Temporal variation in phytoplankton beta diversity patterns and metacommunity structures across subtropical reservoirs

Abstract

Summary Analysing the beta diversity components and spatial patterns of species distribution may provide key insights into how local communities respond to human‐induced environmental changes. Indeed, analysing temporal variation in spatial patterns of species distribution may reveal trends of biotic homogenisation and its likely causes. We used an eight‐year phytoplankton data set generated from to 17 reservoirs in Brazil to analyse the temporal variability in beta diversity and its turnover and nestedness components. This data set was also used to determine metacommunity structures across these subtropical reservoirs. We tested whether there was any long‐term temporal trend in beta diversity that might be indicative of biotic homogenisation or differentiation. We also tested how temporal variation was related to climatic conditions, environmental heterogeneity, productivity and cyanobacterial dominance. Lastly, we verified whether the phytoplankton metacommunity showed non‐random structure. We did not find supporting evidence for biotic homogenisation among the reservoirs. Rather, we did find that nestedness decreased during the study. Environmental heterogeneity was the main variable positively related to phytoplankton beta diversity, while high ammonium concentration and cyanobacterial abundance were negatively correlated with spatial variation among the reservoirs. Despite the noticeable temporal variation in metacommunity structures, the phytoplankton species responded similar to latent environmental gradients. Clementsian patterns found mainly in fall and winter were consistent with differences in species composition between sites, reinforcing the role of environmental filtering in driving changes in these metacommunities. Our results suggest that eutrophication control is essential to prevent biotic homogenisation, at least in our study system. We also highlight that metacommunity patterns detected in snapshot surveys should not be extrapolated temporally.