Received November Accepted February Published online May Only six kingdoms of life - PDF document

13November20033February2004 Publishedonline17May2004 OnlysixkingdomsoflifeThomasCavalier-SmithDepartmentofZoology,UniversityofOxford,SouthParksRoad,OxfordOX13PS,UKTherearemanymorephylaofmicrobesthanofmacro-organisms,butmicrobialbiodiversityispoorlyunderstoodbecausemostmicrobesareuncultured.PhylogeneticanalysisofrDNAsequencesclonedafter 1252T.Cavalier-SmithOnlysixkingdomsoflife Table1.Thesixkingdomsoflifeandthe34microbialphyla(basedonCavalier-Smith1998,2002,2003 empirePROKARYOTA(Cavalier-Smith2002kingdomBacteriasubkingdomNegibacteria(phylaEobacteria,Sphingobacteria,Spirochaetae,Proteobacteria,Planctobacteria,subkingdomUnibacteria(phylaPosibacteria,Archaebacteria)empireEUKARYOTA(Cavalier-Smith1998)kingdomProtozoa(Cavalier-Smith2002,2003subkingdomSarcomastigota(phylaAmoebozoa,Choanozoa)subkingdomBiciliatainfrakingdomRhizaria(phylaCercozoa,Foraminifera,Radiozoa)infrakingdomExcavata(phylaLoukozoa,Percolozoa,Euglenozoa,Metamonada;thelatternowincludesParabasaliaandAnaeromonadea;Cavalier-Smith2003infrakingdomAlveolata(phylaMyzozoa(Cavalier-Smith&Chao2004),Ciliophora)Biciliataincertaesedis:phylumApusozoa(maybesistertoExcavata);phylumHeliozoakingdomAnimalia(Myxozoaand21otherphyla)(Cavalier-Smith1998;Cavalier-Smith&Chao2003kingdomFungi(phylaArchemycota,Microsporidia,Ascomycota,Basidiomycota)(Cavalier-Smith2000kingdomPlantaesubkingdomBiliphyta(phylaGlaucophyta,Rhodophyta)subkingdomViridaeplantae(Chlorophyta,Bryophyta,TracheophytakingdomChromistasubkingdomCryptista(phylumCryptista:cryptophytes,goniomonads,katablepharids)subkingdomChromobiotainfrakingdomHeterokonta(phylaOchrophyta,Pseudofungi,Opalozoa(comprisingsubphylaOpalinata,Sagenista)infrakingdomHaptista(phylumHaptophyta) Nomicrobialmembers.All34phylathatcontainmicrobesarelisted.AlthoughcentrohelidHeliozoamightbeChromista,theyprobablybelonginProtozoa(Biciliata)(Cavalier-Smith&Chaodeepsuspicionunlessitiscorroboratedbyindependentevidence.Suchevidenceisnotavailableforpurelyenvironmentalsequences.Withineukaryotesbothsingle-geneproteintreesandproteintreescombiningdatafromseveralormanygeneshaverevealedthatmostpublishedrRNAtreeshavefundamentallyincorrecttopologiesforseveralmajorlong-branchclades;thisisparticularlytruefortreespublishedbeforecorrectionsforintramolecularvariationinevolutionaryratesbecamederigueurandthosewithundulysparsetaxonsampling.Despitetheseproblems,rRNAphylogenyremainsavaluabletoolforinitialexplorationsofbiodiversity.Suchstudiesonculturedprotists(unicellulareukaryotes)havecontributedtomajorrecentimprovementsineukaryoticphylogenyandhigh-levelclassification(Silbermanetal2002;Simpsonetal.2002;Cavalier-Smith2003;Cava-lier-Smith&Chao2003,2004)andthroughsynthesiswithmorphologicalandotherdatatoaconsiderablereductioninthenumberofprotistphyla(Cavalier-Smith)comparedwithadecadeago(Cavalier-Smith).PCRofDNAextractsfromenvironmentalsampleswouldaprioribeexpectedtorevealmanyfewerdeeplybranchingnovellineagesthanhasbeenthecaseforbacteria.Thisisbecausemosteukaryotecellsaremuchlargerandmorphologicallymuchmorecomplexthanmostbacteria,socenturiesofmicroscopicalstudyhaveprobablyalreadyrevealedmostofthemajortypes,evenoflineagesthathaveneverbeencultured.Thishasbeenborneoutbythefirstsuchstudiesofeukaryotesfromaerobichabitats,wherealmostalltheso-called‘novel’lin-eagesdiscoveredcanbeeasilyplacedwithintheknownphyla(Loetal.2001;Moon-vanderStaay2001).Thisistrueevenforthealveolateprotozoa,wherethemajorgroupsof‘novellineages’almostcertainlyProc.R.Soc.Lond.B(2004)belongwithinthephylumMyzozoa(Miozoa)assisterstothedinoflagellates(Cavalier-Smith&Chao2004).Becausewithinmostphylatherearemanyknownandwell-describedprotists(whetherofclearorobscuretaxo-nomicposition)thathavenotyetbeenculturedorhadtheirrRNAsequenced,anystudyofenvironmental-DNAlibrariesisboundtocomeupwithunidentifiablelineagesthatarenotclosetoknownsequences.Butineukaryotes,unlikebacteria,mostofthesearelikelytobefromgroupswithpreviouslyknownrepresentatives;suchlineagesaresimplyunidentified,nottotallynewtoscience.Anaerobicaquatichabitatsareecologicallyimportant(Fenchel&Finlay1995)buthavebeenundersampledbyculturingandecologicalstudies.Itwasformerlythoughtthatearlyeukaryotesmighthavebeenanaerobic(Cavalier-Smith1983).However,allwell-characterizedgroupsofanaerobeshavenowbeenshowntohavehadaerobicancestors(Embley&Hirt1998;Roger1999;Silbermanetal.2002;Simpsonetal.2002),makingithighlyprob-ablethattheancestraleukaryotewasaerobic(Cavalier-Smith2002).Nonetheless,theremotepossibilityremainsthatsomelittle-knownorentirelyunknownprimitivelyanaerobiceukaryotegroupmightstillexist,andanaerobicenvironmentscouldstillholdearly-branchingeukaryoticlineagesimportantforunderstand-ingtheoriginofthenucleated(eukaryotic)cell.TherecentstudiesofDawson&Pace(2002)andStoeck&Epstein(2003)ofDNAextractedfrommarineandfresh-wateranoxicsitesarethereforeofparticularinterest.Theyfoundthatover90%oftheirsequencescouldeasilybeidentifiedasbelongingtoknownkingdomsandphyla,butwerenotabletoidentifyrelativesfor13andninenovelsequences,respectively.Theirconclusionsthatthesenewsequencesrepresenteightandsixnovelkingdoms, OnlysixkingdomsoflifeT.Cavalier-Smith1253 Glaucophyta Viridaeplantae(green plants)kingdom PlantaekingdomChromistaRadiozoaApusozoa CercozoaChoanozoa kingdomAnimaliakingdomFungiAmoebozoa ancestral heterotrophic aerobic uniciliate eukaryote ca. 900 Myr ago bikonts cortical alveoli (corticates) **DHFR–TS fusion gene kingdom Bacteria unikonts (ancestrallyuniciliate eukaryotes) plastid GAPDH replacement polyubiquitininsertionPosibacteriaancestral neomuranArchaebacteria origin of life and photosynthesis ? 2.9–3.5 Gyr ago eubacteria pyrimidine synthesistriple gene fusion Eukaryotaempire Prokaryota posterior cilium; flatmitochondrial cristae;EF1- insertion membrane losshyperthermophily;isoprenoid ether lipids;reverse gyraseN-linkedglycoproteins;histones H3/4 cytoskeleton; nucleus; cilium LoukozoaEuglenozoaPercolozoaMetamonada Heliozoa