Angiosperm Phylogeny Inferred from 18S Ribosomal DNA Sequences

Volume84 Number1 1997

Annals of the Missouri Botanical Garden

ANGIOSPERM PHYLOGENY INFERRED FROM 18S RIBOSOMAL DNA

Douglas E. Soltis,2Pamela S. Soltis,2

SEQUENCES'

Daniel L. Nickrent,3Leigh A. Johnson,2 William J. Hahn, Sara B. Hoot,5 JenniferA. Sweere,4RobertK. Kuzoff,2 Kathleen A. Kron,,6Mark W. Chase,7 Susan M. Swensen,8Elizabeth A. Zimmer,4Shu-Miaw Chaw,9LynnJ. Gillespie,'0 W. JohnKress," and KennethJ. Sytsma12

ABSTRACT Parsimony analyseswereconductedfor223 species representing all majorgroupsofangiosperms usingentire18S ribosomalDNA (rDNA) sequences. Althoughno search swappedto completion, the topologiesrecoveredare highly concordant withthoseretrieved via broadanalysesbased on thechloroplast of18S generbcL.The generalcongruence rDNAand rbcLtopologiesfurther In all analyses,thefirst-branching clarifiesthebroadpictureofangiosperm phylogeny. are Amborellaceae,Austrobaileyaceae, These taxa angiosperms Illiciaceae,and Schisandraceae,all woodymagnoliids. are alwaysfollowedby the paleoherbfamilyNymphaeaceae.This same generalorderof early-branching taxa is preIn mostsearches,the remaining servedwithseveralsuitesofoutgroups. taxa represent early-branching Piperalesand otherordersof subclass Magnoliidae(sensu Cronquist).Withthe exceptionofAcorus,the monocotsare supportedas In mostanalyses,taxawithuniaperturate monophyletic and typicallyhave as theirsisterCeratophyllum. pollenforma gradeat the base ofthe angiosperms; a largeeudicot lade is composedprimarily oftaxa havingtriaperturate pollen. Two large subclades are presentwithinthe eudicots,one consistinglargelyof Rosidae and a second corresponding These data sets of closelyto Asteridaesensu lato. Subclasses Dilleniidaeand Hamamelidaeare highlypolyphyletic. 18S rDNA sequences also permitan analysisof the patternsof molecularevolutionof thisgene. Problemsderiving fromboththe prevalenceofindelsand uncertainalignment of 18S rDNA sequences have been overstated in previous studies.Withthe exceptionof a fewwell-defined regions,insertionsand deletionsare relativelyuncommonin 18S rDNA;sequencesare therefore easilyalignedbyeye acrosstheangiosperms. Indeed,severalindelsin highlyconserved informative. Initialanalysessuggestthatbothstemand loop bases are important regionsappearto be phylogenetically sources of phylogenetic Of the stem information, althoughstempositionsare proneto compensatory substitutions. changesanalyzed,only27% destroya base-pairing couplet;73% maintainor restorebase pairing. 'This researchwas supportedin partby grantsfromthe NationalScience Foundation(DEB 9307000 to DES, DEB 9407984 to DLN, DEB 9303266 to WJH,DEB 9407350 to KAK, DEB 9306913 to SMS and LorenH. Rieseberg),'the NationalScience Council,RepublicofChina(2818F) to SMC, theMellonFoundation grant(toEAZ, PSS, and DES), the Trust(Department of Botany,Washington StateUniversity, to LAJand RKK), and the Scholarly BettyW. Higinbotham StudiesProgram oftheSmithsonian Institution foraccess to PAUP* 4.0. and (to WJKand EAZ). We thankD. Swofford S. Farrisforconducting a parsimony D. Olmstead, jackknifeanalysis.Wealso thankB. Alverson, J.Doyle,M. Hershkovitz, J.Palmer,Y.-L. Qiu,J.Rodman,and Q.-Y. Xiangforproviding plantmaterial and DNAs used in thisstudy.Weappreciate thevaluableadviceofD. Swofford on analyzing in usingtheUNIX version largedata setsand thehelpofM. Hershkovitz ofPAUP*. We also thankD. Olmsteadand M. Sandersonforhelpfulcomments on themanuscript. ANN. MISSOURI BOT. GARD.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

84: 1-49. 1997.

2

Annals of the MissouriBotanical Garden

Althoughtheangiosperms are almostuniversally presentdifferent viewsofrelationships amongthese consideredto be monophyletic, manybasic ques- groups. tionsofangiosperm phylogeny remainunanswered, In the largestphylogeneticanalysis of angioincluding:(1) whatare the first-branching angio- sperms,Chase et al. (1993) presentedthe results sperms?(2) whatis the ancestorof the monocots? oftwoparsimony analysesofDNA sequencesfrom (3) whatare the majorgroupsof angiosperms and the chloroplastgene rbcLfor475 and 499 species the relationships amongthesegroups?Despite in- of seed plants. More recently,Rice et al. (1997) tensivestudy,thesequestionshave been difficult to have reanalyzedthe 499-taxonrbcLdata matrixto answerfora varietyof reasons.Mostnotable,per- searchforshorter trees.The benefits to thesystemhaps, is the inadequacyof the fossilrecordalone atics community of performing these large phyloto answerthesequestionsconclusively. In addition, geneticanalysesofseed plantsin general,and antheapparentrapidradiationoftheangiosperms fol- giospermsin particular,have been considerable. lowingtheiroriginresultedin few morphological These studiesprovidecomprehensive, explicitphysynapomorphies amonglineagesat the base ofthe logenetichypothesesof higher-level relationships angiospermtree,hinderingattemptsto resolvere- in theangiosperms. Furthermore, theneed forsimlationshipsamong major groups (Crane et al., ilar studies of angiospermsbased on otherchar1995). Finally,the angiospermspresentrelatively acter sets has been recognized,and such studies few morphologicalcharactersfor comparisonat have been encouraged(e.g., Chase et al., 1993). higherlevels. For example,recentcladisticanaly- Particularly is the comparisonof chloimportant ses of morphologicalcharactersfor angiosperms roplast-basedphylogenetic estimates(Chase et al., (Donoghue& Doyle, 1989a, b) and all seed plants 1993) withtopologiesderivedfromanalysesofnu(Doyle et al., 1994) includedonly54 and 82 charclear genes. As recentlydemonstrated acters,respectively. by For reasons reviewedelsewhere,phylogenetic Doyle et al. (1994), carefulanalysisof bothmoranalyses based on nuclear DNA have largelyinphologicaland moleculardata is requiredto unvolvedportionsof the rDNA cistron(e.g., Mindell derstandangiosperm phylogeny. Duringthe past decade, several attemptshave & Honeycut,1990; Hillis & Dixon, 1991; Hamby & Zimmer,1992; Sanderson& Doyle,1993a; Nickbeen made to reconstruct the phylogeny oftheanrent & Soltis, 1995). Analysesof,18S rDNA and giosperms.Morphologicaland molecularanalyses rRNA sequences have been used forphylogenetic usually identifythe Gnetalesas the extantsister in animals groupto the angiosperms,in eitherthe shortest inferenceat highertaxonomic"'levels trees or those slightlylonger(e.g., Crane, 1985, (e.g., Sogin et al., 1986; Field et al., 1988; Wain1988; Donoghue& Doyle,1989a, b; Doyle & Don- rightet al., 1993; Wada & Satoh,1994), protozoa oghue, 1986, 1992; Loconte & Stevenson,1991; (Schlegel et al., 1991), algae (Buchheimet al., Hamby& Zimmer,1992; Chase et al., 1993; Doyle 1990; Huss & Sogin, 1990; Kantz et al., 1990; et al., 1994; Nixonet al., 1994; butsee Goremykin Hendrickset al., 1991; Chapman & Buchheim, et al., 1996; Chaw et al., 1997). Molecularphylo- 1991; Bakkeret al., 1994; Ragan et al., 1994; Olgeneticanalysesinclude those based on rbcL se- sen et al., 1994), fungi(Forsteret al., 1990; Swann quences (Chase et al., 1993), partial18S and 26S & Taylor,1993; Hinkleet al., 1994), lichens(Gar(Mishleret al., 1994; ribosomalRNA sequences (Hamby & Zimmer, gas et al., 1995), bryophytes 1992), and rbcS aminoacid sequences (Martin& Capesius, 1995; Kranzet al., 1995), gymnosperms Dowd,1991). These analysestendto identify many (e.g., Chaw et al., 1993, 1995, 1997), and even of the same majorgroupsof taxa, but theyoften amongthedeepestbranchesoflife(Wolters& Erd2 Department of Botany,Washington StateUniversity, Pullman,Washington 99164, U.S.A.

3Department of PlantBiology,SouthernIllinoisUniversity, Carbondale,Illinois62901, U.S.A. 4Laboratory of MolecularSystematics, Smithsonian Institution, Washington, D.C. 20560, U.S.A.

