A new predictive model (MINASPACS) for spatially extensive biological assessments of southeastern Brazilian streams

Abstract

Freshwater ecosystems are threatened by flow regulation, sedimentation, habitat degradation, non-native species, and water pollution. These disturbances have led to global losses of biodiversity and habitats. Therefore, it is essential to evaluate the ecological condition of freshwater ecosystems to promote effective management practices. Quantitative predictive models based on multivariate analyses of taxa richness are recognized ecological tools that can facilitate the monitoring and managing of freshwater ecosystems worldwide. However, few studies have used this approach to assess tropical rivers and streams. By evaluating predictive models, we can assess their usefulness for determining water-body taxonomic richness. We built a RIVPACS-type model based on macroinvertebrate assemblages (MINASPACS), for spatially extensive taxa richness assessments of Minas Gerais state streams, southeast Brazil. As a second objective, we assessed the sensitivity of the MINASPACS to human-induced disturbances affecting Minas Gerais streams through the relative risk (RR) approach. The MINASPACS model was trained with biological and environmental data from 78 reference sites and showed good accuracy (R² ​> ​0.6, SD O/E ​= ​0.16). We found that percent of urban infrastructure, percent of catchment anthropogenic land use, Turbidity, Total Nitrogen, and Total Phosphorus represented significant risks to the taxa richness of Minas Gerais streams. Because of its accuracy, sensitivity, and use of map-level predictor variables, our model provides a clear, simple, and defensible measure of stream macroinvertebrate taxa richness across diverse biomes.