Chapter 15, this volume), whereas both the relative and absolute divergence times in molecular phylogenies can be controversial (Pulquério and Nichols, 2007). Because of these reservations, we use both approaches.

Taxonomically, the ratio of species to genera should indicate the regional diversification rate over the past several million years, if genera are approximately equal in age. This ratio correlates strongly with log(species richness) among equal-area grid cells (Pearson’s r = 0.61; Fig. 14.2a). This correlation remains highly significant (corrected F = 78.49, corrected P < 0.001, n = 4,152: based on a subsample of cells and excluding single species occurrences) when degrees of freedom are reduced to account for spatial autocorrelation (Clifford et al., 1989). Moving to phylogeny, places where a high proportion of species are on short terminal branches in the tree are likely to have rapid diversification, turnover, or immigration in their recent history (Roy and Goldberg, 2007). However, analyses are complicated by the low resolution (uncertain relationships) at the tips of the phylogeny, which introduces overestimates of the respective branch lengths. We ameliorated this problem by reducing ages of terminal polytomies using the correction suggested by Nee in Mooers and Atkins (2003) and by assuming that the descendants from each polytomy diversified under a Yule process (Nee, 2006). Fig. 14.2b highlights the Andean and Himalayan diversity peaks, but not the African great lakes, as recent evolutionary crucibles. Much of the temperate north stands out more than much of the tropics in this map, and there is a negative overall correlation between the proportion of short branches and log(species richness) (r = −0.38, corrected F = 21.01, corrected P < 0.001, n = 4,210, analyses as above), although this depended on how we corrected for terminal polytomies. This result partially echoes recent findings of higher recent speciation and extinction rates in temperate than in tropical mammals (Weir and Schluter, 2007). These maps also imply that some regions have seen marked shifts in net diversification rate, whereas others may have remained steady.

Areas of “old” and “young” diversity can be identified from the residuals of a loess regression across cells of total evolutionary history (i.e., total branch length in the phylogeny of a cell’s species) on species number. The African diversity peak emerges as old, whereas much of Andean diversity is young (Fig. 14.2c). Character disparity—among-species variation in morphology—also shows geographic pattern: Fig. 14.2d maps one index of disparity, the variance in log(body mass). Disparity tends to be high where diversity is old (r = 0.29, corrected F = 5.10, corrected P < 0.05, n = 4,210, analyses as above), although tropical regions drive this relationship.

Mammalian biodiversity, then, shows complex geographic and phylogenetic patterns of richness, recent diversification, and character variation. The African diversity peak is old and disparate, that in Asia is young and disparate, and the Andean peak is young with low disparity. These