Untitled - Department of Biology

Vol 452 | 10 April 2008 |doi:10.1038/nature06614LETTERSBroad phylogenomic sampling improves resolution ofthe animal tree of lifeCasey W. Dunn 1 {, Andreas Hejnol 1 , David Q. Matus 1 , Kevin Pang 1 , William E. Browne 1 , Stephen A. Smith 2 ,Elaine Seaver 1 , Greg W. Rouse 3 , Matthias Obst 4 , Gregory D. Edgecombe 5 , Martin V. Sørensen 6 ,Steven H. D. Haddock 7 , Andreas Schmidt-Rhaesa 8 , Akiko Okusu 9 , Reinhardt Møbjerg Kristensen 10 ,Ward C. Wheeler 11 , Mark Q. Martindale 1 & Gonzalo Giribet 12,13Long held ideas regarding the evolutionary relationships amonganimals have recently been upended by sometimes controversialhypotheses based largely on insights from molecular data 1,2 . Thesenew hypotheses include a clade of moulting animals (Ecdysozoa) 3and the close relationship of the lophophorates to molluscs andannelids (Lophotrochozoa) 4 . Many relationships remain disputed,including those that are required to polarize key features ofcharacter evolution, and support for deep nodes is often low.Phylogenomic approaches, which use data from many genes, haveshown promise for resolving deep animal relationships, but arehindered by a lack of data from many important groups. Here wereport a total of 39.9 Mb of expressed sequence tags from 29 animals belonging to 21 phyla, including 11 phyla previously lackinggenomic or expressed sequence tag data. Analysed in combination with existing sequences, our data reinforce several previouslyidentified clades that split deeply in the animal tree (includingProtostomia, Ecdysozoa and Lophotrochozoa), unambiguouslyresolve multiple long standing issues for which there was strongconflicting support in earlier studies with less data (such as velvetworms rather than tardigrades as the sister group of arthropods 5 ),and provide molecular support for the monophyly of molluscs, agroup long recognized by morphologists. In addition, we findstrong support for several new hypotheses. These include a cladethat unites annelids (including sipunculans and echiurans) withnemerteans, phoronids and brachiopods, molluscs as sister to thatassemblage, and the placement of ctenophores as the earliestdiverging extant multicellular animals. A single origin of spiralcleavage (with subsequent losses) is inferred from well supportednodes. Many relationships between a stable subset of taxa findstrong support, and a diminishing number of lineages remainrecalcitrant to placement on the tree.Expressed sequence tags (ESTs) provide opportunities to samplediverse genes from a large number of taxa 6 . Several recent phylogenomic studies, based largely on EST data, analysed matrices containing more than 140 genes from up to 34 metazoans (multicellularanimals) 7 9 . However, the included species were not well sampledacross extant metazoan diversity. These analyses also relied on eitherribosomal proteins or a list of target genes identified from a small(1,152 ESTs) choanoflagellate data set 10 , limiting the possibilities ofEST studies to inform gene selection and homology assignment.Rather than look for predefined sets of genes in our data, we presentan explicit procedure for gene selection (see Methods and Supplementary Fig. 2).Our complete matrix includes data from 77 taxa (of which 71 aremetazoans) and 150 genes. On average, taxa in our matrix include50.9% of the 150 genes, and overall matrix completeness is 44.5%.Maximum likelihood (WAG model of sequence evolution; Figs 1 and2) and bayesian (CAT 11 and WAG models of sequence evolution;Fig. 2) analyses of our matrix support the major groups of the ‘newanimal phylogeny’ 2 . These groups have also been supported by otherEST based analyses 9 , but not by phylogenomic studies that consider asmall number of animal taxa 12 . Primary analyses of the 77 taxonmatrix recover Metazoa, Bilateria and Protostomia with strong bootstrap support (.90%). This is an improvement compared to someprevious phylogenomic studies that did not recover Protostomia,which in part led one study to conclude that it may not be possibleto reconstruct the relationships of several major clades of