Assessing endemism at multiple spatial scales, with an example from the Australian vascular flora

Aim To develop an approach for assessing the spatial scale of centres of endemism among species level data. Location Australia. Methods Endemism is inherently scale dependent. Therefore, the Corrected Weighted Endemism (CWE) index used by Crisp et al. [J. Biogeogr. (2001)28:183] is extended to account for species samples in local neighbourhoods as a Spatial CWE index. This then allows an analysis of how the degree of endemism of a location (cell) changes with spatial scale. The quality of the Spatial CWE index results are assessed using three spatial randomizations at the species level with and without preserving species richness and distributional patterns. We show that CWE is equivalent to beta diversity and predict that it should show high rates of change around centres of endemism. Results Similar patterns to those found by Crisp et al. using a data set of vascular flora from Australia are retrieved, but the extent to which they are scale dependent is more easily identified. For example, the Central Australian centre discounted by Crisp et al. is identified when a three-cell radius neighbourhood is used. However, the level of endemism in this centre is no greater than in the margins of many of the coastal centres of endemism. Most of the identified centres of endemism are better than random at all scales and are increasingly so as the spatial scale increases. As predicted, the highest rate of change in Spatial CWE (beta diversity) is most often between zero- and one-cell radius neighbours in most centres of endemism. Main conclusions The explicit incorporation of geographical space in analyses allows for a greater understanding of the scale-dependence of phenomena, in this case endemism and beta diversity.