Strengths and limitations of molecular sequence comparisons for inferring the phylogeny of the major groups of fishes

Although the phylogenetic relationships of the major groups of fishes have been extensively studied with morphological characters, not all have been convincingly resolved. Analyses of molecular sequences from these groups may provide additional insights into problematical relationships, but are only just beginning to appear. We compare our own results from analyses of 18s ribosomal RNA sequences with those of other studies using globins, parvalbumins, insulin, 28s ribosomal RNA, and portions of two mitochondria1 genes (12S ribosomal RNA and cytochrome b). Our evaluation of these studies reveals some of the difficulties encountered in reconstructing ancient divergences within the fishes, including unequal rates of evolution (among regions of a molecule as well as among lineages), gene duplication, extinction of lineages, and a possible rapid radiation of gnathostome higher taxa. The importance of evaluating the robustness of particular phylogenetic hypotheses is stressed. Some molecules appear to be inappropriate for investigating higher level divergences within the fishes; others are more promising, but must be examined in more taxa to allow an adequate evaluation of their utility. Convincing support for particular hypotheses of relationship will ultimately require congruence of trees generated from independent molecular data sets.