Dissecting phylogenetic fuzzy weighting: theory and application in metacommunity phylogenetics

Summary Metacommunity phylogenetics aims at evaluating environmental and/or historical factors driving clade distribution. Phylogenetic fuzzy weighting (PFW) describes clade distribution across metacommunities based on fuzzy sets defined by phylogenetic relatedness among species. The method enables analysing environmental and/or biogeographic determinants of clade distribution. PFW also offers an exploratory tool for visualizing clade distribution via Principal Coordinates of Phylogenetic Structure (PCPS). In this article, we describe the theoretical properties and biological backgrounds of PFW and evaluate its statistical performance (type I error and statistical power) in assessing environmental and phylogenetic determinants of species distribution in comparison with other phylobetadiversity methods (COMDIST, COMDISTNT, Rao's H and UniFrac). The statistical performance of PFW and the other phylobetadiversity metrics was tested by (i) simulating metacommunities under different species assembly scenarios (species distribution influenced or not by environment and/or phylogeny), niche breadth tolerance and species pool sizes; (ii) submitting community matrices to PFW and deriving pairwise phylogenetic dissimilarities between communities (DP) and PCPS; (iii) submitting these metrics and the other phylobetadiversity methods to different analytical approaches (Mantel test, regression on dissimilarity matrices – ADONIS, and GLM) to evaluate the influence of environment and phylogeny on metacommunity phylogenetic structure; and (iv) estimating type I error and power estimates via alternative permutation procedures. Results demonstrated that PFW provides robust assessment of environmental and phylogenetic drivers of species distribution across metacommunities. Although all methods had acceptable type I error for both Mantel test and ADONIS, only PFW showed acceptable power for both tests. Rao's H had acceptable power only for Mantel test, while COMDIST had acceptable power only for ADONIS. COMDISTNT and UniFrac showed poor statistical performance for both tests. Conversely, GLM had acceptable power only for the first PCPS. Performing ADONIS on DP provides a robust overall assessment of environmental and phylogenetic drivers of species distribution. On the other hand, performing PCPS analysis after rejecting the null hypotheses via ADONIS allows identifying the phylogenetic nodes mostly associated with environmental gradients. PFW enables synthesizing and analysing