RECOVERING PHYLOGENETIC SIGNAL FROM DNA-SEQUENCES - RELATIONSHIPS WITHIN THE CORVINE ASSEMBLAGE (CLASS AVES) AS INFERRED FROM COMPLETE SEQUENCES OF THE MITOCHONDRIAL-DNA CYTOCHROME-B GENE

Phylogenetic analysis of cytochrome-b sequences and cranial osteologic al characters for nine genera of corvine passerine birds supports the hypothesis that the two major groups of birds of paradise, the manucod ines and paradisaeinines, constitute a monophyletic group and that the ir postulated sister group is the Corvidae (crows, jays, and allies). The data are also consistent with the hypothesis that the bowerbirds a re not closely related to the birds of paradise but instead lie near t he base of the corvine assemblage. The corvine radiation exemplifies a case of multiple star phylogenies embedded within a major clade, with the branching pattern characterized by very short internodal divergen ce times. Such histories are difficult to resolve no matter what type of data is employed, because little change accumulates between branchi ng events. With respect to sequence data, reconstructed tree topologie s are sensitive to the choice of outgroup and to the method of analysi s ( e.g., transversion vs. global parsimony). In such cases, assessing the ''reliability'' of a best-fit or most-parsimonious tree inferred from any particular data set becomes problematic. Statistical tests of tree topologies that depend on random sampling of characters will gen erally be inconclusive in that all cladistic components will tend to b e poorly supported because relatively few character-state changes will be recorded between branching events. It is suggested, on the other h and, that congruence in cladistic signal across different data sets ma y be a potentially more useful method for evaluating the reliability o f the signal of any one data set. Resolution of star phylogenies will probably be possible only if DNA sequence and morphological characters are combined in a single analysis.