387489a0Nature3876632199705294894930028-0836199710.1038/387489a01476-467919977 November 19963 April 199729 May 1997ukNatureNatureNATUREnatureNature is a weekly international journal publishing the finest peer-reviewed research in all fields of science and technology on the basis of its originality, importance, interdisciplinary interest, timeliness, accessibility, elegance and surprising conclusions. Nature also provides rapid, authoritative, insightful and arresting news and interpretation of topical and coming trends affecting science, scientists and the wider public./nature/journal/v387/n6632issueJournal homeArchiveCurrent issueAdvance online publicationPrivacy policySubscribeNature Publishing GroupSupplementsCurrent issue387489a0Evidence for a clade of nematodes, arthropods and other moulting animals
AU  - Aguinaldo, Anna Marie A.
AU  - Turbeville, James M.
AU  - Linford, Lawrence S.
AU  - Rivera, Maria C.
AU  - Garey, James R.
AU  - Raff, Rudolf A.
AU  - Lake, James A.[ast] Molecular Biology Institute and MCD Biology, University  
 
of California, Los Angeles, California 90095, USA[dagger] Department of Biological Sciences, University of Arkansas, Fayetteville, Arkansas 72701, USA[Dagger] Department of Biological Sciences, Duquesne University, Pittsburgh, Pennsylvania 15282, and Department of Biology, University of South Florida, Tampa, Florida  
 
33620, USA[sect] Department of Biology and Indiana Molecular Biology Institute, Indiana University, Bloomington, Indiana 47405, USAThe arthropods constitute the most diverse animal group, but, despite their rich fossil record and a century of study, their phylogenetic relationships remain unclear1. Taxa previously proposed to be sister groups to the arthropods include Annelida, Onychophora, Tardigrada and others, but hypotheses of phylogenetic relationships have been  
 
conflicting2,3. For example, onychophorans, like arthropods, moult periodically, have an arthropod arrangement of haemocoel1,4, and have been related to arthropods in morphological and mitochondrial DNA sequence analyses4,5. Like annelids, they possess segmental nephridia and muscles that are a combination of smooth and  
 
obliquely striated fibres6. Our phylogenetic analysis of 18S ribosomal DNA sequences indicates a close relationship between arthropods, nematodes and all other moulting phyla. The results suggest that ecdysis (moulting) arose once and support the idea of a new clade, Ecdysozoa, containing moulting animals: arthropods, tardigrades, onychophorans,  
 
nematodes, nematomorphs, kinor-hynchs and priapulids. No support is found for a clade of segmented animals, the Articulata, uniting annelids with arthropods. The hypothesis that nematodes are related to arthropods has important implications for developmental genetic studies using as model systems the nematode Caenorhabditis elegans and the arthropod Drosophila  
 
melanogaster, which are generally held to be phylogenetically distant from each other.We have analysed relationships of arthropods to other taxa by sequencing complete 18S rDNAs from representative taxa, aligning them with existing 18S sequences from other metazoan taxa, and analysing them by using standard phylogenetic techniques7. This study confirms the suspected relationships between arthropods and other taxa, such as tardigrades and  
 
onychophorans. But by careful consideration of rates of evolution, we find the surprising result that nematodes are also closely related to arthropods. An outstanding problem with the molecular phylogeny of nematodes is that their 18S sequences evolve too rapidly to be useful for phylogenetic reconstruction. Previously published sequences of nematodes have a  
 
substitution rate 2-3 times greater than those of most other Metazoa. Hence special efforts were made to include only the slowest evolving sequences from representative taxa, because errors due to unequal rate effects and alignment artefacts are compounded by including rapidly evolving sequences8. To obtain three slowly evolving nematode sequences, 10-20 nematode 18S  
 
genes were sequenced (J.R.G., unpublished results). Marked differences are observed, depending upon whether rapidly or slowly evolving sequences are present (Fig. 1). When both rapidly and slowly evolving nematode sequences (bold type) are included, all nematodes branch from the base of the bilateral animals (Fig. la), whereas, when only the slowest nematode sequence  
 
