Endemism patterns are scale dependent.

Areas of endemism are important in biogeography because they capture facets of biodiversity not represented elsewhere. However, the scales at which they are relevant to research and conservation are poorly analysed. Here, we calculate weighted endemism (WE) and phylogenetic endemism (PE) separately for all birds and amphibians across the globe. We show that scale dependence is widespread for both indices and manifests across grain sizes, spatial extents and taxonomic treatments. Variations in taxonomic opinions-whether species are treated by systematic 'lumping' or 'splitting'-can profoundly affect the allocation of WE hotspots. Global patterns of PE can provide insights into complex evolutionary processes but this congruence is lost at the continental to country extents. These findings are explained by environmental heterogeneity at coarser grains, and to a far lesser extent at finer resolutions. Regardless of scale, we find widespread deficits of protection for endemism hotspots. Our study presents a framework for assessing areas for conservation that are robust to assumptions on taxonomy, spatial grain and extent.