J Mol Evol (1996) 42:552-559 JO U R N A L O F MOLECULAR EVOLUTION

© Springer-Verlag New York Inc. 1996

18S rRNA Suggests That Entoprocta Are Protostomes, Unrelated to Ectoprocta

L.Y. Mackey,1 B. Winnepenninckx,2 R. De Wachter,2 T. Backeljau,3 P. Emschermann,4 J.R. Garey1

1 Department of Biological Sciences, Duquesne University, Pittsburgh, PA 15282, USA 2 Departement Biochemie, Universiteit Antwerpen (UIA), Universiteitsplein 1, B-2610 Antwerpen, Belgium 3 Koninklijk Belgisch Instituut voor Natuurwetenschappen, Vautierstraat 29, B-1040 Brussel, Belgium 4 Institut für Biologie III, Albert-Ludwig-Universität-Freiburg, Schanzlestrasse 1, D-79104 Freiburg, Germany

Received: 27 May 1995 / Accepted: 8 January 1996

Abstract. The Ento- and Ectoprocta are sometimes Introduction placed together in the Bryozoa, which have variously been regarded as proto- or deuterostomes. However, En­ The phylum Entoprocta (Kamptozoa) comprises about toprocta have also been allied to the pseudocoelomates, 150 species of small, sessile, solitary or colonial species while Ectoprocta are often united with the Brachiopoda (Emschermann 1985). In their life cycle and larval form and Phoronida in the (super)phylum Lophophorata. they show some similarities to the Ectoprocta, which Hence, the phylogenetic relationships of these taxa are prompted Nielsen (1977) to place both taxa together in still much debated. We determined complete 18S rRNA the Bryozoa (moss animals). According to Jägersten sequences of two entoprocts, an ectoproct, an inarticulate (1972) the Entoprocta can be derived from mollusc-like brachiopod, a phoronid, two annelids, and a platyhel- ancestors, i.e., coelomate protostomes. Nielsen (1994), minth. Phylogenetic analyses of these data show that (1) on the other hand, sees a definite affinity between the entoprocts and lophophorates have spiralian, protosto- Entoprocta/Ectoprocta and the Platyhelminthes (flat mous affinities, (2) Ento- and Ectoprocta are not sister worms) and Nemertini (ribbon worms). Other authors taxa, (3) phoronids and brachiopods form a monophy- (Clark 1921; Cori 1936) have denied any close relation­ letic clade, and (4) neither Ectoprocta or Annelida appear ship between Ecto- and Entoprocta and consider the lat­ to be monophyletic. Both deuterostomous and pseudo- ter as neotenic annelid trochophorae (Emschermann coelomate features may have arisen at least two times in 1982, 1985). All these postulated relationships have been evolutionary history. These results advocate a Spiralia- questioned because the Entoprocta possess only a small Radialia-based classification rather than one based on mesenchymous body cavity, which points to a pseudo- the Protostomia-Deuterostomia concept. coelomate or even acoelomate origin (e.g., Hyman 1951; Brusca and Brusca 1990). Moreover, the phylum Ecto­ Key words: Ectoprocta — Entoprocta — Phoronida procta (moss animals sensu stricto) is also often united — Brachiopoda — Lophophorata — 18S rRNA — Mo­ with Brachiopoda (lamp shells) and Phoronida (horse­ lecular phylogeny — Oligochaeta — Hirudinida — shoe worms) in the (super)phylum Lophophorata or Ten­ Polychaeta taculata (e.g., Valentine 1973; Zimmer 1973; Gutmann 1978; Emig 1984). These three groups are mainly joined on the basis of their overall tripartite body plan and the presence of a lophophore, which is a horseshoe-shaped, Abbreviations: NJ: neighbor-joining; MP: maximum parsimony ciliated, tentacular feeding organ containing coelomic Correspondence J.R.to: Garey extensions. Yet, the phylogenetic relationships of the lo- 553 phophorates remain controversial since they show an position 111, was determined on a restriction fragment cloned into the amalgam of deuterostome and protostome features. On plasmid pBluescript SK+. The major part was determined on a PCR amplification product obtained by means of primers binding to con­ the basis of their oligomerous body plan, the three lo- served areas of the 18S and 28S rRNA gene (Winnepenninckx et al. phophorate groups have traditionally been allied to the 1994; Van der Auwera et al. 1994). The 18S rRNA gene ofLingula deuterostomes (e.g., Zimmer 1973; Emig 1984). Nielsen lingua and Haemopis sanguisuga was amplified in two overlapping ( 1977) however, concluded that ectoprocts have evolved parts using primers complementary to the 5' and 3' of the 18S rRNA from an entoproct-like ancestor and that both groups gene (Winnepenninckx et al. 1994) and primers complementary to a conserved part of the 18S rRNA gene and the 5' end of the 28S rRNA have protostome affinities. The Phoronida and Bra­ gene (Winnepenninckx et al. 1994; Van der Auwera et al. 1994). All chiopoda seem to him more related to Deuterostomia PCR amplification products were ligated into T-tailed PSK+ vector (Nielsen 1977, 1994). According to Gutmann (1978), the (Biorad, USA) and cloned into DH5 E. coli bacteria. To avoid an monophyletic lophophorates have been derived from a enhancement of the error rate by cloning the PCR products prior to protostome annelid-like metamerous ancestor. Some sequencing (e.g., Bevan et al. 1992), plasmids were extracted from ten different clones and pooled. Dideoxynucleotide sequencing of both workers even consider lophophorates as an intermediary DNA strands of the pooled plasmids was performed with USB (Cleve­ group between protostomes and deuterostomes (e.g., land, OH, USA) and Pharmacia (Uppsala, Sweden) kits using 16 oli­ Siewing 1976, 1980; von Salvini-Plawen 1982). An gonucleotide primers (Winnepenninckx et al. 1994). The 18S rRNA analysis of partial 18S rRNA sequences, (Field et al. gene of Phoronis architecta. Pedicellina cernua. Barentsia benedeni, 1988) suggested that the brachiopod Lingula reevi be­ and Nereis limbata was PCR amplified in two overlapping fragments spanning nearly the complete gene, corresponding to nucleotides 130- longs to a cluster of Protostomia. More robust analyses 1965 of the human sequence (EBI [EMBL] accession number were carried out by Halanych et al. (1995) which indi­ M10098). PCR products were cloned into M13 mpl8 nondirectionally cated that the lophophorates are polyphyletic, with Bra­ with an appropriate restriction enzyme. DNA sequencing was carried chiopoda + Phoronida forming a clade separate from out completely in both directions from the M13 templates of a single Ectoprocta. They found that all three lophophorate phyla clone, with the dideoxynucleotide sequencing method using Sequenase (USB, Cleveland, OH, USA), commercial M13 primers, and conserved were protostomes, forming a clade with Annelida and internal primers. Additional sequencing reactions were carried out us­ Mollusca, which they named Lophotrochozoa as a varia­ ing inosine mixes as needed to resolve some sequencing artifacts (Au- tion of the Eutrochozoa taxon (Ghiselin 1988; Eemisse et subel et al. 1995). Since for Phoronis architecta, Pedicellina cernua, al. 1992). Barentsia benedeni, and Nereis limbata, only a single clone was se­ We further assessed the phylogenetic relationships of quenced, their sequence may contain minor amplification errors. Esti­ mations of the error rate of Taq polymerase range from 2 x 1CT4 to less the “ lophophorate” phyla Ectoprocta, Brachiopoda, and than 1 x IO-5 according to Eckert and Kunkel (1991) and 2.75 x lCT3 Phoronida by determining new complete or nearly com­ according to Bej et al. (1991). plete 18S rRNA sequences of representatives of these three taxa as well as those of two annelids and a platy- Alignment and Tree Construction. Sequences were fitted into an helminth. We also present the first 18S rRNA gene se­ alignment of small-subunit rRNA sequences (Van de Peer et al. 1994) quences from two entoprocts in order to explore the evo­ using a computer program developed by De Rijk and De Wachter lutionary relationships of Entoprocta to Ectoprocta and (1993). Subsequent manual adjustments were made on the basis of secondary-structure features (Van de Peer et al. 1994). Previously pub­ other protostome taxa. lished sequences used for the phylogenetic analyses have the following EBI accession numbers: Acanthopleura japonica, X70210; Anemonia sulcata, X53498; Artemia salina, X01723; Branchiostoma floridae, M97571; Eisenia fetida, X79872; Eurypelma californica, X I3457; Materials and Methods