Metacommunities are dynamic systems, but the influence of time independently of environmental change in their configuration has been rarely considered. Temporary ponds are excellent ecosystem models, as they have well-defined boundaries in time and space; their communities are relatively isolated through a landscape matrix, and the progress of time leads to major changes through ecological succession and in habitat suitability related to hydroperiod dynamics. Therefore, strong temporal effects are expected to influence their metacommunity structure. We surveyed 30 temporary ponds along the dry tropical region of western Costa Rica and Nicaragua at three different moments of their hydroperiod: shortly after the infilling of the water bodies, at the middle of the hydroperiod and just before desiccation. We obtained data on 56 environmental variables (including limnological, climatic, landscape, hydrogeomorphological and conservation state), and used geographic coordinates to build spatial variables (Moran Eigenvector Maps). We collected biological samples and estimated the specific abundance of phytoplankton, zooplankton and benthic invertebrates. To evaluate the relative role of environmental, spatial and temporal (sequential sampling season) effects in metacommunity organization of the 646 species found, we used variation partitioning with distance-based redundancy analyses for each group of organisms. We repeated the analyses for selected taxonomic groups, separately for each sampling season, to check how the variance explained by spatial and environmental factors varied between sampling periods. The inclusion of time in the analysis highlighted that pure temporal effects explained part of metacommunity variance in almost every group, being as important as spatial or even environmental effects for some groups of organisms. In contrast to the assumed low environmental constraints in tropical areas (i.e. high and stable temperatures), we found strong environmental effects. Passive dispersers were more influenced by environmental factors than active ones. We also found a positive relationship between the body size of the different groups of organism and the magnitude of the temporal effect, interpreted as related to generation time. Finally, when analyzing each sampling period separately, we found differences in the relative role of environment and space at different sampling periods, showing that snapshot surveys may not be representative of highly dynamic metacommunities.