Competitive interactions between felid species may limit the southern distribution of bobcats Lynx rufus Vı ´ctor Sa ´nchez-Cordero, David Stockwell, Sahotra Sarkar, Hawoei Liu, Christopher R. Stephens and Joaquı ´n Gime ´nez V. Sa´nchez-Cordero (victor@ibiologia.unam.mx) and J.Gime´nez, Depto de Zoologı´a, Inst. de Biologı´a, Univ. Nacional Auto´noma de Me´xico, Apartado Postal 70-153, Me´xico D. F. 04510, Me´xico.  D. Stockwell and H. Liu, San Diego Super Computer Center, Univ. of California, 9500 Gilman Drive, La Jolla, CA 92093, USA.  S. Sarker, Section of Integrative Biology, Univ. of Texas at Austin, Waggener Hall 316, Austin, TX 78712 1180, USA.  C. R. Stephens, Inst. de Ciencias Nucleares, Univ. Nacional Auto´noma de Me´xico, Apartado Postal 70-153, Me´xico D. F. 04510, Me´xico. Bobcats are opportunistic felids occurring in a diverse range of habitats and with a widespread distribution from southern Canada to southern Mexico. To explore why the bobcat’s distribution stops at the Isthmus of Tehuantepec, we modelled the ecological niches, projected as potential distributions, of the felid community (bobcat Lynx rufus, puma Puma concolor, jaguar Panthera onca, margay Leopardus wiedii, jaguarundi Herpailurus yagouaroundi, and ocelot Leopardus pardalis) in southern Mexico, using occurrence data, environmental maps, the computer algorithm GARP, and a GIS platform. The resulting geographical projection of the ecological niche of bobcats extends south of the Isthmus of Tehuantepec, suggesting that ecological conditions exist for the establishment of populations. The overlap of the modelled distribution of the bobcat was large with that of the puma (97%), but low with that of the ocelot (44%), margay (46%), jaguar (49%), and jaguarundi (52%), the latter three having relatively similar size and feeding habits to bobcats. Moreover, an independent analysis computing a geographic co-occurrence index showed a similar trend of geographic avoidance (values B4.60 showing a statistically significant anti-correlation) between bobcats with the margay, ocelot, jaguarundi, and jaguar, but not with the puma (values 0.15), while all felids, except bobcats, showed a geographic co-occurrence in southern Mexico (values ranging from 1.91 to 4.71). The Isthmus of Tehuantepec, a lowland region with subtropical habitat, is unlikely to serve as a geographic and ecological barrier to bobcats. As mammal inventories have been conducted for over a century in this region with no records of bobcats, it is unlikely that bobcats are present but have just not been seen. Fossil records also provide no support for the presence of bobcats in that region in the past. Thus, competitive interactions with other felid species appear important in limiting the southern distribution of bobcats, preventing dispersal to a suitable but geographically reduced area south of the Isthmus of Tehuantepec. A fundamental question in biogeography is what limits species’ distributions (Lomolino et al. 2005). One hypoth- esis is that, in temperate regions, abiotic factors are more important in limiting species’ distributions, whereas in tropical regions, distributions are predominantly limited by biotic factors, such as competition, commensalism, parasit- ism, and plant-animal interactions (Lomolino et al. 2005). Recent research that uses tools for modelling ecological niches provides insight into these questions. By combining known species occurrences with relevant environmental parameters in digital maps and using a GIS platform, one can characterise species’ ecological niches projected as potential distributions (Stockwell and Peters 1999, Peterson et al. 1999, Anderson et al. 2002, Sevenning and Skov 2004, Sobero ´n and Peterson 2005). These distribution hypotheses can serve as a framework for identifying factors affecting species distributions (Sobero ´n and Peterson 2005). For example, Svenning and Skov (2004) used climatic variables to model European tree distributions and found that temperature limited dispersal capability in postglacial expansion. They also observed a strong association between ecological niche occupancy and latitude: species at higher latitudes showing a higher occupancy (Svenning and Skov 2004). Ecological niche modelling using environmental variables has proven robust for invasive species because such species tend to disperse to geographic areas where similar environmental conditions exist (Peterson 2003). Further, Anderson et al. (2002) modelled species ecological niches of South American rodents, and found a geographic replace- ment of one species for another within the potential distribution of the latter species. These authors invoked competitive exclusion as an important factor affecting the distributions of these rodents. The bobcat Lynx rufus is a widespread Nearctic species ranging from Canada to southern Mexico (Hall 1981, Lariviere and Walton 1997). It is a vagile and opportunistic Ecography 31: 757764, 2008 doi: 10.1111/j.1600-0587.2008.05327.x # 2008 The Authors. Journal compilation # 2008 Ecography Subject Editor: John Linnell. Accepted 19 June 2008 757