A probabilistic model characterizing fish assemblages of French rivers: a framework for environmental assessment

1. Management of running waters and assessment of water quality trends requ ire the use of biological methods. Among potential indicators, fish assembl ages are of particular interest because of their ability to integrate envir onmental variability at different spatial scales. 2. The French Water Agencies and the Ministry of the Environment initiated a research programme to develop a fish-based index that would be applicable nation-wide. Such an index should encompass the relative importance of geo graphic, ecoregional and local factors influencing the distribution of rive rine fishes. 3. An effective way of using the information available from fish assemblage s to establish such an index is through the use of the 'reference condition approach' which involves testing a fish assemblage exposed to a potential stress against a reference condition that is unexposed to such a stress. 4. Logistic regression procedures were applied, using a fish data set of 65 0 reference sites fairly evenly distributed across French rivers and define d by some easily measured regional and local characteristics, to elaborate the simplest possible response model that adequately explains the observed patterns of occurrence for each species of a fish assemblage at a given sit e of any given river. This allows us to predict a 'theoretical' assemblage at a site. 5. The models were validated using a second independent data set of 88 refe rence sites. Using a third data set of 88 disturbed sites, the observed ass emblages were then compared against the reference condition as defined by t he 'theoretical' assemblages. The amount of deviation between the expected and observed assemblages within these sites is used as a measure of degrada tion. 6. This approach could be used as a framework for adapting and calibrating a multi-metric index, thereby serving as a practical technical reference fo r conducting cost-effective biological assessments of lotic systems.