The Effect of Phylogenetic Uncertainty and Imputation on EDGE Scores

Faced with the challenge of saving as much diversity as possible given financial and time constraints, conservation biologists are increasingly prioritizing species on the basis of their overall contribution to evolutionary diversity. Metrics such as EDGE (Evolutionary Distinct and Globally Endangered) have been used to set such evolutionarily-based conservation priorities for a number of taxa, such as mammals, birds, corals, amphibians, and sharks. Each application of EDGE has required some form of correction to account for species whose position within the tree of life are unknown. Perhaps the most advanced of these corrections is phylogenetic imputation, but to date there has been no systematic assessment of both the sensitivity of EDGE scores to a phylogeny missing species, and the impact of using imputation to correct for species missing from the tree. Here we perform such an assessment, by simulating phylogenies, removing some species to make the phylogeny incomplete, imputating the position of those species, and measuring (1) how robust ED scores are for the species that are not removed and (2) how accurate the ED scores are for those removed and then imputed. We find that the EDGE ranking for species on a tree is remarkably robust to missing species from that tree, but that phylogenetic imputation for missing species, while unbiased, does not accurately reconstruct species evolutionary distinctiveness. On the basis of these results, we provide clear guidance for EDGE scoring in the face of phylogenetic uncertainty.