| Abstract: | Ecologists have long realized that stable species richness values can mask rapid turnover in species composition. Because turnover occurs as a consequence of both local and regional processes, understanding the responsible factors provides insight on processes influencing community structure at different scales. Despite the insights to be gained from data on species turnover, they remain relatively uncommon. We present data on the interannual turnover in species composition of larval amphibian communities in 37 ponds over seven years. Species composition of a given pond community was highly dynamic; about half of the species that could be found breeding in a particular pond were actually present in a given year. All species participated in this community turnover, but to different degrees. Using a model selection approach, we show that a statistical model including local environmental factors (pond area, hydroperiod, and canopy cover) and pond connectivity on the landscape provided the best predictions of turnover. Averaged parameter estimates were significant for area, hydroperiod, and connectivity and these same variables were identified by hierarchical partitioning as having significant independent effects on turnover. Area and hydroperiod were negatively related to turnover, whereas connectivity was positively related to turnover. Additionally, the average fraction of years a species was present in a pond was positively correlated with average local population size, but even more strongly correlated with regional population size, suggesting both local and regional influences on turnover. Of the measured biotic factors (biomass of fish, invertebrate predators, anuran and caudate larvae), presence of fish was the only factor that significantly affected rates of turnover. Several mechanism could be responsible for changes in species composition (species extinctions, skipped breeding and movement of choruses), but extinctions appear to be the major cause of turnover. These results have important implications to understanding long‐term persistence of species on landscapes and the causes of patterns in species richness on environmental gradients. |
