Unraveling the ecological factors that control variation in local community structure in space and time is fundamental to metacommunity ecology. In this scenario, environmental filtering and spatial processes are recognized as important drivers of community assembly, yet their relative importance is anticipated to vary for biological communities in different seasons, network positions and organisms with distinct dispersal modes. In this study, we used a dataset (macroinvertebrate communities and environmental variables) collected in different seasons from the Ganjiang River in China to test the above ideas. We divided the whole metacommunity in each season into mainstream communities, tributary communities, strictly aquatic dispersers and aquatic/aerial dispersers, and subsequently used variation partitioning to examine the relative contribution of environmental and spatial factors separately for the overall and decomposed components of the metacommunity. Our results showed that both environmental filtering and spatial processes were important drivers of variation in community structure, yet their explanatory powers varied considerably among seasons. Environmental filtering was the primary driver of metacommunity organization in most scenarios, while the effects of spatial processes surpassing environmental filtering occurred only sporadically. For communities in different network positions, tributary communities were structured by both strong environmental filtering and profound effects of spatial processes via dispersal limitation. However, communities in mainstream sites were mainly determined by environmental filtering, and the effects of spatial processes were almost negligible. Moreover, environmental filtering was clearly more important for aquatic/aerial dispersers, while spatial processes were more influential for strictly aquatic dispersers. We thus concluded that environmental filtering, spatial processes, network position and dispersal mode can interact to regulate metacommunity organization of riverine macroinvertebrates. Considering that the relative contribution of these factors varied among seasons, we strongly uphold the idea that community ecology research should go beyond one-season snapshot surveys in river networks.