Glottidia pyramidata, U12647; Glycera americana, U19519; Gordius aquaticus, X87985;Herdmania momus, X53538; Homo sapiens, Biological Materials and DNA Extraction. The animals examined X03205; Lanice conchilega, X79873; Limicolana kambeul, X66374; in the present study are the ectoproct Alcyonidium gelatinosum (Oos­ Ochetostoma erythrogrammon, X 79875; Opisthorchis viverrini, tende, Belgium), the brachiopod Lingula lingua (Hong Kong), the an­ X55357; Phascolosoma granulatum, X79874;Phoronis vancouveren­ nelids Nereis limbata (Polychaeta) (Panacea, Florida) and Haemopis sis, U l2648; Placopecten magellanicus, X53899; Plumatella repens, sanguisuga (Hirudinida) (Vrouwenpolder, The Netherlands), the platy- U12649; Saccharomyces cerevisiae, V01335; Schistosoma mansoni, helminth Bipalium sp. (Sao Miguel, Azores), the phoronid Phoronis X53498; Tenebrio molitor, X07801; Terebratalia transversa, U12650; architecta (Panacea, Florida), and the entoprocts Pedicellina cernua and Tripedalia cystophora, L10829. and Barentsia benedeni from laboratory cultures (Emschermann 1987). Distances between sequences were calculated using the method of DNA of Alcyonidium gelatinosum, Lingula lingua, HaemopisKimura san­ (1980), modified to take gaps into account (Van de Peer et al. guisuga, and Bipalium sp. was extracted as described by Winnepen- 1990). Neighbor-joining (NJ) analyses (Saitou and Nei 1987) were ninckx et al. (1993). DNA ofPhoronis architecta. Pedicellina cernua, performed with the program TREECON (Van de Peer and De Wachter Barentsia benedeni, and Nereis limbata was prepared according to 1993) using all aligned sites. Maximum Parsimony (MP) trees were Hempstead et al. (1990). constructed using the PAUP (Swofford 1993) heuristic search option with branch swapping (tree bisection reconnection), treating gaps as missing data. TREECON was used to identify 756 informative parsi­ Amplification and Sequencing of the 18S rRNA The Genes. Alcy­ mony sites utilized in the MP analysis. onidium gelatinosum DNA was partially degraded and therefore the The confidence of all trees was estimated via bootstrapping, run­ 18S rRNA gene was PCR amplified in six overlapping fragments with ning 1,000 replicates. Only nodes having bootstrap values higher than primer combinations in the gene itself and in the 5.8S rRNA gene 70% were considered to be reliable (Hillis and Bull 1993). PAUP (Wilmotte et al. 1993; Winnepenninckx et al. 1994). The minor part of (Swofford 1993) was used to carry out the decay index analysis (Breme the 18S rRNA gene of Bipalium sp., up to a HindlU site at nucleotide 1988; Donoghue et al. 1992) and to examine alternate MP trees. AÍ- 554 temate topologies using minimum evolution criteria were examined form a clade, we analyzed trees in which each of the two with PHYLTEST (Kumar 1995) using four-cluster analyses (Rzhetsky ectoproct sequences were used as clusters, in which dip- et al. 1995) with a Kimura two-parameter distance model and a cor­ loblasts were used as an outgroup, and in which the taxa rection for unequal rates of substitution at different sites (Winnepen­ ninckx et al. 1995). used as the fourth cluster varied in each of eight trees (Table 1A). The two ectoproct sequences only grouped together in three of the eight best trees, but with low to high confidence that the best tree was better than alter­ Results nate trees with Alcyonidium + Plumatella (CP values 20-91). Similarly, we tested if Plumatella and Bra­ Newly obtained 18S rRNA sequences were submitted to chiopoda + Phoronida form a clade (Table IB) and found EBI or GenBank under the following accession numbers that in two of eight tests, Brachiopoda + Phoronida were (numbers of nucleotides determined for each sequence more closely allied to other phyla (Echiura and Mol­ are indicated in parentheses): Alcyonidium gelatinosum lusca) than to Plumatella but with low confidence (CP (1813), X91403; Barentsia benedeni (1762), U36272; values 49 and 61). We also tested if Alcyonidium and Bipalium sp. (1805), X91402; Haemopis sanguisuga Brachiopoda + Phoronida form a clade (Table 1C) and (1874), X91401;Lingula lingua (1813), X81631; Nereis found that in three of eight tests, Brachiopoda + Pho­ limbata (1818), U36270; Pedicellina cernua (1807), ronida were closer to other phyla (Annelida, Echiura and U36273 and Phoronis architecta (1734), U36271. Mollusca) than to Alcyonidium with moderate to high Figure 1 shows the NJ tree obtained on the basis of confidence (CP values 73, 94, and 92). We found that in distances calculated from an alignment of complete or five of eight tests, Entoprocta are more closely related to nearly complete 18S rRNA sequences of 32 Metazoa other phyla (Annelida, Arthropoda, Echiura, Mollusca, using the formula of Kimura (1980) modified by Van de and Sipuncula) than to the ectoproct Alcyonidium (Table Peer et al. (1990) to take gaps into account. The MP tree ID) with low to high confidence (CP values 84, 55, 97, obtained on the basis of this alignment is shown in Fig. 71, 75, respectively). Similarly, in two of eight tests, 2. It is a strict consensus of two trees with a length of Entoprocta are more closely related to other phyla (An­ 4,127 obtained by heuristic search. The consistency in­ nelida and Echiura) than to the ectoproct Plumatella dex was 0.368; there were 17 trees which were one step (Table IE) with low confidence (CP values 11 and 67). longer and 85 trees that were two steps longer. Both the Finally, we tested if Entoprocta groups with Brachiopoda NJ and MP trees suggest that Entoprocta, Ectoprocta, + Phoronida (Table IF) and found with high confidence Brachiopoda, and Phoronida are protostome spiralians. that in three of eight tests, other phyla (Annelida, However, the branching pattern within this clade is un­ Echiura, and Mollusca) were more closely related to Bra­ stable and differs in the NJ and MP analyses, usually chiopoda + Phoronida than Entoprocta to Brachiopoda + with low bootstrap values. The decay indices in Fig. 2 Phoronida (CP values 84, 98, and 98). illustrate the instability of the MP tree. The lophopho­ rates do not appear as a monophyletic group, and an MP tree with the lophophorates as a monophyletic group would be at least 12 steps longer than the shortest tree. Discussion The Annelida appear to be paraphyletic in both trees. An Oligochaeta-Hirudinida cluster is seen in both trees, and the Polychaeta also appear paraphyletic in the two trees. This study provides further evidence that the three “lo- There is strong support for the monophyly of Pho­ phophorate” phyla are clearly protostomes, allied with ronida and Brachiopoda in the NJ tree but there is no Mollusca, Annelida, and other protostome phyla. The NJ evidence for monophyly of Ectoprocta because the two tree supports the Lophotrochozoa taxon of Halanych et ectoprocts, Plumatella repens and Alcyonidium gelatino­al. (1995), but there is less support with the MP tree. The sum fail to form a clade in either tree. Alcyonidium gela- results also confirm that the three ‘ ‘lophophorate’ ’ phyla tinsoum groups with a sipunculid in both trees while are not a monophyletic assemblage, but that Brachiopoda Plumatella repens appears near the arthropods in the MP + Phoronida are a monophyletic group. An MP tree con­ tree but within the Eutrochozoa in the NJ tree. The En­ taining a monophyletic clade of the three “lophopho­ toprocta form their own clade among the Eutrochozoa rate” taxa would be at least 12 steps longer than the and do not appear closely linked to any of the lophopho­ shortest tree. Although there is high statistical support for rate phyla. An MP tree with a polyphyletic Lophophorata Brachiopoda + Phoronida in the NJ tree, both trees (Figs. and a monophyletic Ectoprocta with Entoprocta as a sis­ 1, 2) show Phoronida and Brachiopoda as paraphyletic ter group would be at least 20 steps longer than the taxa with uncertain branching within the Brachiopoda + shortest tree. Phoronida clade. Four-cluster analysis was used to test several hypoth­ Both the NJ and MP trees show the Ectoprocta as a eses using minimum-evolution criteria. To test the hy­ polyphyletic taxon. Alcyonidium gelatinosum is an ecto­ pothesis that the ectoprocts Alcyonidium and Plumatella proct of the class Gymnolaemata, considered to be the 555

