ecoregion reference streams (Bennett et al. 1987). Our second criteria for reference classification required each site to have at least 75% forested watersheds. Exceptions to our reference-criteria rule were four fish collections from 1963. We initially classified these sites as reference sites, but appropriate temporal land-use data were not available to determine land-use percentages; we classified all four as reference sites. Contributing watersheds for each site were delineated by using a geographical information system (GIS), and we determined watershed land use by using the Arkansas Gap Analysis Program (AR-GAP) landcover data layer (Smith et al. 1998, Weih 2001). The final AR-GAP data layer included 36 land cover classes with a 100 ha resolution. We grouped the land-cover classifications into agriculture, forested, urban, and water land-use classes. We omitted 19 (~20%) of the sites from use in metric selection and scoring procedures to later validate the IBI and assess the consistency of IBI site classifications (i.e., reference versus non-reference). To ensure that the removed data set included a range of stream sizes, we stratified all sites into watershed size-groups of 0–100 km2, > 100–300 km2, and > 300 km2. We randomly removed 20% of the sites from each watershed size-group; one reference site was removed per group. Candidate IBI Metrics Candidate IBI metrics were selected from a variety of sources (see Simon and Lyons 1995). We also considered some novel metrics that seemed potentially useful for the Ozark Highlands ecoregion. Only sites that had non-zero values for most sites were considered as candidate IBI metrics. Zero values indicated that no individuals or species incorporated into a metric were commonly collected. Age-0 individuals were excluded from fish collections, and were identified by visually comparing individual fish sizes for each species. An exception was lamprey ammocoetes (Ichthyomyzon and Lampetra juveniles). Juvenile lampreys were considered as one species because of difficulties identifying them to species. They are easily identified as lamprey juveniles and are potentially important in determining stream-site conditions because they indicate the presence of a migratory adult in a stream. All observed fish species were placed into taxonomic, functional, reproductive, and trophic categories (Etnier and Starnes 1993, Giese et al. 1987, Jenkins
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
451
and Burkhead 1993, OEPA 1987, Pflieger 1997, Robison and Buchanan 1988, Smogor 1996, Smogor and Angermeier 1999). Individual metrics were developed in the categories of taxonomic richness and composition, trophic composition, reproductive richness and composition, and fish abundance and condition. Taxonomic richness and composition metrics were developed with their respective taxa. The trophic classification of Smogor and Angermeier (1999) was used whereby food types were grouped into detritus (DET), algae-vascular plants (AH), invertebrates (INV), and fish (P). Classifications were: DET/AH = 1, DET/AH/INV = 2, DET/AH/INV/P = 3, AH/INV = 4, AH/ INV/P = 5, INV = 6, INV/P = 7, and P = 8. We classified fish reproductive behaviors similar to Smogor and Angermeier (1999). Reproductive classifications were: simple (no site preparation or parental care), lithophilic (sand or gravel substrate) spawners = 1, lithophilic, site-prep spawners = 2, lithophilic, parental-care spawners = 3, lithophilic, siteprep, parental-care spawners = 4, miscellaneous, site-prep spawners = 5, miscellaneous, parental-care spawners = 6, miscellaneous, site-prep, parental-care spawners = 7, and simple, miscellaneous spawners = 8. Nest associates were considered site-prep spawners because they required a modified substrate to spawn. Metric Selection We used two criteria to choose the final IBI metrics from a list of candidate metrics. The first criterion was the ability of a metric to discriminate between reference and non-reference sites. Metrics that met the first criterion were tested for metric redundancy, our second criterion. To test each candidate metric for its ability to discriminate between reference and non-reference sites, we performed an analysis of covariance (ANCOVA) (Smogor and Angermeier 1999); watershed size, a surrogate measure of stream size, was used as the covariate and was considered significant at α = 0.05. For this analysis, all metrics consisting of proportional data were arcsin transformed to meet normality assumptions (Zar 1999). Metrics showing significant differences between reference and non-reference data continued to be considered as candidate metrics. We tested for metric redundancy by conducting simple correlations between all combinations of remaining candidate metrics (Barbour et al. 1992). High and low correlation coefficients (-0.90 > r > 0.90) indicated redundant metrics (Angermeier et al. 2000). The metric from each redundant group having the lowest P-value from the ANCOVA was included in the IBI. Metric Scoring We standardized metrics to score from 0 to 10 similar to Minns et al. (1994) by developing threshold limits and linear equations. Threshold limits were minimum, 50th, and 95th percentile values of reference and
452
Southeastern Naturalist
Vol. 2, No. 3
non-reference data ranges, and they were dependent on metric relationships with stream-site quality. For metrics where watershed size was a significant covariate, we further investigated the relationship between watershed size and metric values for reference sites only by using simple regression. Metrics significantly related to watershed size at reference sites had their thresholds adjusted for watershed size. After determining threshold limits, we adjusted each metric to score from 0 (very poor condition) to 10 (reference condition) by using the equation: MS = A + B*(MR)
where MS = Metric Score, MR = Raw Metric Value, A = the y-intercept in the regression of MS versus MR, B = the slope in the regression of MS versus MR. Regressions were computed from two known points; upper and lower thresholds were assigned scores of 0 or 10 depending on a metric’s relationship with stream site quality. In the above metric scoring equation the following conditions must be in place: If MR < LT, then MR = LT If MR > UT, then MR = UT
where LT equals the lower threshold limit and UT equals the upper threshold limit. Thus, the equation calculates a metric score from a raw metric value and upper and lower threshold limits. Threshold limits define the maximum and minimum values a raw metric value may have when included in the equation. Raw metric values above the upper threshold limit and below the lower threshold limit take the value of each respective threshold. All metrics affected by watershed size had their threshold limit values rounded to the nearest integer. As a result, all metrics were standardized to score from 0 to 10. IBI Scoring and Qualitative Classifications We used individual metric scores to standardize the IBI to score from 0 to 100. We calculated IBI scores as follows:
where IBI = IBI score, MS = metric score of the ith metric, and N = the number of metrics. The number of IBI metrics may potentially change if particular metrics are determined to perform poorly, or other useful metrics are added. The qualitative site classifications for IBI scores are > 0 - < 20 (very poor), 20 - < 40 (poor), 40 - < 60 (fair), 60 - < 80 (good), and 80 - 100 (reference). An IBI score of zero is assigned if no fish are collected. IBI Precision Because a single digit threshold between qualitative site classifications is arbitrary, Karr (1981) recommended that a range of IBI scores
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
453
be centered around each threshold and professional judgement be used to determine which classification is assigned. To recommend such scoring ranges, we estimated within-site precision of the IBI using a jackknife technique, similar to a bootstrap method used by Fore et al. (1994). Fifteen stream sites were used for the jackknife analysis, and each was divided into 15 contiguous segments 5 MSWs in length (Dauwalter 2002). Fish were collected by backpack electrofishing from individual stream segments. Jackknife samples were obtained by randomly selecting, without replacement, 10 of 15 stream segments (equal to approximately 51 MSWs, and recommended for IBI calculations) 25 times per stream site. We calculated IBI scores from these jackknife samples, and computed 95 percent confidence limits for each site. Relations between confidence-limit lengths and mean IBI scores of jackknife samples, total number of individuals, and species richness of all 15 sites were assessed using simple correlations at α = 0.05. Total number of individuals and species richness of all 15 sample segments of each sample site were used for analyses. Metric Contributions We used a metric-remainder correlation coefficient (Hughes et al. 1998) to evaluate relative metric contributions to IBI scores in all five qualitative site classifications combined and for IBI scores in each exclusive classification. The metric-remainder correlation coefficient correlates a specific metric score with an IBI score calculated without the specific metric under investigation. A lower correlation coefficient is indicative of a higher metric contribution to the IBI. IBI Metric Relations with Environmental Variables We also determined relations between our IBI metrics and selected land-use and physicochemical variables at each site by using a Spearman rank correlation. Land-use variables were correlated with IBI metrics by using the AR-GAP data layer used in selecting reference sites (Smith et al. 1998, Weih 2001). Riparian (forested) zone width was measured from digital orthographic quadrangles (DOQs; 1-meter resolution) with GIS software. Water samples for associated fish-collection sites were processed by the ADEQ water-quality laboratory (ADPCE 1993, ADPCE 1995, APHA 1998). Relations among physicochemical and land-use variables were assessed using a principal components analysis (PCA) with a varimax rotation; outliers were removed and appropriate data transformations were made (Gauch 1982, Sokal and Rohlf 1995). Percent agriculture land-use was not included because proportional variables summing to unity can confound analyses (Tabachnick and Fidell 1996), but it was almost always inversely related to the percent forested because very little urban and water land-use existed in the watersheds of study sites.
454
Southeastern Naturalist
Vol. 2, No. 3
RESULTS Data The fish-collection database used for IBI development included 96 fish collections. Random removal of 32% of individuals from each fish collection removed at least one species from most collections. The average number of species removed per collection was 1.3 (SD = 1.22; range = 0–5); gars (Lepisosteidae) were most often removed (38%) from samples in which they were collected, while temperate basses (Moronidae) were never removed (0%). Site Classifications Our reference-site criteria selected 13 of 96 sites as reference sites. The randomly removed sites used for site-classification and IBI validation included 11 sites in the 0–100 km2 watershed size group, 3 in the > 100– 300 km2 group, and 5 in the > 300 km2 group, for a total of 19 sites, including 3 reference sites; one reference site was removed per size group. Metric Selection and Scoring When we tested 39 candidate metrics, 16 showed a significant difference (P ≤ 0.10) in raw metric values between reference and nonreference sites (Table 1). Correlation coefficients between differentiating metrics resulted in 17 pairs of highly correlated metrics (-0.90 > r > 0.90), indicating metric redundancy. The selection of one metric from all redundant groups, with one exception, resulted in seven IBI metrics being selected for the IBI. The exception was that we included the metrics number of darter, sculpin, and madtom species and number of lithophilic spawning species instead of number of intolerant species. Although number of intolerant species differentiated stream conditions best, it was redundant with the two included metrics, which were not redundant themselves. Because they also differentiated stream conditions well, they incorportated benthic and reproductive ecology, and they allowed more metrics to be included thereby increasing IBI robustness, we retained number of darter, sculpin, and madtom species and number of lithophilic spawining species metrics and excluded number of intolerant species from the IBI. Thus, the Ozark Highland IBI metrics were: percent (of individuals) as algivorous/herbivorous, invertivorous, and piscivorous; percent with black spot Neascus spp. or an anomaly; percent green sunfish Lepomis cyanellus, bluegill L. macrochirus, yellow bullhead Ameiurus natalis, and channel catfish Ictalurus punctatus; percent invertivorous; percent top carnivores; number of darter Etheostoma and Percina, sculpin Cottus, and madtom Noturus species; and number of lithophilic spawning species (Table 2). Of the selected metrics, percent top carnivores; number of darter, sculpin, and madtom species; and number of lithophilic spawning spe-
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
455
