Species richness on islands in time: variation in ephemeral pond crustacean communities in relation to habitat duration and size

Ephemeral ponds are doubly insular habitats in that they are discrete in time as well as in space. Predicting species richness based on pond size has been attempted using measures of both spatial extent and habitat duration, but habitat qualities alone only drive community composition under the species-sorting metacommunity paradigm. We tested the hypothesis that community composition in temporary ponds is driven by species sorting due to pond duration. In order to eliminate bias due to under-sampling, we sampled 34 pools distributed among 3 complexes in every ponding event over a period of six years, and identified every individual for the microcrustacean taxa. Our data were consistent with most of the predictions of the species-sorting hypothesis. There was a clear pattern of difference between ponds in species richness as well as higher species richness in years with higher rainfall. The set of crustacean species that we found in the pools was highly significantly nested across the region, but not necessarily within localized groups of ponds (complexes). We also found differences in community composition among complexes. Pond depth was the best predictor of species richness when data were summarized over the whole study, but in one year with unusually high rainfall, pond area and hydroperiod were significant but depth was not. We did find some species in all ponds. It is likely that given their short development time, ponds do not differ in habitat quality for these few species. These results taken together emphasize the variability inherent in ephemeral pond ecology, with detectable differences in crustacean communities and the factors influencing them between years as well as between ponds, and at scales of meters as well as kilometers. Although our data provide further evidence that species sorting on pond permanence is an important factor structuring temporary pond crustacean communities, our assumption that dispersal is not limiting still needs to be tested.