Historical (phylogenetic) biogeography and community ecology were once integrated as part of the broader study of organismal diversity, but in recent decades have become largely separate disciplines. This is unfortunate because many patterns studied by community ecologists may originate through processes studied by historical biogeographers and vice versa. In this study, we explore the causes of a geographic pattern of community structure (habitat use) in the emydid turtle assemblages of eastern North America, with more semi-terrestrial species of the subfamily Emydinae in the north and more aquatic species of Deirochelyinae in the south. Specifically, we address the factors that prevent northern emydines from invading southern communities. We test for competitive exclusion by examining patterns of range overlap, and test for the role of niche conservatism using analyses of climatic and physiological data based on a multilocus molecular phylogeny. We find no support for competitive exclusion, whereas several lines of evidence support the idea that niche conservatism has prevented northern emydines from dispersing into southern communities. Our results show how understanding the causes of patterns of historical biogeography may help explain patterns of community structure.