Deterministic and stochastic processes in natural and modified floodplain habitats: consequences for fish community temporal dynamics

Abstract

Habitat degradation is expected to alter community structure and consequently, ecosystem functions including the maintenance of biodiversity. Understanding the underlying abiotic and biotic assembly mechanisms controlling temporal and spatial community structure and patterns is a central issue in biodiversity conservation. In this study, using monthly time series of benthic fish data collected over a three-year period, we compared the temporal community dynamics in natural and modified habitats in one of the largest river-lake floodplain ecosystems in China, the Dongting Lake. We found a prevailing strong positive species covariance, i.e. species abundance changes in the same way, in all communities that was significantly negatively impacted by water nutrients levels. The positive species covariance, which was consistent for both wet and dry years and among habitat types, had significantly negative effects on community stability, which was measured by the average of aggregated abundance divided by temporal standard deviation. In contrast to species covariance, community stability was significantly higher in modified habitats than in natural habitats. Furthermore, our results demonstrated that the ecological stochasticity (i.e. community assembly processes generating diversity patterns that are indistinguishable from random chance) was significantly higher at natural sites than at the modified sites, suggesting that deterministic processes might control the community composition (richness and abundance) at the modified habitat through reducing species synchrony and positive species covariance observed in the natural habitats. When combined, our results suggest that human habitat modification creates environmental conditions for the development of stable benthic fish community in the highly dynamic floodplains, leading to niche-based community and decrease of temporal β-diversity.