Additive Distances, Rate Variation, and the Perfect-Fit Theorem

Recent critics of distance data have argued that a molecular clock is a necessary assumption for the use of distance data in phylogenetic reconstruction. In fact, several pairwise tree-construction algorithms have been developed that make no such assumptions. When distances are additive, these algorithms efficiently recover the correct tree in spite of any rate-disparity. Furthermore, the correct tree will be unique in exhibiting a perfect-fit to a matrix of additive distances. Finally, net amounts of shared derived change can then be identified if the tree includes an unambiguous outgroup.