Index of Biotic Integrity Applied to a Flow-regulated River System

Zachary H. Bowen
Mary C. Freeman
David L. Watson
U.S. Geological Survey
2150 Centre Avenue, Bldg C
Fort Collins, CO 80526

Citation: Bowen, Z. H., M. C. Freeman, and D. L. Watson, 1996, Index of Biotic Integrity Applied to a Flow-Regulated River System IN Proceedings of the 50th Annual Conference of the Southeastern Association of Fish and Wildlife Agencies, 1996: U.S. Fish and Wildlife, p. 26-37.

Introduction

The IBI offers resource managers an ecologically based method for assessing the health of aquatic ecosystems. The original IBI developed for midwestern streams (Karr 1981, Karr et al. 1986) consisted of 12 fish community parameters, or metrics, divided into categories of species richness, trophic structure, and fish abundance and condition. The 12 metrics were selected to evaluate different aspects of the health of stream ecosystems, and were therefore used to reflect changes in community structure or function that might not be assessed by measures of water chemistry or contaminant levels alone. The IBI provided a tool for quantifying changes in ecosystem health as a result of habitat degradation or flow alteration, in addition to chronically poor chemical water quality (Karr and Dudley 1981).

The IBI was first used to assess the biological quality of streams in the central United States (e.g., Berkman et al. 1986, Osborne et al. 1992). As the IBI gained popularity new versions were developed for use in other regions throughout the United States (reviewed by Miller et al. 1988, Simon and Lyons 1995), France (Oberdorff and Hughes 1992), and Canada (Steedman 1988). Versions of the IBI developed for the central United States retained most of the metrics used in the original IBI with minor modifications. However, versions developed for the eastern and western United States and for lakes and estuaries used substantially different metrics that reflected faunal differences among systems (Simon and Lyons 1995). Simon and Lyons (1995) provided a review of changes in IBI metrics for versions developed for use in different regions, and noted the need for development of IBI's for the species-rich streams of the Southeast.

The Tallapoosa River, a major tributary to the Alabama River system, has been extensively altered for production of electricity, flood control, water supply, and recreation. Natural flow patterns exist in only about 25% (93 km) of the uppermost Piedmont portion of the river, in west-central Georgia and east Alabama. Harris Dam, located near the town of Wedowee, Alabama, is the most upstream impoundment on the Tallapoosa River. Peaking hydropower production at Harris Dam (with hourly discharges that vary from leakage flow, near 0, to >200 m³/s during generation near capacity) alters the flow regime in the 80-km reach of the river between the dam and the backwaters of Lake Martin, the next downstream impoundment. Thurlow Dam is the downstream-most impoundment on the Tallapoosa River and regulates flow in the lower 75-km Coastal Plain section of the river, to its confluence with the Coosa River near Montgomery, Alabama. Although a minimum flow requirement at Thurlow Dam of 34 m³/s (mandated by the Federal Energy Regulatory Commission in 1991) reduces the magnitude of hourly discharge fluctuations, daily peak flows are typically about 660% (225 m³/s) of the minimum flow.

The physical and biological diversity of the Tallapoosa River system presents an opportunity to evaluate the sensitivity of an IBI to faunal community changes corresponding to various degrees of streamflow alteration. At least 125 species of fishes historically occurred in the Tallapoosa River (Pierson et al. 1986, Swift et al. 1986), including at least 3 endemic species. The river continues to support a significant portion of the native fish fauna in the unregulated headwater section as well as in the flow-regulated Piedmont and Coastal Plain reaches (Travnichek and Maceina 1994, National Biological Service, unpubl. data). Thus, although extensively managed, the Tallapoosa system represents an important natural resource. If an IBI adapted for use in the Tallapoosa system is sensitive to faunal changes associated with hydrologic alteration, it could be used to monitor effects of future development in this basin and, with appropriate modification, in other flow-regulated rivers in the southeastern U.S.

Our objectives were to 1) develop an IBI for application to a flow-regulated, species-rich river system in Alabama, 2) quantify fish assemblage differences among sites variously affected by flow alteration, and 3) provide a baseline for evaluating the effects of future changes in the watershed.

