The relative importance of environmental variables at different spatial scales in structuring a stream fish metacommunity

Abstract

The use of variables from different spatial scales has been recommended to increase the explanatory power of environmental models that seek to explain community structure. In this study, using electrofishing data, we tested the relative importance of local (stream habitat characteristics) and landscape (percentage cover of pasture, agriculture, natural vegetation, and water bodies) variables on the structure of fish communities in 13 streams located in veredas, which are small wetlands in the Brazilian savannahs. We used ordination methods to reduce the dimensionality of the environmental data matrices and used the first two axes from these analyses as explanatory variables in variation partitioning models. In these models, we also used the geographic coordinates of the sampling sites to control for spatial effects. The matrix of species abundance by site was our response variable. We found that only local and spatial variables significantly explained the structure of fish communities in wetlands streams. These results suggest that the effect of landscape variables should occur mainly when the variation in land use is more pronounced (when streams are distributed over watersheds with markedly contrasting land uses). Local variables, such as macrophytes, coarse woody debris, and leaf litter were the main predictors in our study, indicating the importance of habitat structure on fish communities. We emphasise that the small wetlands of the Cerrado biome are of great value for biodiversity conservation and are currently highly threatened by human activities. Currently, deforestation for the development of agriculture and livestock activities is the main threat to these ecosystems.