Department of BiologicalSciences,University ofWisconsin,Milwaukee,Wisconsin53201, U.S.A. Department of Biology,WakeForestUniversity, Winston-Salem, NorthCarolina27109, U.S.A. 7Laboratory ofMolecularSystematics, RoyalBotanicGardens,Kew,Richmond,SurreyTW9 3AB, UnitedKingdom. 8 Department of Biology,Ithaca College,Ithaca,New York14850, U.S.A. 9 Institute ofBotany,AcademiaSinica, Nankang,Taipei,Taiwan,Republicof China. 10 ResearchDivision,CanadianMuseumof Nature,Ottawa,OntarioKIP 64P, Canada. "1Department of Botany,NationalMuseumof NaturalHistory,SmithsonianInstitution, Washington, D.C. 20560, U.S.A. 12 Department of Botany,University ofWisconsin,Madison,Wisconsin53706, U.S.A. '5

6

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

3

mann,1986; Olsen, 1987; Woese,1987; Embleyet accordwithexistingclassifications, butsamplingof al., 1994; Bhattacharya & Medlin,1995). nonmagnoliidtaxa was sparse and may explain Despitethiswide usage in othermajorgroupsof someoftheunusualrelationships suggestedamong organisms, the 18S rRNA gene has receivedcom- morederivedangiosperms.Furthermore, manyof parativelylittleattentionin angiosperms.In large thenodeswerepoorlysupported.As a resultofthe partthisreflects theenormousinterest in,and dem- unusualrelationships suggestedforsome taxa and onstratedutilityof, rbcL sequences forinferring the poor resolutionobtainedin this study,angiophylogeny,particularlyat the familylevel and spermsystematists remainedunsureof the utility above (e.g., Chase et al., 1993; Morgan& Soltis, of 18S and 26S rRNA (and rDNA) sequences for 1993; Olmsteadet al., 1993; Kron& Chase, 1993; inferring phylogeny. Qiu et al., 1993; Rodmanet al., 1993). In addition, More recently, Nickrentand Soltis (1995) comskepticismapparentlyexists amongmanyangio- pared the rateof evolutionand phylogenetic resosperm systematists regardingthe utilityof 18S lutionofentire18S rDNAsequenceswiththosefor rDNAsequencesforinferring plantphylogeny. Ear- the chloroplastgene rbcL using a taxonomically ly analyses of 18S rDNA or rRNA sequences in similarsuite of 59 angiosperms.Pairwisecomparangiosperms(e.g., Nickrent& Franchina,1990; isons showed that rbcL is generallyabout three Hamby& Zimmer,1992; Nickrent& Starr,1994), timesmorevariablethan18S rDNA. However,bewhilein generalpointingto the possibilephyloge- cause of the longerlengthof 18S rDNA,the ratio neticutilityofthesedata,raisedconcernsthat18S ofthe numberofphylogenetically informative sites rDNA maybe tooevolutionarily conservative toad- per molecule is only about 1.4 times greaterfor dressphylogenetic questionsat thefamilyleveland rbcL than for 18S rDNA. Exploratory parsimony above and thatinsertions and deletionevents(in- analysesofangiosperms showedthatseveralclades dels) occurfrequently in at least some portionsof were stronglysupportedby both rbcL and 18S 18S rDNA, makingalignmentof sequences diffi- rDNA data sets. Nickrentand Soltis (1995) concult. In addition,otherbasic backgroundinforma- cludedthatcomplete18S rDNAsequencesare suftion regardingthe molecularevolutionof the 18S ficiently variableto conductphylogenetic studiesat rRNA gene is not available forangiosperms.For higherlevels withinthe angiosperms. example,giventhat18S rRNA,as well as rRNAs Here we explorefurther the higher-level phyloin general,have inherentsecondarystructure that geneticrelationships withinthe angiosperms using includescharacteristic loop (non-paired)and stem entirenuclear18S rDNA sequences. Morespecif(paired) stretchesof RNA, should changesin the ically,we providephylogenetic hypotheses forflowencodingstemand loop bases be consideredequal- eringplantsbased on analysesof four18S rDNA ly informative in phylogenetic analyses?Modelsof data sets,differing-in boththe numberoftaxa and rRNA evolutionsuggestthatpaired (stem)bases theinclusionofindelsas additionalcharacters.We willundergocompensatory changesto maintainthe also comparethe phylogenetic estimatesbased on base pairing.However,empiricalstud- 18S rDNAsequenceswiththoseobtainedfromphyappropriate ies ofangiosperm rRNAstructure are few(e.g.,Se- logeneticanalysisofrbcLsequences (Chase et al., necoff& Meagher,1992), and available data sets 1993). Using the phylogeneticestimates,we exhave notbeen used to evaluatepatterns ofevolution aminepatternsofmolecularevolutionof18S rDNA ofthe 18S rRNAgene in angiosperms. by assessingthe frequencyof insertionsand deleThe historyof the use of 18S rRNA and rDNA tions,theprevalenceofcompensatory changes,and sequences forphylogenyreconstruction in angio- therelativephylogenetic ofstemversus importance spermswas recentlyreviewed(Nickrent& Soltis, loop changesin angiosperm18S rDNA. 1995). To date,thelargeststudiesof18S sequences are thoseof Hambyand Zimmer(1992) and NickMATERIALS AND METHODS rentand Soltis (1995). Zimmerand collaborators conductedphylogenetic analyses using directse- SPECIES SAMPLED AND SOURCES OF PLANTMATERIAL quencingof rRNAfromapproximately 60 species ofvascularplants,ofwhich29 weredicotsand 17 The species includedin this analysisare given monocots(Zimmeret al., 1989; Hamby& Zimmer, in Table 1, alongwithfamilymembership, general 1992). These investigators sequenced portionsof collection information, and GenBank accession both 18S and 26S rRNA,yieldinga totalof 1701 numbersforthe 18S sequences. In Table 1, and base positionsper taxon,1097 base positionsfrom throughout the text,we generallyfollowthe taxothe 18S gene and 604 fromthe 26S gene. The nomiccircumscriptions of Cronquist(1981) fordishortest treesobtainedhad a numberoffeaturesin cots and Dahlgrenet al. (1985) formonocots.This

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Table 1. Species analyzedfor18S rDNA sequence variation.Species are arrangedalphabeticallyby families(dicotsaccord Withinfamilies,species are arrangedalphabetically by genus.t indicatesthoseangiosperms presentin the228-taxondata sets, Species

Family

Voucher/source

Pachystachys lutea Nees Acerrubrum L. Actinidiasp. TetragoniaexpansaMurr. L. Sagittariatrifoliata Alliumthunbergii G. Don Eucharisgrandiflora Planch & Linden Hippeastrum sp. Amborellatrichopoda Baill. Isolona sp. Mkiluafragrans Chlorophytum nepalenseBaker Lomatiumtriternatum (Pursh) Coult.& Rose Tabernaemontana divaricata (L.) R. Br. Acoruscalamus L.

Acanthaceae Aceraceae Actinidiaceae Aizoaceae Alismataceae Alliaceae Amaryllidaceae

Johnson95-003, WS Soltis& Soltis2515, WS Morgans.n., WS Hershkovitz 111, WS Chaw 1371, HAST NA 55049, US Hahn 6868, WIS

Amaryllidaceae Amborellaceae Annonaceae Annonaceae Antheriaceae Apiaceae

Hahn 6875, WIS Suh 44, US Chase 542, K Schatz3364, WIS Kress92-3434, US Soltis2266, WS

Apocynaceae

Nickrent 2978, SIU

Araceae

Nickrent 2941, SIU

Calla palustrisL. VeitchiasessilifoliaBurret HederahelixL. tomentosa Aristolochia Sims.

Araceae Arecaceae Araliaceae Aristolochiaceae

Hahn 6959, WIS Hahn, US Plunkett1368, WS Nickrent 2922, SIU

AsarumcanadenseL.

Aristolochiaceae 1

Nickrent 2888, SIU

t SarumahenryiOliver

Aristolochiaceae

Qiu 91018, NEU

Tagetessp. Tragopogon dubiusScop. Austrobaileya scandensC. T. White

Asteraceae Asteraceae Austrobaileyaceae

Nickrent 3061, SIU Soltis2472, WS Nickrent 2953, SIU

ImpatienswalleranaHook. Batis maritimaL.

Balsaminaceae Bataceae

Johnson95-071, WS Iltis30500, WIS

t t

t

t

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Literature citation

Nickrent& Soltis, 1995

Nickrent& Soltis, 1995 Nickrent& Starr, 1994 Nickrent& Soltis, 1995

Table 1. Continued. Species

Family

Bauera rubioidesAndrews Begonia metallicaX sanguinea Symbegoniasp. thalictroides Caulophyllum (L.) Michx. t Podophyllum peltatumL. AlnusglutinosaL. Gaertn.

Voucher/source

Baueraceae Begoniaceae Begoniaceae Berberidaceae

Kew 1977-6377, K Chase 225, NCU Kew U012, K Hoot 925, UWM

Berberidaceae Betulaceae

Nickrent 2891, SIU unknown

Parmentiera ceriferaSeem. Bombaxceiba Burm. BourreriasucculentaJacq. thaliana (L.) Heyn. Arabidopsis BrassicahirtaMoench

Bignoniaceae Bombacaceae Boraginaceae Brassicaceae Brassicaceae

Johnson95-005, WS Alverson s.n., WIS J. Miller6421, MO unknown unknown

Glomeropitcairnia penduliflora (Grisebach)Mez Berzelialanuginosa(L.) Brongn. BuddlejadavidiiFranch. L. Buxussempervirens Byblisgigantea floridusL. Calycanthus CampanularamulosaWall. LobeliaerinsL. Canna coccineaMill. CleomehaslerianaChodat Koeberlinia spinosaZucc. t Loniceramaackii(Rupr.)Max-

Bromeliaceae

Kress92-3466, US

Bruniaceae Buddlejaceae Buxaceae Byblidaceae Calycanthaceae Campanulaceae Campanulaceae Cannaceae Capparaceae Capparaceae Caprifoliaceae

Prices.n., IND Johnson95-031, WS Hoot 921, UWM Palmengarten B. G. Nickrent 2893, SIU Jansen984, MICH Jansen989, MICH Chaw 1371, HAST Al-Shehbaz,s.n., MO Al-Shehbaz, s.n., MO Nickrent 3060, SIU

Caprifoliaceae Caricaceae Casuarinaceae Celastraceae

Olmsteads.n., COLO MissouriB. G., MO Nickrent 2977, SIU Nickrent 2894, SIU

Cephalotaceae Ceratophyllaceae

Chase 147, NCU Qiu 91027, NCU

im.

albus (L.) Blake Symphoricarpos Caricapapaya L. L. Casuarinaequisetifolia Euonymusalatus (Thunb.)Siebold Labill. Cephalotusfolicularis L. demersum Ceratophyllum

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Literature citation

Savard& Lalonde, 1991

Unfriedet al., 1989 Rathgeber& Capesius, 1990

Nickrent& Franchina, 1990

Table 1. Continued. Species

Family

Voucher/source

Cercidiphyllum japonicumSiebold & Zucc. Spinacia oleraceaL.