TOP OF PAGE

METHODS

Study Sites

Sampling was conducted at 7 study sites located in the Tallapoosa River system in Alabama (map provided in Travnichek and Maceina 1994). Five sites were located in the Piedmont subregion. Three of the Piedmont study sites (RP1, RP2, and RP3) were in the regulated portion of the Tallapoosa River downstream from Harris Dam. One of the 2 unregulated Piedmont sites (UP1) was located on the Little Tallapoosa River, a major tributary flowing into Harris Reservoir. The second unregulated Piedmont site (UP2) was located in the Tallapoosa River upstream from Harris Reservoir. Two sites (RCP1 and RCP2) were located below Thurlow Dam in the Coastal Plain subregion of Alabama. The 7 sites correspond to those sampled by Travnichek and Maceina (1994) during 1990-1992 and were chosen to reflect a variety of physical habitat conditions and to continue long-term data collection efforts.

Fish Sampling

Fish were collected during June - September 1994 and July - August 1995 using 1.5 x 6.0 m prepositioned area electrofishers (PAE, Bain et al. 1985). The length of individual study sites ranged from 2.5 - 4.0 km with each site divided into sampling sections of about 180 m in length. At the start of each day a section was randomly selected and 10 - 20 PAE samples were collected in the section depending on the availability of shallow-water habitat. Visibility limitations and safety concerns restricted sampling to areas < 1.5 m deep. If a section was selected where there were no such areas, another section was randomly selected. Sampling continued until evening when low light limited visibility. Usually 2 or 3 sections were sampled in a day. We collected 100 PAE samples at each site in both 1994 and 1995 (total N = 1400).

Selection and Analysis of Raw IBI Metrics

We reviewed metrics used in published versions of the IBI (Simon and Lyons 1995). Karr (1981) and subsequent authors grouped metrics into 5 categories: (1) species richness and composition; (2) indicator species; (3) trophic function; (4) reproductive function; and (5) abundance and condition. We used these categories as a framework for selecting metrics. Because of sampling limitations associated with PAE's, we based the metrics on small-bodied fishes (usually <150 mm TL) that were typically collected in shallow-water habitats. In order to examine redundancy in potential metrics, we calculated Pearson correlation coefficients for all raw metric scores based on our 1994-1995 collection data.

Development of Metric Expectations

Using distributional data (Lee et al. 1980, M. Pierson, Alabama Power Company, unpubl. data) and collection records from 1990-1995 (Alabama Cooperative Fish and Wildlife Research Unit, unpubl. data, USGS, unpubl. data, B. Freeman, University of Georgia Museum of Natural History, unpubl. data) we estimated the number of species common enough to be collected in a thorough sampling effort at Piedmont and Coastal Plain study sites. This was done to establish expected values for species richness and composition metrics. We did not include historically rare species (those that occurred in the basin but were represented in few collections) or large-bodied fishes that were unlikely to be captured by PAE's in our estimates in order to minimize negative bias in IBI scores. Because of differences in stream size and faunal composition, we established different expectations for total species richness at UP1, UP2, the remaining Piedmont sites. For the Coastal Plain sites, where fewer collection data were available and distributional data indicated as many as 70 common fish species, 70 was averaged with the highest number of species collected at either site in 1994 or 1995 to estimate expected total species richness. We used our 1994-1995 collection data to establish expectations for metrics that were based on relative abundances by selecting the best value observed in both years at Piedmont and at Coastal Plain sites. The best value approach to establishing metric expectations has been widely used (e.g., Fausch et al. 1984, Karr et al. 1986, Lyons 1992, Osborne et al. 1992) and typically works well when unimpaired reference sites are difficult to identify (Simon and Lyons 1995). For metrics based on relative abundances, we chose to use the same expected values at each Piedmont site because there was not sufficient evidence in historical collection records to suggest appropriate values that would reflect possible longitudinal differences in relative abundances among sites.

Metric Standardization, Formulation, and Sensitivity Analysis

We followed the approach of Minns et al. (1994) for standardizing metric scores, formulating the IBI, and analyzing effects of individual metrics on the overall IBI score. Raw metrics were standardized to a scale of 0 to 10. A value of 10 would be assigned if the raw metric equaled or exceeded the expected value. Otherwise the standardized score was calculated as B x raw score, where B = 10/expected value. Standardized IBI metrics were summed and multiplied by 10/N_m (N_m, number of metrics) to obtain an IBI score that varied continuously from 0 to 100. A score of 0 would indicate that sampling produced no fish and a score of 100 would indicate raw scores > the expected value for each of the metrics. IBI scores were divided into 5 categories: >0-20, >20-40, >40-60, >60-80, and >80.