animalsbecause the metazoan radiation was too rapid 13 . It now seems thatthose findings were largely caused by limited taxon sampling, a resultconsistent with reanalyses 14 . Bootstrap support for Lophotrochozoaand Ecdysozoa is low in the 77 taxon consensus tree, but this iscaused by the instability of a relatively small number of taxa (seebelow). Whereas Deuterostomia had poor support in recentphylogenomic analyses 15 , in analyses of our 77 taxon matrix maximum likelihood bootstrap support for Deuterostomia is .80%.Within Deuterostomia, Xenoturbella was found to be sister toAmbulacraria (echinoderms and hemichordates) in a study thatincluded 1,372 Xenoturbella ESTs 7 . Our inclusion of 3,840 additionalXenoturbella ESTs is consistent with this previous analysis (Figs 1, 2).None of our results are congruent with Coelomata, a group consisting of taxa that have a coelomic body cavity, which was favouredbefore molecular data became available. Coelomata has been recovered in some studies using many genes from a very small number oftaxa 12,16 , but it now seems clear that this is an artefact of poor taxonsampling.Low support values on consensus trees can be caused by largescale structural rearrangements or by the instability of particular taxa.If, for instance, a taxon is only placed within a particular clade 50% of1 Kewalo Marine Laboratory, PBRC, University of Hawaii, 41 Ahui Street, Honolulu, Hawaii 96813, USA. 2 Department of Ecology and Evolutionary Biology, Yale University, PO Box208105, New Haven, Connecticut 06520, USA. 3 Scripps Institution of Oceanography, University of California San Diego, 9500 Gilman Drive 0202, La Jolla, California 92093, USA.4 Kristineberg Marine Research Station, Kristineberg 566, 450 34 Fiskebackskil, Sweden. 5 Department of Palaeontology, The Natural History Museum, Cromwell Road, London SW75BD, UK. 6 Ancient DNA and Evolution Group, Biological Institute, University of Copenhagen, Universitetsparken 15, DK 2100 Copenhagen, Denmark. 7 Monterey Bay AquariumResearch Institute, 7700 Sandholdt Road, Moss Landing, California 95039, USA. 8 Zoological Museum, University of Hamburg, Martin Luther King Platz 3, 20146 Hamburg, Germany.9 Biology Department, Simmons College, The Fenway, Boston, Massachusetts 02115, USA. 10 Zoological Museum, University of Copenhagen, Universitetsparken 15, DK 2100Copenhagen, Denmark. 11 Division of Invertebrate Zoology, American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024, USA. 12 Department ofOrganismic and Evolutionary Biology, 13 Museum of Comparative Zoology, Harvard University, 26 Oxford Street, Cambridge, Massachusetts 02138, USA. {Present address:Department of Ecology and Evolutionary Biology, Brown University, 80 Waterman Street, Providence, Rhode Island 02912, USA.©2008 Nature Publishing Group745

NATURE | Vol 452 | 10 April 2008LETTERSbe that Platyzoa is an artefact of attracting unstable long branchspecies to their vicinity.Analyses of the 64 taxon matrix (Fig. 2 and Supplementary Fig. 9)show strong support for several important clades. To test if confidence in the relationships between stable taxa is overestimated in theabsence of unstable taxa, we pruned away the 13 unstable taxa fromeach of the 1,000 bootstrap trees inferred from the 77 taxon matrix.This generated a set of trees containing only stable taxa, but for whichrelationships had been inferred in the presence of unstable taxa.Clade frequencies were calculated from this pruned tree set andmapped onto the most probable 64 taxon tree (Fig. 2). Thesereduced tree support values are very similar to bootstrap supportvalues calculated from the 64 taxon matrix, indicating that unstabletaxa do not affect the inference of most relationships between stabletaxa, only obscure these affinities.The 64 taxon matrix strongly supports a sister group relationshipbetween Platyhelminthes and the remaining lophotrochozoans. Asimilar result, although uniting gastrotrichs with platyhelminths,was proposed recently 19 . Consistent with recent findings 20 , Urechiscaupo, an echiuran, is placed as sister to the annelid Capitella sp., andthe sipunculan