cies were significantly influenced by watershed size as indicated by the ANCOVA (Table 1). When the relation between watershed size and metric values was examined using reference sites only, percent top carnivores was not affected (P = 0.190), whereas number of darter, sculpin, and madtom species (P = 0.001) and number of lithophilic spawning species (P = 0.002) were and had their thresholds adjusted for watershed size (Fig. 2). Selected metrics were scored from 0 to 10 using linear equations and metric thresholds (Table 3). Qualitative Classifications For removed sites, the percent agreement between qualitative classifications generated by the IBI and those generated by the reference criteria was 73.7%. When all sites were included, the IBI and referenceTable 1. Ozark Highland candidate IBI metrics, and analysis of covariance results between metric values of reference and non-reference sites. Percent metrics represent the percent of individuals sampled. Metrics were considered to differentiate significantly at P ≤ 0.10. Watershed size was considered a significant covariate at P ≤ 0.05, as indicated by an asterisk (*). For all metrics reference N = 10 and non-reference N = 67, except black spot and anomaly metrics where reference N = 6 and non-reference N = 13. Metric Number of individuals per second electrofishing Percent algivorous/herbivorous, invertivorous, and piscivorous Percent algivorous/herbivorous and invertivorous Percent with an anomaly Percent benthics Percent 3+ black spot cysts Percent with black spot or an anomaly Percent with black spot Percent cyprinids Percent darters and sculpins Percent darters, sculpins, and madtoms Percent detritivorous and algivorous/herbivorous Percent green sunfish, bluegill, yellow bullhead, and channel catfish Percent generalists Percent green sunfish Percent green sunfish and yellow bullhead Percent intolerants Percent invertivorous Percent invertivorous and piscivorous Percent lithophilic, site-prep spawners Percent shiners Percent simple, lithophilic spawners Percent simple, miscellaneous spawners Percent stonerollers Percent top carnivores Percent miscellaneous, site-prep, parental-care spawners Number of benthic species Number of cyprinid species Number of darter and sculpin species Number of darter species Number of darter, sculpin, and madtom species Number of generalist species Number of intolerant species Number of lithophilic spawning species Number of shiner species Number of simple, lithophilic spawning species Number of species Number of sucker species
P-value 0.438 0.054 0.309 0.266 0.390 0.452 0.037 0.081 0.198 0.735 0.489 0.550 0.012 0.115 0.017 0.044 0.124 0.055 0.776* 0.738 0.465* 0.631 0.225* 0.165 0.010* 0.018 0.013* 0.219* 0.013* 0.026* 0.013* 0.735* 0.001* 0.002* 0.117* 0.224* 0.028* 0.171
456
Southeastern Naturalist
Vol. 2, No. 3
Table 2. Fish classifications used to derive Ozark Highland metric values. Number designations are described in the Candidate IBI Metrics section. Family/Species Petromyzontidae Ammocoete Lepisosteidae Lepisosteus osseus Clupeidae Dorosoma cepedianum Cyprinidae Campostoma anomalum Campostoma oligolepis Ctenopharyngodon idella Cyprinella galactura Cyprinella whipplei Cyprinus carpio Erimystax dissimilis Erimystax harryi Erimystax x-punctatus Hybopsis amblops Luxilus cardinalis Luxilus chrysocephalus Luxilus pilsbryi Luxilus zonatus Lythrurus umbratilis Nocomis asper Nocomis bigutattus Notemigonus crysoleucas Notropis boops Notropis greenei Notropis nubilus Notropis ozarcanus Notropis rubellus Notropis telescopus Phoxinus erythrogaster Pimephales notatus Pimephales promelas Pimephales tenellus Semotilus atromaculatus Catostomidae Carpiodes cyprinus Carpiodes velifer Catostomus commersoni Erimyzon oblongus Hypentelium nigricans Ictiobus bubalus Ictiobus niger Minytrema melanops Moxostoma carinatum Moxostoma duquesnei Moxostoma erythrurum Moxostoma macrolepidotum Ictaluridae Ameiurus melas Ameiurus natalis Ictalurus punctatus Noturus albater Noturus exilis Noturus flavater Pylodictis olivaris Salmonidae Oncorhyncus mykiss Salmo trutta Aphredoderida Aphredoderus sayanus Fundulidae Fundulus catenatus Fundulus notatus Fundulus olivaceus Poecilliidae Gambusia affinis Atherinidae Labidesthes sicculus
Top carnivore
Intolerant
Benthic
x
x x x x x x x x
x x x x
x x x x x x x x x
x x
x x
x x x x x x
x
x x x x x x x x x x
x x x
Trophic
Reproductive
1
2
8
8
2
8
4 1 4 6 6 2 2 2 2 6 4 6 4 6 4 4 4 4 6 6 1 6 4 6 1 4 2 4 7
2 2 8 8 8 8 1 1 1 1 2 2 2 2 2 2 2 8 1 1 2 1 1 1 2 7 6 3 2
2 2 6 4 4 2 4 2 6 6 4 6
8 8 1 2 1 8 8 1 2 1 1 1
3 5 5 6 6 7 8
4 7 7 4 4 4 7
7 7
2 2
6
8
6 4 6
3 6 8
4
8
6
8
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
457
criteria site classifications agreed 66.7% of the time. All but two of the disagreeing classifications from all sites resulted from the IBI classifying sites as reference sites when they were originally pre-classified as non-reference sites by the reference criteria. When IBI scoring ranges for professional judgement of site quality were incorporated (see IBI Precision section below), the percent consistent classifications increased as expected (Table 4). IBI Precision Ninety-five percent confidence-interval lengths of IBI scores for our jackknife samples, for each site, averaged 2.0 (SD = 0.88) IBI units. The minimum confidence interval was 0.7, while the maximum was 3.7. We rounded-up the maximum confidence-interval length to use as a range for professional judgement of qualitative stream conditions at or near classification thresholds. This range of 4 IBI units should be centered on each respective threshold. For example, without this range a stream site with an IBI score of 59 may be classified as being in “fair” condition when it is actually in “good” condition. No significant correlations were detected between confidence-limit lengths and mean IBI scores (r = -0.469; P = 0.078), number of individuals (r = -0.142; P = 0.613), and species richness (r = 0.135; P = 0.631). Table 2, continued. Family/Species Cottidae Cottus carolinae Cottus hypselurus Moronidae Morone chrysops Centrarchidae Ambloplites ariommus Ambloplites constellatus Ambloplites rupestris Lepomis cyanellus Lepomis gulosus Lepomis macrochirus Lepomis megalotis Lepomis microlophus Lepomis punctatus Micropterus dolomieu Micropterus punctulatus Micropterus salmoides Pomoxis annularis Pomoxis nigromaculatus Percidae Etheostoma blennioides Etheostoma caeruleum Etheostoma euzonum Etheostoma flabellare Etheostoma juliae Etheostoma punctulatum Etheostoma spectabile Etheostoma stigmaeum Etheostoma zonale Percina caprodes Percina evides Percina nasuta Stizostedion vitreum
Top carnivore
Intolerant
Benthic
Trophic
Reproductive
x x
x x
7 7
4 4
8
8
7 7 7 7 7 5 6 6 4 7 7 7 7 7
4 4 4 7 7 7 2 7 7 4 7 7 7 7
6 6 6 6 6 6 6 6 6 6 6 6 7
8 1 1 4 1 1 1 1 8 1 1 1 1
x x x x
x x x
x x x x x
x
x x x x x x x x
x
x x x
x x x x x x x x x x x x
458
Southeastern Naturalist
Vol. 2, No. 3
No. of Darter, Madtom, and Sculpin Species
Metric Contributions The relative contribution of each metric to the IBI varied for all scoring ranges (Table 5). At sites where black spot and anomaly data were collected (i.e., where the percent of fish containing black spot cysts or other external anomalies was recorded), number of lithophilic spawning species and percent with black spot or an anomaly metrics contributed most in reference streams, while number of darter, sculpin, and madtom species and percent with black spot or an anomaly contrib-
No. of Lithophyllic Spawning Species
Watershed Size (km2)
Watershed Size (km2) Figure 2. Upper threshold limits and raw metric values for two Ozark Highland IBI metrics requiring threshold limit adjustments for watershed size, a surrogate measure of stream size. Linear functions from 0–800 km2 are best-fit lines for reference-site data. All sites N = 77.
