Metazoan relationships on the basis of 18S rRNA sequences: A few years later...

Metazoan Relationships on the Basis of 18S rRNA Sequences: A Few Years Later...'

BIRGITTA M. H. WINNEPENNINCKX,2 YVES VAN DE PEER,* AND THIERRY BACKELJAU

SYNOPSIS. The 18S rRNA database is continuously growing and new tree construction methods are being developed. The present study aims at assessing what effect the addition of recently determined animal 18S rRNA sequences and the use of a recently developed distance matrix calculation method have on the results of some previously published case studies on metazoan phylogeny. When re-assessing three case studies, part of their conclusions was confirmed on the basis of the present 18S rRNA data set: 1) the monophyly of Arthropoda; 2) the monophyly of the Vestimentifera-Pogonophora and their protostome character; 3) the doubt about the monophyletic origin of Echiura-Sipuncula and 4) the coelomate character of Nemertea. Yet, it is also demonstrated that some of the previous results are no longer warranted when updating the analyses: 1) the monophyly of both the Annelida and the Eutrochozoa; 2) the sister-group relationship of Echiura to Vestimentifera-Pogonophora and 3) the polyphyly of the Mesozoa and their close relationship to Myxozoa and Nematodes. In addition, some new very preliminary evidence is provided for: 1) a common ancestry of Platyhelminthes and Mesozoa and the monophyly of the latter group and 2) the monophyly of Clitellata, Hirudinida and Oligochaeta. Finally, doubt is casted on the monophyly of the Polychaeta and the polychaete orders Spionida, Phyllodocida, and Sabellidae. Of course, these hypotheses also need further testing.

INTRODUCTION

Despite several decades of morphological and anatomical research, numerous aspects of metazoan relationships have remained uncertain (for an overview see Brusca and Brusca, 1990; Willmer, 1990; Meglitsch and Schram, 1991). In the 80s, molecular data were introduced as independent markers to trace phylogenetic relationships. Several properties make small ribosomal subunit RNA (SSU rRNA) an excellent tool for this purpose, such as its ubiquity, homology, length and functional constancy (e.g., Raff, 1988; Raff et al., 1989; Hillis and Dixon, 1991; Solignac et al., 1991; Woese, 1991). Moreover, as there is a difference in variability between sites, the molecule can be used to trace phylogenetic relationships at a broad range of taxonomical levels. Finally, several very conservative regions in the molecule allow the development of "universal" PCR and sequencing primers. Therefore, SSU rRNA sequences have had an overwhelming impact on phylogenetic studies in general and have yielded several new insights in evolutionary biology (e.g., Woese and Fox, 1977; Cedergen et al., 1988; Gouy and Li, 1989; Sogin et al., 1989; Hasegawa et al., 1993; Doolittle and Brown, 1994; Halanych et al., 1995).

A decade ago, mainly due to the work of Field et al. (1988), SSU rRNA (18S rRNA) was also introduced to trace metazoan relationships. Since then there has been an "18S rRNA sequence-boom," 413 complete or nearly complete metazoan 18S rRNA sequences were available this year in the database of Van de Peer et al. (1998). Nearly all metazoan phyla are now represented and for some phyla tens of species have been studied (e.g., Arthropoda, Mollusca, Annelida, Nematoda; for an overview of sequences available see Van de Peer et al., 1998).

Most of the 18S rRNA studies focus on only one or a few aspects of metazoan phylogeny and very often include only a small part of the available 18S rRNA sequence data. Relevant taxa are sometimes even not considered without giving a valid explanation. Arbitrarily, we regard a particular taxon to be relevant for a certain problem if there is any previous indication that this taxon is closely related to the organisms under consideration or that this taxon belongs to the clade which is likely to contain one of the species under consideration.

Not only the number of available 18S rRNA sequences has increased dramatically but also alignments, tree construction methods and stability tests have improved over the course of time. Yet, earlier results are seldomly re-assessed using more recent alignments and more advanced tree construction methods (e.g., Adoutte and Philippe, 1993). In a previous paper (Winnepenninckx and Backeljau, 1996), we reported on the effect that alignment may have on phylogenetic results. The present paper assesses the impact of broader taxon sampling and/or the use of a more advanced distance matrix calculation method on previously published analyses. In particular we intend to: 1) examine the hypotheses of Kim et aL (1996) on annelid, molluscan, and arthropod monophyly by adding relevant sequences that were originally omitted from their analyses; 2) give an update of our previously published paper on the relationships of protostome worms (Winnepenninckx et al., 1995a) by adding recently determined sequences to the data set; 3) assess the effect of using a more advanced, recently developed distance matrix calculation method and of including additional sequences on the results on mesozoan relationships (Katayama et al., 1995, Hanelt et al., 1996; Pawlowski et al., 1996).