is included (Fig. Ib), the nematode branches high within the protostomes as the sister taxon of the arthropods. Furthermore, analysis of the slowly evolving protein-synthesis elongation factor EF-la also place nematodes within the protostomes (J.R.G., A.M.A.A. and J.A.L., unpublished results), suggesting that other evolutionary processes9 are not responsible. These  
 
results are consistent with unequal rates artefactually placing rapidly evolving, long-branched nematode sequences adjacent to the long branch that joins the outgroup to the tree. Molecular sequence analysis, using the available fast-evolving 18S rRNA nematode sequences or faster evolving molecules, has demonstrated a similarly deep placement10, whereas some  
 
morphological studies have predicted a placement similar to that found using only the slowly evolving nematode sequence11. To exclude rapidly evolving taxa, all complete 18S rDNA sequences relevant to this study were systematically surveyed. An alignment of about 50 of the most useful complete sequences was constructed and the distances from each taxon to the last  
 
common ancestor of protostomes was calculated using the paralinear/LogDet method12'13 (Table 1). Guided by these distances, the slowest evolving protostome and outgroup taxa were selected (shown in bold). These included the slowest evolving sequences from the following taxa: a cnidarian as an outgroup to triploblastic animals7'14, a deuterostome as an outgroup to the  
 
protostome animals, a polychaete, an oligochaete, a brachiopod, a mollusc, a non-moulting aschelminth, representatives of the six phyla of non-arthropod moulting animals, and four major arthropod groups (a chelicerate, a crustacean, a myriapod and an insect). Figure 1 Phylogenetic analysis of 18S rDNA sequence data illustrating the effects of unequal rate biases on  
 
nematode placement. In a, both rapidly and slowly evolving nematode sequences (Caenorhabditis and Strongyloides, and Trichinella) are included in the analysis; the nematodes branch from the bottom of the tree, even before the deuterostome-protostome divergence. In b, only the slowly evolving Trichinella sequence is included and this nematode now branches from within  
 
the protostome clade, as the sister taxon to the arthropods. The majority-rule consensus tree derived from phylogenetic reconstructions is shown in Fig. 2. Four reconstruction methods were used, including paralinear (LogDet) distances12'13, maximum parsimony9, Kimura two-parameter distances28, and Jukes-Cantor distances28. Paralinear (LogDet) distances12'13 were  
 
emphasized because of their generality (most distance methods are special cases of paralinear distances). As preliminary calculations indicated an excess of constant sites (see Methods), all distance methods were corrected for site-to-site variation. Bootstrap values for these four methods, respectively, are shown adjacent to the interior nodes. In all of the  
 
reconstructions, the protostome taxa are clustered into two monophyletic groups. One clade, containing all the moulting animals (kinorhynch, priapulid, nematomorph, onycho-phoran, nematode, tardigrade, crustacean, insect, myriapod and chelicerate) is present in 95, 78, 85 and 79% of the trees derived through paralinear distances, maximum parsimony, Kimura two-parameter  
 
and Jukes-Cantor distances, respectively. The other protostome clade, containing the articulate brachiopod, mollusc, oligochaete, polychaete and rotifer, is present in 98,80,99 and 100% of the bootstrap replicates. A monophyletic protostome clade is also supported in 97, 83, 96 and 95% of the bootstrap replicates. Interpreted using the empirical results of Hillis and  
 
Bull15 as a guideline, these data provide significant support (P  0.05) for a clade of arthropod-related moulting animals within the protostomes. This conclusion is further supported by topology-dependent cladistic permutation tail probability tests confirming the significance of the arthropod-related clade (P  0.01). Figure 2 Phylogenetic analysis of 18S rDNA  
 
sequence data to determine relationships among the moulting metazoans. The moulting animals are present as the top ten taxa, the Lophotrochozoa are shown in the middle, and outgroups are shown at the bottom. The topology shown here is a majority-rule consensus combining the results from four individual majority-rule consensus trees derived using the following methods:  
 
paralinear/LogDet distances, maximum parsimony, Kimura two-parameter distances and Jukes-Cantor distances. All distance methods are corrected for site-to-site variation. The numbers next to the central branches represent the percentage of bootstrap replicates supporting the clades for these methods, respectively (from top to bottom). We initially found that flatworm  
 