Figure1.Thetreeoflifebasedonmolecular,ultrastructuralandpalaeontologicalevidence.Contrarytowidespreadassumptions,therootisamongtheeubacteria,probablywithinthedouble-envelopedNegibacteria,notbetweentheeubacteriaandarchaebacteria(Cavalier-Smith2002).Establishingitsprecisepositionwithconfidenceisoneofthemostchallengingquestionsinevolutionarybiologyandisnotyetsolved;itmayliebetweenEobacteriaandotherNegibacteria(Cavalier-Smith);eventheinternalbranchingorderofeubacterialphylaismoreuncertainthanforeukaryotes.Thepositionoftheeukaryoticroothasbeennearlyascontroversial,butislesshardtoestablish:itprobablyliesbetweentheunikontsandthebikonts(Langetal.2002;Stechmann&Cavalier-Smith2002,2003).ForclaritythebasaleukaryotickingdomProtozoaisnotlabelled;itcomprisesfourmajorgroups(Alveolata,cabozoa,AmoebozoaandChoanozoa)plusthephylumApusozoa,whichmaybesistertotheotherbikontsor,possiblymorelikely,sistertoexcavatesalone(Cavalier-Smith2003),andprobablyHeliozoa.Heliozoaareofuncertainphylogeneticpositionwithinthebikonts,andmightbechromists(Cavalier-Smith&Chao2003)notearlierbranchingasshownhere.Symbiogeneticcellenslavementoccurredfourorfivetimes:intheoriginofmitochondriaandchloroplastsfromdifferentNegibacteria,ofchromalveolatesbytheenslavingofaredalga(Cavalier-Smith1999,2003;Harper&Keeling2003)andintheoriginofthegreenplastidsofeuglenoid(excavate)andchlorarachnean(cercozoan)algae—agreenalgalcell(G)wasenslavedeitherbytheancestralcabozoan(dashedarrow)or(lesslikely)twiceindependentlywithinexcavatesandCercozoa(asterisks)(Cavalier-Smith2003).Theagesofeukaryotesandarchaebacteriaarealsocontroversial;reasonsforscepticismoversubstantiallyearlierdatesthanshownareexplainedbyCavalier-Smith(2002)andCavalier-Smith&Chao(2003).Thethumbnailsketchesshowthefourmajorkindsofcellinthelivingworld,plusthemostcomplexeukaryotecellsofall:thecryptophytes(uppersketch,nucleusandnucleomorphinblue,mitochondrioninpurple,chloroplastingreen,periplastidspaceinred).Themiddleonesshowinredthecontrastingancestralmicrotubularcytoskeleton(ciliaryroots)ofunikonts(aconeofsinglemicrotubulesattachingthesinglecentrioletothenucleus)andbikonts(twobandsofmicrotubulesattachedtotheposteriorcentrioleandananteriorfanofmicrotubulesattachedtotheanteriorcentriole);ciliaandplasmamembraneareblackandthenucleusblue.Theloweronesshowthesingleplasmamembraneofunibacteria(Posibacteriaplusarchaebacteria),whichwereancestraltoeukaryotes,andthedoubleenvelopeofNegibacteria,whichwereancestraltomitochondriaandchloroplasts(whichretainedtheoutermembrane,showninred).EubacteriaareaparaphyleticgradecomprisingNegibacteriaandPosibacteria.respectively,areverysurprisingindeed,giventhatonlyfiveeukaryotickingdomsarecurrentlyrecognized(figureHavewereallymissedasmuchdeepeukaryoticbiodiv-ersityastheseauthorsclaim?No!Theirconclusionsareinvalid,notonlybecauseonecannotassessrankfromdivergencedepthonamoleculartree,butalsobecausethekeyanalysesshownintheirfiguresincludefartoofewknowneukaryotestobeabletoidentifytheirlikelyProc.R.Soc.Lond.B(2004)relativesoryieldanevolutionarilyreliabletopology.Inadditionbothstudieswereseriouslyflawedbyasystematicmisrootingcausedbyincludingtheverydistantbacterialoutgroups.Ihavethereforereanalysedtheirdatausing167,notjust37(Dawson&Pace2002)oronly8–35(Stoeck&Epstein2003),knowneukaryotictaxarep-resentingallthemajorgroups.Asrecentmicroscopicstudieshaveshownthatthedeepseaharboursmanynovelprotists(Hausmannetal.),sometimesquite 1254T.Cavalier-SmithOnlysixkingdomsoflife extraordinaryones(Hausmannetal.),Ihavealsoincludedfourenvironmentalsequencesfromthedeepseanotpreviouslyassignedtophylathatweremorecautiouslysuggestedaspossiblyrepresentingnovelkingdomsetal.2003).Myanalysisof193sequencesis,Ithink,themosttaxonomicallycomprehensiveandbal-ancedphylogeneticstudytodateofeukaryotic18SrRNA.Unsurprisingly,itdoesnotsupporttheideathatanaerobichabitatsorthedeepoceancontainuncharacterizedeukar-yotickingdoms.All26mysterylineagesdiscoveredbythethreestudiescanbeplacedwithintheestablishedsix-kingdomclassification(table1)inestablishedphylaandusuallyalsoinclasseswithinthekingdomProtozoa,exceptforonethatprovestobeanartefactualchimeraofsequencesfromtwophyla.Dawson&Pace(2002)alsointerpretedtheirtreeintermsofmyformerhypothesis(Cavalier-Smith1983thattheearliesteukaryoteslackedmitochondria,assertingthatsevencladeswere‘deepbranching’,asdidStoeck&Epstein(2003)andanotherstudythatappearedafterIcompletedthepresentanalysis(Stoecketal.2003).How-ever,recentevidenceindicatesthattherootoftheeukary-otetreeliesamongaerobiceukaryoteswithmitochondriaetal.2002;Simpson&Roger2002;Stechmann&Cavalier-Smith2002,2003),notanaerobiconesasthoseauthorsmistakenlyassumed(figure1).Therefore,asIshallexplain,thederivedpositionsofalltheirsequencesonmytreegiveaddedsupportforthenowprevailingviewthatthelastcommonancestorofeukaryoteswasaerobicandthatmitochondriaoriginatedimmediatelyfollowingorevenduringtheoriginofthenucleus.2.METHODSThe12unknownenvironmentalsequencesfromfig.4ofDawson&Pace(2002),claimedtorepresentnewkingdoms,plusanotheroftheirsequencesthattheydidnotassigntoaknowngroup(BOLA868),10phyleticallyunassignedsequencesfromStoeck&Epstein(2003)andfourfromLo´pez-Garc(2001,2003)werealignedwith167sequencesofknownorganisms,representingallthemajoreukaryoticlineages,obtainedfromGenBank;whereverpossibleshort-branchrep-resentativeswerechosentominimizephylogeneticartefacts.Asthe26environmentalsequenceswereincomplete,theterminalregionsandobviouslyambiguouslyalignedregionswereomittedfromthephylogeneticanalysis.Bacterialoutgroupswerealsoomitted,asthesearesodistantthattheywouldsimplyhavearte-factuallyjoinedthelongesteukaryoticbranch(diplomonads)andcausedanartefactualrooting,ashappenedinthetreesofDawson&Pace(2002)andStoeck&Epstein(2003);intheformerthebacterialbranchwasabouttwiceaslongasthepartofthediplomonadone.Omittingbac-teriaallowedtheinclusionof1044nucleotidepositionsnotjust789asincludedbyDawson&Pace(2002);Stoeck&Epstein(2003)didnotspecifyhowmanytheyused.Neighbour-joining(BioNJ),weightedleast-squares(power2)distanceanalysesImodel:0.120037;parameterscal-culatedviav.3.06)andmaximumparsimonywerecarriedoutusingPv.4.0b10.Inadditiontothetreeshowninfigure2,alargenumberofothertreeswerealsocalcu-latedwithdifferingtaxonsamples,includingtheadditionoflonger-branchtaxasuchasdiplomonads,retortamonads,Para-basalia,PercolozoaandForaminiferatoexcludethepossibilityProc.R.Soc.Lond.B(2004)thatanyofthesequencesgroupedwiththemandtotesttherobustnessofthegroupingsfound.Althoughtheseothertreesarenotshown,thegeneralizationsaboutgroupingarebasednotonlyonfigure2butalsoonthesedozensofothertrees.AsrRNAtreescannotbereliablyrootedusingbacterialout-groupsbecauseoflong-branchartefacts(Philippe2000),anindependentmethodisneeded.Figure1showsthatthepositionoftheeukaryoterootcanbeunambiguouslyspecifiedbytheuseoftwocomplementarygenefusions.Oneisthederivedfusionofdihydrofolatereductase(DHFR)andthymidinesynthetase(TS)firstnotedbyPhilippeetal.