In order to determine relative contributions of individual metrics to the IBI score, we calculated a reduced IBI for each metric: reduced IBI = 10(N_m x IBI/10 - test metric)/(N_m - 1) (Minns et al. 1994). We then computed the difference between reduced and complete IBI's for each site and year, for each metric. The total variances (i.e., across sites and years) of the differences were used to assess the sensitivity of the complete IBI to individual metrics. Because each metric did not contribute uniformly to the complete IBI score across the range of possible IBI scores (Angermeier and Karr 1986), we also calculated the variances of the differences for IBI scores grouped by score category (i.e., >40-60, >60-80). The ratio of within-category variance to total variance for a metric indicated range sensitivity. For example, a metric with a high ratio affected IBI scores in a particular category more than a metric with a low ratio (Minns et al. 1994).

Application of the Tallapoosa River IBI

Travnichek and Maceina (1994) sampled the 7 study sites during 1990-1992 and found reduced fish species richness, diversity, and density at the most strongly regulated sites compared to unregulated sites. To assess the ability of the IBI to detect faunal changes that were reflected in species richness, diversity, and density, we calculated IBI scores using Travnichek and Maceina's data. Because sampling methods were identical and total sample sizes were similar between their study and ours, we also included IBI scores for Travnichek and Maceina's data in our sensitivity analysis.

TOP OF PAGE

RESULTS AND DISCUSSION

IBI Metrics and Expectations for the Tallapoosa River System

We collected a total of 20,686 individuals in 78 species from 1400 PAE samples during 1994 and 1995. Sixty-two species were collected in both years and 8 species were collected in 1994 but not in 1995, and 8 different species were collected in 1995 but not in 1994. Cyprinids and percids (primarily Etheostoma and Percina spp.) dominated collections in terms of relative abundances in both years. We collected very few individual fish >200 mm TL. Sampling gear bias against larger individuals was evident in the relatively low proportion (0.005 - 0.06) of top predators collected at all sites in both years.

The 9 metrics selected for the IBI and expected values for each of the metrics are presented in Table 1. Metric categories found in published IBI's were used when possible and substitutions or omissions were made only when metrics did not discriminate among any of the sites or when metrics were highly intercorrelated. Species richness and composition metrics were similar to those in Karr's (1981) original IBI with the exception that we included Micropterus spp. in the number of sunfish species and we did not include a top predator trophic category. These changes were made based on the low proportion of top predators in collections.

a. This value applies to sites RP1, RP2, and RP3 only; because of smaller stream size, 40 and 44 were used as expected values for species richness (based on historical collection data) at sites UP1 and UP2, respectively.

Indicator species metrics included percentage of intolerant species but did not include percentage of green sunfish. Based on published accounts of habitat requirements (Lee et al. 1980, Etnier and Starnes 1993) we identified 8 species as intolerant, including the endemic Tallapoosa shiner (Cyprinella gibbsi) and Tallapoosa darter (Etheostoma tallapoosae) (Table 2). We chose to use percentage of individuals as intolerant species rather than number of intolerant species to reduce redundancy with species richness and composition metrics. We based our decision on the assumption that sites with high biological integrity support larger populations of intolerant species compared to degraded sites. Because green sunfish accounted for <1% of the total number of individuals at all sites the percentage of green sunfish metric was omitted. We substituted evenness, calculated as Shannon diversity (H') / ln (H_max or total number of species), for the green sunfish metric as proposed for streams in the Midwest (Simon and Lyons 1995).

Trophic function metrics included percentage of individuals as insectivorous cyprinids and percentage as benthic fluvial specialists (Table 1). Benthic fluvial specialists were small species that forage on the stream bottom in areas with moderate to swift current. Fishes in the insectivorous cyprinids and benthic fluvial specialists groups were selected based on published accounts of habitat use and behavior (Lee et al. 1980, Etnier and Starnes 1993). We chose to omit the omnivore metric because it did not discriminate among sites and was highly correlated with the percentage of individuals as cyprinids (R = 0.76, P = 0.002) which resulted in reduced scores at species-rich Coastal Plain sites where cyprinids dominated the fish assemblage.

Reproductive function is often assessed by the percentage of hybrids (Simon and Lyons 1995). Because only 7 hybrids (all Lepomis spp.) were collected in both years, the percentage of hybrids metric was not useful for comparing sites and therefore not included in the IBI. However, if warranted by future collection data, the percentage of hybrids metric could be used in the IBI. We used density expressed as mean N/ PAE sample as an abundance metric. We did not include a metric based on fish condition because we did not collect the required data.

