You are here:
EMPIRICAL RISK RELATIONSHIPS BETWEEN INVERTEBRATES, HABITAT AND WATER QUALITY IN MAIA DATA SETS
Citation:
Benjamin, R. B., F A. Fulk, AND S M. Cormier. EMPIRICAL RISK RELATIONSHIPS BETWEEN INVERTEBRATES, HABITAT AND WATER QUALITY IN MAIA DATA SETS. Presented at Society of Environmental Toxicology and Chemistry, Baltimore, MD, November 11-15, 2001.
Impact/Purpose:
The purpose of this research project is to provide methods, tools and guidance to Regions, States and Tribes to support the TMDL program. This research will investigate new measurement methods and models to link stressors to biological responses and will use existing data and knowledge to develop strategies to determine the causes of biological impairment in rivers and streams. Research will be performed across multiple spatial scales, site, subwatershed, watershed, basin, ecoregion and regional/state.
Description:
A technique for developing a non-weighted risk index, originally developed for use with Ohio fish assemblage data, was applied to invertebrate, habitat and water quality data collected from Mid-Atlantic streams of the U.S. during 1997-98. Multiple habitat and water quality variables were treated as predictors of individual invertebrate assemblage variables. Conventional multivariate methods were poorly suited for analyzing the observed MAIA data distrubutions. Predictor and response variables were rank ordered and sorted into high, moderate or low categories. Most responses were significantly associated with six to nine predictors which were selected using a combination of discriminant analyses and Chi-Square tests. For each response variable, the Maximum Response Profile (MRP) was defined as the combination of significant predictor categories resulting in the greatest frequencies of high-response categories (one-on-one predictor and response comparisons). A Risk Index (RI) was computed for each response variable at each sampling location. Each RI was the total number of significant, categorized predictors at each time/location not falling into the MRP. Increased RIs were consistently associated with increased frequencies of low-response categories. Results were different between data collected from riffles or pools, and some response variables could not be satisfactorily explained using this approach. Successful application of this empirical technique to both fish and invertebrate assemblages suggests it has broad applicability.