- Homo sapiens -ffir VERTEBRATA DEUTEROSTOME 98 L ■ Branchiostoma floridae CHORDATA < : COELOMATA o ------Herdmania momus UROCHORDATA < —i DC D istance 0.1 - Tenebrio molitor INSECTA Artemia salina CRUSTACEA ARTHROPODA ■ Eurypelma californica CHELICERATA — Eisenia fetida OLIGOCHAETA HIRUDINIDA - Lanice conchilega ANNELIDA - G lycera am ericana POLYCHAETA - Nereis limbata PROTOSTOME ■ Ochetostoma erythrogrammon ECHIURA COELOMATA Limicolana kambeul GASTROPODA Acanthopleura japonica POLYPLACOPHORA MOLLUSCA — Placopecten magellanicus BIVALVIA I Phoronis vancouverensis PHORONIDA a. Phoronis architecta C/D Ungula lingua INARTICULATA Glottidia pyramidata : BRACHIOPODA Terebratalia transversa ARTICULATA inoi Pedicellina cernua ENTOPROCTA PSEUDOCOELOMATA ‘------Barentsia benedeni ■ Plumatella repens PHYLACTOLAEMATA ECTOPROCTA PROTOSTOME ■ Alcyonidium gelatinosum GYMNOLAEMATA COELOMATA Phascolosoma granulatum SIPUNCULA _ ] - Gordius aquaticus NEMATOMORPHA PSEUDOCOELOMATA - Opistorchis viverrini TREMATODA - Schistosoma mansoni : PLATYHELMINTHES ACOELOMATA - Bipalium sp. TURBELLARIA

