PHYLOGENETIC TREES: AN ANALYSIS OF THE TREENESS TEST

Abstract.—We have examined the behavior of the "treeness " test for evolutionary indepen-dence (Cavalli-Sforza and Piazza, 1975) using "evolutionary histories " from simulated and real data. For each set of characters, we have evaluated the treeness statistic (T) for the 945 topologies possible with six populations, or a random sample of them, and have analyzed the resulting distributions. All the distributions studied show common features: (1) Generally, a small fraction (less than 6%) of all topologies gave T values compatible with the phylogenetic hypothesis. (2) This fraction became smaller as the number of characters increased. (3) The T values obtained for the simulated and reconstructed trees are always found at the far left tail of the distribution. [Treeness test; phylogenetic analysis; phylogenetic trees.] The genetic relationships among a group of populations can be represented graph-ically with tree structures (Cavalli-Sforza and Edwards, 1964, 1967). In some in-stances, it is important to assess the phy-logenetic meaning of such structures. Th