Invertebrate communities delineate hydro-ecoregions and respond to anthropogenic disturbance in East-Amazonian streams

Abstract

Many tropical regions lack models predicting the biological and environmental conditions expected in any given area, thus precluding the implementation of reference condition-based water policies. We focused on streams of French Guiana, and tested two predictions: geomorphology determines ecological sub-regions that have typical invertebrate communities, and diversity declines as anthropogenic pressure increases. Sixty-five stream sites were sampled for benthic invertebrates and physical–chemical variables across various watersheds. We used the Self-Organizing Map algorithm (neural network) to model relationships between invertebrate communities and environmental variables. Sites characterized by invertebrate communities clustered into two major subsets matching French Guiana’s hydro-ecoregions: the coastal alluvial plain characterized by recent sediment and low elevations, and the Guiana Shield characterized by an eroded rocky substrate and dense rainforests. Changes in community composition, and to a lesser extent taxonomic richness within each sub-region revealed ecological impacts of gold mining and logging, further clustering hydro-ecoregions into subsets of reference and impaired sites. Further analyses would, however, be needed to identify tipping points between natural and disturbed states, especially in remote headwater streams where gold mining had the harsher impact upon freshwater diversity, making upstream communities resembling the most downstream impacted ones.