Cercidiphyllaceae

Soltis2540, WS

Chenopodiaceae

Nickrent 2896, SIU

arborescens Sw. Hedyosmum icaco L. Chrysobalanus Licania tomentosa Fritsch ClethraalnifoliaL. CoichicumautumnaleL. Elasis sp. IpomoeahederaceaJacq. Aucubajaponica Thunb. Raoul Corokiacotoneaster Heiwingialaponica-(Thunb.)F. Dietr. CostusbarbatusSuess. CrassulamarnieranaHuber& Jacobsen Dudleyaviscida(S. Watson) Moran HaKalanch~be diagremontana met& Perrier Sedumrubrotintum Clausen Crossosoma Nutt. californicum Abobratenuiflora Naudin. gummiferum Ceratopetalum Small CuscutagronoviWilld.

Chloranthaceae Chrysobalanaceae Chrysobalanaceae Clethraceae Colchicaceae Commelinaceae Convolvulaceae Cornaceae Cornaceae Cornaceae

Nickrent 3022, SIU FairchildTrop.G 76-311 FairchildTrop.B. G. 64-734 Kron1884s, NCU Hahn 6864, WIS Evans s~n.,WIS Coiwells.n., MO U.S. Natl.Arb. Arb. 74211 Strybing ArnoldArb.912

Costaceae Crassulaceae

Kress94-3710, US, Morgan2152, WS

Crassulaceae

B. C. 62801 Huntington

Crassulaceae

Morgan2151, WS

Crassulaceae Crossosomataceae Cucurbitaceae Cunoniaceae

Morgan2153, WS RanchoSanta Ana B. C. Chase 915, K Keller2135, CAS

Cuscutaceae

Nickrent 3015, SIU

Cyperaceae Cyrillaceae Daphniphyllaceae Datiscaceae

Kress92-3463, US Krons~n.,NCU Qiu 91026, NCU RanchoSanta Ana B. C.

Schrader Cyperusalbostriatus L. tCyrillaracemifolia Daphniphyllum sp. Datisca glomerata(Presl)Baill.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Literature citation

Nickrent& Soltis,

1995

Nickrent& Soltis,

1995

Kron,1996

Table 1. Continued.

t t

t

t

Species

Family

Tetrameles nudifloraR. Br.

Datiscaceae

Diapensia lapponicaL. Galax urceolataL. Dillenia alata (DC) Mart. Dipsacussp. Droseracapensis Diospyros virginianaL. Elaeagnus umbellataThunb. Sloanea cf. tatifolia K. Sch. Ephedrasinica Stapf S. Wats. Ephedratorreyana Arctostaphylos uva-ursi(L.) Spreng. Ait. Vaccinium macrocarpon EucommiaalmoidesOliv. Eucryphialucida Druce Drypetes roxburghii(Wall.) Hurus. Euphorbiapulcherima EupteleapolyandraSiebold & Zucc. AlbiziajulibrissinDurazz. Bauhinia sp. Glycinemax. (L.) Merrill

Diapensiaceae Diapensiaceae Dilleniaceae Dipsacaceae Droseraceae Ebenaceae Elaeagnaceae Elaeocarpaceae Ephedraceae Ephedraceae Ericaceae

CentralB. G., Sara Bori, Thailand;Philbrick& Wongpraseri 2272 Hills 89018, NCU Kron163, NCU Nickrent 2956, SIU Jansen931, MICH Palmengarten B. G. Kron3004, NCU Nickrent 2898, SIU Alverson s.n., WIS TI-9297, TI Gillespie4236, US Johnson94-085, WS

Ericaceae Eucommiaceae Eucryphiaceae Euphorbiaceae

Kron2937TNCU Qiu 91024, NCU Strybing Arb.86-0250 FTG-83463A,K

Euphorbiaceae Eupteleaceae

Soltis& Soltis2541, WS Qiu 9001, NCU

Fabaceae Fabaceae Fabaceae

Doyle 1526, BH Doyles.n., MSU unknown

PisumsativumL.

Fabaceae

CarolinaBiologicalLaboraN.C. tory,Burlington Alverson 2172, WIS Reznicek9756, MICH MissouriB. G. 860162, MO RanchoSanta Ana B. G. 13280 Prices.n., IND

* Muntingiacalabura Griseb.

DicentraeximaTorrey FouquieriasplendensEngelm. GarryaellipticaDouglas ex Lind. GeraniumcinereumCav.

Flacourtiaceae Fumariaceae Fouquieriaceae Garryaceae Geraniaceae

Voucher/source

Literature citation

Eckenrodeet al.,

1985

*A recentdetailedphylogenetic relat analysisof FlacourtiaceaeusingrbcLsequences indicatesthatMuntingiais onlydistantly

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Table 1. Continued. Species

Family

Aeschynanthus radicansJack. Gnetum gnemonL. Gnetumnodiflorum Brongn. Gnetumurens(Aubl.) Blume Greyiaradlkoferi Szyszyl. Brexiamadagascarensis Thouarsex Ger Gawl. Escallonia coquimbensis Reamy Itea virginicaL. Phyllonomalaticuspus(Turcz.) Engl. Pterostemon Ramirotundifolius rez RibesaureumPursh RousseasimplexJ. E. Smith tasmanicaHook E Tetracarpaea GunneramanicataLinden Haloragiserecta(Banks ex Murr.)Eichler Altingiasp. L. Liquidambarstyraciflua Heliconiaindica Lamark Bowiea volubilisHarveyex. Hook f. Torr. Hydrangeamacrophylla PhiladelphuslewisiiPursh. Phacelia bicolorTorr.ex Wats. Illiciumparviflorum Michx.ex Vent. Gladiolusbuckevildii (L. Bol.) Goldblatt t Isophysistasmanica(Hook.) T. Moore Phil. Lactorisfernandeziana

Voucher/source

Gesneriaceae Gnetaceae Gnetaceae Gnetaceae Greyiaceae Grossulariaceae

Nickrent 2979, SIU Gillespie4212, US Gillespie4246, US Kresset al. 91-3271, US Arb.640406 Strybing Kew 1977-14901,K

Grossulariaceae Grossulariaceae Grossulariaceae

U. Calif.B. G. 52-1333 Ware9401, WS Morgan2124, WS

Grossulariaceae

Sanchez259, TEX

Grossulariaceae Grossulariaceae Grossulariaceae Gunneraceae Haloragaceae

Soltis& Soltis2220, WS Herbarium,MauritiusSugar Ind. Res. Inst. Jordans.n., HO Kruckeberg s.n., WTU Chase 453, K

Hamamelidaceae Hamamelidaceae Heliconiaceae Hyacinthaceae

RBG, Edinburgh, Qiu 93006 Soltis& Soltis2516, WS Kress80-1118, US Hahn 6882, WIS

Hydrangeaceae Hydrangeaceae Hydrophyllaceae Illiciaceae

Morgan2150, WS Soltis& Soltis2411, WS Johnson92-005, WS Naczi 2784, MICH

Iridaceae Iridaceae

Goldblatt& Manaing 9504, MO J. Bruhls.n.,TAS

Lactoridaceae

Steussyet al. 11,784, OSU

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Literature citation

Table 1. Continued. Species

Family

Voucher/source

LamiumamplexicauleL. Akebiaquinata (Houtt.)Decne. Sassafrasalbidum(Nutt.)Nees Wallace t Liliumformosanum Floerkeaproserpinicoides Willd. LinumperenneL. MalpighiacoccigeraL. hirsutum L. Gossypium MarantabicolorVell. Menispermum canadensisL. t Tinospora caifraMiers Mollugoverticillata L. t MonotropaunifloraL.

Lamiaceae Lardizabalaceae Lauraceae Liliaceae Limnanthaceae Linaceae Malpighiaceae Malvaceae Marantaceae Menispermaceae Menispermaceae Molluginaceae Monotropaceae

Johnson95-001, WS Nickrent 2945, SIU Soltis& Soltis2518, WS H0962, HAST Reznicek8609, MICH Nickrent 2900, SIU Nickrent 2905, SIU Alverson s.n., WIS Kress94-3724, US Naczi 2837, MICH Jaarsveld2131, NBG Hershkovitz 37, WS Nickrent 3018, SIU

Morusalba L. MoringaoleiferaLam. Musa acuminataColla Nelumbolutea (Willd.)Pers. Nepenthes sp. NymphaeatuberosaPaine

Moraceae Moringaceae Musaceae Nelumbonaceae Nepenthaceae Nymphaeaceae

Nickrent 2924, SIU Ntis30501, WIS U.S. Bot. Gard.s.n., US Hoot 9212, 1713 Nickrent 3056, SIU Nickrent 2906, SIU

Mirabilisjalapa L. acuminataDecne. Camptotheca Olea europaeaL. t ClarkiaxantianaA. Gray Opilia amentaceaRoxb. OncidiumexcavatumLindl. Paeonia suffructicosa HypecoumimberbaSm. Helmholtzia sp. PhytolaccaamericanaL. Peperomeaserpens(Swartz) Loud.

Nyetaginaceae Nyssaceae Oleaceae Onagraceae Opiliaceae Orchidaceae Paeoniaceae Papaveraceae Philydraceae Phytolaccaceae Piperaceae

Hershkovitz 60, WS Arb.74-180 Strybing 95-004, WS Jkhnson Gottlieb7436, DAV Nickrent 2816, SIU Chase 83427, K Chase 486, K Chase 528, K Kress92-3505, US Hershkovitz 38, WS Nickrent 2907, SIU

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Literature citation

Nickrent& Soltis, 1995

Nickrent& Soltis, 1995

Nickrent& Soltis, 1995

Table 1. Continued. Species

Family

Voucher/source

Literature citation

Pittosporum japonicumHort.ex C. Presl. PlatanusoccidentalisL.

Pittosporaceae

Riesebergs.n., RSA

Platanaceae

Soltis& Soltis2514, WS

PlumbagoauriculataLam. Oryzasativa L.

Plumbaginaceae Poaceae

Nickrent s.n., SIU unknown

Zea maysL.

Poaceae

Acanthogiliagloriosa(Brandegee) A. G. Day & Moran Cobaea scandensCav. Gilia capitataSims.

Polemoniaceae

(GenomicDNA) Clontech Laboratories, Inc., Palo Alto,CA Johnson& Mort,95-070, WS

Polemoniaceae Polemoniaceae

Pattersons.n., WS Johnson92-15, WS

PolygalapaucifloraWilld. CocolobauviferaL.

Polygalaceae Polygonaceae

Doyle 1567, BH Nickrent 2927, SIU

Primulasp. KnightiaexcelsaR. Br.

Primulaceae Proteaceae Proteaceae

Johnson95-006, WS Univ.of California, Santa Cla- Nickrent& Soltis, 1995 ra, B. G. Douglas 110, MEL

Punicaceae Pyrolaceae

Nickrent 2931, SIU Colwell,s.n., MO

Ranunculaceae Ranunculaceae

Voss& Howard,MICH Nickrent 2932, SIU

Ranunculaceae

Qiu 91030, NCU

Rhamnaceae

Morgan2155, WS

coriaceumC. T. t Placospermum White& W. D. Francis t Punica granatumL. t Pyrolapicta Sm.

t Coptistrifolia(L.) Salisb.

RanunculussardousCrantz

MarXanthorhiza simplicissima shall CeanothussanguineusPursh

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Nickrent& Soltis, 1995

Takaiwaet al., 1984

Nickrent& Soltis, 1995

Nickrent& Soltis, 1995

Table 1. Continued. Family

Voucher/source

Rosaceae Rosaceae Rosaceae Rubiaceae Rutaceae Sabiaceae Santalaceae

Morgan2131, WS E. E. Dicksons.n., BH Morgan2130, WS Xiang s.n., OSU Nickrent 2977, SIU Wagner6518, HAST Nickrent 2731, SIU

Sapindaceae Sapotaceae Sargentodoxaceae

Nickrent 2915, SIU Chase 129, NCU Qiu s.n., PE

Sarraceniaceae Saururaceae

Morgans.n., WS Nickrent 2940, SIU

SaururuscernuusL.

Saururaceae

Suh 128, US

(Piper) Boykiniaintermedia Jones pectinataEndl. Eremosyne

Saxifragaceae

Grable11638, WS

Saxifragaceae

Annels& Hearn4795,

Francoa sonchifoliaCav.

Saxifragaceae

Soltis& Soltis2479, WS

HeucheramicranthaDouglas spathulatum Lepuropetalon (Muhl.)Elliott Montiniacaryophyllacea Thunb. Banks Parnassiafimbriata sedoidesL. Penthorum Hook. Saxifragaintegrifolia VahliacapensisThunb. Schisandrachinesis(Turcz.) Baill. LinariavulgarisP. Mill

Saxifragaceae Saxifragaceae

Soltis& Soltis1949, WS Thomass.n., NLU

Saxifragaceae Saxifragaceae Saxifragaceae Saxifragaceae Saxifragaceae Schisandraceae

Williams2833, MO Soltis& Soltis s.n., WS Hayden2232, WS Soltis& Soltis2253, WS Van Wyk10-579, PRU Rezniceks.n., MICH

Scrophulariaceae

ColwellMO CAI, SIU

Species Dress Photiniafraseri Prunuspersica(L.) Batsch Spiraea vanhoutei(Briot)Zubel MitchellarepensL. t CitrusaurantiumL. Sabia swinhoei OsyrislanceolataHochst.& Steud. t Koelreuteria paniculataLaxm. Manilkarazapota (L.) Royen t Sargentodoxacuneata(Oliv.) Rehdes & Wil. SarraceniapurpureaL. cordataThunb. Houttuynia

t

PERTH

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

citation Literature

Kron,1996

Nickrent& Soltis,

1995

Nickrent& Soltis, 1995

Table 1. Continued. Species

Family

Voucher/source

erianthusBenth. t Orthocarpus

Scrophulariaceae

ColwellCA SVDlO, SIU

t PedicularislanceolataMichx.

Scrophulariaceae

ColwellMO SFPl, SIU

Brunfelsia pauciflora(Cham.& Schlechtend.)Benth. t Sparganiumeurycarpum Engl.

Solanaceae

Johnson95-002, WS

Sparganiaceae

Nickrent 2943, SIU

t StyraxamericanaLam. paniculataMiq. Symplocos Tacca sp.

Styracaceae Symplocaceae Taccaceae

CyanellacapensisL. Tetracentron sinensisOliv. Cameliajaponica L. LueheaseemanniiCuatr. Trochodendron araijoides Siebold & Zucc. Tropaeolum majis L.

Tecophilaeaceae Tetracentraceae Theaceae Tiliaceae Trochodendraceae

Kron3002, NCU Kron3005, NCU MissouriBot. Garden894904, US Hahn 6966, WIS Qiu 90009, NCU Nickrent 2929, SIU Alverson 2184, WIS Qiu 90026, NCU

Tropaeolaceae

Chase 113, NCU

Turneraulmifolia Celtisyunnanensis C. K. Schneid. Zelkovaserrata Pilea cadiereiGagnep.& Guillam. ViscumalbumL. Drimysaromatica Drimyswinteri J. R. & G. Forster Xanthorrhoea sp. Zamia pumila L. Noronha Zingibergramineum GualacumsanctumL.

Turneraceae Ulmaceae

Chase 220, NCU Qiu P90002, NCU

Ulmaceae Urticaceae

Soltis& Soltis2517, WS Nickrent 2972, SIU

Viscaceae Winteraceae Winteraceae

Nickrent 2253, SIU Suh 9, US Suh 47, US

Xanthorrhoeaceae Zamiaceae Zingiberaceae Zygophyllaceae

Kress92-3422, US unknown Kress91-3266, US Anderson s.n., MICH

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Literature citation

Nickrent& Soltis, 1995 Kron,1996 Kron,1996

Nickrent& Soltis, 1995

Nairn& Ferl, 1988

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

13

approachparallelsthatin Chase et al. (1993) and highlyconservednatureofthe 18S rRNAgene,but shouldfacilitatecomparison. also to the factthatmostlengthmutations involve As withthe broad analysesof rbcL sequences, insertionsor deletions of a single base pair. close examinationof the genera included in this Straightforward ofsequenceswas further alignment studywillrevealan uneventaxonomic distribution. facilitated bythefactthatmostindelsin 18S rDNA Some groupsare relativelywell represented(e.g., are confinedto a fewspecificregionsthatare parSaxifragaceaesensu stricto[Saxifragaceaes. str.] ticularlyprone to variationin primarysequence and allies, ranunculids,Asteridaesensu lato [As- and length,such as the terminiof helices E10-1, teridaes.l.]), whereasothersare notas thoroughly 17, E23-1, and 43 (see also Nickrent& Soltis, sampled (e.g., portionsof Dilleniidae and Magno- 1995). Because theywere difficult to align overa liidae). However,ourselectionoftaxa was notran- broad taxonomicscale, no attemptwas made to dom.We attempted to includesamplesfromall ma- align foursmall regionsof 18S rDNA over all of jor angiosperm orders and subclasses sensu the taxa analyzed: positions230-237; 496-501; Cronquist(1981). Furthermore, in selectinggenera 666-672; 1363-1369 (see Appendix).These base forsequencing,we triedto sample representatives positionscorrespondto the sequence of Glycine fromeach of the major clades recoveredin the max (Eckenrodeet al., 1985), which providesa analysesofChase et al. (1993) (e.g.,rosidI, asterid convenientreferencesequence because of the I, asteridIII, etc.),as well as fromthevarioussub- availabilityof a proposedribosomalRNA secondclades presentwithinthosemajorclades. We also ary structuremodel (Nickrent& Soltis, 1995). used, whenavailable,the same DNAs used previ- These fourregionsof ambiguousalignmentwere ously for the sequencingof rbcL (Chase et al., subsequentlyeliminatedfrom the phylogenetic 1993). If a givenDNA was no longeravailable,we analyses,followingSwofford and Olsen (1990). In attemptedto obtainleaf materialof the same spe- addition,the extreme5' and 3' ends of the secies, and if thatfailed,froma congenericspecies. quences were not includedin the analyses.PosiAnotherfactorthatinfluenced ourchoiceoftaxa tions1-20 wereexcludedbecause theycorrespond was sequence quality.As discussedin detailbelow, to the forwardPCR primer.Because mostof the one outcomeof this studywas the discoverythat sequences wereclearlyreadableat, or just before, many available sequences are erroneous,some base position41, we begananalysisofourdata set highlyso. We therefore to eliminateany at position42. At the3' end ofthe 18S sequences, attempted dubioussequences fromour data sets. In addition, base positions1751-1808 (on Glycine)wereoften some available sequences were not included be- difficult to read and also wereeliminatedfromthe cause theywereincompleteor containednumerous analysis.Some sequences are incompleteat the 3' ambiguitiesor extensivegaps. end and are approximately 1700 base pairs in Several laboratorieswere involvedin this proj- length.The totallengthof the aligned 18S rDNA ect; hence,severaldifferent protocolswereused to sequences was 1850 base pairs. generatethe sequences analyzed. Althoughboth Twoindels,each ofa singlebase pair,weredeautomatedand manualsequencingstrategieswere tectedin highlyconservedregionsofthe18S rRNA employed,70% of the sequences analyzedwere gene notproneto insertion-deletion (Table 2). One generatedvia automatedsequencing.The general indel (A), an apparentdeletionbased on outgroup methodsused for PCR amplification and subse- comparison, is presentin all higherdicots(i.e., the quentmanualsequencingof 18S rDNA are provid- large lade consistingof Rosidae and Asteridae ed in Nickrent(1994), Nickrentand Starr(1994), s.l.). A secondindel (B), an apparentinsertion, ocand Bultet al. (1992). Generalmethodsfortheau- curs in all membersof the saxifragoidlade (also tomatedsequencingapproachfor 18S rDNA are referredto as Saxifragales;D. Soltis & Soltis, givenin D. Soltisand Soltis(1997). The base com- 1997). These twoindels wereincludedas characpositionof the oligonucleotideprimersused for ters in two of the phylogeneticanalyses,as dePCR and sequencingare providedin Nickrentand scribedbelow. Starr(1994) and Bult et al. (1992). PHYLOGENETICANALYSIS ALIGNMENTOF THE 18S rDNASEQUENCES