(2000)andshownbyStech-mann&Cavalier-Smith(2002,2003)tocharacterizeallgroupsofbikonteukaryotes;bycontrastallthreeopisthokontgroupsandAmoebozoahaveseparateDHFRandTSgenesliketheirbacterialancestors(Stechmann&Cavalier-Smith2002,2003ThisfusionshowsthatbikontsareacladeandthattherootcannotliewithinthemasitdoesonrRNAtreesrootedbybac-terialoutgroups.Thesecondisafusionofthefirstthreegenesofthepyrimidinebiosynthesispathway,whichisfoundinani-mals,fungiandAmoebozoa,butnotinbacteriaorbikonts,showingthatunikontsareacladeandthattherootcannotlieamongthem.Thereforetheeukaryoterootmustliepreciselybetweenthebikontsandtheunikonts(Stechmann&Cavalier-Smith2003).PreciselythesamepositionfortherootisshownbythecytosolicHsp90treewherethebranchlengthsofalltheeukaryotegroupsarerelativelyuniform,inmarkedcontrasttothegrosslynon-clock-likerRNA,ifitisrootedusingtherela-tivelycloselyrelatedendoplasmicreticulumparalogueandnotthemuchmoredistantbacterialoutgroups(Stechmann&Cavalier-Smith2003).Concatenatedtreesbasedonmitochon-drialproteins,whichalsohaverelativelyuniformbranchlengths(exceptfortheaberrantEuglenozoa,whichmustbeexcluded),alsoplacetherootpreciselybetweenbikontsandunikonts(Langetal.2002)whenrootedusingbacterialoutgroups,whicharerelativelymuchlessdistantthanforrRNA.IdeallyoneshouldalsoroottherRNAtree(figure2)betweentheunikontsandthebikonts.Infigure2thisisnotpossible,astheAmoebozoaarenotcleanlyseparatedfromthebikonts,soitisrooted(onlyapproximatelycorrectly)betweentheopisthokontsandthe3.IDENTIFYINGRELATIVESOFTHEPUTATIVELYANAEROBICMYSTERYCLADESFigure2indicatesthat22outofthe24sequencesthatpreviousauthors(Dawson&Pace2002;Lo2003;Stoeck&Epstein2003)couldnotevenplaceinestablishedkingdomsformonly17distinctgroups,everyoneofwhichisrelatedtoknowngroups,sevenbelonginginasingleorderofeugregarineApicomplexa.Therankoftheknowngroupstowhichtheycanbeassignedwithreasonableconfidenceishighlyvariable,rangingfromgenustophylum.Ishalldiscussthemintheordershownbythenumberedarrowsinfigure2.ThreeoftheselineagesbelongintheprotozoanphylumAmoebozoa.One(BOLA187/366)isveryrobustlysistertotheanaerobicuniciliateamoeboflagellate‘’(thequotessignifythatthestrainsequencedwasmisidentifiedandisnotevena)—itmightevenbelonginthesamegenus;thiscladeisnowtreatedastheclassBreviatea(Cavalier-Smithetal.2004).Thesecond(BOLA868)belongsintheorderEuamoebida(moderatesupport);mosttreesplaceitassistertothe OnlysixkingdomsoflifeT.Cavalier-Smith1255 familyAmoebidae,butinfigure2itissistertotheLeptomyxidaeinstead(itwassistertotheAmoebidaeinthecorrespondingbootstrapconsensustrees,bothdis-tanceandparsimony,withjustover50%support);itmaybelongintheAmoebidae.Thethirdlineage(LEMD267)issistertoonmosttrees(e.g.thebootstrappedconsensustreescorrespondingtofigure2;distance62%andparsimony84%support)orlessoftentoplusthemyxogastrids(figure2)—itmightevenbelongtothegenusorinthefamilyFilamoebidae,anditiscertainlyamemberoftheorderVaripodidaandtheclassVariosea(Cavalier-Smithetal.2004).Thusnoneofthesesequencesrepresentsanewkingdom,phylum,classorevenorder;assuchasmallfractionoftheAmoebozoahasbeensequenced,theymaynotevenbenewfamilies,generaorspecies.Theyaresimplyunidentifiedamoe-bozoansequences.Itcannotsafelybeassumedthatall20novelunculturedlineagesofDawson&Pace(2002)andStoeck&Epstein(2003)areanaerobic,asaerobiccystsorothercellsmustsometimesenteranaerobichabitats.Itislikely,however,thattheBOLA187/366cladeatleastisanaerobiclikeitssister‘M.invertens’.Thefactthatthisshort-branchcladedoesnotgroupwithotherAmoebozoa(itdoesinsometreesbasedonlongerrRNAsequences;etal.2001)maybejustaconsequenceoftheverypoorresolutionatthebaseoftheribosomalrRNAtree,probablycausedbyacombinationofrapidearlyradiationandsaturationeffects.Infigure2eventhecladedoesnotgroupwiththerestoftheAmoebozoa,althoughitdoeswithslightlydifferenttaxonsamplesorThefailureofDawson&Pace(2002)toidentifyanyofthesesequencesasAmoebozoaprobablypartlyarosebecausetheirfig.4includedonlyfiveAmoebozoa,not40ashere.Theirfig.4specificallydidnotincludeM.invertens,myxogastridsoranyAmoebidaeorLeptomyxidae,theverygroupstowhichtheirsequencesarerelated.Furthermore,theirfiveAmoebozoaformedthreeapparentlyunrelatedclades(fourcountingLEMD267)ratherthanasingleoneashereandinanotherrecentanalysiswithgoodtaxonsampling(Bolivaretal.2001).Thisemphasizestheimportanceofbroadtaxonsamplingforobtainingsoundtreesandmakingsweepingconclusionsaboutthenon-affinityofenvironmentalclonestoknownlineages.TheirtreeandthoseofStoeck&Epstein(2003)wereprobablyalsodistortedbyincludingtheverydistantbacterialout-groups,assuchextremeoutgroupstendtopulltheMyce-(andoftenalsotheArchamoebae)towardsthemandawayfromtheotherAmoebozoa,asseeninthetreesofCavalier-Smith),Milyutinaetal.(2001)andSilbermanetal.(2002)comparedwiththebetteronesofBolivaretal.(2001)andfigure2.However,thedistortionoffig.4ofDawson&Pace(2002)andtheartefactuallywidedisper-saloftheAmoebozoawerefarworsethaninthemaximum-likelihoodtreesofMilyutinaetal.(2001)thatalsoincludedbacterialoutgroups.Thereasonablywell-supportedgroupingtogetheroftheArchamoebae,andtheclade(mislabelledasacan-thamoebaeinDawson&Pace(2002):isnotanacanthamoebid—itdoesnotevenbelonginthesameclass;Cavalier-Smithetal.2004)inMilyutinaetal.Proc.R.Soc.Lond.B(2004)(2001),unlikeinDawson&Pace(2002)wheretheyappearedasthreeapparentlyunrelatedclades,isalsonote-worthyasitsuggeststhattheirdispersalintheDawson&Pace(2002)treecannotbeattributedsolelytotheinclusionofeubacterialoutgroups.However,Milyutina(2001)included10Amoebozoaandused1209align-mentpositions,comparedwithonly789inDawson&Pace(2002).BothfactorswouldhaveimprovedtheirLineage4(LEMD052/CCI78)belongstotheproto-zoanphylumCercozoaandisadeepbranchwithinthesubphylumEndomyxa(Cavalier-Smith&Chao2003onmosttreesitissistertoPhytomyxeawithmoderatebootstrapsupport(67%ontheparsimonytreecorre-spondingtofigure2),butinfigure2itissistertoallotherCercozoa.TheDawson&Pace(2002)fig.4includedonlytwocercozoans(notthemostdivergentpossible),not12ashere.Thispoortaxonsamplingcoupledwiththeuseofonly789not1044positionsprobablyexplainswhyLEMD052didnotgroupwithintheCercozoaontheirtree(thoughitwasnearby).Althoughthebootstrapsup-portforCercozoaislowinfigure2(probablybecauseonlypartialsequencescouldbeusedbecauseoftheincom-pletenessoftheenvironmentalones),bothLEMD052andCCI78possessthealmostuniquesignaturedeletionthatcharacterizesallCercozoa(Cavalier-Smith&Chao).Inseparateanalyseswithover80cercozoansequencestheLEMD052/CCI78cladebrancheswellwithinthemassistertoPhytomyxeawithgoodsupport.Atpresentonlyoneanaerobiccercozoanisknown(Cavalier-Smith&Chao2003),andthisnovelcladedoesnotgroupwithit.Weknowfromourownstudiesofenviron-mentalcercozoansequencesthathundredsofcercozoansequencescanbefoundthatarenotclosetoidentifiedstrains,andwehaveidentifiedacladefromaerobichabi-tatsinthesamepositionontherRNAtreeasLEMD052/CCI78(Bass&Cavalier-Smith2004).Thus,inadditiontonotbeinganewkingdomorevenphylum,thiscercozoanclademaynotevenbeanaerobic;itcouldbeanewclassorjustdeep-branchingPhytomyxea.ItisalsoquestionablewhethersequenceCCA32(Stoeck&Epstein2003)isfromananaerobe.Stoeck&Epstein(2003)didnotclaimthatitrepresentsanovelkingdom,asitgroupedwithstrongsupportontheirtreeswiththeapusozoan,butitwasincludedherebecausenosequencewaspreviouslyknowntogrouprobustlywith.Ondistancetreesitisweaklysistertothezooflagellate,notincludedinthetreeofStoeck&Epstein(2003),butinsomeparsimonyanalysesitgroupsweaklywithAT4-68inanunresolvedpositiondeepamongthebikonts.Thetaxonomicpositionhasitselfbeenproblematic,butitisnowclassifiedintheclassDiphyllateawithintheprotozoanphylumApusozoa(Cavalier-Smith2003).Figure2doesnotresolvethepositionoftheclade,showingitasaverydeepbikontbranchnotsistertoanyothergroup.However,veryminorchangesinthegammacorrectionparametersorintaxonsamplingcancauseittogroupwithApusomonadidaandtheotherApusozoaonthetree,aswellaswiththebrevi-ates.ThisisconsistentwithCCA32beingamemberoftheclassDiphyllateaofthephylumApusozoa.CurrentlynoanaerobicApusozoaareknown;althoughthereisno 