Invertebrate diversity in relation to chemical pollution in an Umbrian stream system (Italy).

We used self-organizing maps (SOM, neural network) to bring out patterns of benthic macroinvertebrate diversity in relation to river pollution. Fourteen stations were sampled over various seasons in the Nestore drainage basin (Central Italy) and characterized for macroinvertebrate communities, nutrient and heavy metal concentrations. Physicochemical variables were introduced into a SOM previously trained with macroinvertebrate data. Patterns of communities matched spatial and seasonal changes in environmental conditions, including water chemistry related to economic activities in the catchment. Although our analyses did not allow us to establish the specific effect of any given environmental parameter upon macroinvertebrate community composition based on the field study, they enabled us to map the ecological health of river ecosystems in a readily interpretable manner.