A permutation test of host-parasite cospeciation.

We introduce a statistical method that explores host-parasite coevolution by testing the null hypothesis that hosts and their associated parasites evolved independently. This test is simple and intuitive and involves only suitable randomization of the observed data. It is not even necessary to construct host and parasite phylogenetic trees, as the test can be performed directly on distance matrices. Statistical power of the test was evaluated using simulated data consistent with the alternative hypothesis of cospeciation. Results were compared with the method of Mantel (1967) and the ParaFit method of Legendre et al. (2002). We observed that our method has greater power overall and thus a higher ability to detect cospeciation in closely related host-parasite systems. Our test was also successful when applied to the pocket gopher and chewing lice system.