Mechanisms that drive species co-occurrence are poorly documented for intermittent rivers of semiarid regions. Here, we investigated fish community assembly in response to habitat types and physicochemical conditions ...Mechanisms that drive species co-occurrence are poorly documented for intermittent rivers of semiarid regions. Here, we investigated fish community assembly in response to habitat types and physicochemical conditions in intermittent rivers of the lower Okavango Delta, Botswana. Using Joint Species Distribution Models, we inferred relative influences of environmental filtering and species interactions on patterns of species co-occurrence. Fishes were surveyed from multiple locations during drought and flood phases of the annual hydrological cycle. Species were classified into trophic guilds to facilitate inference about the types of species interactions that influence community structure. Water physicochemistry (depth, temperature, and dissolved oxygen concentration) was significantly associated with species distributions, whereas habitat type (ephemeral vs. permanent) was not significantly associated with species distribution. Controlling for the influence of environmental covariates resulted in three negative residual correlations, of which two involved non-predatory fishes from different trophic guilds, suggesting a behavioral selection of, or enhanced survival within, habitats with different environmental conditions. There was a negative residual correlation between a predator and a potential prey species, suggesting a role for predation mortality or threat in species segregation. Our results demonstrate that using trophic guilds in conjunction with JSDMs can enhance inferences about mechanisms of community assembly.展开更多
文摘Mechanisms that drive species co-occurrence are poorly documented for intermittent rivers of semiarid regions. Here, we investigated fish community assembly in response to habitat types and physicochemical conditions in intermittent rivers of the lower Okavango Delta, Botswana. Using Joint Species Distribution Models, we inferred relative influences of environmental filtering and species interactions on patterns of species co-occurrence. Fishes were surveyed from multiple locations during drought and flood phases of the annual hydrological cycle. Species were classified into trophic guilds to facilitate inference about the types of species interactions that influence community structure. Water physicochemistry (depth, temperature, and dissolved oxygen concentration) was significantly associated with species distributions, whereas habitat type (ephemeral vs. permanent) was not significantly associated with species distribution. Controlling for the influence of environmental covariates resulted in three negative residual correlations, of which two involved non-predatory fishes from different trophic guilds, suggesting a behavioral selection of, or enhanced survival within, habitats with different environmental conditions. There was a negative residual correlation between a predator and a potential prey species, suggesting a role for predation mortality or threat in species segregation. Our results demonstrate that using trophic guilds in conjunction with JSDMs can enhance inferences about mechanisms of community assembly.