sequences, like rapidly evolving nematode sequences, branched below the base of the bilateral animals. Hence multiple flatworm taxa were sequenced in order to obtain slowly evolving 18S sequences. In experiments similar to those shown in Fig. 1 (with flatworm sequences substituted for nematode sequences), flatworms were shown to branch artefac-tually deep. Given the  
 
importance of the phylogenetic position of the platyhelminthes to theories of the evolution of bilateral animals16'17, a tree containing slowly evolving lophotrochozoal taxa and the most slowly evolving flatworm, Stenostomum, was reconstructed (Fig. 3). Bootstrap support for the clade consisting of the flatworm and other lophotrochozoans is high (91, 88, 83 and 80%,  
 
for paralinear distances, maximum parsimony, Kimura two-parameter and Jukes-Can tor distances, respectively), consistent with the placement of the flatworms within the Lophotrochozoa18. Figure 3 Phylogenetic analysis of 18S rDNA sequence data to illustrate relationships of the flatworm to other protostome animals. The Lophotrochozoa are present as the top eight taxa,  
 
the Ecdysozoa are shown in the middle, and outgroups are shown at the bottom. The topology shown here is a majority-rule consensus combining the results from four individual majority-rule consensus trees derived using the following methods: paralinear/LogDet distances, maximum parsimony, Kumura two-parameter distances and Jukes-Cantor distances. All distance methods  
 
are corrected for site-to-site variation. The numbers next to the central branches represent the percentage of bootstrap replicates supporting the clades for these methods, respectively (from top to bottom). Divisions within the protostom es have long been a major point of contention among zoologists. Conventional wisdom supports the existence of a clade, the  
 
Articulata, that includes the segmented animals, chiefly the arthropods and the annelids. This concept has a long tradition, but has been called into question by analysis of morphological and palaeontological data3'19 and of 18S rRNA* sequence data7'20'21. Eernisse et al. characterized two clades within the protostomes, the arthropods and the Eutrochozoa (annelids,  
 
molluscs and other protostomes developing from a trochophore larva) with morphological data3. A number of studies using 18S data7'14'20'21'30 identified two clades within the protostomes, the arthropods and the coelomate protostomes of Field et al.y now called Lophotrochozoa7. The lophotrochozoans include the annelids, molluscs, rotifers, phoronids, brachiopods,  
 
bryozoans, platy-helminthes and related phyla. Our data indicate that the sister clade to the lophotrochozoans contains the remaining protostomes, which all develop by moulting. Segmentation does not seem to be a synapomorphy uniting annelids and arthropods. Our analyses, which include four aschelminths (pseudocoelomates), do not support aschelminth monophyly, in  
 
agreement with molecular studies2'10'30. Our studies are consistent with the clade cephalorhyncha16. Table 1 Substitution rates of 18S rDNA sequences Phylum Genus Chaetognatha Sipuncula Pogonophora Platyhelminthes Nemertea Echiura Vestimentifera Mollusca Aschelminthes: Acanthocephala Gastrotricha Rotifera Lophophorates: Phoronida Ectoprocta Brachiopoda Annelida  
 
Lophotrochozoa Sagitta Phascolosoma Siboglinum Bdelloura Fasciolopsis Stenostomum Lineus Ochetostoma Ridgeia Lymnaea Placopecten (bivalve) Acanthopleura (polyplacophoran) Moniliformis Lepidodermella Brachionus Phoronis Plumatella Glottidia Terebratalia Eisenia La nice Enchytreus (oligochaete) Stylaria (oligochaete) Glycera (polychaete) Nematoda Onychophora Tardigrada  
 
Kinorhyncha Nematomorpha Arthropoda Priapula Arthropods and relatives Strongyloides Caenorhabditis Trichuris Trichinella Euperipatoides Milnesium Macrobiotus Pycnophyes Gordius Artemia Panulirus (crustacean) Drosophila Crossodonthina Tenebrio (insect) Scolopendra (myriapod) Androctonus Eurypelma (chelicerate) Priapulus Chordata Echinodermata Ctenophora Cnidaria  
 