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
459
Table 3. IBI metrics, their intercept (A) and slope (B) scoring coefficients, and their upper (UT) and lower (LT) threshold limits. Metrics requiring a scoring adjustment for watershed size (km2; WS) have linear functions to derive upper threshold limits. Metric scores (MS) are calculated from raw metric values (MR) using the equation: MS = A + B*(MR). Raw metric values above or below threshold limits take the value of their respective threshold limit. See Metric Scoring section for scoring instructions. Metric Coefficients Metric Percent algivorous/herbivorous, invertivorous, and piscivorous Percent with black spot or an anomaly Percent green sunfish, bluegill, yellow bullhead, and channel catfish Percent invertivorous Percent top carnivores Number of darter, sculpin, and madtom speciesa Number of lithophilic spawning speciesa Number of darter, sculpin, and madtom speciesb Number of lithophilic spawning speciesb
Threshold Limits
A
B
LT
UT
10.24
-10/9.00
0.22
9.22
11.53 10.15
-10/0.57 -10/24.99
0.09 0.38
0.66 25.37
0.00 0.00 0.00 0.00 0.00 0.00
10/40.00 10/2.80 10/12.00 10/26.00 10/UT-LT 10/UT-LT
0.00 0.00 0.00 0.00 0.00 0.00
40.00 2.80 12.00 26.00 4.909+0.0091(WS)c 14.336+0.0150(WS)c
a
Metric coefficients and threshold values for sites with watershed sizes (WS) ≥ 800 km2 Metric coefficients and threshold values for sites with watershed sizes (WS) < 800 km2 c Round upper threshold to nearest integer b
Table 4. Percent consistency values for reference and non-reference site classifications, as classified for IBI development and by the IBI. Removed sites represent 19 randomly chosen sites not included in IBI development. All sites included those used to develop the IBI and removed sites (N = 96). Adjusted indicates classifications incorporating IBI scoring ranges (4 IBI units) for professional judgement of stream-site quality. Removed Sites Outcomes
All Sites
Reference criteria Adjusted Reference criteria Adjusted
Consistent Classifications IBI reference, pre-classified as non-reference IBI non-reference, pre-classified as reference
73.7 21.1 5.2
78.9 15.9 5.2
66.7 31.3 2.0
71.9 26.1 2.0
Table 5. Metric-remainder correlation coefficients (and contribution ranks) from correlations between metric scores and IBI scores without the particular metric under investigation. A low correlation coefficient indicates a high metric contribution to the IBI. Data for the percent with blackspot or an anomaly metric were not available for all sites; therefore, metric contribution analyses were conducted separately for sites where data were and were not available. No sites were in Very Poor condition, and no sites containing percent with black spot or an anomaly data were in Fair or Poor condition. Metric
All sites Reference
Without Percent with black spot or an anomaly Percent algiv./herb., invert., and pisc. Percent grn sf, bluegill, yel. bull., ch. cfish Percent invertivorous Percent top carnivores Number of darter, sculpin, and madtom species Number of lithophilic spawning species
Good
Fair
Poor
0.333 (1) 0.525 (3) 0.426 (2) 0.526 (4) 0.727 (5) 0.738 (6)
0.080 (3) 0.227 (4) -0.303 (1) -0.082 (2) 0.368 (6) 0.338 (5)
-0.652 (1) -0.320 (2) -0.290 (3) -0.065 (6) -0.269 (4) -0.156 (5)
-0.587 (1) -0.440 (3) -0.580 (2) 0.169 (5) 0.233 (6) 0.005 (4)
-0.547 (3) 0.589 (6) -0.956 (1) -0.635 (2) -0.131 (4) 0.390 (5)
With Percent with black spot or an anomaly Percent algiv./herb., invert., and pisc. Percent with black spot or an anomaly Percent grn sf, bluegill, yel. bull., ch. cfish Percent invertivorous Percent top carnivores Number of darter, sculpin, and madtom species Number of lithophilic spawning species
0.365 (3) 0.318 (2) 0.639 (6) 0.048 (1) 0.505 (5) 0.394 (4) 0.797 (7)
0.230 (6) -0.487 (2) 0.177 (5) -0.177 (4) 0.240 (7) -0.204 (3) -0.535 (1)
-0.204 (4) -0.425 (2) -0.055 (5) -0.225 (3) 0.095 (6) -0.438 (1) 0.709 (7)
460
Southeastern Naturalist
Vol. 2, No. 3
uted most in good stream conditions. When black spot and anomaly data were not available, percent invertivorous and percent top carnivores contributed most in reference and poor stream conditions, whereas other metrics contributed most in good and fair conditions. Trophic metrics appeared to be most influential on IBI scores. IBI Metric Relations with Environmental Variables Spearman rank correlations indicated that from 2 to 15 physicochemical and land-use variables were significantly correlated with all Table 6. Significant (α = 0.05) Spearman rank correlations (rs) between IBI metrics and tested physicochemical and land-use variables. Metrics correspond to: percent (of individuals) as algivorous/ herbivorous, invertivorous, and piscivorous (1); percent with black spot or an anomaly (2); percent green sunfish Lepomis cyanellus, bluegill L. macrochirus, yellow bullhead Ameiurus natalis, and channel catfish Ictalurus punctatus (3); percent invertivorous (4); percent top carnivores (5); number of darter Etheostoma and Percina, sculpin Cottus, and madtom Noturus species (6); and number of lithophilic spawning species (7). N indicates sample sizes for each variable tested. Variable Aluminum Ammonia nitrogen Arsenic Barium Biochemical oxygen demand Bromide Boron Cadmium Calcium Chloride Chromium Cobalt Copper Fluoride Hardness Iron Magnesium Manganese 48-hour dissolved oxygen 48-hour temperature Nickel Nitrate nitrogen Orthophosphorus pH % forested %100m buffer forested % agriculture % 100m buffer agriculture % urban % 100m buffer urban Potassium Road density 100m buffer road density Sedimentation Silicate Sodium Sulfate Total dissolved solids Total Kjeldahl nitrogen Total phosphorus Total suspended solids Turbidity Vanadium Zinc