1256T.Cavalier-SmithOnlysixkingdomsoflife Chromobiotakingdom Chromista Jakoba incarcerataBlepharisma americanumOxytricha nova Colpoda inflataParamecium tetraureliaHeterocapsa triquetra Colpodella ponticaPerkinsus sp. Toxoplasma gondii Oxymonas ‘Mastigamoeba invertens’Rhizamoeba Goniomonas BAQA65divBOLA48 Monocystis agilisLEMD119Parvilucifera infectansLeidyana migratorGregarina polymorpha Lecudina sp. Selenidium vivaxparasite of Ammonia beccariiCCI73CCI31CCA38 Ophriocystis elektroscirraNoctiluca scintillansColpodella sp.parasite of Selenidium terebellaeCryptosporidium serpentisCryptosporidium parvaHepatozoon catesbianaeMattesia geminataAmoebophrya Furgasonia blochmanniProtocruzia adherensCryothecomonas longipes Excavata kingdom FungiMyzozoaAlveolataLoukozoaRadiolariaPhytomyxea CercozoaRhizariakingdom AnimaliaRetariaopisthokonts 17 4 6Euglenozoa Excavata JakobeaMetamonada MycetozoaArchamoebeaLoukozoaAmoebozoaBreviateaApusozoa Varipodida Apusomonadida EuamoebidaVanellidae 12 51583 Malawimonadea Rhodophyta kingdom Plantae Anaeromonadea Gregarinea Apicomplexa AmoebozoaExcavata AncyromonadidaThecomonadea Diphyllatea Amoebidae Leptomyxidae137121116 97/99 69-51/5396 99 3530 62/ 648451 96759276 98 15/- 279797979995 -898259 8491 584720 635969620/- 59- 881981528585867295998315-86 /28549697275599/97-81 818784 10% nucleotide substitutionMyxogastria Figure2.(Captionopposite OnlysixkingdomsoflifeT.Cavalier-Smith1257 Figure2.Phylogeneticanalysis(BioNJ)of193eukaryotic18SribosomalRNAsequencesusing1044nucleotidepositions.The26environmentalsequencesnotpreviouslyassignedtoestablishedgroupsareinbold,andthe17cladestheyformaremarkedbyrednumberedarrows.Onlybootstrappercentagesof95%ormoreareshownexceptforthosedirectlyrelevanttothepositionsoftheenvironmentalclades:distance(1000pseudoreplicates:BioNJ)upperorleft;parsimony(1000pseudoreplicates)lowerorright;asthetreeiscrowded,someareshownbythenamedcladesnotthebipartitionpointsonthetreeitself;bipartitionswith100%supportbybothmethodsaremarkedbyasingleblackdisconthebranchitself.Thescalebarrepresents10%sequencedivergence.Thegreatdisparityofbranchlengthsindicatesthat18SrRNAisgrosslynon-clock-likeinitsevolutionaryrates;thiswouldbeseveraltimesgreaterthanshownhadnotthelongest-branchtaxabeenexcludedtoreduceartefacts.Alltaxanotincludedinthefourderivedkingdoms(Animalia,Fungi,Plantae,Chromista;shownincolour)belongintheunlabelledbasalkingdomProtozoa(inblack,seealsotable1). reasonwhysomeshouldnotexist,itseemsmorelikelythatCCA32isanaerobicflagellate(apartfrombreviates,anaerobicprotists,unlikeCCA32,havelongorverylongbranchesonrRNAtreesmakingthemveryhardtoplace;itseemsthatlossofthemitochondrialgenomegenerallycausesdramaticandlong-sustainedaccelerationofnuclearrRNAgeneevolution;Cavalier-Smith2002Sequenceccw88ofStoeck&Epstein(2003)turnsouttobeanartefactualchimerathatwasprobablyformedduringthePCRprocess.The3partofthemoleculeisclearlyfromadeep-branchingheterokontasithasthecharacteristicheterokontsignaturesequence(Cavalier-etal.1994)andbrancheswithorcloseto.The5partgroupsweaklywithanotheroftheirunidentifiedenvironmentallineages,CCW8(significantlyfromthesameDNAsample).Thischimerawasomittedfromtheanalysisshownbecauseoftheriskofitssystem-aticallydistortingthetree.SequenceCCW8mightbegenuinelyanaerobic.ItgroupsassistertothejakobidLoukozoaplustheEugleno-zoainfigure2;whenPercolozoaareaddedtothetreetheygroupwithEuglenozoa,notCCW8.ThephyleticpositionofCCW8istheleastclearofallsequencesfromanaerobichabitats;itislikelytobeamemberoftheLou-kozoa,apossiblyparaphyleticphylumatthebaseoftheexcavates(Cavalier-Smith2003).Theotherloukozoan,,isinanisolatedpositioninfigure2,butonmosttreesthatalsoincludePercolozoaitgroupsassistertoMetamonada,whicharehereinanunusualartefactualpositionwithintheAmoebozoa.Thebootstrappedcon-sensustreeforthefigure2datasetshowedassistertoAnaeromonadeawith19%supportandthiscladeplusAT4-68assistertothediscicristate/jakobidclade;excavateswouldbeholophyleticonthattreeexceptforthemisplacementofalonewithintheAmoebozoa.EstablishingtheunityandbasalbranchingorderforexcavatesisnotoriouslydifficultonrRNAtreesbecauseoftheirtremendousratevariations(Cavalier-Smith2003).Avarietyofanaerobicflagellateshasbeenobservedmicroscopicallybyecologistsbutnotstudiedsystematically;onesuchthatappearstohavejusttherightmixofcharacterstobranchpreciselyasdoesCCW8isProc.R.Soc.Lond.B(2004)flagellate2ofFencheletal.(1995),whichhasagroove(likeLoukozoaandPercolozoa)andtwoposteriadcilia(unlikethesetwophylaorEuglenozoa,butlikethepostu-latedmissinglinkbetweenLoukozoaanddiscicristates;Cavalier-Smith2003TheremainingeightlineagesofDawson&Pace(2002)andStoeck&Epstein(2003)allapparentlybelonginthephylumMyzozoainthesubphylumApicomplexa,classGregarineaandorderEugregarinida.ThisismostobviousforCCI7,whichissoclosetosp.thatitalmostcertainlybelongstothatverygenus.TheLEMD145/003cladeissistertoGregarinapolymorphawith100%support,whileLEMD119issistertowithveryhighsup-port.BOLA267/CCI73has100%supportassistertotheparasiteof,identifiedbyLeanderetal.asagregarine.BOLA458/212groupsweaklyassistertoSelenidiumvivax.CladeCCI31/CCA38groupswithmod-eratesupportwiththeparasiteclade.Twocladesthatconsistentlybranchwithinthegregarines(BAQA65/BOLA48andDH145(ofLoetal.2001)/CCW75/100)are,however,suchlongbranchesthattheiridentificationasgregarinesismoreopentoques-tion,butthereisnoreasonfromthepresentanalysistoregardanyofthesesequencesasrepresentingnovelking-domsorphylaorevenclassesororders.ItisnotsurprisingthatDawson&Pace(2002)andStoeck&Epstein(2003)failedtoidentifyanyofthesesequencesasgregarines,asneitherincludedanyknowngregarinesintheirpublishedtrees,andDawson&Pace’s(2002)hadonlyoneapicomplexan.WhenIomitgregar-inesaltogetherfromthepresentdatasetandaddPercol-ozoa(asinDawson&Pace2002),noneofthesesequencesgroupwithinApicomplexa,Myzozoaor(usually)eventheAlveolates,butformasinglelargeclade(ortwoclades)thatisverydistantlyrelatedandveryweaklysistertoPercolozoa;thislong-branchartefactiscorrectedonlybytheadditionofnumerousgregarines.GregarinesthemselveshavehighlyvariablerRNAevol-utionaryratesandbranchlengths;coupledwiththeexplosivebasalbranchingofMyzozoa,thismeansthatbootstrapsupportfortheunityoftheMyzozoaitselfisverylowwhentheyareallincluded.Itishardtoknowjusthowfig.4ofDawson&Pace(2002)wasproduced.Thetextcallsit‘onetypicaltree’withoutspecifyingmethod,butimplyingthatonlyonewasused;thefigurelegendcallsit‘aconsensustree’withoutexplaininghowafullyresolvedconsensustreewithbranchlengthswascalculated.Thetreeshowssupportvaluesfordifferentmethods,threethatdidnotallowforintersitevariationandone(Bayesian)thatdid;forthefirstthreetherewasnosupportatallforfourbipartitionsthatsepar-atedmostofthesequencesclaimedtorepresentseparatekingdoms.Thetreeitselflooksverydifferentfromtheonepresentedhere.Itispectinatewiththeknownandenvironmentalsequencesinpositionsfromtoptobottominapproximateproportiontothebranchlengths.Suchatreehasallthehallmarksofbeingdominatedbylong-branchartefacts,andisjustthekindoftreethatoneusedtogetbeforecorrectionsforintersiteratevariationwereintroduced(Cavalier-Smith1993,1995).Correctionforintersiteratevariationisimportanttoreducelong-branchartefacts,thoughsuchcorrectioncanneverbetotallysuc-cessfulformoleculesthatareasgrotesquelynon-clock- 1258T.Cavalier-SmithOnlysixkingdomsoflife likeasrRNA.ThereadercaneasilycomparetheoveralleukaryotetreesproducedbybothmethodsforrRNAbyconsultingBolivaretal.