Relationships Among Metrics

Correlations of raw metric scores across sites and years (N = 14) suggested some overlap between metrics. Overall species richness was correlated with number of darter species (R = 0.62, P = 0.016) and number of sucker species (R = 0.72, P = 0.004), highlighting the contribution of percids and catastomids to total species richness. Similarly, the number of darter species was positively correlated with number of sucker species (R = 0.57, P = 0.033). Percentage of benthic fluvial specialists was correlated with percentage of intolerant species (R = 0.59, P = 0.027). This relationship was expected because 6 of the 8 intolerant species were also in the benthic fluvial specialist category. Other correlations that were difficult to explain suggested possible site effects or spurious correlations: the percentage of insectivorous cyprinids was inversely correlated with number of sunfish species (R = -0.70, P = 0.006) and percentage of intolerant species (R = -0.55, P = 0.043). All other correlations were not significant (P >0.05)

Sensitivity Analysis

The highest total variances of the differences between reduced and complete IBIs were associated with percentage of insectivorous cyprinids and percentage of intolerant species (Table 3). Metrics with high total variances of the differences contribute more to the complete IBI score than metrics with low variances. Other high variance values were associated with the number of darter species and density. The lowest variance (i.e., least contribution to complete IBI score) was associated with species richness. Because the 4 metrics with the highest variances were distributed among the 4 metric categories (Table 3), the IBI should be sensitive to changes in assemblage composition, trophic structure, or fish abundance.

IBI sensitivity to metrics was not the same in different score categories. The IBI was most sensitive to species richness and composition metrics, and percentage of benthic fluvial specialists when IBI score was in the >40-60 category. IBI values in the >60-80 category were most sensitive to trophic function metrics, percentage of intolerant species, and density. Because only 2 IBI scores were in the >80 category, we were not able to assess IBI metric sensitivity for scores > 80.

Application of the IBI

IBI scores for the 7 study sites based on Travnichek and Maceina's (1994) data were consistent with their original observations of reduced species richness, diversity, and density at regulated Piedmont sites in the Tallapoosa River compared to unregulated sites (Table 4). Similarly, IBI scores for the Coastal Plain sites accurately reflected differences observed by Travnichek and Maceina (1994). The IBI score for the downstream Coastal Plain site (RCP2) was considerably

higher than the score for the upstream site (RCP1), located about 2 km downstream from Thurlow Dam. These results suggest that the IBI is sensitive enough to accurately summarize differences in fish assemblages in the Tallapoosa River.

IBI scores from the 3 available data sets were variable among years but patterns were evident (Table 4). IBI scores for UP2 were consistently higher than scores for RP1 and RP3 and, in 2 of the 3 data sets (1990-1992 and 1995), than the score for RP2. This result was consistent with our subjective ranking of site UP2 as least impacted by human activity (i.e., flow is unregulated and local riparian disturbance is minimal at this site, whereas UP1 is largely bordered by pasture). The rank order of IBI scores for regulated Piedmont sites was inconsistent across years but the range of IBI scores among regulated sites was consistently low (4.16-5.83). At Coastal Plain sites the IBI score for RCP2 was higher than the score for RCP1 in 1990-1992 and 1995. The low IBI score for RCP2 in 1994 was largely attributable to a lower percentage of intolerant species.

IBI scores were higher in 1995 than in 1994 at 6 of the 7 sites. We speculate that the observed differences in IBI scores were related to differences in discharge between years. Mean hourly discharge for the Tallapoosa River measured near Montgomery, Alabama, was higher in 1994 than in 1995 during spring (136 m³/sec vs. 75 m³/sec) and summer (233 m³/sec vs. 46 m³/sec). High spring and summer discharges may have resulted in reduced reproductive success and survival during 1994 compared to 1995.

TOP OF PAGE

Conclusions

Simplification of biological data is the goal in the use of indices such as the IBI (Gerritsen 1995). Thus, at sites where low IBI scores suggest possible environmental degradation, additional sampling and analyses may be required to identify the nature and extent of changes in the ecosystem. The IBI represents a potentially useful tool for the resource manager because it summarizes information about fish communities into a single value that is more easily interpretable, especially by non-biologists, than more complex analyses. The index may also present a more accurate assessment of system function than individual measures such as species richness. For example, in 1994 we collected more species at RP1 than at any other site although abundances and overall IBI were low. This suggests species richness alone would be a poor indicator of biotic integrity at a site experiencing severe flow fluctuations and possibly influenced by variable colonization and recruitment processes.