N 17 54 12 12 40 16 12 12 17 55 12 13 17 16 28 12 17 17 56 53 12 57 54 57 96 96 96 96 96 96 12 96 96 26 12 17 57 55 5 57 55 48 12 18
1
2
3
0.382
0.502
4
-0.511
5
6
7
0.323
0.346
-0.378
-0.581
-0.574
-0.456
-0.566
-0.627
0.556 0.510 0.384
-0.348 -0.316 0.352 0.306 0.282 0.231
-0.415
0.268 0.579
0.332 0.320
-0.591 -0.402
-0.395 -0.297 0.327 0.263 0.253
0.467 0.354 -0.377 -0.299 -0.359 -0.281
0.408 0.256 0.620
-0.554 -0.367
-0.292 -0.281
-0.413 -0.283 -0.409
-0.434
-0.416
-0.570
0.319
-0.312 0.287 0.491
0.310 0.213 -0.211
-0.309 0.380 -0.263 -0.208 -0.536 -0.379
0.543
-0.422
0.470
0.639
-0.264 -0.498 -0.370 0.410
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
461
seven IBI metrics (P < 0.05). Chloride, nutrients, land use, road densities, and sedimentation appeared to be the most significant and strongly related variables to Ozark Highland IBI metrics (Table 6). Water-quality and land-use data for 19 variables from 31 sample sites were included in the PCA (Table 7). This constituted a variable-to-sample size ratio of 0.61. Principle components (PC) represented composite chemical, land-use, nutrient, water clarity, and dissolved oxygen-temperature variables. A plot of PC 1 versus PC 2 scores indicated that reference sites were correctly classified relative to other analyzed sites (Fig. 3). DISCUSSION We used existing and novel methods and metrics to develop an IBI for wadeable streams in the Ozark Highlands ecoregion of Arkansas. Our metric-selection criteria selected 7 of 39 candidate metrics. Some metrics we selected are commonly used in IBI development (e.g., percent invertivorous), while others are unique to this study (e.g., percent green sunfish, yellow bullhead, bluegill, and channel catfish). Our IBI metrics included taxonomic, trophic, reproductive, and fish health characteristics of fish-assemblages. Although we only selected 7 metrics, we believe our reference-site classification and metric-selection processes selected IBI metrics with proven abilities to differentiate stream-site quality. For these reasons, and the fact that the data used represented all three major Arkansas drainages in the ecoregion, we believe that the Ozark Highlands IBI is a robust index. Comparing reference and non-reference site classifications with IBI scoring classifications revealed that 73.7% of removed sites were classiTable 7. Correlation coefficients between physicochemical and land-use data and principal components from a principal components analysis (PCA) conducted using data from 31 Ozark Highland stream sites. Percent of variance accounted for by each component is indicated. Principal Component Variable Ammonia nitrogen Chlorides 48-hour dissolved oxygen 48-hour temperature Nitrate nitrogen Orthophosphorus Percent forested Percent 100m buffer forested Percent urban Percent 100m buffer urban pH Road density 100m buffer road density Sulfates Total dissolved solids Total organic carbon Total phosphorus Total suspended solids Turbidity Percent of variance
1
2
3
4
5
0.302 0.847 0.171 -0.283 0.640 0.875 -0.335 -0.316 0.116 0.450 0.143 0.195 0.418 0.829 0.392 0.450 0.859 -0.018 -0.048 41.2
0.187 0.302 0.351 -0.134 0.359 0.233 -0.853 -0.833 0.890 0.552 0.109 0.848 0.621 0.141 -0.033 0.441 0.256 0.151 -0.068 16.5
0.847 -0.103 -0.024 0.068 -0.202 -0.010 -0.119 0.237 0.129 0.146 -0.729 -0.086 0.330 0.128 -0.670 0.469 -0.031 0.156 0.325 9.0
0.059 -0.055 0.017 -0.060 0.029 -0.079 -0.081 -0.067 0.115 -0.191 -0.280 -0.015 -0.251 0.024 -0.246 0.122 0.025 0.933 0.918 7.8
-0.127 0.099 0.814 -0.829 0.323 0.090 -0.198 -0.195 0.011 0.247 -0.050 0.188 0.303 0.238 0.003 -0.263 0.120 0.030 0.036 5.7
462
Southeastern Naturalist
Vol. 2, No. 3
fied consistently. Two removed sites pre-classified as reference sites received IBI scores less than 80. Therefore, all inconsistent classifications, except those two, were from sites originally pre-classified as non-reference sites being classified as reference sites by the IBI, which indicates that fish assemblages at those sites reflected reference conditions. The two preclassified reference sites not classified as such by the IBI may be outliers that really did not reflect reference conditions. Even so, both sites received scores of 76, which were close to the scoring threshold for reference classification (i.e., 78 to 82). In addition, when IBI scoring ranges for professional judgment of site quality were accounted for, percent consistent classifications between IBI classifications and our criteria increased to 78.9% for the 19 removed sites and 71.9% for all 96 sites. If there was error in classifying and selecting reference sites, it was in a conservative direction. It is preferable to leave out some sites scoring as reference sites from IBI development procedures than to include sites not scoring as reference sites. Classifying non-reference stream sites as reference sites in development procedures would defeat
Figure 3. Plot of composite chemical variable (PC 1) versus composite land-use variable (PC 2) from a PCA of environmental variables (N = 31) from select Ozark Highland stream sites. R = reference sites, and N = non-reference sites.
