INFERRING CAUSES OF BIOLOGICAL IMPAIRMENT IN APPALACHIAN STREAMS: WATERSHED-BASED PROBLEM FORMULATION AND INTEGRATION OF MULTIPLE LINES OF EVIDENCE

Human activities such as mining, logging, agriculture and residential development have caused significant biological degradation to many streams of West Virginia, USA. Employing benthic macroinvertebrates as biological indicators of stream health, the West Virginia Department of Environmental Protection (WVDEP) has identified streams across the state that do not meet aquatic life use designations. Therefore, these streams are considered biologically impaired. The development of Total Maximum Daily Loads (TMDLs) is required for all biologically-impaired streams within the state and mandates the identification of stressors to the biological community, so that pollutants can be controlled in each watershed. EPA's Stressor Identification guidance was used to identify and rank physical, chemical, and biological stressors that may have caused impairments to the aquatic community. This process involved the analysis of all available water quality, habitat, physical, biological, historical, anecdotal, and observational data to infer the likely causes of impairment for each stream. A comprehensive conceptual model was developed that provides the linkage between potential impairment causes, their sources, and the pathway by which each stressor can impact the benthic macroinvertebrate community. Data were analyzed using established water quality standards and stressor-response threshold values were developed based on statistical analysis and reference population data. Quantitative data were plotted and analyzed spatially using a “geo-order” format of assigning relative positions to sampling locations from downstream to upstream for each impaired stream and its tributaries within a subwatershed. Watershed characteristics (e.g. land use and soils), point source inventories, site observations, and other lines of evidence were included in the analysis to identify sources of the stressors.Stressor Identification required the integration of watershed-based conceptual models of impairment, field biological and chemical monitoring databases, empirical models of biological impairment, and ecotoxicological principles in a strength-of-evidence approach to infer causes of impairment. Candidate causes included known toxic contaminants (metals), conventional pollutants (organic and nutrient enrichment), sedimentation, habitat degradation, and ionic concentration (conductivity). Analysis of some candidate causes was modified by the measures available that documented them. Candidate causes were screened to eliminate those shown not to co-occur with effects. Remaining candidate causes were ranked according to considerations of evidence within each watershed, as well as from statewide empirical models and from other published sources. Strongest inferences were obtained where the independent predictive model agreed with within-watershed observations of stressor measures. Final stressor determinations for each watershed will be used for the development of management plans (TMDL implementation).