Likelihood Analysis of Mitochondrial Restriction-Cleavage Patterns for the Human-Chimpanzee-Gorilla Trichotomy

In spite of considerable work by many evolutionary biologists, it has not been possible to separate convincingly the human (H) - chimpanzee (C) - gorilla (G) trichotomy into a pair of sequential dichotomies. There are three possible phylogenetic groupings (HC)G, (HG)C, and (CG)H, and each has its proponents. The evidence remains ambiguous, and instead of choosing among the available phylogenies, it might be better to provide a statement of their relative probabilities. We develop a likelihood analysis of mtDNA restriction-pattern data that can be used to make such probability statements and illustrate it with data on humans, chimpanzees, gorillas, orangutans, and gibbons. The results of our analyses suggest that the best fitting model is that of the (CG)H grouping, but with a very short time span between the first and second splits. For either the (HC)G or (HG)C hypothesis, a true trichotomy is the best model (no elapsed time between the two splits). Chi-square tests indicate no compelling resolution among the three models, however, and all three retain nontrivial posterior probabilities. We also compare each model with an alternative allowing for rate heterogeneity among lineages, but there is no convincing evidence for such heterogeneity. Our results suggest that, while it may eventually be possible to resolve the trichotomy into a pair of unambiguously ordered (but very close) dichotomies, it is possible that the ancestors of all three taxa (H, C, and G) were still conspecific subsequent to the second split, perhaps no more different than the "major races" of extant Homo sapiens.