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
463
the purpose of calibrating metrics to reference condition, which is an important component in IBI development and the “integrity” concept. Our jackknife sample data from the IBI precision analysis indicated no significant relation between site quality (i.e., mean IBI score) and 95% confidence-limit lengths of those sites. Fore et al. (1994) reported similar findings for their bootstrap data. Conversely, they found leastdisturbed site IBI scores varied least across years, a result also reported by Karr et al. (1987). Hughes et al. (1998) reported a weak increase in IBI score standard deviations with increasing IBI scores, although variance was greatest at intermediate IBI scores. There was a negative and almost-significant, but variable, trend between mean IBI scores and confidence-limit lengths of those scores for our jackknife data. We also found no relation between IBI-score confidence-interval lengths and number of individuals and species richness, results contrary to findings reported by Fore et al. (1994). All seven metrics were shown to have significant relations with at least two of the selected physicochemical and land-use variables. Although some of these significant correlations were likely due to chance alone because of the many pairwise comparisons, these established relations may help managers discern causes of stream impairment. For example, chloride, nutrient, land use, road density, and sedimentation levels were most consistently correlated with IBI metrics. Similar relations have been shown in other U.S. regions and are thought to be the most important factors contributing to stream degradation in the U.S. (Lammert and Allan 1999, Lenat and Crawford 1994, Matthews et al. 1992, Waite and Carpenter 2000, Wang et al. 1997). We also found that land-use of entire watersheds was more strongly related to the IBI metrics than land-use in close proximity to Ozark Highland streams. Mechanisms influencing these relations may be similar to those discussed by Omernik et al. (1981) in which land use adjacent to streams may act as a nutrient sink for a period of time, but eventually nutrient inflow must equal nutrient outflow. In addition, unmapped waterways and storm runoff may contribute nutrients and sediments from areas away from lands bordering streams. The PCA revealed some predictable relations among physicochemical and land-use variables. For example, phosphorus variables, as well as sulfate and chloride concentrations, were strongly correlated with PC 1, which accounted for 41.2 % of the data variance. These variables should be positively related, if a relation exists, because they all may occur from high nutrient inputs, organic decay, and possibly watertreatment effluents. The ADPCE (1993) reported high coincident concentrations of phosphorus, chlorides, and total dissolved solids from an untreated wastewater-treatment-facility discharge, and higher chloride and sulfate levels below, when compared to above, treated effluent discharges. Although data from that particular untreated discharge event
464
Southeastern Naturalist
Vol. 2, No. 3
were not used in our analyses, it demonstrates that untreated effluents may have caused the relations observed between phosphorus, chlorides, and sulfates. In addition, PC 2 indicated a positive relation between road density and percent urban land-use and a negative relation between percent forested land-use and percent urban land-use and road density. Nutrient concentrations also showed a slightly positive relation with PC 2. This may result from all agricultural land-use identified in the Ozark Highland watersheds being pastureland. Nutrients from fertilizers or livestock excretory products may be absorbed or buffered by pasture vegetation. Also, nutrients may make their way into streams predominantly during episodic events such as storms producing runoff. Because water samples were collected during base-flow conditions, we may not have detected the true nutrient load resulting from agricultural land use. Observed relations between ammonia nitrogen, pH, and total dissolved solids (PC 3) may be signals of natural watershed conditions, but otherwise remain unexplained. Relations between turbidity and total suspended solids (PC 4), and between temperature and dissolved oxygen (PC 5), conform to water-quality expectations (Allan 1995). Plotting PC 1 versus PC 2 scores showed that reference sites included in the PCA grouped together (Fig. 3). The observed grouping indicated that landuse intensity and chemical variables were important attributes of analyzed reference sites, which is characteristic of useful IBIs. The groupings also indicated that at least three appropriate reference sites were selected and that many Ozark Highland sites were in relatively good condition, which was concordant with the distribution of IBI scores for Ozark Highland stream sites (Fig. 4). All stream sites were sampled with the common objective of collecting most fish species and sampling all available habitats. Despite this
Figure 4. Frequency histogram of IBI scores for stream sites sampled in the Ozark Highlands ecoregion; Mean = 73.2, SD = 17.98, and N = 96.
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
465
common objective, electrofishing efforts may have varied among sample sites. Even so, sampling efforts at sites for which data were available indicated that the linear stream distances sampled were greater, on average, than what is recommended for some biomonitoring purposes (Meador et al. 1993). Furthermore, we developed the IBI for a specific sampling effort (51 MSWs), which nearly perfectly estimated relative species abundances, and collected, on average, 95% of all species collected in a stream reach 75 MSWs in length (Dauwalter 2002). This is important because differences in sampling effort can affect IBI scores (Angermeier and Karr 1986), and, therefore, electrofishing effort should be standardized when applying the Ozark Highlands IBI. Finally, despite the potential for somewhat inconsistent sampling effort for IBI development, our results suggest that the Ozark Highlands IBI can differentiate at a minimum reference and non-reference stream conditions. For example, our classification consistency was high, the PCA suggested that at least three good reference sites were chosen, and IBI metrics correlated with water-quality variables and land-uses believed to be major causes of stream impairment. Therefore, the Ozark Highlands IBI should be a useful tool for determining stream-site quality, identifing impaired streams, prioritizing stream-rehabilitation efforts, and meeting CWA requirements. ACKNOWLEDGMENTS This project was funded by the U.S. Environmental Protection Agency, Region 6, under contract #68-C98-111, Work Assignment #1–10 to Tetra Tech, Inc., which subcontracted to the University of Arkansas at Pine Bluff. We thank all ADEQ personnel involved in their fish collections and data management. We are grateful to J. Wise, B. Posey, and C. Davidson for classifying stream conditions based on their professional judgements. R. Smogor provided ideas on IBI development. P. Crocker, M. Barbour, J. Jackson, N. Stone, A. Goodwin, E. Buckner, and two anonymous reviewers commented on earlier drafts of this manuscript. We also thank C. Okiror, J. Poole, T. Hungerford, L. Lackey, and S. Mutagyera for help with data collection and entry.