Withthe exceptionof a few,small,well-defined regions,alignmentof 18S rDNA sequences was easily accomplishedby eye across all taxa. This generalease of alignmentis due not only to the

We constructedfourdata sets forphylogenetic analysis:(1) a data set of223 angiosperms plus five membersofGnetalesas outgroups, without thetwo indels notedabove; (2) taxonsamplingas in (1), butwithindelsA and B (see also Table 2) included;

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

14

Annals of the MissouriBotanical Garden

Table 2. Potentiallyphylogenetically informative indels locatedin conservedregionsof 18S rNDA. Indel A is one-bpdeletionthatcharacterizesall highereudicots (i.e., the Rosidae and Asteridaes.l. clades). Indel B is a one-bpinsertion thatunitesall membersofthesaxifragoid lade. Base positionscorrespond to thelastpositiongiven in the sequence of Glycinemax.

giosperms),bothwithand withoutindels,we used as outgroups: (1) thefiveGnetalesand Zamia pumila; (2) thefiveGnetales,Zamia pumila and Cycas revoluta.Similarly, forthelargedata sets (223 angiosperms)we used as outgroups: (1) thefiveGnetales and Zamia pumila; (2) thefiveGnetalesand a recentlyacquiredsequenceofWelwitschia mirab7is. Indel A Because ofthelargenumberoftaxainvolved,we 1529 used twobasic searchstrategies.The firstmethod Glycine CCGGGTAATCYJTTG was a heuristicsearchperformed usingPAUP* 4.0 Trochodendron CCGGGTAATCYJTTGA (Swofford, pers. and to comm.) a lesser extent IndelB PAUP 3.1.1 (Swofford,1993) with MULPARS, 1406 RANDOM taxonaddition,and TBR branch-swapGlycine TATGGCCGCYFA -GGC ping. These searcheswere permitted to run fora Heuchera TATGGCGATITAAGGC weekor moreusingeithera MacintoshQuadra650 or Sun Sparc Server600P. These searchesdid not producetrees as shortas those producedby the (3) a data set of 194 angiosperms, plus fiveGne- methodbelowand will notbe discussedfurther. tales as outgroups,withoutthe two indels; (4) taxon The primarysearch strategywas inspiredby samplingas in (3) above,withindelsA and B in- Maddisonet al. (1992) and suggestedby D. Swofcluded. For data sets 2 and 4, the indelsA and B ford(pers.comm.).For each of the fourdata sets, were added to the data matrix,and the position we used 50-100 consecutive searches without scoredas eitherpresent(1) or absent(0). MULPARS using RANDOM taxon additionand Fourdatasetswereused forseveralreasons.First, NNI branch-swapping. We thenperformed multiple the approachused permitted an assessmentof the searches(300-500 replicates;a Sun Sparc Server phylogenetic informativeness ofthetwoindels.Sec- 600P typicallyrequired19-25 hoursto complete ond, our goal in constructing the twosmallerdata five replicates)using RANDOM addition,MULsetswas to improvethephylogenetic analysisbyre- PARS, and TBR branchswapping,whereonlytwo movingincompleteand/orpossiblyerroneousse- trees (NCHUCK = 2) of a specified length quences and by reducingthe size of the matrixto (CHUCKLEN) or longerwere saved per replicate. make the problemmoretractable.The twosmaller The lengthoftheshortest treesobtainedin theNNI data sets (3 and 4) differ fromthe largerdata sets searcheswas used as theinitialCHUCKLEN value. (1 and 2) in theremovalofseveraltaxahavinglong As shortertrees were found,additionalsearches branchlengths(e.g., Dillenia,Acorus)and several were conductedwith lower CHUCKLEN values. taxaforwhichthesequenceswereincomplete (e.g., This approachpreventedthe searchesfrombeing severaloftheranunculids). In addition,representa- overwhelmed withtrees. tivesfromsomeofthelargerclades (e.g.,monocots, The finalportionofthissearchstrategy involved Asteridaes.l.) andfromsomefamiliesforwhichmul- use ofthe shortesttreesobtainedabove as starting tiple sequences were available (e.g., Annonaceae, pointsforsubsequentsearches,again withMULAristolochiaceae, Proteaceae)wereremovedto con- PARS and TBR branchswapping.These searches structdata sets 3 and 4. were permitted to run forweeks or monthsusing The outgroupswere fivemembersof Gnetales: MacintoshQuadra/Centris 650 or PowerMac6100 Ephedra sinica, E. torreyana, Gnetum nodiforum, or 7100 computers. Typicallyno morethan2000G. gnemon, and G. urens. Gnetales were the logical 5000 treesweresaved in anyofthesesearches.We choice of outgroupbecause theyappear as the ex- used starting treesofseveraldifferent lengthswhen tant sister to the angiosperms in most recent phy- implementing thisfinalportionofthe searchstratlogeneticanalyses(e.g., Crane,1985, 1988; Doyle egy to exploretree space frommultipleperspec& Donoghue, 1986, 1992; Donoghue & Doyle, tivesand to preventtheanalysisfromstallingwhile 1989a, b; Loconte& Stevenson,1991; Chase et al., swappingon suboptimaltrees (P. Soltis & Soltis, 1993; Doyle et al., 1994; Nixon et al., 1994). In 1997). For data set 1, 94 startingtreesof lengths addition,to ascertainthetopologicalimpactofoth- 3929, 3930, 3934, 3936, 3937, 3938, 3939, 3940, er outgroups, particularly withregardto the first- and 3941 were used (shortesttree ultimately obwe conductedseveralother tainedhad a lengthof3923 steps).For data set 2, branching angiosperms, searches.Usingthe smaller18S data sets (194 an- 78 startingtrees of lengths3938, 3939, 3940,

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1

Soltis et al.

3941, 3942, 3944, 3946, and 3947 were used (shortesttree ultimatelyobtainedhad a lengthof 3930 steps). For data set 3, 192 startingtreesof lengths3506, 3508, 3509, 3512, 3514, 3515, and 3517 were used (shortesttreeultimately obtained had a lengthof 3501 steps). For data set 4, 96 startingtreesof lengths3513, 3514, 3515, 3516, 3517, 3520, and 3521 (shortest treeultimately obtainedhad a lengthof 3507 steps).Manyof these searchesresultedin treesone to severalstepslonger than the shortesttrees ultimatelyobtained; these longertreeswere also examinedto help ascertainthe generalsupportforsomebranches.Implementing the above search strategy forthe data sets described ultimatelyentailed well over two yearsofcomputertime. Implementingdecay analyses (Bremer,1988; Donoghueet al., 1992) is impractical withdata sets ofthissize. To obtainan estimateofsupportforthe 18S rDNA topologies,we applied the parsimony jackknifeapproach(Faris et al., 1997) to data set 1 (thisanalysiswas kindlyconductedbyS. Farris). The jackknifeis a resamplingapproach,similarto in whichthecharactersofa data set thebootstrap, are resampledto generatereplicatedata sets.Each replicateis analyzed,and the proportion of replia givenconclusion(in thiscase a cates supporting clade) is considereda measureof support.Jackknifepercentagescan therefore be interpreted similarly to bootstrappercentages.In this analysis, 1000 replicateswere conducted,and a minimum jackknifevalue of 50 (CUT = 50) was used (i.e., onlyd1adessupportedbyjackknifevalues of 50% or greaterwereretained).If a node is supportedby one uncontradicted character,the jackknifevalue is 63% (Farriset al., 1997). Thus,clades withvalues of 63% or moreare strongly supported;nodes withvaluesof51-62% are less wellsupported, and thosewithvaluesof50% orless receiveno support.

18S rDNA have, in large part,been greatlymisunderstood.In particular,insertionand deletion eventshave longbeen consideredcommonin 18S rDNA; concomitantly, alignmentwas considered highlyproblematic. Untilnow,theexistingdatabase of angiosperm18S rDNA sequences was insufficientto assess theseviews.

