Geographic characters---traits relating to the spatial distribution of a species---may both affect and be affected by processes associated with lineage birth and death. This is potentially confounding to comparative analyses of species distributions because current models do not allow reciprocal interactions between the evolution of traits and the growth of phylogenetic trees. Here we introduce a likelihood-based approach to estimating region-dependent rates of speciation, extinction, and range evolution from a phylogeny, using a new model in which these processes are interdependent. We demonstrate the method with simulation tests that accurately recover parameters relating to the mode of speciation and source-sink dynamics. We then apply it to the evolution of habitat occupancy in Californian plant communities, where we find higher rates of speciation in chaparral than in forests and evidence for expanding habitat tolerances.