LITERATURE CITED ADPCE (Arkansas Department of Pollution Control and Ecology). 1993. Water quality survey of the Huntsville STP effluent on Town Branch and Holman Creek. Arkansas Department of Pollution Control and Ecology, Little Rock, AR. WQ93-03-1. ADPCE (Arkansas Department of Pollution Control and Ecology). 1995. Water quality, macroinvertebrate, and fish community survey of the Upper White River watershed, northwest Arkansas. Arkansas Department of Pollution Control and Ecology, Little Rock, AR. WQ95-08-1. ADPCE (Arkansas Department of Pollution Control and Ecology). 1996. Arkansas water quality inventory report. Arkansas Department of Pollution Control and Ecology, Little Rock, AR. Alder, R.W., J.C. Landman, and D.M. Cameron. 1993. The Clean Water Act 20 Years Later. Island Press, Washington, DC. 320 pp.
466
Southeastern Naturalist
Vol. 2, No. 3
Allan, J.D. 1995. Stream Ecology: Structure and Function of Running Waters. Chapman and Hall, New York, NY. 388 pp. Angermeier, P.L., and J.R. Karr. 1986. Applying an index of biotic integrity based on stream-fish communities: considerations in sampling and interpretation. North American Journal of Fisheries Management 6:418–429. Angermeier, P.L., R.A. Smogor, and J.R. Stauffer. 2000. Regional frameworks and candidate metrics for assessing biotic integrity in Mid-Atlantic Highland streams. Transactions of the American Fisheries Society 129:962–981. APHA (American Public Health Association). 1998. Standard Methods for the Examination of Water and Wastewater. American Water Works Association and the Water Pollution Control Federation, American Public Health Association, Washington, DC. Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid bioassessment protocols for use in wadeable streams and rivers: periphyton, benthic macroinvertebrates, and fish, 2nd edition. Office of Water, U.S. Environmental Protection Agency, Washington, DC. EPA 841-B-99-002. Barbour, M.T., J.L. Plafkin, B.P. Bradley, C.G. Graves, and R.W. Wisseman. 1992. Evaluation of EPA’s rapid bioassessment benthic metrics: Metric redundancy and variability among reference sites. Environmental Toxicology and Chemistry 11:437–449. Bennett, C., J. Giese, B. Keith, R. McDaniel, M. Maner, N. O’Shaughnessy, and B. Singleton. 1987. Physical, chemical and biological characteristics of least-disturbed reference streams in Arkansas’ ecoregions: Volume I - Data collection. Arkansas Department of Pollution Control and Ecology, Little Rock, AR. Dauwalter, D.C. 2002. An Index of Biotic Integrity for fish assemblages in Arkansas’ wadeable Ozark Highland streams. Master’s Thesis. University of Arkansas at Pine Bluff, Pine Bluff, AR. 153 pp. Didier, J., and P. Kestemont. 1996. Relationships between mesohabitats, ichthyological communities, and IBI metrics adapted to a European river basin (The Meuse, Belgium). Hydrobiologia 341:133–144. Etnier, D.A., and W.C. Starnes. 1993. The Fishes of Tennessee. The University of Tennessee Press, Knoxville, TN. 681 pp. Fausch, K.D., J.R. Karr, and P.R. Yant. 1984. Regional application of an Index of Biotic Integrity based on stream fish communities. Transactions of the American Fisheries Society 113:39–55. Fore, L.S., J.R. Karr, and L.L. Conquest. 1994. Statistical properties of an Index of Biological Integrity used to evaluate water resources. Canadian Journal of Fisheries and Aquatic Sciences 51:1077–1087. Gauch, H.G., Jr. 1982. Multivariate Analysis in Community Ecology. Cambridge University Press, New York, NY. 298 pp. Giese, J., B. Keith, M. Maner, R. McDaniel, and B. Singleton. 1987. Physical, Chemical, and Biological Characteristics of Least-Disturbed Reference Streams in Arkansas’ Ecoregions: Data Analysis. Arkansas Department of Pollution Control and Ecology, Little Rock, AR. 148 pp. Hlass, L.J., W.L. Fisher, and D.J. Turton. 1998. Use of the Index of Biotic Integrity to assess water quality in forested streams in the Ouachita Mountains ecoregion, Arkansas. Journal of Freshwater Ecology 13:181–192. Hughes, R.M., P.R. Kaufmann, A.T. Herlihy, T.M. Kincaid, L. Reynolds, and D.P. Larsen. 1998. A process for developing and evaluating indices of fish assemblage integrity. Canadian Journal of Fisheries and Aquatic Sciences 55:1618– 1631.
