Geographic range shifts do not erase the historic signal of speciation in mammals.

Many evolutionary analyses assume that the positions of species geographic ranges are sufficiently phylogenetically conserved that current ranges reflect ancestral ranges and retain the historic signal of speciation. The validity of this assumption has been challenged, because there is evidence that ranges can shift rapidly and extensively. Here I test the assumption of range conservatism using simulations and empirical tests of phylogenetic signal in geographic positions of ranges within mammal orders, families, and genera. In most taxa, range positions show strong phylogenetic signal, quantified using Pagel's λ, Mantel tests, and a novel method to measure phylogenetic signal near the tips of a phylogeny. Taxa with highly labile range positions are exceptions to the general pattern and include very young groups such as Sciurus that may still be in the early, rapid-expansion phase of adaptive radiation. In two orders containing many species with large distributions (Artiodactyla and Carnivora), temporal patterns of range evolution are consistent with large instantaneous shifts in range position associated with allopatric speciation. In most other taxa, range evolution is better described by models that allow ranges to evolve along branches of the phylogeny. The results point to a common pattern of phylogenetically conserved ranges where the current position of species ranges reflects their position at the time of speciation, modified by gradual drift of range boundaries through time.