|
2B - Natural Stressors and Toxicants (TUP/76) Ohio1: Using Predictive Models to Quantify Biological Condition in Ohio Rivers. Hawkins, Charles1, Dyer, Scott2, Posthuma, Leo3, De Zwart, Dick3, 1 Utah State University, Logan, Utah, USA2 The Procter & Gamble Company, Cincinnati, Ohio, USA3 RIVM, Bilthoven, The Netherlands, The Netherlands ABSTRACT- Comparison of the observed fauna with that expected to occur in the absence of human-caused stress provides a basis for quantifying the biological condition of potentially stressed ecosystems. Predictive models provide one means of specifying the taxa expected to occur at a site by estimating the probabilities of each taxon in the species pool occurring at each site. These probabilities can be used to calculate the ratio of the number of expected taxa observed at a site to the number expected to occur (O/E), which is an intuitive and easily interpretable measure of biological impairment. RIVPACS-type predictive models have been used to assess biological condition in streams based on invertebrate samples but have not been applied to fish assemblages. Here, we describe development of a fish predictive model for streams and rivers in Ohio. The model was constructed from data collected at 114 reference sites of high biological quality. The model was applied to over 2,400 sampling events to quantify biological impairment. Although predictive models do not provide information regarding the cause of impairment, they do identify the taxa missing from a site that should occur. When coupled with Species Sensitivity Distributions for different stressors, predictive models provide a means of identifying those stressors most likely to be causing biological impairment at a site. Key words: SSD, RIVPACS, Multiple Stressors, Eco-epidemiology |