Additional impacts on aquatic resources are inevitable given the rapid growth of the human population in the Southeast. For example, additional water development projects are being considered in the upper portion of the Tallapoosa River system to meet water supply demands for the metropolitan Atlanta area. Long-term sampling programs will be necessary to detect and respond to declining biological integrity in our rivers; an IBI similar to that formulated in this study, and refined by incorporation of additional data, could prove useful for monitoring biotic changes in southeastern river systems resulting from watershed and streamflow alteration.

TOP OF PAGE

LITERATURE CITED

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 Am. J. Fish. Manage. 6:418-429.

Bain, M. B., J. T. Finn, and H. E. Booke. 1985. A quantitative method for sampling riverine microhabitats by electrofishing. North Am. J. of Fish. Manage. 5:489-493.

Berkman, H. E., C. F. Rabeni, and T. P. Boyle. 1986. Biomonitors of stream quality in agriculture areas: fish vs. invertebrates. Environ. Manage. 10:413-419.

Etnier, D. A. and W. C. Starnes. 1993. The fishes of Tennessee. Univ. of Tenn. Press, Knoxville, Tenn. 681pp.

Fausch, K. D., J. R. Karr, and P. R. Yant. 1984. Regional application of an index of biotic integrity based on stream fish communities. Trans. of the Am. Fish. Soc. 113:39-55.

Gerritsen, J. 1995. Additive biological indices for resource management. J. North Am. Benthol. Soc. 14:451-457.

Karr, J. R. 1981. Assessment of biotic integrity using fish communities. Fisheries 6:21-27.

Karr, J. R. and D. R. Dudley. 1981. Ecological perspective on water quality goals. Environ. Manage. 5:55-68.

Karr, J. R., K. D. Fausch, P. L. Angermeier, P. R. Yant, and I. J. Schlosser. 1986. Assessing biological integrity in running waters: a method and its rationale. Ill. Nat. Hist. Surv. Spec. Publ. 5, Urbana, Ill. 28pp.

Lee, D. S., C. R. Gilbert, C. H. Hocutt, R. E. Jenkins, D. E. McAllister, and J. R. Stauffer, Jr. 1980. Atlas of North American freshwater fishes. N. C. State Mus. of Nat. Hist., Raleigh, N.C. 867pp.

Lyons, J. 1992. Using the index of biotic integrity (IBI) to measure environmental quality in warmwater streams in Wisconsin. Gen. Tech. Rep. NC-149. N.C. For. Exp. Sta., U. S. 51pp.

Miller, D. L., P. M. Leonard, R. M. Hughes, J. R. Karr, P. B. Moyle, L. H. Schrader, B. A. Thompson, R. A. Daniel, K. D. Fausch, G. A. 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. Can. J. of Fish. and Aquat. Sci. 51:1804-1822.

Oberdorff, T. and R. M. Hughes. 1992. Modification of an index of biotic integrity based on fish assemblages to characterize rivers of the Seine-Normandie basin, France. Hydrobiologia 228:117-130.

Osborne, L. L., S. L. Kohler, P. B. Bayley, D. M. Day, W. A. Bertrand, M. J. Wiley, and R. Sauer. 1992. Influence of stream location in a drainage network on the index of biotic integrity. Trans. of the Am. Fish. Soc. 121:635-643.

Pierson, J. M., R. S. Krotzer, and E. J. Tyberghein. 1986. Fishes from the upper Tallapoosa River and three of its major tributaries in Alabama. J. of the Ala. Acad. of Sci. 57:1-18.

Simon, T. D. and J. Lyons. 1995. Application of the index of biotic integrity to evaluate water resource integrity in freshwater ecosystems. Pages 243-260 in W. S. Davis and T. D. Simon, eds. Biological assessment and criteria: tools for water resource planning and decision making. Lewes Publishers, Chelsea, Mich.

Steedman, R. J. 1988. Modification and assessment of an index of biotic integrity to quantify stream quality in southern Ontario. Can. J. of Fish. and Aquatic Sci. 45:492-501.

Swift, C. C., C. R. Gilbert, S. A. Bortone, G. H. Burgess, and R. W. Yerger. 1986. Zoogeography of the freshwater fishes of the United States: Savannah River to Lake Pontchartrain. Pages 213-324 in C. H. Hocutt and E. O. Wiley, editors. The zoogeography of North American freshwater fishes. John Wiley and Sons, New York.

Travnichek, V. H. and M. J. Maceina. 1994. Comparison of flow regulation effects on fish assemblages in shallow and deep water habitats in the Tallapoosa River, Alabama. J. of Freshwater Ecol. 9:207-216.

TOP OF PAGE