Influence of non-point source pollution on riverine fish assemblages in South West France.

The relationship between non-point source pollution (NSP) and fish assemblages in the Garonne basin, SW France was studied. Two independent data sets were coupled, one containing 20 physico-chemical variables and another containing 40 fish species in 84 study sites. Species were classified in guilds according to their feeding habitat and their diet composition. The physico-chemical variables were log-transformed and standardized for a factor analysis in which they were grouped into four factors which accounted for 80% of the total variability. These were named according to factor loadings (i.e. a measure of the variance of a given variable) whose absolute values were larger than 0.5. Hence, the first factor (F1) was formed by variables linked to NSP, most notably by sodium, chloride, potassium, orthophosphates, nitrites and chemical oxygen demand. The second factor (F2) was related to alkalinity (i.e. bicarbonates, calcium, conductivity and pH). The third factor (F3) included oxygen saturation rate and dissolved oxygen, and F4 combined both temperature and flow. Factor scores (i.e. weighted sums of the original variables) were then introduced in stepwise multiple regression models as explanatory variables of log-transformed fish species richness of trophic guilds. The NSP factor was significant (p < 0.05) for the following models: benthic omnivores (r2 = 0.66), all species (r2 = 0.65), total benthic species (r2 = 0.63), total water-column species (r2 = 0.57), benthic invertivores (r2 = 0.32) and water-column invertivores (r2 = 0.16). The guilds for which NSP was not significant were water-column omnivores, water-column piscivores and benthic detritivores. Thus, there was evidence of an inversely proportional association, though not causation, of NSP with species richness of riverine fish trophic guilds on a large spatio-temporal scale. Fish assemblages may respond in different ways to NSP depending on their species composition, on the region and on the scale, and not only to physico-chemical properties of water.