The Mongolian share of the Palaearctic steppe biome hosts some of most intact grassland regions, these are traversed by high peaks creating diverse and rich habitats. We sampled 100 randomly located vegetation plots in a 2 × 2 km² mountain summit region that is isolated by surrounding drier lowlands in the Gobi Gurvan Saykhan mountain range in southern Mongolia. All sampled plants were assigned to phytogeographic groups using new data on their geographic distribution ranges. Relationships between local scale and global scale distribution were analysed with fourth-corner analysis. To this end, we investigated if statistical methods developed for functional trait analysis can be used in a biogeographical context to test whether plot-level vegetation patterns mirror range-level distribution data. Furthermore, we determined if range distributions are reflected in habitat preferences, and how this link could be applied to decision-making in conservation biology. The fourth-corner analysis revealed that phytogeographical groups showed significant preferences for certain local habitat types in the heterogeneous mountain range. The local, Central Asian endemics showed, however, no clear preference along the altitudinal gradient. Fourth-corner analysis proofed useful for testing relations between local scale habitat preferences and large-scale distribution. In the Gobi Altay, species at their distribution limits where much more strongly restricted in their local distribution, while endemics tended to cover a wider range of local habitat conditions. Given that biogeographical responsibility has become a major topic in conservation biology, the approach can also be helpful for applied studies elsewhere.