Abstract
A recently described test for a consistent pattern of timing of cospeciation between host and parasite taxa uses the correlation between observed genetic distance matrices for the hosts and parasites. In this paper I argue that this method can be improved by comparing trees rather than distance matrices. In the special case of a molecular clock, the test compares cluster heights in dendrograms for the hosts and the parasites. Because tree topology alone can cause a significant correlation between the cluster heights, a randomization test is described that allows for this autocorrelation. The degree of autocorrelation is a function of tree topology, increasing as the trees become more unbalanced. Balanced trees offer more sensitive tests of hypotheses concerning timing of cospeciation. [Bootstrap; cladistics; cospeciation; dendrograms; molecular clock; randomization tests; rates of evolution.]

This publication has 0 references indexed in Scilit: