Molecular phylogeny of the hominoids: inferences from multiple independent DNA sequence data sets

Consensus on the evolutionary relationships of humans, chimpanzees, andgorillas has not been reached, despite the existence of a number of DNAsequence data sets relating to the phylogeny, partly because not all genetrees from these data sets agree. However, given the well-known phenomenonof gene tree-species tree mismatch, agreement among gene trees is notexpected. A majority of gene trees from available DNA sequence data supportone hypothesis, but is this evidence sufficient for statistical confidencein the majority hypothesis? All available DNA sequence data sets showingphylogenetic resolution among the hominoids are grouped according togenetic linkage of their corresponding genes to form independent data sets.Of the 14 independent data sets defined in this way, 11 support a human-chimpanzee clade, 2 support a chimpanzee-gorilla clade, and one supports ahuman-gorilla clade. The hypothesis of a trichotomous speciation eventleading to Homo; Pan, and Gorilla can be firmly rejected on the basis ofthis data set distribution. The multiple-locus test (Wu 1991), whichevaluates hypotheses using gene tree-species tree mismatch probabilities ina likelihood ratio test, favors the phylogeny with a Homo-Pan clade andrejects the other alternatives with a P value of 0.002. When theprobabilities are modified to reflect effective population size differencesamong different types of genetic loci, the observed data set distributionis even more likely under the Homo-Pan clade hypothesis. Maximum-likelihoodestimates for the time between successive hominoid divergences are in therange of 300,000-2,800,000 years, based on a reasonable range of estimatesfor long-term hominoid effective population size and for generation time.The implication of the multiple-locus test is that existing DNA sequencedata sets provide overwhelming and sufficient support for ahuman-chimpanzee clade: no additional DNA data sets need to be generatedfor the purpose of estimating hominoid phylogeny. Because DNA hybridizationevidence (Caccone and Powell 1989) also supports a Homo-Pan clade, theproblem of hominoid phylogeny can be confidently considered solved.