Outgroups Lampetra Branchiostoma Strongylocen trotus Antedon Mnemiopsis Anemonia Tripedalia Substitutions per site 0.143 0.079 0.070 0.147 0.083 0.063 0.061 0.058 0.055 0.060 j 0.042 H 0.040 1 0.111 :! 0.070 1 0 058 : 0.053 d 0.049 d 0.044 H 0044: 0.057 n 0.056 j 0.052 : 0 042 H 0 033 j 0.111 0.007 0.008 0.012 0.009 0.063 0.007 0.007 0.007 0.006 0.007 0.006 0.009 0.007  
 
0.007 0.007 0.006 0.006 0006 0.007 0.006 0006 0006 0005 0.192: 0.187: 0.141 : 0.110: 0.090 : 0.079 : 0.079 : 0 075 : 0 068 : 0.068 : 0 065 : 0.121 : 0.056 : 0 048 : 0043: 0.046 : 0 038 : 0.040 : 0.014 0.013 0.012 0010 0009 0.008 0.009 0.007 0.007 0.007 0.008 0.011 0.007 0.006 0006 0.006 0005 0005 0.065 : 0.059 : 0.043 : 0 040 : 0.130: 0.101 : 0.100: 0.007 0.006 0.006  
 
0.005 0.111 0.009 0009 Distances are calculated by paralinear/LogDet distances and the [plusmn] s.d. estimated from bootstrap replicates. The number of substitutions per position from the last common ancestor of protostomes was calculated with respect to three slowly evolving reference taxa. Distances to protostome taxa were calculated using Tripedalia and Antedon as  
 
outgroup taxa and either Glycera ex Priapulus, depending upon which ingroup taxon was being examined. Distances to outgroup taxa were calculated using Glycera, Priapulus and Acanthopleura as reference taxa. Our interpretation of these results is shown in Fig. 4. The most obvious feature of this phylogeny is that it separates the protostomes into two groups, an  
 
arthropod-related clade exclusively composed of animals that moult, and a lophotrochozoal clade exclusively containing non-moulting animals. All members of the arthropod-related clade undergo ecdysis22. In addition, all members lack locomotory cilia, although other groups (for example, chaetognaths and acanthocephalans) also lack them16. Given the observed tree  
 
topology and these common structural features, this raises the possibility that ecdysis and the cellular modifications associated with it may have been derived only once within this clade. Because these 18S rDNA analyses support the hypothesis that all moulting animals (arthropods, tardigrades, onychophorans, nema-todes, nematomorphs, kinorhynchs and priapulids) share  
 
a common ancestor to the exclusion of deuterostomes and the lophotrochozoans, we have chosen the node-based name23 Ecdyso-zoa. This group is defined as these taxa plus their last common ancestor and all of its descendants. The name reflects the property that all members of this group, and only members of this group, undergo ecdysis during at least part of their life  
 
cycles. Figure 4 As inferred from 18S rDNA, the Protostomia consists of two major groups. The Lophotrochozoa includes the lophophorates, molluscs, annelids, rotifers and other groups7. The Ecdysozoa includes the arthropods, tardigrades, onychophorans, nematomorphs, nematodes, kinorhynchs, priapulids and probably the loriciferans. (So far, no living specimens and fewer  
 
than 200 preserved loriciferans (which moult) have been collected. Morphological evidence, however, suggests a close relationship to kinorhynchs and priapulids16'29.) The common ancestors of these clades are indicated. It was unexpected to find nematodes contained within the Ecdysozoa because in previous molecular studies they diverged deep in the protostome tree, even  
 
before the deuterostome-proto-stome bifurcation10. Boore et a/.24, in their pioneering study using mitochondrial gene order, assumed that nematodes were an outgroup to the protostomes. We realized the results of previous molecular studies could be unequal rate artefacts caused by the extremely rapid nucleotide-substitution rates found in previously published rhabditid  
 
nematode sequences, and therefore sequenced numerous nematode species to identify slowly evolving representatives. Unequal rate effects are well documented in theory15 but are usually ignored. Morphological studies also support the inclusion of nematodes with many ecdysozoans, although not with arthropods11'16. One thoughtful analysis groups nematodes, nematomorphs,  
 
