Definition of the biodiversity hotspots

We overlaid the distributions of mammals and breeding ranges of birds onto a 100 km by 100 km grid to define hotspots of biodiversity (fig. S1, A and B). WE in each grid cell was calculated by weighting each species’ occurrence by the inverse of its corresponding range and then summing values across all species in a cell. We then defined the hotspots as the cells with the top 20% of WE values. This definition covered 19.9% of our grid surface, roughly equivalent in size to Myers’ hotspots that comprised 17.0% of global land surface (6), and resulted in 3826 and 3304 cells for mammals and birds, respectively. There was a 74.4% overlap between the mammal and bird hotspots and a 69.7 and 71.3% overlap between the Myers’ hotspots and the ones we defined for mammals and birds, respectively (fig. S1). We also found an overlap between our WE-based hotspots and hotspots defined simply with total SR of 56.9 and 62.4% for mammals and birds, respectively. Overlap was expected since WE and SR were positively correlated (Spearman’s ρ = 0.74 in our dataset).

We also assessed whether alternative ways of delineating biodiversity hotspots changed our results. First, we defined hotspots based on SR. Hotspots were defined as the cells with the top 20% values of SR and, as expected, were concentrated in the tropics (fig. S1, C and D). Two realms, Australasia and Nearctic, had very few or no hotspot cells or showed a very small overlap with the WE hotspots (table S4). Therefore, we were only able to carry out further analyses within the Afrotropics, Indo-Malay, Neotropics, and Palearctic. Second, we delineated alternative hotspots as “centers of endemism” where NRS were found. These regions have been proposed to provide many opportunities for past speciation while enabling the survival of narrow-ranged endemics due to stable environments (31). Following Jetz et al. (31), we defined NRS as species with a range of ≤100,000 km² (10 cells in our grid) and hotspots as all the cells where these species were found (n_mammals = 2723 cells, n_birds = 2041 cells). The resulting hotspots substantially overlapped with the WE-based hotspots, particularly for mammals (fig. S1, E and F, and table S4). We repeated the diversification rate and historical biogeography analyses described in the Results and Discussion for the SR- and NRS-based hotspots and found that the results were largely congruent with the WE-based hotspots (see Supplementary Text and figs. S7 and S8).