potheses, defining well-resolved nodes of the phylogeny, and identifying unresolved relationships, thereby focusing our efforts more efficiently.Here, we employ a statistical approach, using both parsimony and likelihood analyses of molecular sequence data to test a suite of previously proposed hypotheses of phylogenetic relationships in marmots (Marmota). A sequential phylogenetic estimation procedure was used that culminated in a maximum likelihood analysis utilizing a model of sequence evolution with parameters estimated from the data.Marmots are large terrestrial rodents found today throughout much of northern Eurasia and North America, including the Bering Strait region of western Alaska and eastern Siberia (Fig. 1). Their ecology and ethology has been studied extensively (e.g., There is a growing movement in systematics from simply making estimates of phylogeny to hypothesis testing and reliability estimation (Huelsenbeck and Rannala, 1997). This statistical perspective allows more precise delineation of which relationships are well understood and which need additional investigation. It also promotes the generation of explicit evolutionary and biogeographic models while providing the tools to reject hypotheses. These developments promise to accelerate our understanding of evolution by improving our hy- Syst. Biol. 48(4):715-734, 1999 Abstract.-There are 14 species of marmots distributed across the Holarctic, and despite extensive systematic study, their phylogenetic relationships remain largely unresolved. In particular, comprehensive studies have been lacking. A well-supported phylogeny is needed to place the numerous ecological and behavioral studies on marmots in an evolutionary context. To address this situation, we obtained complete cytochrome (cyt) b sequences for 13 of the species and a partial sequence for the 14th. We applied a statistical approach to both phylogeny estimation and hypothesis testing, using parsimony and maximum likelihood-based methods. We conducted statistical tests on a suite of previously proposed hypotheses of phylogenetic relationships and biogeographic histories. The cyt b data strongly support the monophyly of Marmota and a western montane clade in the Nearctic. Although some other scenarios cannot be rejected, the results are consistent with an initial diversification in North America, followed by an invasion and subsequent rapid diversification in the Palearctic. These analyses reject the two major competing hypotheses of M. broweri's phylogenetic relationships-namely, that it is the sister species to M. camtschatica of eastern Siberia, and that it is related closely to M. caligata of the Nearctic. The Alaskan distribution of M. broweri is best explained as a reinvasion from the Palearctic, but a Nearctic origin can not be rejected. Several other conventionally recognized species groups can also be rejected. Social evolution has been homoplastic, with large colonial systems evolving in two groups convergently. The cyt b data do not provide unambiguous resolution of se...