priapulids, kinorhynchs and Iqricifera (but not arthropods, onychophorans and tardigrades) using the synapo-morphies, "loss of locomotory cilia, cuticle moulted, introvert with spines, teeth or scalids"16. (These first two synapomorphies also serve to unite the ecdysozoa.) Another recent cladistic analysis of morphological characters supports a clade of moulting  
 
animals excluding the priapulids3, although nematomorphs were not included in that analysis. Given the tremendous interest in the nematode Caenorhabditis elegans and the arthropod Drosophila melanogaster as model systems, the hypothesis that both are closely related has important implications for developmental and genomic studies. For example, it has been assumed that  
 
developmental mechanisms common to Caenorhabditis and to Drosophila originated before the protostome-deuterostome divergence and hence should also be found in Homo sapiens. Our results imply that mechanisms found in both nematodes and fruitflies will not necessarily be found in humans. The inclusion of the priapulids within an arthropod-containing clade was not  
 
anticipated because most morphological studies had not indicated a close priapulid, arthropod phylogenetic relationship2'3. Both arthropods and priapulids are numerically prominent members of the Burgess shale faunas25, indicating the early success (and successful preservation) of ecdysozoans in the Cambrian radiation. These studies provide evidence that the nematodes  
 
are not primitive metazoans but are protostomes related to arthropods. They also support a monophyletic protostome clade. Considering the greatly differing morphologies, embryological features and life histories of the moulting animals, it was initially surprising that the ribosomal RNA tree should group them together. However, given that all moulting taxa sampled are  
 
in this clade, and given the significant anatomical modifications associated with moulting, such as the lack of locomotory cilia, ecdysis appears to be a defining synapomorphy for this group, although additional molecular data from other molecules are necessary to test further or confirm the monophyly of the moulting animals. Methods DNA isolation. Total genomic DNA  
 
was isolated by standard techniques and amplified by the polymerase chain reaction (PCR). PCR fragments or complete sequences were then cloned into a plasmid vector before sequencing. Replicates of the PCR amplification were sequenced in both directions. A list of the PCR and sequencing oligonucleotides and PCR reaction conditions is available from J.A.L. or J.R.G.  
 
(garey@chuma.cas.usf.edu). Sequences. The following sequences are available in GenBank: Brachionus plicatilis (Rotifer; accession number, U49911), Enchytraeus sp. (Oligochaete; accession number, U95948), Euperipatoides leukartii (Onychophoran; accession number, U49910), Gordius sp. (Nematomorph; accession number, U51005), Macrobiotus sp. (Tardigrade; accession number,  
 
U49912), Milnesium tardigradum (Tardigrade; accession number, U49909), Stenostomum sp. (Platyhelminth; accession number, U95947), Stylaria sp. (Oligochaete; accession number, U95946), and Trichinella spiralis (Nematode, accession number, U60231). Sequence alignments. An alignment of 49 complete sequences was constructed using the star alignment procedure to reduce  
 
biases8, with the slowly evolving Glycera americana sequence used as the reference, and then proofread by hand. Pairwise alignments of nucleotide sequences were performed with the ALIGN program, using a break penalty of 6; nucleotide identities, transversions and transitions were scored as +3, -H and 0, respectively, based on preliminary experiments with EF-la and 18S  
 
rDNA. Regions were excluded from the analysis if extreme length variation existed among sequences, or if many of the sequences contained gaps that could be easily moved with little or no change in alignment score. The alignments are available from JA.L. Phylogenetic reconstruction. The 17-taxon phylogenetic trees shown in Figs 2 and 3 were obtained using PAUP version  
 
