Previsione della fauna ittica mediante reti neurali artificiali.

Predicting the structure of fish assemblages in rivers is a very interesting goal in ecological research, both from a purely theoretical point of view and from an applied one, for instance when river management strategies are to be defined or when the implementation of the Directive 2000/60/EC is taken into account. Models for estimating the abundance or the probability of presence of fish species have been developed using different approaches. Although some conventional statistical tools provided interesting results, the application of artificial neural networks has recently outperformed those techniques in terms of accuracy and ease of development. Artificial neural networks are especially effective in reproducing the complex, non-linear relationships that link fish species to environmental variables. Recent developments of the artificial neural network training procedures, specifically aimed at solving ecological problems, allowed to optimize the prediction of species assemblages. The improvement in prediction involves not only the accuracy of the models, but also their ecological consistency. Some results about models for fish assemblages in the rivers of the Veneto region (Northern Italy) are presented and their potential applications are discussed.