Phylogeny of whales: dependence of the inference on species sampling.

From phylogenetic analyses of the 12s and 16s mitochondrial ribosomal DNA and of myoglobin amino acid sequences, Milinkovitch et al. ( 1993) proposed the hypothesis that one group of toothed whales (Odontoceti ), the sperm whales (Physeteridae) , is more closely related to the baleen whales (Mysticeti) than to other alleged odontocetes such as dolphins. This hypothesis is in conflict with the traditional view (Barnes et al. 1985; Novacek 1993) that the odontocetes form a monophyletic clade. From a maximum parsimony analysis of the cytochrome b gene, Arnason and Gullberg ( 1994) recently challenged Milinkovitch et al.‘s ( 1993) hypothesis as well as the traditional tree, claiming that the mysticetes are closer to the dolphins rather than to the sperm whales. They used the cow as the only outgroup and the giant sperm-whale as the only representative of Physeteridae, but the estimated tree may depend on the sampled species (Lecointre et al. 1993; Cao et al. 1994~). It is now highly likely that artiodactyles are the closest relatives of whales (Novacek 1992; Milinkovitch et al. 1993; Janke et al. 1994; Cao et al. 1994b), and by including many alternative artiodactyl outgroups (Irwin et 1991) in their cytochrome b dataset, we show that nason and Gullberg’s conclusion is shaky and that an argument based on a single species as an outgroup is unstable. Although the maximum parsimony method is used widely in molecular phylogenetics (Stewart 1993), it sometimes gives biased estimate of the phylogeny, particularly when the evolutionary rate differs among lineages (Felsenstein 1978; Hasegawa et al. 1992; Cao et al. 1994a), while the maximum likelihood method that does not assume the rate constancy is robust against the violation of the constancy (Hasegawa and Fujiwara 1993; Kuhner and Felsenstein 1994). For this reason, we used the PROTML program in our package MOLPHY ( Adachi and Hasegawa 1994). This program is for the maximum likelihood inference of protein phylogeny (Kishino et al. 1990)) and it was based originally on the