' Anem onia sulcata ANTHOZOA CNIDARIA — Tripedalia cystophora CUBOZOA n ■ Saccharomyces cerevisiae FUNGI

Fig. 1. Neighbor-joining tree based on an alignment of 32 complete or nearly complete metazoan 18S rRNA sequences using Kimura (1980) distances. Saccharomyces cerevisiae was used as the outgroup. Numbers above the nodes are bootstrap values (percent of 1,000 replicates), shown only when higher than 50%. Branches are drawn to scale.

more “primitive” ectoproct (Nielsen 1995), while Plu­ Entoprocta to Ectoprocta and Brachiopoda + Phoronida. matella repens is a member of the class Phylactolaemata. Table ID shows with high confidence that Entoprocta is Alcyonidium groups with Sipuncula in both trees, but more closely related to a number of phyla than to Alcy­ with low statistical support. Alcyonidium and the sipun- onidium (CP values from 55 to 97). The relationship of culid sequence have the longest branches within the pro­ Entoprocta to Plumatella is less clear, but Entoprocta tostome taxa, and form an outgroup to the nonarthropod appears to be more closely related to Annelida and protostomes, suggesting that the placement of Alcy­ Echiura than to Plumatella but with little statistical sup­ onidium with Sipuncula could be due to long branch port (CP values 11 and 67, respectively). Finally, four- attraction (Felsenstein 1988; Olsen and Woese 1993). cluster analysis indicated that Entoprocta is more closely Additional sequences from other (slow evolving) ecto­ related to Annelida, Echiura, and Mollusca than to Bra­ procts will be needed to determine if Ectoprocta is a chiopoda + Phoronida (CP values 84, 98, and 98, respec­ polyphyletic group. Four-cluster analyses do not support tively). The branch leading to Alcyonidium is one of the the hypothesis that Ectoprocta are monophyletic (Table longest in the NJ tree, and there was concern that it might 1A) or that either Alcyonidium or Plumatella form a affect the placement of Entoprocta and Plumatella, so we clade with Brachiopoda + Phoronida (Table IB, 1C), constructed NJ and MP trees (not shown) without the although in the case of Plumatella, CP values are only 68 Alcyonidium sequence in the analysis and found that the and 49 (respectively) that Brachiopoda + Phoronida relative positions of Entoprocta and Plumatella were un­ grouping with Echiura and Mollusca are better trees than changed. The branches leading to the entoprocts are Brachiopoda + Phoronida grouping with Plumatella. nearly identical in length to the branch leading to Plu­ The two entoprocts form a highly supported clade in matella in the NJ tree (Fig. 1), so unequal evolutionary both NJ and MP trees, well within the protostomes, allied rates between the two sequences are not likely to affect with Mollusca, Annelida, and other phyla but not closely their respective locations. Therefore, based on the data associated with any of the “lophophorate” clades in­ presented here, we are confident that Entoprocta and Ec­ cluding either of the ectoprocts. An MP tree with Ento­ toprocta are not closely related. procta allied with Ectoprocta would be at least 20 steps The inclusion of the Entoprocta within the protostome longer than the shortest tree. Extensive four-cluster clade agrees with some larval similarities (Nielsen 1977; analyses were used to further explore the relationship of Emschermann 1985). It also supports the hypothesis that 556

Homo sapiens VERTEBRATA DEUTEROSTOME CHORDATA Branchiostoma floridae : COELOMATA O Herdmania momus UROCHORDATA

