Cross-scale variation in species richness–environment associations

Aim  The scale dependence of many ecological patterns and processes implies that general inference is reliant on obtaining scale-response curves over a large range of grains. Although environmental correlates of richness have been widely studied, comparisons among groups have usually been applied at single grains. Moreover, the relevance of environment–richness associations to fine-grain assemblages has remained surprisingly unclear. We present a first global cross-scale assessment of environment–richness associations for birds, mammals and amphibians from 2000 km down to c. 20 km. Location  World-wide. Methods  We performed an extensive survey of the literature for well-sampled terrestrial vertebrate inventories over clearly defined small extents. Coarser grain richness was estimated from the intersection of extent-of-occurrence range maps with concentric equal-distance circles around fine-grain assemblage location centroids. General linear and simultaneous autoregressive models were used to relate richness at the different grains to environmental correlates. Results  The ability of environmental variables to explain species richness decreases markedly toward finer grains and is lowest for fine-grained assemblages. A prominent transition in importance occurs between productivity and temperature at increased grains, which is consistent with the role of energy affecting regional, but not local, richness. Variation in fine-grained predictability across groups is associated with their purported grain of space use, i.e. highest for amphibians and narrow-ranged and small-bodied species. Main conclusions  We extend the global documentation of environment–richness associations to fine-grained assemblages. The relationship between fine-grained predictability of a group and its ecological characteristics lends empirical support to the idea that variation in species fine-grained space use may scale up to explain coarse-grained diversity patterns. Our study exposes a dramatic and taxonomically variable scale dependence of environment–richness associations and suggests that environmental correlates of richness may hold limited information at the level of communities.