2003
D.C. Dauwalter, E.J. Pert, and W.E. Keith
467
Hughes, R.M., D.P. Larsen, and J.M. Omernik. 1986. Regional reference sites: a method for assessing stream potentials. Environmental Management 10:629–635. Jenkins, R.E., and N.M. Burkhead. 1993. Freshwater Fishes of Virginia. American Fisheries Society, Bethesda, MD. 1079 pp. Jones, J.R., B.H. Tracy, J.L. Sebaugh, D.H. Hazelwood, and M.M. Smart. 1981. Biotic index tested for ability to assess water quality of Missouri Ozark streams. Transactions of the American Fisheries Society 110:627–637. Karr, J.R. 1981. Assessment of biotic integrity using fish communities. Fisheries 6(6):21–27. Karr, J.R., K.D. Fausch, P.L. Angermeier, P.R. Yant, and I.J. Schlosser. 1986. Assessing biological integrity in running water: a method and its rationale. Illinois Natural History Survey, Champaign, IL. Special Publication 5. Karr, J.R., P.R. Yant, K.D. Fausch, and I.J. Schlosser. 1987. Spatial and temporal variability of the Index of Biotic Integrity in three midwestern streams. Transactions of the American Fisheries Society 116:1–11. Keith, W.E. 1987. Distribution of fishes in reference streams within Arkansas’ ecoregions. Proceedings of the Arkansas Academy of Science 41:57–60. Lammert, M., and J.D. Allan. 1999. Assessing biotic integrity of streams: effects of scale in measuring the influence of land use/cover and habitat structure on fish and macroinvertebrates. Environmental Management 23:257–270. Lenat, D.R., and J.K. Crawford. 1994. Effects of land use on water quality and aquatic biota of three North Carolina Piedmont streams. Hydrobiologia 294:185–199. Lyons, J., S. Navarro-Perez, P.A. Cochran, E. Santana, and M. Guzman-Arroyo. 1995. Index of Biotic Integrity based on fish assemblages for the conservation of streams and rivers in west-central Mexico. Conservation Biology 9:569–584. Lyons, J., L. Wang, and T.D. Simonson. 1996. Development and validation of an Index of Biotic Integrity for coldwater streams in Wisconsin. North American Journal of Fisheries Management 16:241–256. Matthews, W.J., D.J. Hough, and H.W. Robison. 1992. Similarities in fish distribution and water quality patterns in streams of Arkansas: congruence of multivariate analyses. Copeia 1992:296–305. Matthews, W.J., and H.W. Robison. 1988. The distribution of the fishes of Arkansas: a multivariate analysis. Copeia 1988:358–374. Meador, M.R., T.F. Cuffney, and M.E. Gurtz. 1993. Methods for sampling fish communities as part of the National Water Quality Assessment program. United States Geological Survey Open File Report, Raleigh, NC. USGS/OPR 93–104. Miller, D.L., P.M. Leonard, R.M. Hughes, J.R. Karr, P.B. Moyle, L.H. Schrader, B.A. Thompson, R.A. Daniels, K.D. Fausch, G.H. Fitzhugh, J.R. Gammon, D.B. Halliwell, P.L. Angermeier, and D.J. Orth. 1988. Regional applications of an Index of Biotic Integrity for use in water resource management. Fisheries 13(5):12–20. Minns, C.K., V.W. Cairns, R.G. Randall, and J.E. Moore. 1994. An Index of Biotic Integrity (IBI) for fish assemblages in the littoral zone of Great Lakes’ areas of concern. Canadian Journal of Fisheries and Aquatic Sciences 51:1804–1822. OEPA (Ohio Environmental Protection Agency). 1987. Biological criteria for the protection of aquatic life: volume II: users manual for biological field assessment of Ohio surface waters. State of Ohio Environmental Protection Agency, Division of Water Quality Monitoring and Assessment, Surface Water Section, Columbus, OH.
468
Southeastern Naturalist
Vol. 2, No. 3
Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association of American Geographers 77:118–125. Omernik, J.M., A.R. Abernathy, and L.M. Mate. 1981. Stream nutrient levels and proximity of agricultural and forest land to streams: some relationships. Journal of Soil and Water Conservation 36:227–231. Pflieger, W.L. 1997. The Fishes of Missouri. Missouri Department of Conservation, Jefferson City, MO. 372 pp. Robison, H.W., and T.M. Buchanan. 1988. Fishes of Arkansas. The University of Arkansas Press, Fayetteville, AR. 536 pp. Rohm, C.M., J.W. Giese, and C.C. Bennett. 1987. Evaluation of an aquatic ecoregion classification of streams in Arkansas. Journal of Freshwater Ecology 4:127–141. Simon, T.P., and J. Lyons. 1995. Application of the Index of Biotic Integrity to evaluate water resource integrity in freshwater ecosystems. Pp. 245–262, In W.S. Davis and T.P. Simon (Eds.). Biological Assessment and Criteria: Tools for Water Resource Planning and Decision Making. CRC Press, Inc., Boca Raton, FL. 671 pp. Smith, K.G., R.S. Dzur, D.G. Catanzaro, M.E. Garner, and W.F. Limp. 1998. The Arkansas GAP Analysis Project: Final Report. Center for Advanced Spatial Technologies (CAST), Fayetteville, AR. 97 pp. Smogor, R.A. 1996. Developing an Index of Biotic Integrity (IBI) for warmwater wadeable streams in Virginia. Master’s Thesis. Virginia Polytechnic Institute and State University, Blacksburg, VA. 126 pp. Smogor, R.A., and P.L. Angermeier. 1999. Effects of drainage basin size and anthropogenic disturbance on relations between stream size and IBI metrics in Virginia. Pp. 249–272, In T.P. Simon (Eds.). Assessing the Sustainability and Biological Integrity of Water Resources Using Fish Communities. CRC Press, Inc., Boca Raton, FL. 671 pp. Sokal, R.R., and F.J. Rohlf. 1995. Biometry: the Principles and Practice of Statistics in Biological Research. W. H. Freeman and Company, New York, NY. 887 pp. Tabachnick, B.G., and L.S. Fidell. 1996. Using Multivariate Statistics. HarperCollins Publishers, New York, NY. 880 pp. U.S. Census Bureau. 2001. Census 2000. U.S. Census Bureau, Population Division, U.S. Deparment of Commerce, Washinton, DC. Waite, I.R., and K.D. Carpenter. 2000. Associations among fish assemblage structure and environmental variables in Willamette Basin streams, Oregon. Transactions of the American Fisheries Society 129:754–770. Wang, L., J. Lyons, and R. Gatti. 1997. Influences of watershed land use on habitat quality and biotic integrity in Wisconsin streams. Fisheries 22(6):6–12. Ward, R.C., J.C. Loftis, and G.B. McBride. 1986. The “data-rich but informationpoor” syndrome in water quality monitoring. Environmental Management 10:291–297. Weih, R. 2001. Natural State Digital Database: Version 1.0 Beta 5, UTM Zone 15, NAD27. Arkansas Forest Resource Center, University of ArkansasMonticello, Monticello, AR. Zar, J.H. 1999. Biostatistical Analysis. Prentice Hall, Upper Saddle River, NJ. 663 pp.