Themiste lageniformis is allied with annelids ratherthan molluscs. All analyses place Annelida as sister to a novel groupthat we call Clade A (Fig. 2), consisting of the nemerteans, a phoronidand a brachiopod, with variable support across analyses. Bayesiansupport for a group consisting of Annelida 1 Clade A (Clade B,Fig. 2) is strong (100% posterior probability in CAT and WAG analyses), whereas bootstrap support is moderate (84%). Although abrachiopod annelid relationship is supported by the shared presenceof chitinous chaetae, this new relationship implies that chaetae havebeen lost in nemerteans and phoronids (as in sipunculans, leeches•/•87/93PP(WAG)/PP(CAT)•/88BS(WAG,64 taxon)/BS(WAG,77 taxon)65/53 00/8723/30• = 100%•/••/• •/••/• •/99•/9388/9276/73Clade C•/••/••/•98/91•/9978/85 00/7334/21 99/60•/•56/62•/••/•Lophotrochozoa•/••/•99/9884/64•/•98/97•/•Clade B36/67•/99 29/2167/58 •/••/•Clade A99/59•/•45/53•/•Protostomia•/••/• •/••/••/••/•Cycloneuralia96/5148/3700/86•/•09/06Ecdysozoa98/84 98/99•/•60/6099/95•/••/•00/8604/02Bilateria•/94•/••/•95/9699/99•/••/•Panarthropoda•/• •/••/••/•92/92Arthropoda•/•93/93•/9577/81 98/8269/68 •/99Myriochelata99/99Metazoa•/••/•40/•68/58•/•99/9789/7941/46 •/••/• •/••/67 •/•85/87•/••/•Deuterostomia Ambulacraria•/••/• •/••/• 83/7881/76•/•91/9472/77Figure 2 | Cladogram of the 64-taxon PhyloBayes bayesian analysesconducted under the CAT model. Posterior probabilities (PP) estimatedunder the CAT (15 PhyloBayes runs of 6,000 generations each; 1,200generation burn in) and WAG1I1C (8 MrBayes runs of two milliongenerations each; 125,000 generation burn in; 4 chains per run) models.Maximum likelihood bootstrap support was calculated for the 64 taxon data60/60•/••/• •/••/••/••/••/••/••/••/••/••/•©2008 Nature Publishing GroupEuprymna scolopes MolluscaChaetopleura apiculataChaetoderma nitidulumMytilus galloprovincialisCrassostrea virginicaArgopecten irradiansBiomphalaria glabrataAplysia californicaChaetopterus sp. AnnelidaThemiste lageniformis SipunculaPlatynereis dumeriliiLumbricus rubellusHaementeria depressaUrechis caupoCapitella sp.EchiuraPhoronis vancouverensis PhoronidaTerebratalia transversa BrachiopodaCerebratulus lacteus NemerteaCarinoma mutabilisParaplanocera sp. PlatyhelminthesMacrostomum lignanoEchinococcus granulosusSchmidtea mediterraneaDugesia japonicaPriapulus caudatus PriapulidaEchinoderes horni KinorhynchaXiphinema index NematodaTrichinella spiralisSpinochordodes tellinii NematomorphaRichtersius coronifer TardigradaHypsibius dujardiniEuperipatoides kanangrensis OnychophoraDrosophila melanogaster TetraconataDaphnia magnaFenneropenaeus chinensisCarcinus maenasScutigera coleoptrata MyriapodaAnoplodactylus eroticus ChelicerataCarcinoscorpius rotundicaudaAcanthoscurria gomesianaBoophilus microplusXenoturbella bocki XenoturbellidaStrongylocentrotus purpuratus EchinodermataAsterina pectiniferaSaccoglossus kowalevskii HemichordataPtychodera flavaHomo sapiensChordataGallus gallusCiona intestinalisBranchiostoma floridaeHydractinia echinata CnidariaHydra magnipapillataCyanea capillataNematostella vectensisAcropora milleporaOscarella carmela PoriferaMnemiopsis leidyi CtenophoraMertensiid sp.Monosiga ovataOutgroupsCapsaspora owczarzakiSphaeroforma arcticaAmoebidium parasiticumSaccharomyces cerevisiaeCryptococcus neoformansset (2,000 replicate RaxML runs) and for the relationship of these 64 taxa inthe 77 taxon analysis (by pruning all other taxa from the bootstrap replicatessummarized in Fig. 1). Taxa for which we collected new data are shown ingreen. Support values, as specified at the top left of the figure, are shown inblue.747

1. Polychaete: Polynoidae - Steve Haddock2. Acorn worm: Ptychodera flava - Steve Haddock3. Pycnogonid: Anoplodactylus eroticus - Amy Maxmen4. Nemertean - Steve Haddock5. Ctenophore: Beroe forskalii - Steve Haddock6. Cephalopod: Abraliopsis sp. -Steve Haddock7. Scyphomedusa: Atolla vanhoeffeni - Casey Dunn8. Polychaete: Chaetopterus sp.- Greg Rouse9. Amphipod: Cystisoma sp. - Steve Haddock10. Chaetognath: Eukrohnia fowleri - Steve Haddock11. Onychophoran: Peripatoides novaezealandiae - Gonzalo Giribet