3.1.1 for maximum parsimony analyses and Bootstrappers gambit26 for distance analyses. For both methods, 200 bootstrap trees were calculated to determine the 50% majority-rule consensus tree; each search was initiated with 100 replicates of random taxon addition, and positions with gaps were excluded. For parsimony, the following heuristic search options were used:  
 
starting trees were obtained by stepwise addition (starting seed was 1) with one tree held at each step; and tree-bisection-reconnection branch-swapping was performed with the MULPARS, but not the steepest descent, option. For paralinear/LogDet, Kimura two-parameter and Jukes-Cantor distances, four-point metrics were used to assess quartet values; the quartet  
 
consistency value26 (53.46%) was selected to ensure that the probability of finding the best solution was >99.9%. A cnidarian and an echinoderm were used as the outgroups, except in Fig. 3 where the two slowest echinoderms were used for parsimony to further reduce unequal rate effects. As site-to-site variation was judged to be significant, distances were corrected  
 
for this artefact by estimating nine site categories from the data, calculating distances from the eight non-categories, and estimating trees from the sums of the distances12. Site-to-site variation. Site-to-site variation was considered significant when estimated using a diagnostic statistical test for the number of constant sites27. Maximum-likelihood trees were  
 
calculated using the DNAML program (version 3.4) in PHYLIP. Parameters necessary for the test were calculated for a variety of substitution models using both single and double rate categories determined by the hidden Markov model28. An excess of observed constant sites (overpredicted sites) was found for all models, indicating that even two-site categories could not  
 
fully explain the data. (Using empirical base frequencies and a transition/transversion ration of 2.0, the best single-site model (rate ratio, 2:1; probability of each rate, 0.5, 0.5) predicted 787 [plusmn] 39 site versus 1,081 observed sites, and the best two-category model (rate model, 10:1; probability of each rate, 0.8, 0.2) predicted 963 [plusmn] 38 sites versus  
 
1,081 observed sites. All choices of parameters reconstructed trees with monophyletic ecdysozoal and lophotrochozoal clades, although long computation times prevented bootstrap analysis.) Bootstrap interpretations. Based on empirical studies of bootstrap analyses, they represent highly conservative estimates of phylogenetic accuracy. Typically for maximum parsimony,  
 
bootstrap proportions of =70% correspond to a probability of 2=95% that the respective clade is a historical lineage. For Gambit, the probabilities are slightly less conservative. T-PTP test. The topology-dependent cladistic permutation tail probability (T-PTP) test determines whether the difference in length between the shortest tree supporting the monophyly of this  
 
clade and the shortest tree not supporting monophyly (5 steps difference) is significantly different from the difference in length expected from randomized data. If the difference in length between the monophyly and non-monophyly trees-is outside 95% of the distribution based on randomized data, it can be concluded that the data significantly support monophyly of the  
 
clade. We used 200 randomized data sets that were analysed by maximum parsimony. Acknowledgements. We thank C. Marshall for insight and comments, M. Blaxter for nematode DNA; L. Mackey for technical assistance; D. R. Nelson for identifying Macrobiotus sp.; and J. Felsenstein for suggestions on how to implement the Goldman Statistical tests. One of us (J.M.T.) did part  
 