Tenebrio molitor INSECTA PROTOSTOME Artemia salina CRUSTACEA ARTHROPODA COELOMATA

Eurypelma californica CHELICERATA

Gordius aquaticus NEMATOMORPHA PSEUDOCOELOMATA

Plumatella repens PHYLACTOLAEMATA ECTOPROCTA

Eisenia fetida OLIGOCHAETA PROTOSTOME c Haemopis sanguisuga HIRUDINIDA COELOMATA Lanice conchilega POLYCHAETA Nereis limbata

Pedicellina cernua ENTOPROCTA PSEUDOCOELOMATA 51s____ p Barentsia benedeni

Glycera americana POLYCHAETA ANNELIDA _ -HZ Ochetostoma erythrogrammon ECHIURA < ÇÇ Limicolana kambeul GASTROPODA Q_ Placopecten magellanicus BIVALVIA MOLLUSCA CO

Acanthopleura japonica POLYPLACOPHORA PROTOSTOME COELOMATA Phoronis architecta PHORONIDA Phoronis vancouverensis : Terebratalia transversa ARTICULATA

Lingula lingua BRACHIOPODA INARTICULATA Glottidia pyramidata : Alcyonidium gelatinosum GYMNOLAEMATA ECTOPROCTA

Phascolosoma granulatum SIPUNCULA ___

Opistorchis viverrini TREMATODA Schistosoma mansoni PLATYHELMINTHES ACOELOMATA Bipalium sp. TURBELLARIA

Anemonia sulcata ANTHOZOA

Tripedalia cystophora CUBOZOA

Saccharomyces cerevisiae FUNGI

Fig. 2. Strict consensus of the two equally most parsimonious trees (length = 4,127) based on 756 informative sites of an alignment of 32 complete or nearly complete 18S rRNA sequences. Branches are not drawn to scale. Bootstrap values higher than 50% (percent of 1,000 replicates) are indicated above the branches and decay indices are shown below the nodes.

some pseudocoelomate phyla evolved independently tain specializations of the reproductive system and the from protostome coelomate ancestors, possibly by neo- presence of a clitellum (e.g., Clark 1969; Brusca and teny (Emschermann 1982, 1985; Lorenzen 1985). On the Brusca 1990). basis of our findings, the similarities on which Nielsen The protostome affinity of the lophophorate phyla is (1977) based the close relationship between Ento- and in agreement with the results of more limited 18S rRNA Ectoprocta (e.g., life cycles, metamorphosis, budding) studies (Field et al. 1988; Halanych et al. 1995). These should be ascribed to convergent adaptations, probably results suggest that the supposed deuterostomous char­ due to a common sessile colonial life style. The mono­ acters (Zimmer 1973; Emig 1984) of the Ectoprocta, phyly of the Annelida, which is broadly accepted (e.g., Brachiopoda, and Phoronida (e.g., tripartite coelom, Clark 1969), is not confirmed by our analyses. The pre­ blastopore and mesoderm formation, cleavage type and sent results suggest that the two main synapomorphies of nervous system) are either incorrectly interpreted or have the Annelida viz. the annelid head and the epidermal, been derived independently in different lineages. The paired setae (e.g., Brusca and Brusca 1990), would be latter possibility could support the hypothesis of Berg­ due to convergent adaptations. However, their non- ström (1986) that protostome characters are bilaterian monophyly is only weakly supported and more annelid synapomorphies, which when lacking must have been 18S rRNA sequences are required. Unequal evolutionary lost secondarily. The protostomous relationships of the rates among the annelid sequences presented here could lophophorates are also supported by some biochemical affect their placement. For example, the branch leading (Jeuniaux 1982; Willmer 1990) and morphological (Gus- to Nereis limbata is over twice as long as the branch tus and Cloney 1972; Gutmann 1978; Specht 1988) data, leading to Glycera americana. The monophyly of the whereas the close affinity between Phoronida and Bra­ Clitellata ( = Oligochaeta and Hirudinida) is in accor­ chiopoda agrees with most embryological and morpho­ dance with several morphological features such as cer­ logical studies (e.g., Human 1959; Zimmer 1973; 557

Table 1. Comparison of alternate branching orders of metazoan groups using four-cluster analysis3

A. Relationship of Plumatella (PI), Alcyonidium (AÍ), and Diploblasts (D) with:

Best tree Alternate I CP, Alternate II CP, Annelida (A) ([PI, A], [Al, D]) ([PI, AÍ], [A, D]) 73 ([PI,D],[A,Al]) 92 Arthropoda (Ar) ([Pl,Ar],[Al,Dj) ([Pl,Al],[Ar,D]) 65 ([Pl,D],[Ar,Al]) 80 Deuterostomia (U) ([P1,AI],[U,D]) ([P1,U],[A1,D]) 92 ([P1,D],[U,A1]) 99 Echiura (C) ([P1,C],[AI,D]) ([P1,A1],[C,D]) 91 ([P1,C],[D,A1]) 79 Mollusca (M) ([P1,M],[A1,D]) ([P1,A1],[M,D]) 59 ([P1,D|,|M,A1J) 75 Nematomorpha (N) ([P1,AI],[N,D]) ([P1,N],[A1,D]) 40 ([P1,D1,[N,A1]) 97 Platyhelminthes (P) ([P1,A1],[P,D]) ([P1,P],[A1,D]) 65 ([P1,D],[P,A1]) 98 Sipuncula (S) ([P1,D1,[A1,S]) ([P1,A1],[S,D]) 20 ([P1,S],[D,A1]) 32

