Fine Tuning Gene Regulation

in gene regulation. from many phyla is now making the use of alignments of hundreds or even thousands of genes a standard procedure. These studies have led to a widely accepted phylogeny of all animal phyla that has radically changed our views of animal evolution ( 3). Premolecular phylogenies generally envisaged a gradual increase in complexity from the earliest animals without a body cavity or coelom (acoelomate flatworms) via pseudocoelomate worms (such as nematodes and rotifers) to coelomate protostomes (annelids, arthropods, and mollusks) and deuterostomes (echinoderms and chordates) with a sophisticated mesoderm-lined coelomic body cavity. In contrast, today’s tree divides bilaterally symmetrical animals into protostomes and deuterostomes (see the fi gure, panel B). Within the deuterostomes, the simple urochordates (sea squirts) are closer relatives of the vertebrates than the more fi shlike cephalochordates (amphioxus) ( 13); a third phylum of deuterostomes, the hemichordates (acorn worms), are the sister group of echinoderms and not of the chordates ( 14). The acoelomate platyhelminths, as we have seen, are now known to be related to the coelomate annelids, mollusks, and brachiopods within the Lophotrochozoa. A second acoelomate group, the Xenacoelomorphs, although historically linked to the fl atworms, have rather controversially been placed close to echinoderms to form a fourth phylum of deuterostome ( 15). Pseudocoelomate phyla, including nematodes and rotifers, are scattered throughout the protostomes. All these rearrangements suggest that many characters thought to be important— such as the coelomic body cavity—have in fact been gained and lost multiple times. Although much of the animal tree is now resolved, a number of problems remain. These problems tend to involve relationships either of taxa with extreme systematic biases or among groups that seem to have originated in a rapid radiation, resulting in a lack of signal supporting individual nodes. Future progress will depend on increasing useful signal with larger “phylogenomic” data sets from the widest possible taxonomic sample and on continued improvement in the correspondence between real data and the models used when reconstructing trees.