In the Anthropocene, river networks are globally threatened by human activities, creating dramatic consequences for aquatic biodiversity. Permanent and temporary fragmentation by manmade structures and drying events, respectively, are both increasing in rivers worldwide. Although both of these fragmentation types can limit species dispersal, their individual and combined effects in shaping metacommunity dynamics at both local (site-level) and regional (network-level) scales have not been explored. Here, we examined whether processes structuring aquatic invertebrate metacommunities vary through time in response to flow variability in a river network affected by drying and severe fragmentation by manmade structures. We also compared the relative influences of permanent and temporary fragmentation on metacommunity dynamics and hypothesized that permanent fragmentation would be the primary determinant of metacommunity dynamics. We conducted an intensive sampling effort (30 sampling sites  6 dates) and measured 12 local environmental variables and 4 spatial distances to assess the relative importance of niche-and dispersal-based processes on benthic invertebrate metacommunities across hydrological phases. Spatial distances considered here were overland, network, and 2 fragmentation distances integrating the permanent fragmentation by manmade structures and temporary fragmentation by drying. We used Mantel tests to identify relationships between community dissimilarities and environmental and spatial distance matrices. We identified temporal variability in metacommunity processes with a predominant role of dispersal and no effect of niche-based processes. Metacommunities were shaped primarily by permanent fragmentation, whereas we detected no effect of fragmentation by network-scale drying. Dispersal-based metacommunity processes varied over time, following the wet-dry cycles that characterize dynamic river networks. Our results suggest the importance of key hydrological phases that should be incorporated into conservation perspectives. In addition, we highlight the need to quantify context dependency in metacommunity studies to optimize biodiversity conservation strategies in river-network ecosystems.