of this work as a visiting research investigator in the laboratories of R.A.R. and of W. Brown, University of Michigan, Ann Arbor. This research was supported by USDA grants to J.R.G., NSF grants to J.A.L., and NIH grants to R.A.R. Correspondence and requests for materials should be addressed to J.A.L. (e-mail: lake@mbi.ucla.edu).Boudreaux, , H. B.Arthropod Phylogeny with Special Reference to Insects (Wiley, New York, 1979).Raff, , R. A., Marshall, , C. R. & Turbeville, , J. M.Using DNA sequences to unravel the Cambrian radiation of the animal phyla. Annu. Rev. Ecol. Syst.25, 351-375 (1994).ArticleEernisse, , D. J., Albert, , J. S. & Anderson, , F. E.Annelida and Arthropoda are not sister taxa: a phylogenetic analysis of spiralian metazoan morphology. Syst. Biol.41, 305-330 (1992).Ballard, , J. W.et al.Evidence from 12S ribosomal RNA sequences that onychophorans are modified arthropods. Science258, 1345-1348 (1992).PubMedISIChemPortWeygoldt, , P.Arthropod relationships: the phylogenetic-systematic approach. Z. Zool. Syst. Evolut.-forsch.24, 19-35 (1986).Brusca, , R. C. & Brusca, , G. J.Invertebrates (Sinauer, Sunderland, MA, 1990).Halanych, , K. M.et al.Evidence from 18S ribosomal DNA that the lophophorates are protostome animals. Science267, 1641-1632 (1995).ArticlePubMedISIChemPortLake, , J. A.The order of sequence alignment can bias the selection of tree topology. Biol. Evol.8, 378-385 (1991).ChemPortStewart, , C.-B.The powers and pitfalls of parsimony. Nature361, 603-607 (1993).ArticlePubMedChemPortWinnepenninckx, , B.et al.18S rRNA data indicate that aschelminthes are polyphyletic in origin and consist of at least three distinct clades. Mol. Biol. Evol.12, 1132-1137 (1995).PubMedISIChemPortWright, , K. in Microscopic Anatomy of Invertebrates Vol. 4 (eds Harrison, F. W. & Ruppert, E. E.) 111-195 (Wiley-Liss, New York, 1991).Lake, , J. A.Reconstructing evolutionary trees from DNA and protein sequences: Paralinear distances. Proc. Natl Acad. Sci. USA91, 1455-1459 (1994).PubMedChemPortLockhart, , P. J., Steel, , M. A., Hendy, , M. D. & Penny, , D.Recovering evolutionary trees under a more realistic model of sequence evolution. Mol. Biol. Evol.11, 605-612 (1994).ISIChemPortField, , K. G.et al.Molecular phylogeny of the animal kingdom. Science239, 748-753 (1988).PubMedISIChemPortHillis, , D. M. & Bull, , J. J.An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Syst. Biol.42, 182-192 (1993).ArticleISINielsen, , C.Animal Evolution: Interrelationships of the Living Phyla (Oxford University Press, 1995).Gee, , H.Before the Backbone: Views on the Origin of the Vertebrates (Chapman & Hall, London, 1996).Balavoine, , G.The early emergence of plaryhelminths is contradicted by the agreement between 18S rRNA and Hox dataC. R. Acad. Sci. Paris Life Sci.320, 83-94 (1997).ChemPortValentine, , J. W.Bilaterians of the Precambrian-Cambrian transition and the annelid-arthropod relationship. Proc. Natl Acad. Sci. USA86, 2272-2275 (1989).petterson, , C.The Hierarchy of Life (eds Fernholm, B., Bremer, K. & Jornvall, H.) 471-488 (Elsevier, Amsterdam, 1989).Lake, , J. A.Origin of the Metazoa. Proc. Natl Acad. Sci. USA87, 763-766 (1990).PubMedChemPortRuppert, , E. E. in Microscopic Anatomy of Invertebrates Vol. 4, (eds Harrison, R. W. & Ruppert, E. E.) 1-17 (Wiley-Liss, New York, 1991).de Queiroz, , K. & Gauthier, , J.Phylogeny as a central principle in taxonomy: Phylogenetic definitions of taxon names. Syst. Zool.39, 307-322 (1990).Boore, , J. L., Collins, , T. M., Stanton, , S., Daehier, , L. L. & Brown, , W. M.Deducing the pattern of arthropod phylogeny from mitochondrial DNA rearrangements. Nature376163-165 (1995).ArticlePubMedISIChemPortConway Morris, , S.The fossil record and the early evolution of the Metazoa. Nature361, 219-225 (1993).ArticleLake, , J. A.Calculating the probability of multitaxon evolutionary trees: Bootstrappers GambitProc. Natl Acad. Sci. USA92, 9662-9666 (1995).PubMedChemPortGoldman, , N.Simple diagnostic statistical tests of models for DNA substitution. J. Mol. Evol.37, 650-661 (1993).PubMedChemPortFelsenstein, , J. & Churchill, , G. A.A hidden Markov model approach to variation among sites in rate of evolution. Mol. Biol. Evol.13, 93-104 (1996).PubMedISIChemPortWallace, , R. L., Ricci, , C. & Malone, , G.A cladistic analysis of pseudoceolomate (aschelminth) morphology. Invert. Biol.115, 104-112 (1996).Garey, , J. R., Krotec, , M., Nelson, , D. R. & Brooks, , J.Molecular analysis supports a Tardigrade-Arthropod association. Invert. Biol.115, 79-88 (1996).