nervous system, trunk, and limbs of various members of that phylum, activating Hox genes (Riddle et al., 1993; Echelard et al., 1993; Krauss et al., 1993). It is an obvious hypothesis that a gene ancestral to hh and Shh was present and mediated polarizing activity in the last common protostome—deuterostome ancestor and is probably much more ancient.
It is possible to speculate on aspects of the general course of evolution of metazoan body plans, assuming that the preceding picture of early metazoan evolution is approximately correct. Homeobox genes presumably arose within protistans, but took on a role in the specification of cell fates, movements, and patterns as cell differentiation accompanied the rise of multicellular organisms. During the first few tens of millions of years of metazoan history a regulatory cascade was assembled, mediating the transcription of a growing morphological complexity. If the rise of complexity was as rapid as hypothesized here, it seems likely that many of these regulatory pathways were already present in protistans. The number of transcriptional regulators may have grown hand in hand with the upper bound of metazoan complexity. Bilaterians became increasingly differentiated along their anteroposterior body axes, and this trend may be reflected in the growth of the Hox/HOM cluster. The numbers of homeobox genes besides those in this cluster (and of other regulatory gene types) must have increased as well. At some point during this trend, bilaterians became able to displace sediment and thus to produce trace fossils, and they entered the fossil record of the Late Vendian. By the time that the last common ancestor of deuterostomes and protostomes evolved, presumably still during the Vendian, a large Hox/HOM cluster was present in that lineage. The Cambrian explosion, then, may have occurred largely or entirely within organisms that had a single Hox/HOM cluster. However, the major increase in the upper bound of complexity during the Phanerozoic may be associated with a series of duplications of the cluster.
The postulated branching of lineages at proto-molluscan grades that led eventually to a variety of descendants with distinctive body plans requires that the details of pattern formation responsible for the derived features of each of those body plans evolved independently. In this view the Hox/HOM cluster had a responsibility for anteroposterior differentiation in the proto-mollusk (and earlier) and retained that role, coming to be associated with the development of segmentation in arthropods, in annelids, and with anteroposterior structures in vertebrates, even though those features are quite different and evolved independently. Many early developmental steps also were greatly modified after the ancestral branching that led to separate phyla (Davidson, 1990, 1991). In other words, nearly all of the body architectures, and many of the