|
2B - Natural Stressors and Toxicants (TUP/66) Ohio 2: Using Species Centered Regression Modeling for Multi-Stressor Risk Evaluation. de Zwart, Dick1, Posthuma, Leo1, Dyer, Scott2, Hawkins, Charles3, 1 RIVM, Bilthoven, The Netherlands2 The Procter & Gamble Company, Cincinnati, Ohio3 Utah State University, Logan, Utah ABSTRACT- US State and federal agencies provided data on fish species composition in Ohio rivers, combined with information on habitat characteristics and chemical exposure. By means of Poisson General Linear Modeling (GLM) regression, the abundance of individual fish species is expressed as a function of a variety of different habitat and chemical stressors (predictors). In order to reduce the number of predictors, modeled or measured concentrations of individual toxic chemicals were groupwise (industrial origin, household products and eventually pesticides) converted to a cumulative measure of toxic risk per sampling site by applying Species Sensitivity Distributions (SSD), availability assessment and mixture toxicity calculus. This eco-epidemiological approach eventually quantifies the response of single species with respect to the individual stressors. For every stressor and site, sensitive species can be identified. Up till moderate stress levels, these species can be replaced by more tolerant species. It is also possible to identify species that demonstrate an optimum association to the range of field values of an individual stressor. The method in itself does not yield any information on the absolute magnitude of the ecological effects. Once the species observed in the field are compared with information on the species expected (e.g. RIVPACS, poster Ohio 1), an analysis of the contributions of the individual stressors to the GLM-models allows for absolute quantification and relative attribution of the effects at a site (poster Ohio 3). Key words: mixture toxicity, ssd, multiple stressors, eco-epidemiology |
