Species and their geographical distributions, tabulated either from regional faunal and floral monographs or directly from natural history collections, often are used as the basic units of analysis by ecologists and biogeographers. It has been argued that in order for species to be operationally useful units for evolutionary and ecological studies, they need to be recognizable and identifiable as distinct entities. A growing body of molecular phylogeographic studies demonstrates that currently recognized species often are unreliable in their approximation of fundamental evolutionary and geographical units, leading, for example, to proposed usage of molecular-based evolutionarily significant units in lieu of species in conservation biology. We argue that ecologists and bio- geographers should likewise employ evolutionarily significant units as basic units of analysis when evidence clearly indicates that a formally recognized species either fails to convey important evolutionary and geographical information (i.e. includes multiple geographically distinct evolu- tionary lineages) or fails to delineate a natural entity (i.e. does not represent a monophyletic set of populations). We demonstrate the limitations of current species as evolutionary, geographical, and conservation units within the ecologically well-studied North American desert rodent assemblage. We suggest that biotic surveys should be designed to allow the efficient assembly and dissemination of molecular phylogeographic data from ecologically and biogeographically representative systems.