B. Relationship of Plumatella (PI), Brachiopoda + Phoronida (B), and Diploblasts (D) with:

Best tree Alternate I CP, Alternate II CP, Annelida (A) ([P1,B],[A,D]) (IP1,J,[B,D]) 92 ([P1,D],[A,B]) 45 Arthropoda (Ar) ([Pl,B],[Ar,D]) ([Pl,Ar],[B,D]) 82 ([Pl,D],|Ar,B]) 99 Deuterostomia (U) ([P1,B],[U,D]) ([P1,U],[B,D]) 99 ([P1,D],[U,B]) 99 Echiura (C) ([P1,D],[B,C]) ([P1,B],[C,D]) 68 ([P1,C],[D,B]) 91 Mollusca (M) ([P1,D],[B,M]) ([P1,B],[M,D]) 49 ([P1,M],[D,B]) 99 Nematomorpha (N) ([P1,B],[N,D]) ([P1,N],[B,D]) 99 ([P1,D|,|N,B]) 99 Platyhelminthes (P) ([P1,B],[P,D]) (IP1,P],[B,D)) 99 ([P1,D],[P,B]) 99 Sipuncula (S) ([P1,B],[S,D]) ([P1,S],[B,D]) 99 ([P1,D],[S,B]) 99

C. Relationship of Alcyonidium (AÍ), Brachiopoda + Phoronida (B), and Diploblasts (D) with:

Best tree Alternate I CP, Alternate II CP, Annelida (A) ([A1,D],[B,A]) ([A1,B],[A,D]) 73 ([A1,A],[B,D]) 97 Arthropoda (Ar) ([Al,B],[D,Ar]) ([Al,Ar],[B,D]) 91 ([Al,B],[D,Ar]) 77 Deuterostomia (U) ([A1,B],[D,U]) ([A1,U],[B,D]) 99 ([A1,D],[B,UJ) 99 Echiura (C) ([A1,D],[B,C]) ([A1,B],[C,D]) 94 ([A1,C],[B,D]) 99 Mollusca (M) ([A1,D],[B,M]) ([A1,B],[M,D]) 92 ([A1,M],[B,D]) 99 Nematomorpha (N) ([AI,B],[D,N]) ([A1,N],[B,D]) 99 ([A1,D],[B,N]) 98 Platyhelminthes (P) ([AI,B],[D,P]) ([A1,P],[B,D1) 99 ([AI,D],[B,P]) 97 Sipuncula (S) ([A1,B],[D,S]) ([A1,S],[B,D]) 49 ([A1,D],[B,S]) 63

D. Relationship of Entoprocta (E), Alcyonidium (AÍ), and Diploblasts (D) with:

Best tree Alternate I CP, Alternate II CP, Annelida (A) ([E,A],[A1,D]) ([E,A1],[A,D]) 84 ([E,D],[A1,A]) 62 Arthropoda (Ar) ([E,D],[Al,Ar]) ([E,Al],[ArJ5]) 55 ([E,Ar],[Al,D[) 79 Deuterostomia (U) ([E,A1],[U,D]) ([E,U],[A1,D]) 88 ([E,D],[A1,U]) 89 Echiura (C) ([E,C],[A1,D]) ([E,A1],[C,D]) 97 ([E,D],[A1,C]) 34 Mollusca (M) ([E,D],[A1,M]) ([E,A1],[M,D]) 71 ([E,M],[A1,D]) 02 Nematomorpha (N) ([E,A1],[N,D]) ([E,N],[A1,D]) 16 ([E,DJ,[A1,N]) 88 Platyhelminthes (P) ([E,A1],[P,D]) ([E,P],[A1,DD 54 ([E,D],[A1,P]) 88 Sipuncula (S) ([E,D],[A1,S]) ([E,A1],[S,D]) 75 ([E,S],[A1,D]) 95

E. Relationship of Entoprocta (E), Plumatella (PI), and Diploblasts (D) with:

Best tree Alternate 1 CP, Alternate II CP, Annelida (A) ([E,D],[A,P1]) ([E,PI],[A,D]) 11 ([E,A],[P1,D]) 86 Arthropoda (Ar) ([E,Pl],[Ar,D]) ([E,Ar],[Pl,D]) 99 ([E,D],[Pl,Ar]) 19 Deuterostomia (U) ([E,P1],[U,D]) ([E,U],[P1,D]) 99 ([E,D],[P1,U]) 99 Echiura (C) ([E,D],[C,P1]) ([E,P1],[C,D]) 67 ([E,C],[P1,D]) 68 Mollusca (M) ([E,P1],[M,D]) ([E,M],[P1,D]) 97 ([E,D],[P1,M]) 32 Nematomorpha (N) ([E,P1],[N,D]) ([E,N],[P1,D]) 99 ([E,D],[P1,N]) 91 Platyhelminthes (P) ([E,P1],[P,D]) ([E,P],[P1,D]) 99 ([E,D],[P1,P]) 97 Sipuncula (S) ([E,P1],[S,D]> ([E,S],[P1,D]) 99 ([E,D],[P1,S]) 67

