restricted to the locomotary system, although it does involve the parapodia and their musculature and its vascularization and innervation. Unlike the arthropods, annelidan gonads and nephridia are also seriated in concert with the coelomic compartmentalization. Annelidan and arthropodan segmentation, though associated with locomotion in each case, operate on different principles and involve some different organ systems including the fluid skeletons; there is little reason to regard them as homologous. Eernisse et al. (1992) have presented a cladistic analysis that arrives at similar conclusions (but see Backeljau et al., 1993).
Brachiopods are commonly regarded as allied to phoronids and bryozoans, as all three phyla possess similarly regionated coeloms and lophophores; this relationship has not been corroborated by molecular techniques at this writing. According to the 18S rRNA tree, the brachiopods may have arisen from the last common arthropod—annelid ancestor and therefore from a proto-molluscan grade in the present scenario. It has been suggested (Valentine, 1992) that an unsegmented coelom was evolved within an ancestral worm clade for locomotion in soft sediments; direct peristalsis, which requires an unsegmented coelom, is a common locomotory technique in flocculent muds, for example (Elder, 1980). Radiation of this clade onto firmer substrates may then have produced sessile forms with regionated coeloms to serve both a trunk and a tentacular crown, the feeding lophophore. Perhaps annelids have also descended from a small-bodied ancestor with an unregionated coelom, with segmentation evolving for efficiency in locomotion.
Mollusks exhibit the body plan that is least derived from that of the postulated protostome ancestor, and most molluscan classes have only organ coeloms. However the cephalopods, which form the most advanced molluscan class and which appear only in the Upper Cambrian, have a well-developed coelomic space (a gonocoel?) that can be regarded as a novel evolutionary feature. The nemertine rhynchocoel, which has been demonstrated to be a coelom histologically (Turbeville, 1991), probably was derived independently.
Late Vendian and Manykaian bilaterian body plans are thus visualized as consisting of an array of vermiform types, including flatworms and "round flatworms" with blood-vascular systems, many with hemocoels or "pseudocoels," and some with seriation of one organ system or another (Valentine, 1989, 1990; Bergström, 1989). Organ coeloms were doubtless present in many lineages. Possibly some of these worms were incipiently segmented in the arthropod style. It seems likely that some worms had unregionated perivisceral coeloms, but it is doubtful that any had annelid-style coelomic segmentation. In this scenario, the body plans of these worms were based primarily upon adaptations to loco-