(2001)andMilyutinaetal.andforproteinsbyconsultingBaptesteetal.(2002).Inallcasesthecorrectedtreesaremorecongruentwithotherbiologicaldatathanaretheuncorrectedtrees.TherRNAtreesshowninfig.2ofMilyutinaetal.(2001)areparti-cularlyinstructiveaslikeinDawson&Pace(2002)andStoeck&Epstein(2003)theywererootedbybacterialoutgroupsandincludedmicrosporidiaandeventhelong-branchForaminifera.Theiruncorrectedfig.2ispecti-nateandgradedinbranchlengthsfromtoptobottomlikethatofDawson&Pace(2002),butisnotasbadasthereisamajoramoebozoancladeapartfromslimemoulds.TheirtreeisalsobetterthanthatofDawson&Pace(2002)inshowingaparabasalid/diplomonadclade,nowsupportedalsobyseveralproteintrees,includingcpn60etal.2002),andsharedlateraltransfers(seereviewbyCavalier-Smith2003).Theirgamma-correctedfig.2isverydifferent.Itislargelynon-pectinateandnon-gradedwithabigbangofrapidlyradiatingmostlyshort-branchlineagesfromwhichnumerouslong-branchsequencesstemlargelyindependently,withageneralappearancesimilartomypresenttree,eventhoughitisrooteddifferentlyandartefactuallyontheparabasalidsequence.Astrikingdifferencebetweenthetwotreesisinthepositionofthesolemicrosporidian:intheuncorrectedtreeitiswelltowardsthebaseamongotherlong-branchtaxa(butnotasfarawayfromitstruefungalpositionasinDawson&Pace(2002)),butinthecorrectedtreeitisbetweenopisthokontsandAmoebozoa,i.e.intheunikontpartofthetree,muchclosertoitstruepositionwithintheFungi.Fig.2wasalsobiologicallyrealisticingroupingtogethertheshort-branchradiolariaandtheultra-long-branchForaminifera,thefirstmolecularsupportforthegroupRetaria,establishedonmorphologicalgrounds(Cavalier-Smith1999).Fig.4ofDawson&Pace(2002)wouldseemtobeanuncorrectedtree,whichwouldarte-factuallyhavedispersedbothnamedandunidentifiedsequences,impedingrecognitionoftheirtrueaffinities.Thefrequentcapacityofgamma-correcteddistancetreestorecover(albeitusuallywithweaksupport)cladeswell-substantiatedbymorphologybutwithamixtureofshort-andvery-long-branchtaxathatusuallydisruptthecladeonuncorrectedtreesisalsoevidentinthepresentfigure2:e.g.theshort-branchjakobidsaregroupedwiththelong-branchEuglenozoa,metamonadsareholophyletic,withtheshort-branchanaeromonadcladegroupedwiththe(ifretortamonadsandtheultra-long-branchdiplomonadsarealsoincludedtheygroupasexpectedwith)andthelong-branchmyxogastridsaregroupedwiththemedium-However,figure2isnotaperfectrepresentationofeukaryotephylogeny.Asinmostpreviousstudies,severalmajorgroupsknowntobeholophyleticfromotherexten-siveevidence(seereviewbyCavalier-Smith&Chao)arenotrecoveredasclades(e.g.thekingdomsPlantaeandChromista,andthechromalveolates(ChromistaplusAlveolata)).Moreover,thereisvirtuallynosupportforthebasal-branchingorders(bootstrapsup-portforthemtypicallyintherangeof0–10%)consistentwithaveryrapid,virtuallyexplosiveradiationimmediatelyProc.R.Soc.Lond.B(2004)followingtheoriginoftheeukaryoticcell(Cavalier-Smith&Chao2003).Nonetheless,thepresenttreeismuchmorecongruentwithothermoleculardata,ultra-structureandcellbiology(Cavalier-Smith2002Cavalier-Smith&Chao2003)thananyrecenttreewithonlylimitedtaxonsamplingand/orincludingbacterialoutgroups(Dawson&Pace2002;Silbermanetal.(includingbacterialeadsmanyauthorstoignorepartsofthemoleculeinformativeforeukaryotephylogeny).Incontrasttosuchtrees,includingthoseofDawson&Pace(2002),thelong-branchAmoebozoagroupwithmostoftheshort-branchaerobicAmoebozoa,notarbitrarilyelse-whereorartefactuallywiththelong-branchexcavates,andthelatterdonotallartefactuallyclustertogether.Itisverydifficultindeedfortree-reconstructionalgorithmstocopewiththeextremevariationsinevolutionaryrateandmodeoftherRNAmolecule,especiallyinthevastlyacceleratedsecondarilyamitochondrial,andthediscicristateandmyxogastridamoebozoanlineages.Nonetheless,includingalargenumberoftaxainthetree,includingashighaproportionofthemoleculeasisreasonableandexcludingtheexcessivelydistantoutgroups(bacteria)arekeystoreducingtheseproblemsandobtainingarelativelyundis-tortedphylogeny.ThesethreefeaturesofthepresenttreemayexplainwhyExcavataandAmoebozoaarebothmuchlessrandomlydispersedinthepresenttreethanintaxo-nomicallysparsebacteriallyrootedtrees.ThepositionsofmanyexcavatesonrRNAtreesarenotoriouslysensitivetotaxonsamplingandthephylogeneticparameters,becauseoftheirexceptionaldisparityinrRNAevolution-aryrateandmode;onlyrarelydoallbranchtogether(Cavalier-Smith2002,20034.IDENTIFYINGRELATIVESOFTHEMYSTERYDEEP-OCEANCLADESThetreeofLoetal.(2003)alreadyweaklysuggestedthattwooftheirmysterysequencesweresisters.Figure2showsthatoneofthem,AT-50,doesindeedbranchwithintheclassThecomonadeaofthephylumApusozoa,actuallywithinthegenus(sistertosp.Millportonbothcon-sensustrees),suggestingthatitissimplyanadditionalspecies.Fromourpreviousworkwecon-cludedthatisprobablyvastlyunder-described(Cavalier-Smith&Chao2003).IfAT-50isan,thatmeansthatwealreadyhavesixdra-maticallydifferentsequencesforthisgenusinwhichonlyninespecieshavebeendescribed(Mylnikov1999).Onlytwomorphospecies(A.mutabilisA.debruynei)werepreviouslyrecordedfromthedeepocean(Arndtetal.2003),andtheirsequencesaremarkedlydifferentfromthatofAT-50.Thelatterneednotnecessarilybe,however,fortherearecurrentlynosequencesavailablefromthethecomonadorderHemimastigida.Asthisorderisconsideredtohaveevolvedfroman-likeancestor(Cavalier-Smith2000),oneofthesesequencesmightbefromahemimastigid;althoughhemimastigidswerepreviouslyknownonlyfromterrestrialenvironments,twogenerawererecentlyrecordedinverydeep(1325–1249m)Mediterraneansediments(Arndtetal.2003).Thus,whilewecanbereasonablyconfidentthatAT-50isathecomonad OnlysixkingdomsoflifeT.Cavalier-Smith1259 sequence,wecannotexcludethepossibilitythatitbelongstotheorderHemimastigidaratherthantheApusomonadida.OnsometreesAT4-11groupsassistertotheApuso-monadida(rarelywithinit),butitcanalsogowithintheAmoebozoaor,asinfigure2,inadeeppositionwithnospecificrelative.ItseemslikelythatitbelongseithertotheApusozoaortotheAmoebozoa,asthesephylaseldomappearstrictlyholophyleticonrRNAtrees,butitmightbelongtoathirdprotozoanphylum.SequenceAT4-68isevenhardertoplace.Ittypicallyoccupiesadeepbutvariablepositionwithinthebikontswithnoclearrelatives.InsomedistancetreesitgroupswithverylowsupportassistertotheGlaucophyceaeorwithintheexcavatesassistertoMetamonada,butwithparsimonyitissistertoCCA32nearthebaseoftheplantkingdom.Glaucophytesaretheonlyoneofthethreemajorgroupsoftheplantkingdominwhichsecondarilyheterotrophicspecieshavenotbeenidentified;bycontrasttherearenon-photosyntheticgreenplantsandredalgaethathavesecondarilylostphotosyn-thesis(butretainedplastids,presumablyforstarchand/orfatty-acidsynthesis;Cavalier-Smith1993).Becauseetal.