F. Relationship of Entoprocta (E), Brachiopoda + Phoronida (B), and Diploblasts (D) with:

Best tree Alternate I CP, Alternate II CP, Annelida (A) ([E,D],[A,B]) ([E,B],[A,D]) 84 ([E,A|,[B,D]) 97 Arthropoda (Ar) ([E,B],[Ar,D]) ([E,Ar],[B,D]) 99 ([E,D],[B,Ar]) 84 Deuterostomia (U) ([E,B],[U,D]) ([E,U],[B,D]) 99 ([E,D],[B,U]) 99 Echiura (C) ([E,D),[C,B]) ([E,B],[C,D]) 98 (|E,C1,[B,D1) 98 Mollusca (M) ([E,D],[M,B]) ([E,B],[M,D]) 98 ([E,M],[B,D1) 99 Nematomorpha (N) ([E,B],[N,D]) ([E,N],[B,D]) 99 ([E,D],[B,N]) 99 Platyhelminthes (P) ([E,B],[P,D]) ([E,P],[B,D]) 99 ([E,D],[B,P]) 99 Sipuncula (S) ([E,B],[S,D]) (LE,S],[B,D[) 99 ([E,D],[B,S]) 80 a CP] and CPn are the confidence probability values supporting the best tree as ronis architecta, Phoronis vancouverensis, Terebratalia transversa; Deuterosto- better than alternate tree I or II, respectively (Kumar 1995). The trees supporting mia: Branchiostoma floridae, Herdmania momus, Homo Diploblasts:sapiens; the hypotheses tested are shown in bold (see text) Anemonia sulcata, Tripedalia cystophora; Echiura: Ochetostoma erythrogram- Definition of clusters: Alcyonidium:Alcyonidium gelatinosum; Annelida: Ei- mon; Entoprocta:Barentsia benedeni, Pedicellina cernua; M ollusca:Acantho- senia fetida, Glycera americana, Haemopis sanguisuga, Lanice conchilega, pleura japonica, Limicolana kambeul, Placopecten Nematomor- magellanicus; Nereis limbata; Arthropoda:Artemia salina, Eurypelma californica, Tenebrio pha: Gordius aquaticus; Plumatella: Plumatella repens molitor; Brachiopoda + Phoronida:Glottidia pyramidata, Lingula lingua, Pho- 558

Nielsen 1977; Emig 1984). All these findings advocate Ghiselen MT (1988) The origin of molluscs in the light of molecular the use of a Spiralia-Radialia-based phylogenetic classi­ evidence. In: Harvey PH, Partridge L (eds) Oxford surveys in evo­ lutionary biology, vol 5. Oxford University Press, Oxford, pp 66- fication instead of the commonly used Protostomia- 95 Deuterostomia concept. Gustus RM, Cloney RA (1972) Ultrastructural similarities between setae of brachiopods and polychaetes. Acta Zool 53:229-233 Acknowledgments. We are indebted to Dr. B. Morton for collecting Gutmann WF (1978) Brachiopods: biomechanical interdependencies some of the animals. The work was performed in the framework of the governing their origin and phylogeny. Science 100:890-893 ISBE of the University of Antwerp and as funded by FJBR grants No. Halanych KM, Bachelier JD, Aguinaldo AMA, Liva SM, Hillis DM, 2.0023.94 and 2.0030.93. Parts of this work were supported by USDA Lake JA (1995) Evidence from 18S ribosomal DNA that the lo­ grants 9304072 and 9502133 to J.R.G. and by the Deutsche Forsch­ phophorates are protostome animals. Science 267:1641-1643 ungs-Gemeinschaft. B.W. holds an IWT scholarship. Hempstead PG, Regular SC, Ball IR (1990) A method for the prepa­ ration of high-molecular-weight DNA from marine and freshwater triclads. DNA Cell Biol 9:57 References Hillis DM, Bull JJ (1993) An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Syst Biol Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith 42:182-192 JA, Struhl K (1995) Current protocols in molecular biology. Hyman LH (1951) The invertebrates: Acanthocephala, Aschelminthes Greene, New York and Entoprocta—the pseudocoelomate Bilateria. McGraw-Hill, Bej AK, Mahbubani MFI, Atlas RM (1991) Amplifications of nucleic New York acids by polymerase chain reaction (PCR) and other methods and Hyman LH (1959) The invertebrates: smaller coelomate groups: Chae­ their applications. Crit Rev Biochem Mol Biol 26:301-334 tognatha, Hemichordata, Pogonophora, Phoronida, Ectoprocta, Bergström J (1986) Metazoan evolution—a new model. Zool Ser 15: Brachiopoda, Sipunculida. McGraw-Hill, New York 189-200 Jägersten G (1972) Evolution of the metazoan life cycle. Academic Bevan IS, Rapley R, Walker MR (1992) Sequencing of PCR-amplified Press, New York DNA. PCR Methods Appl 1:222-228 Jeuniaux C (1982) La chitine dans le règne animal. Bull Soc Zool FR Bremer K (1988) The limits of amino acid sequence data in angiosperm 107:363-386 phylogenetic reconstruction. Evolution 2:795-803 Kimura M (1980) A simple method for estimating evolutionary rates of Brusca RC, Brusca GJ (1990) Invertebrates. Sinauer, Sunderland, MA base substitutions through comparative studies of nucleotide se­ Clark AH (1921) A new classification of animals. Bull Inst Oceanogr quences. J Mol Evol 16:111-120 Monaco 400:1-24 Clark RB (1969) Systematics and phylogeny: Annelida, Echiura, Kumar S (1995) PHYLTEST: phylogeny hypothesis testing using the Sipuncula. In: Florkin M, and Scheer BT (eds) Chemical zoology, minimum evolution method. Pennsylvania State University, State vol IV: Annelida, Echiura and Sipuncula. Academic Press, New College York, pp 1-68 Lorenzen S (1985) Phylogenetic aspects of pseudocoelomate evolution. Cori C (1936) Kamptozoa. In: Bronn HG (ed) Klassen und Ordnungen In: Conway Morris S, George JD, Gibson R, Platt HM (eds) The des Tierreichs 4, II. Abt, 4. Buch Akad Verlagsgesellschaft, origins and relationships of lower invertebrates. Clarendon Press, Leipzig, pp 1-119 Oxford, pp 210-223 De Rijk P, De Wachter R (1993) DCSE, an interactive tool for se­ Nielsen C (1977) The relationships of Entoprocta, Ectoprocta and Pho­ quence alignment and secondary structure research. Comput Appl ronida. Am Zool 17:149-150 Biosci 9:735-740 Nielsen C (1994) Larval and adult characters in animal phylogeny. Am Donoghue MJ, Olmstead RG, Smith JF, Palmer JD (1992) Phyloge­ Zool 34:492-501 netic relationships of Dipsacales based on RbcL sequences. Ann Nielsen C (1995) Animal evolution, interrelationships of the living MO Bot Gard 79:333-345 phyla. Oxford University Press, Oxford Eckert KA, Kunkel TA (1991) DNA polymerase fidelity and the poly­ Olsen GJ, Woese CR (1993) Ribosomal RNA: a key to phylogeny. merase chain reaction. PCR Methods Appl 1:17-24 FASEB J 7:113-123 Eemisse DJ, Albert JS, Anderson FE (1992) Annelida and Arthropoda are not sister taxa: a phylogenetic analysis of spiralian metazoan Rzhetsky A, Kumar S, Nei M (1995) Four-cluster analysis: a simple morphology. Syst Biol 41:305-330 method to test phylogenetic hypotheses. Mol Biol Evol 12:163-167 Emig C (1984) On the origin of the Lophophorata. Z Zool Syst Evolut- Saitou N, Nei M (1987) The neighbor-joining method: a new method forsch 22:91-94 for reconstructing phylogenetic trees. Mol Biol Evol 4:406-425 Emschermann P (1982) Les Kamptozoaires. Etat actuel de nos con­ Siewing R (1976) Probleme und neuere Erkenntnisse in der Gross- naissances sur leur anatomie, leur développement, leur biologie et Systematik der Wirbellosen. Verh Dtsch Zool Ges Stuttgart 1976: leur position phylogénétique. Bull Soc Zool Fr 107:317-344 59-83 Emschermann P (1985) Cladus Kamptozoa. In: Siewing R (ed) Lehr­ Siewing R (1980) Das Archicoelomatenkonzept. Zool Jb Syst 103:439- buch der Zoologie. Band II: Systematik. Gustav Fisher Verlag, 82 Stuttgart, pp 576-586 Specht A (1988) Chaetae. In: Westheide W, Hermans CO. The ultra- Emschermann P (1987) A circulating water tank for culturing sessile or structure of the Polychaeta: Microfauna Marina, vol. 4 pp 45-59 hemisessile aquatic organisms in a continuous water current. Arch Hydrobiol 108:425^138 Swofford DL (1993) Phylogenetic analysis using parsimony, version 3.1. Illinois Natural History Survey, Champaign Felsenstein J (1988) Phylogenies from molecular sequences: inference and reliability. Annu Rev Genet 22:521-565 Valentine JW (1973) Coelomate superphyla. Syst Zool 22:97-102 Field KG, Olsen GJ, Lane DJ, Giovannoni SJ, Ghiselin MT, Raff EC, Van de Peer Y, Neefs JM, De Wachter R (1990) Small ribosomal Pace NR, Raff RA (1988) Molecular phylogeny of the animal king­ subunit RNA sequences, evolutionary relationships among differ­ dom. Science 239:748-753 ent life forms, and mitochondrial origins. J Mol Evol 30:463^-76 559

