What drives global variation of biodiversity, and why are some places richer in species than others? The answer certainly depends on spatial scale (grain or resolution), but data are available only from a limited number of scales, and we lack straightforward ways to integrate them. Here, we aim to resolve this with a new cross-scale model that links together several macroecological patterns. The model estimates the relative importance of environmental and spatial predictors of species richness across a range of grains, spanning 11 orders of magnitude, from local plots to entire continents. We apply the model to a new global dataset on tree species richness in thousands of forest plots and hundreds country-like units, which allows us to predict and map biodiversity patterns at any grain.
Results/Conclusions
When we fitted the model to the data, we found different geographic patterns of richness at different grains, strengthening effects of spatial and biogeographic predictors of richness towards coarse grains, and complex grain-dependent effects of environmental predictors of richness. This offers resolution to the debate surrounding the relative importance of ecological versus historical factors in driving patterns of global biodiversity, it unifies disparate and at times contradicting results from previous studies, and it enables efficient and straightforward integration of heterogeneous data from the entire continuum of spatial scales.