(2003)foundnosequencesfromauth-enticphotosyntheticalgaeintheirdeep-seasamples,CCA32isprobablyalsofromaheterotroph.WhileitmightbethefirstheterotrophicmemberofthephylumGlaucophytaandofgreatpotentialinterestforunder-standingtheearlyevolutionoftheplantkingdomifonlyitcouldbecultured,itsoccasionalgroupingwithCCA32orwithintheexcavatesmakesitlikelythatitissimplyabikontprotozoan.5.ALLANAEROBICEUKARYOTESAREPROBABLYDERIVED:PROBLEMSOFROOTINGTHETREEInaccordancewithpreviousevidencethatthelastcom-monancestorofalleukaryoteshadmitochondriacapableofaerobicrespiration(probablyfacultativelyratherthanobligately;Cavalier-Smith2002)andthatallanaerobiceukaryotesarosesecondarilybyconvertingmitochondriaintohydrogenosomesormitosomes(Silbermanetal.2002;Williamsetal.2002;Tovaretal.2003),alltheputa-tivelyanaerobiclineagesclaimedtorepresentnewking-doms(Dawson&Pace2002;Stoeck&Epstein2003)nestwellwithinaerobiccladesinfigure2.Inthepastdecadedatafromnumerousproteinsandthediscoveryofmito-somesfirmlyestablishedthesecondarynatureofamito-chondrialeukaryotes(Cavalier-Smith2002,2003Keeling2003;Roger1999;Silbermanetal.2002;Will-etal.2002).Withnoevidenceorargumentswhatso-ever,Dawson&Pace(2002)superficiallydismissedas‘lateraltransfer’thedisparateandextensiveevidenceforthismajoradvanceineukaryoticphylogeny.Theirtreeisalsotopologicallyincorrect,asshownbythenon-groupingofmicrosporidiawithfungifromwhichtheyevolved(Cavalier-Smith2000;Keeling2003),andbythethreewidelydispersedamoebozoancladesthatinbetteranaly-sescometogether(Bolivaretal.2001),astheydoinfigure2;thetopologyinDawson&Pace(2002)wasprobablydistortedbylong-branchbacterialoutgroups.UnwiseinclusionofbacteriaandthedrasticallyshortenedmicrosporidialgenesalsomeantthattheyexcludedProc.R.Soc.Lond.B(2004)numerousphylogeneticallyinformativesites,usingonly789nucleotidepositions,comparedwith1044inmyAlthoughtheirtreesaretechnicallyinferiortomanypublishedones,beingtopologicallyincorrectinmajorrespectsandrootedinentirelythewrongplace,Dawson&Pace(2002)andStoeck&Epstein(2003)assumethatbothfeaturesarecorrectwhendiscussingso-calleddeepbrancheswithinthem.Theyrootedtheirtreesusingbac-terialoutgroups(despitethisbeingknowntogiveanincorrectrootamongthelongest-brancheukaryotes;Cavalier-Smith2002;Simpson&Roger2002;Stech-mann&Cavalier-Smith2002).Itisremarkablethateightofthelineagesclaimedtorepresentnovelanaerobicking-domsturnouttobelongtoasinglegregarineorder,asnoanaerobicgregarineshavesofarbeendescribed.Theyareneithernovelkingdomsnorearlydiverging;aretheyevenanaerobic?Thecercozoancladeandtwoofthethreeamo-ebozoancladesmaynotbeanaerobicandarecertainlynot‘earlydiverging’sequences.Onlyoneclade(BOLA187/BOLA366)hasanypotentialtobeearlydiverging,butitisnotreallynovel,beingsistertoM.invertens’.Ifthebreviateclade(‘M.invertensBOLA187/BOLA366)reallybelongswithinaholophyleticphylumAmoebozoa,asseemsmostlikely(Cavalier-Smithetal.2004),thenitwouldnotbeearlydivergingeither.Ifso,asfigure1makesclear,therootingoftheeukaryotetreebetweenbikontsandunikonts(notwithinthemosthighlyderivedbikonts,asinfig.4ofDawson&Pace(2002))wouldmeanthattherearealsonoknownextanteukaryotes,whetheraerobicoranaerobic,thatdivergedpriortothelastcommonancestorofanimalsandplants.Farfromrevealing‘novel’early-diverginglineages,asclaimed,bothstudiesfailedtodetectanynovelanaerobiclineagesthatareearlydiverging(giventhecorrectrootingofthetree),whichconsiderablystrengthensthecurrentinterpretationthatthereareprobablynoprimitivelyami-tochondrialeukaryotes.Onlythebreviatesremainasposs-iblecandidatesforsuchaposition.‘’needstobestudiedforthepresenceorabsenceofrelictHsp70andCpn60chaperonesfromamitochon-drialancestryandforthepresenceofthetwofusiongenes.Ifbothsetsofgenesareunfused,thiswouldprovideevi-dencethatbreviatesareearly-diverginganaerobiceukary-otes,aspostulatedtoexistintheArchezoahypothesis(Cavalier-Smith1983).IftheDHFRandTSgenesareunfusedbutthethreepyrimidine-biosynthesisonesarefused,thiswouldplacethebreviatesclearlyintheAmoe-bozoaandunikonts;theconversewouldplacethemonthebikontsideofthebasaleukaryotebifurcation.WhileIhaveidentifiedtwooutofthefourputativelyaerobiccladesofLoetal.(2001,2003),theidentityoftheothertworemainsunclear.Thisisunsurprisingastherearenumerousaerobicprotozoangeneraofuncertaintaxonomicpositionthathavenotbeencultured,sequencedorstudiedultrastructurally.Untilwehavesuchinformationwecannottellwhethertheyrepresentnewordersorclasses(bothlikely)orevenphyla(unlikely,butpossible).NaiveinterpretationsofrRNAtreesandproteinparalogytreeshavegrosslymisledevolutionarybiology(Cavalier-Smith2002IfbreviatesturnouttobesecondarilyanaerobicAmoebozoaandAmoebozoaprovetobeholophyletic,the 1260T.Cavalier-SmithOnlysixkingdomsoflife rootingoftheeukaryotetreebetweenunikontsandbikontsmeansthatthereisnosuchthingasadeep-branchingeukaryote,inthesenseofonebranchingpriortothelastcommonancestorofanimalsandplants.Thecommonuseoftheterms‘crown’andbasal/deepbymanyrRNAsequencersreflectmultiplemisunderstandings(Cavalier-Smith1999).Thedichotomybetween‘crowngroups’and‘basallineages’isbiologicallytotallymeaning-lessandsimplyreflectstheseverelong-branchexclusionartefactsandthesystematicmisrootingofrRNAtreeswhenusingbacterialoutgroups.Forthisreason,andbecausethecladisticterm‘crown’properlyreferstoallextanteukaryotes(Cavalier-Smith2002),themuddledandmisleadingphrase‘crowneukaryote’shouldnolongerbeusedforanysubsetofextanteukaryotes.ThewidespreadbutfalseassumptionthatrRNAisauniversalmolecularclockhasledtoveryseriousmisinter-pretationsofthetreeoflife,especiallyconcerningtherootsofboththebacterialandeukaryoticpartsofthetree(Cavalier-Smith2002).Toreconstructphylogenysatis-factorilyitisfundamentallyunsoundtorelyonasinglemolecule,asDawson&Pace(2002)appearto.Treesbasedonlargenumbersofmolecules(Baldaufetal.etal.2002),andgenic,biochemicalandultra-structuraldatathatcanbetreatedcladistically(Stechmann&Cavalier-Smith2002,2003;Cavalier-Smith&Chao2003),aswellaspalaeontology(Cavalier-Smith2002)andsingle-genetreesforproteins(e.g.Hsp90,whichappearstobemuchmoreclock-likethanrRNA;Stechmann&Cavalier-Smith2003)aswellasrRNAarealltakennoteofinthesynthesisoffigure1.Dawson&Pace(2002)andStoeck&Epstein(2003)ignorethevastmajorityofsuchdataoneukaryoteevol-ution,includingtheevidencethatthepositionsoftherootsoftheirtreesareprofoundlyincorrect(Simpson&Roger2002;Stechmann&Cavalier-Smith2002;Cavalier-Smith&Chao2003);theirdiscussionintermsofa‘crownradiation’and‘deepbranching’isbasicallyupsidedown.Itisthedivergencesamongtheshort-branch,misnamed,‘crowngroups’thatarebasal,whereastheirso-called‘deep’branches(mostlyexcavates)areactuallyamongthemostderivedgroups.ItisparticularlyastoundingthatDawson&Pace(2002)ignoretheevi-denceforthefungalnatureofmicrosporidia,whichwasthefirstgrouptoshowconclusivelyhowdramaticallymis-leadingtherRNAtreecanbewhennaivelyinterpreted(Embley&Hirt1998;Roger1999;Cavalier-Smith2000etal.2002;Keeling2003).Whennaivelyinter-pretedintheabsenceofotherdata,rRNAtreesarethesinglemostmisleadingsourceofinformationwehaveaboutthehistoryoflife.Whenintegratedwithalltheotherinformationtheyareveryvaluable.Analogousclaimsof14noveldivision(phylum)/kingdom-levelbacteriallineages(Hugenholtzetal.basedonunidentifiedenvironmentalsequencesareprob-ablyalsoill-founded;long-branchproblems,manyexacer-batedbythermophilicbias(Cavalier-Smith2002Gribaldo&Philippe2002),coupledwiththenon-existentresolutionatthebaseoftheeubacterialrRNAtree,impedetheirplacementinoneofthesevenestablishedeubacterialphyla(table1inCavalier-Smith2002).Theeubacterialtreeisequallypoorlyresolvedatitsbase,andtheproblemofitsrootingisevenmoreseverethanforProc.R.Soc.Lond.B(2004)eukaryotes.Contrarytoover-confidentbutdeeplyheldassumptions,wedonotactuallyknowwherethebaseofthetreeoflifeis,butitismuchmorelikelytobeamongtheGram-negativeeubacteria(Cavalier-Smith2002thanbetweenarchaebacteriaandeubacteriaasisoftenassumed.Untilweknowtheanswertothisquestion,anyreferenceto‘deeplybranching’or‘ancient’prokaryoticlineages(e.g.Gaucheretal.2003)ispotentiallymislead-ingandimmenselymorecontroversialthanitisgenerallyrealizedtobe.WhileitmaybethattheEobacteriaaretheearliest-divergingphylum(Cavalier-Smith2002),wecannotcurrentlyexcludethepossibilitythattheirappar-entlyprimitivecharactersaresecondarilysimplifiedandthat,asineukaryotes,therearenoextant‘earlydiverging’lineages.EvenwhenrepresentativesofnovelrRNAlin-eagesarecultivated,asisincreasinglybeingdone(Leadbetter2003),andpartlycharacterized,rankingthemasphylapurelyonthebasisofthedegreeofrRNAdiver-gence(e.g.Zhangetal.2003)istaxonomicallyunsoundasrRNAdivergencecanbegreatlyacceleratedforrela-tivelytrivialreasonsanddoesnotnecessarilycorrelatewellwithbiologicallymoreimportantcharacterdifferences.6.CONCLUSIONSFarfromrevealingnovelkingdoms,thenewrRNAsequencesfromanaerobichabitats(Dawson&Pace2002;Stoeck&Epstein2003)andthedeepocean(Loetal.2003)showthatourunderstandingofeukary-otehigh-leveldiversityisactuallynowrathergood.Whenproperlyanalysedtheytellus,inconjunctionwithnumer-ousstudiesfromaerobichabitats(Loetal2001;Moon-vanderStaayetal.2001;Moreira&Loa2002),thattheremaybeveryfew,ifany,pre-viouslyunknownprotistphyla—andno‘newkingdoms’—remainingtobe‘discovered’.Theseotherstudiesdidfindaverysmallnumberofsequencesthatarelongbranchesandashardtoplaceassomeofthoseincludedhere,e.g.etal.(2002),whoalsomisrootedthetreeandmadesimilarunwarrantedstatementsabout‘earlybranch-ing’anaerobiceukaryotes.Whileitisnotimpossiblethatsomeoftheselineagesmightrepresentnewphyla,itismoreprobablethattheyalsowillturnouttobehard-to-placelong-branchrepresentativesofestablishedones.Idonotwishtobemisunderstoodaspredictingthatnonewphylawillbediscovered.Therearescoresofprotistgenerathathavenotyetbeenstudiedultrastructurally(e.g.the,whichwavesitsfilopodialikesema-phores(Hausmannetal.);itsphylumisunknown;myhunchisthatitbelongsintheCercozoa,whichareunsurpassedinfilopodialwonders(Cavalier-Smith&Chao2003))andtherearemanyothersstillundescribed.Itispossiblethatsomemaybelonginasmallgroupsuf-ficientlydistinctfromknownphylatomeritoneofitsown;Apusozoaisonesuchphylumthatwasonlyrecentlyestab-lished(Cavalier-Smith2002).Theremaybeothers,butprobablyratherfew,andpossiblynone.Phylaandking-domsarenotactuallythingsthatonediscovers,butcon-ceptualentitiesthatsystematistscreatebydeliberatelygroupingtogetherknownorganisms.Novelorganisms,moleculesorlineagescanbediscovered,butonelessonfromthepresentstudyisthatmoleculartreesusedtoclaimnovelmajoreukaryotelineagesshouldinfuture OnlysixkingdomsoflifeT.Cavalier-Smith1261 includethebroadestrangeofprotistdiversityasinfigure2andnotjustatinysubsetofit.RibosomalDNAsequencingisveryusefulindeedforevolutionaryandecologicalstudies,andweourselvesareusingenvironmentalrDNAsequencingtoexploretheimmensehiddendiversityofprotists,butithastobecriti-callyintegratedintotraditionaltaxonomyandnotgivenaspuriouspre-eminence.IthankNERCforresearchgrants,andtheCanadianInstituteforAdvancedResearchandNERCforfellowshipsupport.Archibald,J.M.,O’Kelly,C.J.&Doolittle,W.F.2002Thechaperoningenesofjakobidandjakobid-likeflagellates:implicationsforeukaryoticevolution.Mol.Biol.Evol.Arndt,H.,Hausmann,K.&Wolf,M.2003Deep-seahetero-trophicnanoflagellatesoftheEasternMediterraneanSea:qualitativeandquantitativeaspectsoftheirpelagicandben-thicoccurrence.Mar.Ecol.Prog.Ser.,45–56.Baldauf,S.L.,Roger,A.J.,Wenk-Siefert,I.&Doolittle,W.F.2000Akingdom-levelphylogenyofeukaryotesbasedoncombinedproteindata.,972–977.Bapteste,E.(and10others)2002Theanalysisof100genessupportsthegroupingofthreehighlydivergentamoebae:,andProc.NatlAcad.Sci.USA,1414–1419.Barns,S.M.,Delwiche,C.F.,Palmer,J.D.&Pace,N.R.1996Perspectivesonarchaealdiversity,thermophilyandmonophylyfromenvironmentalrRNAsequences.Proc.NatlAcad.Sci.USA,9188–9193.Bass,D.&Cavalier-Smith,T.2004Phylum-specificenviron-mentalDNAanalysisrevealsremarkablyhighglobalbiodiv-ersityofCercozoa(Protozoa).Int.J.Syst.Evol.Microbiol(Inthepress.)Bolivar,I.,Fahrni,J.F.,Smirnov,A.&Pawlowski,J.2001SSUrRNA-basedphylogeneticpositionofthegenera(Lobosea,Gymnamoebia):theoriginofgymnamoebaerevisited.Mol.Biol.Evol.,2306–2314.Cavalier-Smith,T.1983A6-kingdomclassificationandaunifiedphylogeny.InEndocytobiologyII(ed.W.Schwemmler&H.E.A.Schenk),pp.1027–1034.Berlin:deGruyter.Cavalier-Smith,T.1983Endosymbioticoriginofthemito-chondrialenvelope.InEndocytobiologyII(ed.W.Schwemmler&H.E.A.Schenk),pp.265–279.Berlin:deCavalier-Smith,T.1993KingdomProtozoaandits18phyla.Microbiol.Rev.,953–994.Cavalier-Smith,T.1993Theorigin,lossesandgainsofchlor-oplasts.InOriginofplastids:symbiogenesis,prochlorophytesandtheoriginsofchloroplasts(ed.R.A.Lewin),pp.291–348.NewYork:Chapman&Hall.Cavalier-Smith,T.1995Membraneheredity,symbiogenesis,andthemultipleoriginsofalgae.InBiodiversityandevolution(ed.R.Arai,M.Kato&Y.Doi),pp.75–114.Tokyo:TheNationalScienceMuseumFoundation.Cavalier-Smith,T.1998Arevisedsix-kingdomsystemoflife.Biol.Rev.Camb.Phil.Soc.,203–266.Cavalier-Smith,T.1999Principlesofproteinandlipidtar-getinginsecondarysymbiogenesis:euglenoid,dinoflagellate,andsporozoanplastidoriginsandtheeukaryoticfamilytree.J.EukaryoticMicrobiol.,347–366.Cavalier-Smith,T.2000Flagellatemegaevolution:thebasisforeukaryotediversification.InTheflagellates(ed.J.R.Green&B.S.C.Leadbeater),pp.361–390.London:Taylor&Francis.Proc.R.Soc.Lond.B(2004)Cavalier-Smith,T.2000Whatarefungi?InTheMycota,vol.VIIpartA(ed.D.J.McLaughlin,E.J.McLaughlin&P.Lemke),pp.3–37.Berlin:Springer.Cavalier-Smith,T.2002Thephagotrophicoriginofeukary-otesandphylogeneticclassificationofProtozoa.Int.J.Syst.Evol.Microbiol.,297–354.Cavalier-Smith,T.2002Theneomuranoriginofarchaebac-teria,thenegibacterialrootoftheuniversaltreeandbacterialInt.J.Syst.Evol.Microbiol.,7–76.Cavalier-Smith,T.2003Protistphylogenyandthehigh-levelclassificationofProtozoa.Eur.J.Protistol.,338–348.Cavalier-Smith,T.2003TheexcavateprotozoanphylaMeta-monadaGrasseemend.(Anaeromonadea,Parabasalia,,Eopharyngia)andLoukozoaemend.