GIScience 2016 Short Paper Proceedings Understanding spatial patterns of biodiversity: How sensitive is phylogenetic endemism to the randomisation model? S.W. Laffan 1 , A.H. Thornhill 2 , J.T. Miller 3 , N. Knerr 4 , C.E. Gonzales-Orozco 5 , B.D. Mishler 2 Centre for Ecosystem Science, School of Biological, Earth and Environmental Science, UNSW, Sydney, Australia, 2052 Email: shawn.laffan@unsw.edu.au University and Jepson Herbaria, and Dept. of Integrative Biology, University of California, Berkeley, CA 94720-2465, USA. Email: {andrew.thornhill, bmishler}@berkeley.edu Division of Environmental Biology, National Science Foundation, Arlington, Virginia, USA Email: joe@acaciamulga.net National Research Collections Australia, CSIRO National Facilities and Collections, GPO Box 1600, Canberra ACT 2601, Australia Email: Nunzio.Knerr@csiro.au Corporacion Colombiana de Investigacion Agropecuaria, Corpoica, km 17 Via Puerto Lopez, Meta, Colombia Email: cegonzalez@corpoica.org.co Abstract Mapping spatial patterns of phylogenetic diversity helps identify regions of unique evolutionary history warranting conservation. Randomisations form an integral component of this process. Here we test the sensitivity of a method used to identify unusual concentrations of old and new evolutionary history to the underlying randomisation. The results indicate low sensitivity to models of complete spatial randomness and spatial structure (proximal allocation and random walks). 1. Introduction Knowledge of the spatial distribution of biodiversity is essential for the allocation of scarce conservation resources and for understanding the evolutionary histories of a region’s biota. Biodiversity is many-faceted, and can be measured using components such as species, phylogenetic, and trait diversity (Laffan 2014). Biodiversity indices are typically aggregate measures of the taxon assemblage found in a location, with geographic surfaces of indices commonly generated. The most commonly used index is species richness, calculated as the number of unique species in a sample. However, closely related species represent less unique diversity than do distantly related species (see Fig 1), e.g., a sample comprising a human, a gorilla, and an orangutan has less unique diversity than one comprising a snake, a cow, and a squid. If one is interested in the conservation and analysis of biodiversity at an evolutionary level then one needs to use phylodiversity indices (Laity et al. 2015). Phylogenetic Diversity (PD; Faith 1992) is the simplest phylodiversity measure and is calculated as the sum of a tree’s branch lengths in a sample (Figure 1). Phylogenetic Endemism (PE; Rosauer et al. 2009) is calculated in the same way as PD, but the branches are weighted by the fraction of their geographic ranges found in a location, such that wide- ranged branches contribute less than narrow-ranged branches of the same length. PE is used to identify regions containing lineages that are found in few other places. A more recent development of PE is the CANAPE method (Categorical Analysis of Neo- and Palaeo-Endemism; Mishler et al. 2014). CANAPE uses PE with a randomisation test to identify regions of geographically restricted long or short branches. Regions of palaeo- endemism can be considered as museums of evolutionary history currently found in few other places, while regions of neo-endemism can be considered as cradles of new diversity. CANAPE classifies the remaining cells into three other classes, two that contain some