Metacommunity detectives: Confronting models based on niche and stochastic assembly scenarios with empirical data from a tropical stream network

Metacommunity models predict that species richness and composition patterns in communities are determined predominantly by environmental selection and dispersal, with speciation and drift playing a lesser role. In the tropics, our understanding about these processes comes almost purely from empirical data; there is lack of formal confrontation between models and data. Here, we evaluated if a metacommunity simulation model designed for riverine networks could predict insect diversity patterns observed in a tropical stream network within the Atlantic Forest biome. Our individual-oriented model simulated biological processes, including dispersal of adults, oviposition, dispersal of larvae, colonisation and mortality. We used this model to simulate communities considering three assembly scenarios (niche, niche–stochastic and stochastic) and three flight dispersal ability groups (low, medium and high). To parameterise the model, we used data from 97 riffle communities, distributed along a stream network, and biological data from the literature. For the high dispersal ability group, none of the assembly scenarios significantly predicted the observed richness and community dissimilarity. The niche–stochastic assembly scenario, however, provided a better approximation of observed richness and community dissimilarity for the medium and low dispersal ability groups. Our results show that deterministic and stochastic processes combine to better explain observed aquatic insect diversity patterns. Although simulations were limited to similar explanatory power to studies using pattern-oriented approaches, our findings highlight the promise of a more predictive use of metacommunity simulation models.