1997

18S RibosomalDNAPhylogeny

RESULTS AND DISCUSSION I. THE EVOLUTIONOF THE 18S rRNAGENE

The accumulationof a large data set of entire 18S rDNA sequences has permitted a morethoroughassessmentofthegeneralevolutionofthe18S rRNA gene. Unlikeprotein-coding genes,such as thewidelysequencedrbcL,matK,and ndhF,there is no clearframeofreference foraligningsequences or revealingerrors.For example,withproteincodinggenes, translationof a sequence to amino acids will potentiallyreveal some errorssuch as frameshifts and internalstop codons. No such internalcheck is available,however, forrDNA. As a theevolutionof result,generalfeaturesconcerning

15

INSERTION-DELETIONAND ALIGNMENT

This studyrevealsthatindelsare notwidespread in the 18S rDNA sequences of angiosperms, but insteadare confinedto a few,small regionsofthe gene. Furthermore, with the exceptionof these same small regions,alignmentof 18S rDNA sequences is straightforward. Several possibilities mayexplainthe misconception thatthe 18S rRNA geneis highlypronetoinsertion and deletion.First, the literature containsa numberof erroneous18S rDNA sequences. We have resequencedthe 18S rDNA of over 20 taxa, and have foundthatsome publishedsequencesare in errorby as manyas 33 bases, whichcorresponds to 1.8% ofthetotalgene. In several instances,we discoverednumerouserrorsin the 18S rDNA sequence fora taxonusing the same DNA originallyused to producethe reportedsequence. These errorsin previouslygeneratedsequences are notconfinedto base composition,but also involvethe presence of whatwe referto hereas "false"insertions and deletions.For example, we resequenced 18S rDNA fromZea mays and discoveredthattheoriginalsequenceincluded a largenumberof"false"insertions relative to all otherangiosperms.Integrating our new sequenceforZea intotheangiosperm data matrix and removalof the previouslypublishedsequence resultedin theelimination of 14 indelsfromour 18S rDNA data set,fourofwhichwerealignmentgaps thathad to be added to all otherangiosperms. We were able to removeadditionalalignmentgaps throughthe resequencingof othertaxa forwhich 18S rDNA sequences werereportedpreviously. As a point of comparison,the total length of the aligned sequences in the data matrixof Nickrent and Soltis(1995) for64 taxa was 1853 bp. In contrast,the lengthof the aligned sequences in our 228-taxondata matrixis only 1850 bp. The resequencing of additionaltaxa for which published sequences cause numerousalignmentgaps would likely decrease furtherthe total length of the alignedsequences. The numerouserroneous18S rDNA sequences in the literatureperhapsresultfrominherentdifficultiesin sequencingrDNA. Secondarystructure in therRNAforwhichthisgenecodes is also pres-

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

16

Annals of the MissouriBotanical Garden

Table 3. Areainitiallythought tobe pronetoinsertion and deletion;howeverapparentgaps resultfromsequencing difficulties. The underlinedportionof theHydrangea sequence showstheactual base composition thatwe have determined to be presentformanytaxa forthisarea.

sertionand deletion.As a result,alignment ofthis regionwas initiallydifficult. Alignmentproblems were exacerbatedby the apparentoccurrenceof base substitution in theregion.Furthercompoundof alignmentis the factthatthe ing the difficulty regionjust 3' of this area actuallyis proneto in215 as well as to considerablevariasertion-deletion, * in tion Base positions230-237 primary sequence. Hydrangea AAAGGTTGACGCGGGCTTTGCCC correspond to one ofthevariablehelixtermini notGlycine AAAGGTCAACACAGGCTCTGCCT ed above. Cycleand automatedsequencingofover Heuchera AAAGGCCAAC-- - -GCTTTGCCC Lepuropetalon AAAGGTCAACGCTTGCTTCGGCT 100 taxa, however,revealedno indels in the area Prunus AAAGGCCAAC-- - -GCTCTGCCC of positions215-230. In fact,afterresequencing Francoa AAAGGTCGAC-- - -GCTTTGCCC thisregionin manytaxaforwhichmanualsequencPodophyllum AAAGGTCAACG-- -GCTTTGCCC es initiallysuggestedthepresenceofnumerousinAustrobaileya AAAGGCCGAT--CGGCTCTGCCC dels, we have concludedthatindels in thisregion Caulophyllum AAAGGTCAAC???????CTGCCC are eitherextremely rareornonexistent. Thisregion Buxus AAAGGTCGA---------TGCCis G-C rich;as a result,sequencing"stops" often occurred,yieldingonlya portionofthe base pairs actuallypresentin the region.Alignment of these ent in the gene itselfand may lead to compressions incompletesequences suggestednumerousindels and associated sequencing&-problems.More than in this region,leading to the misconception that one molecular systematistwith substantial experi- indelswerefrequent. Similarsequencingproblems ence in the sequencing of chloroplast genes such were encounteredin other portionsof the 18S as rbcL has referredto the sequencing of 18S rDNA rRNAgene.Takentogether, theseregionshad conas "tricky."We have found thatpreparationof sam- tributed to theviewthatinsertionand deletionare ples via cycle sequencing followed by automated commonin 18S rDNA. ofindelshas been overAlthoughthefrequency sequencing yields reliable 18S rDNA sequences that appear more accurate than most manually gen- stated for the 18S gene, several regionsof 18S erated sequences. The critical procedural step is rDNA are, in fact,proneto variationin primary likely the cycle sequencing reactions, in which secsequence and length.However,these regionsare and ondary structureis reduced or eliminated by high small,easilylocated,and,as notedbyNickrent Soltis (1995), typicallyconfinedto the terminiof temperature.Several specific regions of 18S rDNA are particularlydifficultto sequence. These include helices on theproposedsecondarystructure model for18S rRNA(e.g., Nickrent& Soltis,1995). Four base positions 215-230, 1355-1365, and 17051715 (all positions mentioned in this paper corre- such regions,represented by base positions230spond to those of Glycine max; Eckenrode et al., 237, 496-501, 666-672, and 1363-1369, corre1985), as well as several of those small regions not- spond to the terminiof helices E10-1, 17, E23-1, ed earlier that are prone to insertion and deletion and 43, respectively (see Appendix).These regions are difficult to alignovera broadtaxonomicscale, (positions 230-237; 496-501; 666-672; 13631369) (see Appendix). such as all angiosperms,and were therefore not We will use the firstof these regions (base po- used in our phylogenetic analyses (see Materials sitions 215-230) to illustrate the errors that can and Methodsabove). On a lowertaxonomicscale result in 18S rDNA sequencing. Based on manual (e.g., closelyrelatedfamilies),however, even these sequences (generated by D. Soltis & R. Kuzoff, the highlyvariableregionsare easily aligned,permitbase composition of this region in Saxifragaceae tingtheuse oftheseregionsin morefocusedstudand several other rosid families initially appeared ies (e.g., Polemoniaceaeand relatedAsteridaes.l., to involve a large deletion relative to some other Johnsonet al., unpublished;portionsof Saxifragaavailable sequences (see Table 3). Similarly, the ceae s.l., D. Soltis & Soltis, 1997; Orchidaceae, 18S rDNA sequences generated manually by other Cameron& Chase, unpublished). Because indelsin 18S rDNAare neitheras previnvestigators, representing a diversity of angioalent nor as problematicas previouslythought, sperms, typically were lacking one or more base ofclean 18S rDNAsequencesis straightpairs in this region. Alternatively, researchers alignment scored this region as uncertain, using either "?" or forward. Withtheexclusionofthefewsmallregions "N." Thus, sequences available prior to this study notedabove,alignment ofover200 angiosperm seand easily accomsuggested that this region was highly prone to in- quences was straightforward

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

plishedby eye. This is also trueforthe alignment of 18S rDNA sequences on a broaderscale across vascular plants (P. Soltis et al., unpublished).As notedabove,theresequencingofsometaxagreatly facilitatedthe alignment processin angiosperms.

formative indels. However,the discoveryof such indels dependson the availabilityof a largedatabase of accuratesequences. Previouslypublished sequences containing errors and ambiguities, "false"indels,and incompletely sequencedregions have madeassessmentofthephylogenetic potential of indelsin 18S rDNA impossible.

