RESULTS AND DISCUSSION

The analysis yielded two equally parsimonious trees (length = 993; consistency index = 0.525; retention index = 0.522) in which echiurans and pogonophorans nest within a clade containing polychaete and clitellate annelids (Fig. 1). The echiurans, the pogonophorans, and the clitellates each form a monophyletic clade within a paraphyletic grade of polychaetes. This renders the traditionally defined Annelida paraphyletic and the phylum status of echiurans and pogonophorans obsolete. Decay indices, ranging from 3 to >8 (Fig. 1), support these results, and topology-dependent cladistic permutation tail probability tests (29) corroborate monophyly of all groups in the worm clade (P < 0.01), including the echiurans plus sternaspid polychaete clade and the pogonophorans plus maldanid polychaete clade. Moreover, if the maximum parsimony analysis is constrained to the traditional classification of the Annelida, i.e., polychaetes plus clitellates, the resultant trees are 32 steps longer than the most parsimonious trees; a Wilcoxon signed rank test (30) shows this result to be highly significant (P < 0.01).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Results of the phylogenetic analysis showing that echiurans and pogonophorans are derived groups within the annelid clade. The tree shown is the strict consensus of two equally parsimonious trees obtained with a paup (26) heuristic search (100 replicates, random addition search option) on the EF-1α nucleotide sequence data for 20 taxa (first and second codon positions only; 86 parsimony informative sites; length = 993; consistency index = 0.525; retention index = 0.522). The skewness test statistic for the distribution of tree lengths (27), g1 = −0.53, is significant (P < 0.01). Decay indices and branch lengths are shown above and below each node, respectively. The analysis shown incorporated a 4:1 transition/transversion ratio; ratios ranging from 2:1 to 10:1 yielded the same results. The deuterostome, Xenopus laevis, was designated as the outgroup for this analysis. If molluscs and arthropods are used as outgroups, the same worm clade topology results from maximum parsimony analysis, and neighbor joining analysis using the Kimura two-parameter distance model (28) with a γ correction for among-site rate variation also places echiurans and pogonophorans within a monophyletic worm clade.

In this analysis, the clitellates form a monophyletic group that shares an ancestor with some polychaetes, thus supporting previously proposed theories of clitellate evolution; however, the position of the epifaunal polychaetes Nereis and Harmothoe as basal taxa within the worm clade contradicts the view that the ancestral annelid was a burrowing form (31–34). On the contrary, this result lends support to the proposed ancestral condition of polychaetes as morphologically similar to some extant epifaunal polychaete groups, including aphroditiceans (35). Consideration of the relationships of other metazoan groups awaits a broader taxonomic sampling of EF-1α sequences.

Segmentation has been used as the basis for uniting annelids and arthropods as sister taxa; a rival hypothesis supports a sister relationship between molluscs and annelids, with the exclusion of arthropods, on the basis of the trochophore larva (1, 3, 4, 12, 36). In either scheme, echiurans have been considered as primitively unsegmented. However, the results presented here indicate that the unsegmented echiuran condition is derived from a segmented annelid ancestor. Several morphological and embryological lines of evidence provide additional support for the placement of echiurans within the Annelida. For example, echiuran cleavage patterns, chaetal formation, and sperm ultrastructure closely resemble those of polychaete annelids (6, 37, 38).

Early reports of annelid-like teloblastic development in echiurans, the arrangement of ganglia along the larval nerve cord, rings of larval ectodermal mucous glands, and repeated pairs of nephridia in adults are all characters suggestive of segmentation in echiurans. But they have been dismissed as superficial, transitory features (6, 36). In light of the EF-1α sequence data, further investigation of these features is warranted; the possibility that echiurans represent neotenous annelids, in which development of segmentation is suppressed during growth, must be seriously considered (12). Specifically, comparative studies of the expression patterns of homeobox segmentation genes through the larval development of echiurans and segmented annelids may reveal remnants of an ancestral segmented condition in echiurans.

Throughout their colorful taxonomic history, pogonophorans have been variously allied with annelids and with polychaete annelids in particular (19, 39–41). The results of this EF-1α sequence analysis confirm such an alliance, which is further supported by the similarities between the recently described early embryology of pogonophorans and that of polychaete annelids (14–17). The hooked chaetae of pogonophorans also correspond in detail with those of some tube-dwelling polychaetes, including the sister-taxon of pogonophorans in this analysis (Axiothella rubrocincta), the Maldanidae (42). Moreover, the structural properties and amino acid sequences of pogonophoran and annelid extracellular hemoglobin are very similar (43, 44). Pogonophorans, then, represent a group of annelids in which the post-oral segment is greatly elongated and the gut has been lost in the adult.

In a step toward a classification that reflects phylogeny of protostome worms, the Annelida can be redefined to include the echiurans, pogonophorans, clitellates, and polychaetes. The Annelida can then be identified among the protostomes by the presence of paired chitinous chaetae. The “annelid cross” cleavage pattern of blastomeres 1a¹¹²–1d¹¹² occurs in echiurans, clitellates, and polychaetes (6, 45, 46) and may constitute another synapomorphy for the Annelida; the pattern of cleavage in pogonophorans has not yet been fully described. A traditional trademark of annelids, segmentation, has been lost in echiurans. Retention of the taxon Polychaeta is not logically consistent with the phylogeny presented here, and it is proposed that the term “polychaete” be used only as an informal reference for a grade of marine annelids. As clades within the Annelida, the echiurans revert to their original name, the Echiuridae, and the pogonophorans assume the name erected in the first species description for the group, the Siboglinidae (47, 48).

Fundamentally different animal body plans, or phyla, constitute groups that are assumed to maintain their phylogenetic integrity as far back as they can be traced (49). Understanding the evolution of metazoan development, morphology, and diversity hinges on this assumption, so it is crucial to examine the monophyly of phyla and the relationships among them. The phylogeny reconstructed from EF-1α sequences demonstrates that the phylum Annelida as traditionally defined is not monophyletic; the results support placement of the echiurans and pogonophorans within the Annelida. These results compel recognition that some characters taken to define body plans, such as segmentation, are more evolutionarily labile than has been generally considered and that the traditional classification of metazoan phyla does not reflect the phylogenetic relationships of protostome worms.