(Jakobea,):theirevolutionaryaffinitiesandnewhigherInt.J.Syst.Evol.Microbiol.,1741–1758.Cavalier-Smith,T.2003Genomicreductionandevolutionofnovelgeneticmembranesandprotein-targetingmachineryineukaryote–eukaryotechimaeras(meta-algae).Phil.Trans.R.Soc.Lond.,109–134.(DOI10.1098/rstb.Cavalier-Smith,T.&Chao,E.E.2003Molecularphylogenyofcentrohelidheliozoa,anovellineageofbikonteukaryotesthatarosebyciliaryloss.J.Mol.Evol.,387–396.Cavalier-Smith,T.&Chao,E.E.2003Phylogenyandclassi-ficationofphylumCercozoa(Protozoa).,341–Cavalier-Smith,T.&Chao,E.E.2003PhylogenyofChoano-zoa,Apusozoa,andotherProtozoaandearlyeukaryoteJ.Mol.Evol.,540–563.Cavalier-Smith,T.&Chao,E.E.2004Protalveolatephy-logenyandtheoriginsofSporozoaanddinoflagellates(phylumMyzozoanom.nov.)Eur.J.Protistol.,21–48.Cavalier-Smith,T.,Allsopp,M.T.E.P.&Chao,E.E.1994Thraustochytridsarechromists,notFungi:18SrRNAsigna-turesofHeterokonta.Phil.Trans.R.Soc.Lond.Cavalier-Smith,T.,Chao,E.E.&Oates,B.2004MolecularphylogenyofAmoebozoaandtheevolutionarysignificanceoftheunikontEur.J.Protistol.,21–48.Dawson,S.C.&Pace,N.R.2002Novelkingdom-leveleukar-yoticdiversityinanoxicenvironments.Proc.NatlAcad.Sci.,8324–8329.Edgcomb,V.P.,Kysela,D.T.,Teske,A.,deVeraGomez,A.&Sogin,M.L.2002BenthiceukaryoticdiversityintheGuaymasBasinhydrothermalventenvironment.Proc.NatlAcad.Sci.USA,7658–7662.Embley,T.M.&Hirt,R.P.1998Earlybranchingeukaryotes?Curr.Opin.Genet.Dev.,624–629.Fenchel,T.&Finlay,B.J.1995Ecologyandevolutioninanoxic.OxfordUniversityPress.Fenchel,T.,Bernard,C.,Esteban,G.,Finlay,B.J.,Hansen,P.J.&Iversen,N.1995MicrobialdiversityandactivityinaDanishfjordwithanoxicdeepwater.,45–100.Gaucher,E.A.,Thomson,J.M.,Burgan,M.F.&Benner,S.A.2003Inferringthepalaeoenvironmentofancientbac-teriaonthebasisofresurrectedproteins.,285–Giovannoni,S.J.,Britschgi,T.B.,Moyer,C.L.&Field,K.G.1990GeneticdiversityinSargassoSeabacterioplank-,60–63.Gribaldo,S.&Philippe,H.2002AncientphylogeneticTheor.Popul.Biol.,391–408.Harper,J.T.&Keeling,P.J.2003Nucleus-encoded,plastid-targetedglyceraldehyde-3-phosphatedehydrogenase(GAPDH)indicatesasingleoriginforchromalveolateplas-Mol.Biol.Evol.,1730–1735. 1262T.Cavalier-SmithOnlysixkingdomsoflife Hausmann,K.,Hulsmann,N.,Polianski,I.,Schade,S.&Weitere,M.2002Compositionofbenthicprotozoancom-munitiesalongadepthtransectintheeasternMediterraneanDeepSeaRes.,1959–1970.Hausmann,K.,Weitere,M.,Wolf,M.&Arndt,H.2002Meteorasporadicagen.nov.etsp.nov.(Protista):anextraordinaryfree-livingprotistfromtheMediter-raneandeepsea.Eur.J.Protistol.,171–177.Hugenholtz,P.,Pitulle,C.,Hershberger,K.L.&Pace,N.R.1998NoveldivisionlevelbacterialdiversityinaYellowstonehotspring.J.Bacteriol.,366–376.Keeling,P.J.2003Congruentevidencefromalpha-tubulinandbeta-tubulingenephylogeniesforazygomyceteoriginofmicrosporidia.FungalGenet.Biol.,298–309.Lang,B.F.,O’Kelly,C.,Nerad,T.,Gray,M.W.&Burger,G.2002Theclosestunicellularrelativesofanimals.,1773–1778.Leadbetter,J.R.2003Cultivationofrecalcitrantmicrobes:cellsarealive,wellandrevealingtheirsecretsinthe21stcenturylaboratory.Curr.Opin.Microbiol.,274–281.Leander,B.S.,Clopton,R.E.&Keeling,P.J.2003Phy-logenyofgregarines(Apicomplexa)asinferredfromsmall-subunitrDNAandbeta-tubulin.Int.J.Syst.Evol.Microbiol.,345–354.a,P.,Rodriguez-Valera,F.,Pedros-Alio,C.&Moreira,D.2001Unexpecteddiversityofsmalleukaryotesindeep-seaAntarcticplankton.,603–607.a,P.,Philippe,H.,Gail,F.&Moreira,D.2003Autochthonouseukaryoticdiversityinhydrothermalsedi-mentandexperimentalmicrocolonizersattheMid-AtlanticProc.NatlAcad.Sci.USA,697–702.Milyutina,I.A.,Aleshin,V.V.,Mikrjukov,K.A.,Kedrova,O.S.&Petrov,N.B.2001Theunusuallylongsmallsub-unitribosomalRNAgenefoundinamitochondriateamoe-Pelomyxapalustris:itsrRNApredictedsecondarystructureandphylogeneticimplication.,131–139.Moon-vanderStaay,S.Y.,DeWachter,R.&Vaulot,D.2001Oceanic18SrDNAsequencesfrompicoplanktonrevealunsuspectedeukaryoticdiversity.,607–610.Moreira,D.&Loa,P.2002Themolecularecologyofmicrobialeukaryotesunveilsahiddenworld.,31–38.Morris,R.M.,Rappe,M.S.,Connon,S.A.,Vergin,K.L.,Siebold,W.A.,Carlson,C.A.&Giovannoni,S.J.2002SAR11cladedominatesoceansurfacebacterioplankton,806–810.Mylnikov,A.P.1999Newbrackishwateramoeboidflagellatesofthegenus(Apusomonadida,Protozoa).Zool.Zhurnal,771–777.[InRussian.]Philippe,H.2000Opinion:longbranchattractionandprotist,307–316.Philippe,H.,Lopez,P.,Brinkmann,H.,Budin,K.,Germot,A.,Laurent,J.,Moreira,D.,Muller,M.&LeGuyader,H.2000Early-branchingorfast-evolvingeukaryotes?Ananswerbasedonslowlyevolvingpositions.Proc.R.Soc.,1213–1221.(DOI10.1098/rspb.2000.1130.)Proc.R.Soc.Lond.B(2004)Roger,A.J.1999ReconstructingearlyeventsineukaryoticAm.Nat.(Suppl.4),S146–S163.Sait,M.,Hugenholtz,P.&Janssen,P.H.2002Cultivationofgloballydistributedsoilbacteriafromphylogeneticlineagespreviouslyonlydetectedincultivation-independentsurveys.Environ.Microbiol.,654–666.Silberman,J.D.,Simpson,A.G.,Kulda,J.,Cepicka,I.,Hampl,V.,Johnson,P.J.&Roger,A.J.2002Retortamonadflagellatesarecloselyrelatedtodiplomonads:implicationsforthehistoryofmitochondrialfunctionineukaryoteevol-Mol.Biol.Evol.,777–786.Simpson,A.G.&Roger,A.J.2002Eukaryoticevolution:get-tingtotherootoftheproblem.Curr.Biol.,R691–R693.Simpson,A.G.,Roger,A.J.,Silberman,J.D.,Leipe,D.D.,Edgcomb,V.P.,Jermiin,L.S.,Patterson,D.J.&Sogin,M.L.2002Evolutionaryhistoryof‘early-diverging’eukary-otes:theexcavatetaxonisacloserelativeofMol.Biol.Evol.,1782–1791.Stechmann,A.&Cavalier-Smith,T.2002Rootingtheeukary-otetreebyusingaderivedgenefusion.,89–91.Stechmann,A.&Cavalier-Smith,T.2003Therootoftheeukaryotetreepinpointed.Curr.Biol.,R665–R666.Stechmann,A.&Cavalier-Smith,T.2003analysisofeukaryotesusingheat-shockproteinHsp90.Mol.Evol.,408–419.Stoeck,T.&Epstein,S.2003Noveleukaryoticlineagesinferredfromsmall-subunitrRNAanalysesofoxygen-depletedmarineenvironments.Appl.Environ.Microbiol.Stoeck,T.,Taylor,G.T.&Epstein,S.S.2003Noveleukary-otesfromthepermanentlyanoxicCariacoBasin(CaribbeanAppl.Environ.Microbiol.,5656–5663.Tovar,J.,Leon-Avila,G.,Sanchez,L.B.,Sutak,R.,Tachezy,J.,VanDerGiezen,M.,Hernandez,M.,Muller,M.&Luc-ocq,J.M.2003Mitochondrialremnantorganellesoffunctioniniron–sulphurproteinmaturation.VandePeer,Y.,Ali,A.B.&Meyer,A.2000Microsporidia:accumulatingmolecularevidencethatagroupofamitochon-driateandsuspectedlyprimitiveeukaryotesarejustcurious,1–8.Vossbrinck,C.R.,Maddox,J.V.,Friedman,S.,Debrunner-Vossbrinck,B.A.&Woese,C.R.1987RibosomalRNAsequencesuggestsmicrosporidiaareextremelyancienteuka-,411–414.Williams,B.A.,Hirt,R.P.,Lucocq,J.M.&Embley,T.M.2002AmitochondrialremnantinthemicrosporidianTrachipleistophorahominis,865–869.Zhang,H.,Sekiguchi,Y.,Hanada,S.,Hugenholtz,P.,Kim,H.,Kamagata,Y.&Nakamura,K.2003Gemmatimonasaur-gen.nov.,sp.nov.,aGram-negative,aerobic,poly-phosphate-accumulatingmicro-organism,thefirstculturedrepresentativeofthenewbacterialphylumGemmatimonad-etesphyl.nov.Int.J.Syst.Evol.Microbiol.,1155–1163.Asthispaperexceedsthemaximumlengthnormallypermitted,theauthorhasagreedtocontributetoproductioncosts.