PHYLOGENETICALLYINFORMATIVEINDELS

17

Notonlydo indelsin angiosperm18S rDNA se- STEM VERSUS LOOP CHANGES/COMPENSATORY CHANGES quences not cause alignmentproblems,but some indels may be phylogenetically informative at the of the 18S rRNAtranThe secondarystructure level of ourinvestigation. Here we do notconsider scriptmayhave significant forphylogimplications thoseindelslocatedwithinthevariableregionsnot- enyreconstruction usingrRNAor rDNAsequences ed above, but onlythose indels located in highly (e.g., Mishleret al., 1988; Dixon & Hillis, 1993). conservedregionsnotnormally proneto changesin The questionremainsas towhether bothloop bases length.Twosuch indelsin particular(Table 2) ap- (non-pairing bases) and stembases (pairingbases) informative pear to be phylogenetically across the shouldbe used in phylogeny reconstruction and, if angiosperms. so, whetherbases fromstemsand loops shouldbe One indel involvesan apparentdeletionof one consideredequally informative and independent. base pair thatunitesall highereudicots(Table 2, Assuming near-perfectcompensatorymutation indel A). This base pair-is presentin Gnetales, (substitutions that maintainor restorestem base monocots,paleoherbs,Magnoliales,Laurales,ran- complementarity-e.g.,Noller, 1984; Curtiss & Pro- Vournakis,1984; Wheeler& Honeycutt,1988) in unculids,Trochodendraceae, Tetracentraceae, ofthe teaceae, Sabiaceae, Platanaceae,and Nelumbona- stemregionsto maintainsecondarystructure ceae and is absentfromall higherdicots(i.e., the 18S or 26S (28S) rRNA,Wheelerand Honeycutt large Rosidae clade and Asteridaes.l.). Thus,the (1988) recommendedthat stem bases eitherbe of thisindel agreeswiththe resultsof eliminatedfromphylogenetic distribution analysisor weighted phylogenetic analyses based onlyon base substi- byone-halfrelativetoloop bases. However,in their tutions(Figs. 1, 2, 4; all figures,plus Appendix, analysesof28S rRNAsequencesfromvertebrates, followLit. Cit.).In addition,the distribution ofthis Dixonand Hillis (1993) foundthatcharactersfrom indelalso agreeswithtopologiesbased on analyses both stemsand loops containphylogenetic inforof rbcL sequences. It appears, however,thatthis mation.They also foundthatthe numberof comin pensatorymutationsin stem bases was less than base pairmayalso have been lostindependently twoof the monocotsanalyzedhere(i.e., Calla and 40% ofthatexpectedundera hypothesis ofperfect Dixtomaintainsecondarystructure. Chlorophytum). compensation indel on and Hillis therefore The second phylogenetically informative suggestedthattheweighting involvesan apparentinsertion(Table 2, indel B) of stem charactersbe reduced by no more than that unites all membersof the saxifragoidclade 20% relativeto loop charactersin phylogenetic (Saxifragaceaes. str.and close allies; this is the analyses. The large database of 18S rDNA sequences reSaxifragalesof D. Soltis & Soltis,1997). The saxtheopportunity to addressthese ifragoidclade also is unitedby base substitutions portedhereaffords and representsone of the moststrongly supported and otherissues regardingthe impactof the secfromthe phylogenetic clades resulting on analyses. ondarystructureof the 18S rRNA transcript Additionalexamples of potentiallyinformativephylogeny in angiosperms. reconstruction Although indelscan be foundat lowertaxonomiclevels. For it is notour goal to examinesuch issues in detail each ofa here,we will providesome initialobservations reexample,in Zingiberales,twoinsertions, singlebase pair,are foundat positions117 and 260 gardingthe relativeimportanceof bothstemand in all membersof the Zingiberaceae(Kress et al., loop mutations and theprevalenceofcompensatory 70 mono- mutations. 1995). None of the otherapproximately ofstemand loop bascots forwhich18S rDNA has been sequenced exWe followedthedefinitions ofone es used elsewhere(e.g., Dixon & Hillis, 1993): hibitstheseinsertions. an insertion Similarly, base at position655 appears to unitemembersof stembases are thosethatparticipatein base-pairViscaceae. ing interactions; loop bases do notengagein base additional18S rDNA sequencing pairingin thematurerRNA.Mappingbase substiUndoubtedly, will reveal moreexamplesof phylogenetically in- tutionson theproposed18S rRNAsecondarystruc-

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

18

Annals of the MissouriBotanical Garden

46% tureforGlycinemax (see Appendix),we examined analyzed,19% were"double compensatory"; uracilthat involving 120 positions(in 60 taxa)at whichphylogeneticallyweresinglebase substitutions had occurred(based changedone base-pairingcoupletto another("sinbase substitutions informative 8% changeda pairofnoncomon the resultsof the searches; discussed below). gle compensatory"); (the second that plementarybases to complementary Emphasis was placed on those substitutions change); 27% deforthoseclades thatap- type of "single compensatory" providedsynapomorphies several stroyeda base-pairingcouplet.Of these changes treesand thatrepresent pear in all shortest different taxonomiclevels (e.g.,Asteridaes.l., Car- thatresultin mispairingof nucleotides,overoneyophyllidaes.l., monocots,glucosinolates,santa- thirdare adjacentto loop regions.Hence, theloop loids, Caryophyllales,saxifragoids,celastroids, regionsmaysimplybe expandedin theseinstances. Of these 120 positions, Nearlythreequarters(73%) ofthe stemmutations Parnassia-Lepuropetalon). 70 (58%) werestembases, and 50 (42%) wereloop we analyzedmaintainor restorebase pairingand In theircombases. Althoughthisinitialsurveyconsidersonlya would be consideredcompensatory. it parableanalysisof28 rRNAsequences,Dixonand base substitutions, subset of synapomorphous suggeststhatbothstemand loop regionsappearto Hillis (1993) observedthatonly47% ofthe mutawithperhapsa tionsmaintainedor restoredbase pairing.Our recontainphylogeneticinformation, for5S rRNA sites sultsare moresimilarto observations somewhatgreaterproportionof informative foundin stem,ratherthanloop,regions.This topic (Curtiss& Vournakis,1984), whereapproximately analyzedfromstem The 88% of the base substitutions examination. certainly requiresa morerigorous contentof stem versus loop regionswerecompensatory. relativeinformation A similarpatternofmolecularevolutionis seen levtaxonomic bases may,in fact,varyat-different els. For example,some of the morevariableloop for 18S rDNA withina single family,Polemoniet al., unregions(severalofwhichwereremovedfromthese aceae, and its closestrelatives(Johnson analysesbecause the sequenceswere published),where228 variablenucleotidepositions phylogenetic difficult to align) may hold relativelymoreinfor- were examined.Althoughmostof these 228 posimationat lowerlevels (amongand withinclosely tions are located on stems (133 comparedto 95 relatedfamilies)than at highertaxonomiclevels loop characters),the averagenumberof substitucharinformative (ordinaland above), at whichthe sequences be- tionsper site overthepotentially actersis greaterforthe loop characters(5.0) than forconfident alignment. come too divergent changeswas ex- forthestemcharacters(3.1). Usingone ofthemost ofcompensatory The frequency Johnsonet al. the parsimonioustreesas a framework, aminedin 21 stemregionslocated throughout 18S rRNAgene.Followingthegeneralapproachof (unpublished)also consideredin more detail 67 thateitheruniteor appearwithinPoothers(e.g., Dixon & Hillis, 1993), we considered substitutions 36 (53.7%) ocwithinstem regions. lemoniaceae.Ofthesesubstitutions, two classes of substitutions just the 31 stemsubstithatchange cur in loops. Considering The firstclass involvessubstitutions bases to anotherpairof tutions,23 (74.2%) either maintainor restore one pair ofcomplementary bases. This includes"double com- base-pairing.The remainingeightstem substitucomplementary in whichone pair ofcom- tions (25.8%) resultin mispairingof nucleotides, pensatory"substitutions bases is convertedto another(e.g.,C-G withfoursites locatedadjacentto loops. plementary These initialstudiesof the relativeinformation to A-U). This class also includeschangesthatreevent.That is, be- contentofstemand loop regionsand thefrequency quire onlya singlesubstitution forthe changeshave implications cause uracilcan pair withguanineas well as with ofcompensatory of stemand loop bases adenine,it is possibleto have a singlechangefrom use and relativeweighting These data reinforce reconstruction. one base-pairingcoupletto another(e.g., U-G to in phylogeny one typeof the findingsof others(e.g., Dixon & Hillis, 1993; C-G; U-A to U-G). The latterrepresent substitutions (sensu Dixon Smith,1989) thatbothstemand loop regionspro"single compensatory" for phylogenyrecon& Hillis, 1993). The second class of stemsubsti- vide importantinformation tutionsinvolvesthosethatchangeone pairofcom- struction.In addition,the high frequencyof obchange in angiosperm18S bases to a pair of noncomplementaryserved compensatory plementary bases, or vice versa (e.g., C-G to G-G; or C-C to rDNAsuggeststhatperhapsstemcharactersshould C-G). For example,a change of C-G to G-G de- receiveless weightthanloop charactersin future of stemversusloop a change analyses.However,weighting stroysa base-pairingcouplet.Conversely, coupletand charactersis morecomplexthanit mightseem inifromC-C to C-G createsa base-pairing representsanotherexampleof a "singlecompen- tially.Recent workwith18S rDNA sequences in Of the 216 stemchangeswe Polemoniaceae(Johnsonet al., unpublished)demsatory"substitution.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

