Differential effects of environmental heterogeneity on global mammal species richness

Abstract

Aim Spatial environmental heterogeneity (EH) is an important driver of species richness, affecting species coexistence, persistence and diversification. EH has been widely studied in ecology and evolution and quantified in many different ways, with a strong bias towards a few common measures of EH like elevation range. Here, we calculate 51 measures of EH within grid cells world‐wide across three spatial grains to investigate similarities and differences among these measures. Moreover, we compare the association between species richness of terrestrial mammals and each EH measure to assess the impact of methodological choices on EH–richness relationships found by standard macroecological modelling approaches. Location Global. Methods We derive 51 measures of EH from nine variables related to the five subject areas land cover, vegetation, climate, soil and topography, using nine calculation methods. We first explore differences among these EH measures with correlation and principal components analyses. We then analyse the relationship between mammal species richness and each EH measure alone and while accounting for effects of current climate, regional biogeographic history and human influence. We assess the impact of subject area and method of calculation of EH measures on model support using conditional inference trees. Results Despite some redundancy, correlations (rs = −0.45 to 1.00, median 0.35) and spatial patterns indicate clear differences between the EH measures. We find clear effects of subject area and calculation method on the importance of EH measures for mammal species richness. Measures of climatic and topographic EH and measures calculated as counts and ranges (as against, for example, coefficient of variation) receive particularly high model support across all spatial grains. Main conclusions The outcome of broad‐scale EH–richness studies is greatly determined by methodological decisions on calculation of measures and statistical analysis. These decisions should therefore be made carefully with regard to the hypothesis and mechanism of interest.