Can biological assessments discriminate among types of stress? A case study from the Eastern Corn Belt Plains ecoregion

We investigated the feasibility of using the structure of fish and benthic macroinvertebrate communities to distinguish among major types of stressors (e.g., siltation, nutrient enrichment, and stream structural degradation) using spatially and temporally matched data on stressors and responses. The 19 stressor variables addressed stream chemistry and in-stream habitat and included biological oxygen demand (BOD), total suspended solids, nitrogen, phosphorus, and components of the Qualitative Habitat Evaluation Index. The 42 response variables addressed fish and invertebrate community structure and included many of the component metrics of the Index of Biological Integrity and the Invertebrate Community Index as well as variables specifically calculated for this project. All data were collected between 1988 and 1994 by the Ohio Environmental Protection Agency in the Eastern Corn Belt Plains ecoregion. Prior to analysis, variables were transformed to near normality, and variables significantly correlated with drainage area were fit to regression models and the resulting residuals used in the analyses. Multivariate analyses included factor and discriminant analysis. The first six stressor factors explained 69% of the variation. Discriminant functions formed using the response variables significantly separated site clusters classified into high, medium, and low categories along stressor gradients. Both fish and macroinvertebrate variables were important in distinguishing site categories. For example, percentage Tanytarsini midges and percentage Glyptotendipes were important in distinguishing sites having high and low BOD. Percentage darters was associated with sites having high scores for stream corridor structure and low concentrations of inorganic nutrients, and percentage round-bodied suckers was associated with sites having low BOD and low concentrations of zinc and lead. These results indicate that diagnostic models may be developed that will be useful for site-specific and regional assessments.