Community analysis in stream biomonitoring: what we measure and what we don't

Diatom assemblages from 83 epilithic samples taken from the Mesta River, Bulgaria, were regressed against three sets of predictor variables, i.e. environmental, spatial, and temporal. Redundancy analysis (RDA) of species and environmental data explained 36% of the diatom variance and extracted several important gradients of species distribution, associated with a downstream increase in nutrient levels, pH, temperature, and organic pollution. The inclusion of spatial and temporal variables in the RDA model captured additional 24% of the diatom variance and revealed three more gradients, a spatial gradient represented by higher order polynomial terms of latitude and longitude, and two temporal gradients of annual and seasonal variation. Partial RDAs demonstrated that the unique contribution of each predictor set to the explained diatom variance was the highest in the spatial dataset (16%), followed by the environmental (9%), and the temporal (7%) datasets. The remaining 28% of the variance was explained by the covariance of the predictor sets. This suggests that in biomonitoring of single stream basins, the cheap and simple account of space and time would explain most of the variance in assemblage composition obviating the necessity of expensive and time-consuming environmental assessments. The nature of the underlying environmental mechanisms can be easily inferred from the diatom composition itself.