A METHOD IN MULTIVARIATE-STATISTICS TO ANALYZE ECOSYSTEMS STARTING FROM THEIR SPECIES COMPOSITION

Ecosystems are characterized by many correlated biotic and abiotic variables interacting with the environment and changes in their species composition are considered an early indicator of stress. We present a statistical method, based on maximum entropy techniques, which allows us to characterize ecosystems, choosing the species composition as the set of descriptors. Our method is able to enhance small shifts of descriptors, identifying system changes before they become evident. We present the methodological aspects of the analysis and its applications to two real cases.