Van de Peer Y, De Wachter R (1993) TRECON: a software package for relationships among higher Fungi inferred from small ribosomal the construction and drawing of evolutionary trees. Comput Appl subunit RNA sequence analysis. Syst Appl Microbiol 16:436-444 Biosci 9:177-182 Winnepenninckx B, Backeljau T, De Wachter R (1993) Extraction of Van de Peer Y, Van den Broeck I, De Rijk R, De Wachter R (1994) high molecular weight DNA from molluscs. Trends Genet 9:407 Database on the structure of small ribosomal subunit RNA. Nucleic Winnepenninckx B, Backeljau T, De Wachter R (1994) Small ribo­ Acids Res 22:3488-3494 somal subunit RNA and the phylogeny of Mollusca. Nautilus Suppl van der Auwera G, Chapelle S, De Wachter R (1994) Structure of the 2:98-110 large ribosomal subunit RNA of Phytophtora megasperma, and Winnepenninckx B, Backeljau T, Mackey LY, Brooks JM, De Wachter phylogeny of Oomycetes. FEBS Lett 338:133-136 von Salvini-Plawen, L (1982) A paedomorphic origin of the oligomer- R, Kumar S, Garey JR (1995) 18S rRNA data indicate that the ous animals? Zool Ser 11:77-81 aschelminthes are polyphyletic in origin and consist of at least three Willmer P (1990) Invertebrate relationships. Cambridge University distinct clades. Mol Biol Evol 12:1132-1137 Press, Cambridge Zimmer RL (1973) Morphological and developmental affinities of the Wilmotte A, Van de Peer Y, Goris A, Chapelle S, De Baere R, Nelissen lophophorates. In: Larwood GP (ed) Living and fossil Bryozoa. B, Neefs JM, Hennebert G, De Wachter R (1993) Evolutionary Academic Press, London, pp 593-599