ParaFit

ParaFit is a statistical method used to test for coevolution between two groups of organisms, typically hosts and their parasites or mutualists, using their phylogenies and observed association data. It assesses whether the two phylogenies show congruent patterns given the interaction matrix, indicating potential coevolutionary structure.

The method uses two phylogenetic distance matrices derived from the host and parasite trees, and a bipartite association matrix that records which hosts are linked to which parasites. ParaFit calculates a global statistic, ParaFitGlobal, to test the null hypothesis that host and parasite phylogenies are independent of the observed associations. It can also produce individual link statistics, ParaFitLink, to identify specific host–parasite pairings that contribute to any detected signal.

Computation is based on permutation testing. The observed ParaFitGlobal statistic is compared against a null distribution generated by permuting associations (or labels) to obtain a p-value. The approach is implemented in the R package ade4, where users typically input cophenetic (patristic) distance matrices derived from the host and parasite trees along with the association matrix.

ParaFit has been widely applied in coevolution and cophylogeny studies, including analyses of host–parasite systems, plant–pollinator interactions, and other biotic associations where phylogenetic information is available. While useful for detecting phylogenetic congruence, interpretation requires caution, as ParaFit does not identify the exact mode or direction of coevolution and can be influenced by sampling bias or biogeographic structure.

See also: PACo, cophylogeny, host–parasite coevolution.