Environment and Space Rule, but Time Also Matters for the Organization of Tropical Pond Metacommunities

Abstract

Metacommunities are dynamic systems, but the influence of time independently of environmental change in their configuration has been rarely considered. In temporary ponds, strong temporal effects are expected to influence their metacommunity structure, even in relatively constant environments such as tropical habitats. We therefore expect that time as an independent factor could modulate tropical pond metacommunities, which would be also less affected by niche-related processes than by dispersal-related processes. In addition, good dispersers should be more environmentally structured than bad dispersers, which should be more spatially structured. Finally, the relevance of temporal effects should vary among organisms with different generation times. To test these hypotheses, 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 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 for metacommunity organization, we used variation partitioning with distance-based redundancy analyses for each group of organisms. 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 organisms and the magnitude of the temporal effects, 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.