Aim Our goal was to relate the phytoplankton metacommunity to its possible determinants in a micro watershed: (I) determinants related to landscape-scale filtering, (II) determinants referring to local microhabitat filtering, (III) determinants referring to previous colonization, and (IV) determinants representing three different dispersal routes. Methods Eight sampling stations were selected along the Cascavel River watershed, located in the state of Paraná, Brazil. Samples were collected quarterly for three years. All phytoplankton samples were quantitatively analyzed to determine the density of the metacommunity. In addition, it was characterized the landscape in terms of land use and occupation, and environmental characterization in terms of physical and chemical variables of the water. All data underwent relevant statistical analysis, where variance partitioning was carried out using partial RDA models, with prior selection of predictor variables, to estimate the relative role of each predictor in the community. We also compared three possible dispersal routes: “Asymmetric Eigenvector Map” (AEM), “Overland” and “Watercourse”. Results It was found that the metacommunity was best explained by “asymmetric eigenvector mapping” (AEM), indicating that because it is a small spatial scale the high connectivity between the sampling stations enables species to disperse overland as well. The different filters act together and depend on rainfall variation. Besides fluctuating temporally, the influence of these mechanisms is subject to which dispersal hypothesis is being considered. Conclusions At the watershed scale, we argue that small-scale processes should be considered, since they homogenize the landscape and consequently leave the environmental gradient similar between sampling stations. In addition, the connectivity of colonization patches is essential to understand the behavior of microalgae that have a high dispersal capacity and are not restricted only to the river course.