To improve our understanding on the temporal aspects of metacommunity structure, we focused on benthic macroinvertebrates collected seasonally (i.e., wet, drying, dry and rewetting seasons) in Dongting Lake, a large subtropical floodplain lake in China. We employed the elements of metacommunity structure (EMS) framework and variation partitioning to examine whether metacommunity structure and its underlying mechanisms vary among seasons with distinct biotic and abiotic features. We found gradual shifts in the main assembly mechanisms throughout the year, from apparent spatial structuring (potentially indicative of mass effects) in the wet season to more environmental filtering dynamics in the dry season. When the degree of connectivity was high in the wet season, the benthic metacommunity was characterized by nested structure associated with clumped species loss, and was shaped mainly by spatial processes. However, quasi-Clemensian structure was assigned to metacommunities in the transitional seasons with intermediate connectivity, during which environmental variables were more important than spatial factors in describing community structure. When the degree of connectivity was low in the dry season, the benthic metacommunity displayed Clementsian structure, which was configured solely by environmental variables. The rapid shifts in metacommunity dynamics between seasons mainly result from the considerable changes in the hydrological conditions of Dongting Lake, as the studied system varies from lacustrine to fluvial phases within a single year. Taken together, our results revealed that taking temporal aspects into account gives a better insight into metacommunity organization, especially when the studied systems embrace remarkable variability in hydrological regimes.