19

low),and tanninsand alkaloidsas secondarycompoundsforma lade (labeled eudicot lade). The lattergrouphas been referredto as the eudicots (Donoghue& Doyle,1989b; Doyle& Hotton,1991; Chase et al., 1993). Althoughthetermeudicothas been variouslydefined,we willuse theChase et al. (1993) definition to facilitatecomparisonbetween thetwostudies.A eudicot lade was also recovered in analysesofrbcLsequences (Chase et al., 1993), a grade,as theydo here, but, insteadof forming thoseplantswithuniaperturate pollenforma weakly supportedlade in therbcLtrees. II. PHYLOGENETICRELATIONSHIPS There are two major exceptionsto the general betweentheeudicot lade and the correspondence Each broad phylogenetic analysis yielded distribution of the triaperturate pollentypes(other thousands of most parsimonious trees; it is likely thanthe obviousdeparturesobservedin the trees that shorter trees exist for all four data sets and derivedfromanalysisofdata set 3). First,theWinthat additional classes of most-parsimonioustrees teraceaeand severalfamiliesof paleoherbs(Chlowere not recovered. Nonetheless, we feel that it is ranthaceae, Lactoridaceae, Aristolochiaceae)all significantthat analyses of three of the four data possess uniaperturate pollen,yetappearwithinthe sets suggest the same general topology.The shortest eudicot (triaperturate) lade in the shortesttrees trees obtained fromsearches of data sets 1 and 2 obtainedin analysesof data sets 1 and 2 (Figs. 1, are essentially identical, and differencesbetween 2). These exceptionsmayreflectlow taxondensity the shortest trees from analysis of these two data and/orthe low resolvingpowerof 18S rDNA sesets and data set 4 are minorand weakly supported. quence data (see below);thesetaxa seem to be unAlthough phylogenetic analysis of data set 3 restablein positionin thevarioussearches.In broad vealed many of the same major clades recovered analysesofrbcLsequences,in contrast, thesetaxa by searches of the other data sets, relationships are clearlymembersofthe uniaperturate lade. among some of these clades differ;most notable are The second exceptioninvolvesIlliciaceae and the weakly supported, unusual positions of monoSchisandraceae.Unlikethe examplesabove, howcots and saxifragoids (see below). ever,whichwe suspect representspuriousphyloAll searches revealed the same major clades geneticplacements,Illiciaceae and Schisandraceae (e.g., Rosidae, Asteridae s.l., Caryophyllidae s.l., appearto be trueearly-branching angiosperms (see monocots, saxifragoids), as well as the same suite below),yetpossess triaperturate pollen.These famof taxa as sister to all remaining angiosperms. In ilies similarlyappear as early-branching angiogeneral, the trees obtained in these exploratory, spermsin analysesbased on rbcLsequences(Chase broad analyses of 18S rDNA sequences depict reet al., 1993; Qiu et al., 1993). As reviewedby lationships very similar to those obtained in broad Doyle et al. (1990), however,the tricolpatecondianalyses of rbcL sequences (Chase et al., 1993). tionin Illiciaceae and Schisandraceaeis different The general features observed in the shortesttrees fromthatwhichcharacterizeseudicots.Hence, the obtained fromthe four searches are discussed be18S rDNA analysesfurther supporttherbcL-based low. The several unusual relationshipsamong major inferencesof Qiu et al. (1993) thatIlliciaceae and clades suggested by analyses of data set 3 are disSchisandraceaerepresent an independent evolution cussed in more detail below under "Differences oftricolpatepollen. Among the ShortestTrees." FourfamiliesofwoodyMagnoliidaeconsistently appear as sistertaxa to all remainingangiosperms FIRST-BRANCHING FAMILIES analyzed: Amborellaceaeand a lade of AustroPhylogenetic analyses of three of four data. sets baileyaceae,Illiciaceae, and Schisandraceae.The (data sets 1, 2, and 4) suggest thatthose taxa having latterthreefamiliesformone of the moststrongly uniaperturate pollen (monosulcate and monosul- supportedclades in thisstudy(jackknifevalue of cate-derived) and ethereal oils appear as early- 94%). In searchesofdata sets 1 and 2, a lade of branching angiosperms, forminga grade (labeled Austrobaileyaceae, Illiciaceae,and Schisandraceae monosulcate grade in Figs. 1, 2, and 4). Those is the sistergroupto all otherangiosperms, folplants having triaperturatepollen (tricolpate and lowedsubsequently by Amborellaceae;in analyses tricolpate-derived),with a few exceptions (see be- of data sets 3 and 4, the positionsof these two onstratesthatloop regions evolve more rapidly than do stem regions. Thus, in more focused studies in which it is possible to align and use the entire 18S rDNA sequence, stem and loop regions should perhaps be given equal weight. In broader studies in which the rapidly evolving loops are removed due to alignment difficultiesand only the more conserved loop regions are included in the analysis, stems should be downweighted;however,more detailed analyses are required to estimate appropriate weights.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

20

Annals of the MissouriBotanical Garden

giventheclose amongthe angiosperms."If the 18S rDNA inferlineagesare reversed.Surprisingly, a thatAmborellaceae, relationshipsuggested between Illiciaceae and ence is correctin suggesting Schisandraceaeby others(e.g., Cronquist,1981; familylackingvessel elements,are amongthefirstQiu et al., 1993), Illiciumis sistertoAustrobaileya- branchingangiosperms, this analysismaysupport Schisandrain all fouranalyses.These fourgenera, the hypothesisthat the angiospermswere primiwithNymphaeales,forma lade in the rbcLanal- tivelyvesselless (Bailey, 1957; Cronquist,1981; yses of Chase et al. (1993) and Qiu et al. (1993). Young,1981). In the shortesttreesresultingfromsearchesof In contrastto thisstudy,analysesbased on parall fourdata sets, fromone to severalfamiliesof tial 18S and 26S rRNA sequences suggestedthat paleoherbs(sensu Donoghue& Doyle,1989a) sub- a groupof paleoherbs(Aristolochiales, Piperales, sequentlyfollowAustrobaileyaceae,Illiciaceae, Nymphaeales)is the sistertaxonto all otherflowSchisandraceae, and Amborellaceae. Nymphae- eringplants (Hamby& Zimmer,1992). However, followthesefourfamiliesin all ofthefourwoodyfamiliesofMagnoliidaeappearing aceae immediately shortesttrees. In searches of data sets 1 and 2, as first-branching taxa in our18S rDNAtrees(AmNymphaeaceaeforma lade withPiperaceae and borellaceae,Schisandraceae,Illiciaceae, and Ausby Peperomiaand Hout- trobaileyaceae),only Illiciaceae were sampled by Saururaceae(represented jackknifevalue Hamby and Zimmer(1992). Other phylogenetic tuyniaand Saururus,respectively; of 85%), whereasin searchesof data set 4, these analysessimilarlysupportthepositionofsomepasame threefamiliesforma grade withNymphae- leoherbsas first-branching taxa amongthe angiofol- sperms(e.g.,Doyleet al., 1994; Nixonet al., 1994). aceae as sisterto all remainingangiosperms, lowed by a lade of Piperaceae and Saururaceae. Paleoherbsare sisterto otherangiosperms in trees In searchesofdata set3, Nymphaeaceaealso follow based on a combination and rRNA of morphology Austrobaileyaceae,Illiciaceae, Schisandraceae, sequence data and in thosederivedindependently and Amborellaceae,but Nymphaeaceaeare then frommorphological and rRNA data (Doyle et al., an unusualplacementdisfollowedbysaxifragoids, 1994). However,thistopologyis onlyweaklysupcussed in moredetailbelow. portedby morphologicaldata, with trees rooted Amborellaceae,followedby (1) a lade of Ausnextto Magnolialesonlyone step longer.FurtherIlliciaceae,and Schisandraceae,(2) trobaileyaceae, more,the rRNAdata set employedby Doyle et al. Nymphaeaceae,(3) a lade or gradeofPiperaceae, (1994) is thatofHambyand Zimmer(1992), which, Saururaceae,Aristolochiaceae,and Lactoridaceae as notedabove,lacked severalcriticalwoodymagan(similarto Fig. 4), appearas thefirst-branching noliids. giospermswhenZamia and Cycasare used as adanalysesof rbcL sequences Broad phylogenetic ditionaloutgroups(see Materialsand Methods).In et al., 1993; Rice et al., 1997) place the (Chase analyses of a largerdata set of 271 preliminary as sisterto all reCeratophyllum Gnetum, aquatic genus usingspecies of Welwitschia, angiosperms of CeratoThis placement maining angiosperms. Amborellaceae,a lade and Ephedraas outgroups, been suggestedon morphological also has phyllum of Austrobaileyaceae-Illiciaceae-Schisandraceae, Nixonet al., and Nymphaeaceaeagain appearas successivesis- grounds(Les, 1988; Les et al., 1991; in the latterstudy alternative trees although 1994), tersto all remaining angiosperms. as sister The positionofwoodymagnoliidsas first-branch-place thepaleoherbfamilyChloranthaceae A offlonumber to the flowering plants. remaining ing taxa in these 18S rDNA trees is in general the to also conform features of Ceratophyllum ral reviewsofangiosperm withtraditional agreement a angioview that the primitive represents genus lationships(e.g., Cronquist,1968, 1981; Stebbins, 1974; Takhtajan,1969, 1980) that suggestthat sperm (Endress, 1994). However,Ceratophyllum in any of our woodyMagnoliidaeare the mostprimitiveextant does not appear as first-branching threeof Searches involving analyses. phylogenetic analysesof Donangiosperms.The morphological 2, and the four data sets Ceratophyllum 4) place (1, oghueand Doyle (1989a) and Loconteand Stevenin generalagreea finding son (1991) also supportthe woodyMagnoliidaeas as sistertothemonocots, the most ancestralliving group of angiosperms. mentwithearlierrRNAsequenceanalyses(Hamby Otherdata also pointto the antiquityof at least & Zimmer,1992). Subsequentto theAmborellaceae,Austrobaileysome of these genera.For example,Endress and thatthepollenofAus- aceae, Illiciaceae, Schisandraceae,Nymphaeaceae, Honegger(1980) determined one of the and Piperales,in analysesofthreeoffourdata sets trobaileyaresemblesClavatipollenites, oldestprobableangiosperm fossils,and concluded (1, 2, and 4) are additionalfamiliesand ordersof that Austrobaileyamay be "especially archaic Magnoliidae: Annonaceae, Calycanthaceae,and

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

1997

18S Ribosomal DNA Phylogeny

Lauraceae, all woodyfamiliestraditionally consid- aceae (Aristolochia, Asarum,Saruma), Lactoridaered amongthemostprimitive extantangiosperms. ceae, Sabiaceae, and Chloranthaceae (Hedyosmum). Withtheexceptionoftheshortest treesresulting The latterlade is an unexpectedgrouping (see befromanalysisof data set 3, the monocotsalso ap- low) of paleoherbs (Aristolochiaceae,Lactoridapear as an earlylineageofangiosperms. The mono- ceae, Chloranthaceae), woodyMagnoliales(Wintercotsare monophyletic, withtheexceptionofAcorus, aceae), and eudicots (Sabiaceae). With the whichdoes notappear closelyrelatedto the other exceptionofAristolochiaceae, Lactoridaceae,Chlomemberof Araceae included(Calla). In analyses ranthaceae,and Winteraceae,the presenceof the ofthetwodata sets (1 and 2) thatincludedAcorus, remainingtaxa on branchesat the base of the euAcorus follows Nymphaeaceae-Piperalesas the dicots closely parallels resultsretrievedfromthe subsequentsisterto all remaining In phylogenetic angiosperms. analysesofrbcLsequences (Chase et analysesofrbcLsequences,Acoruswas considered al., 1993). "phylogenetically isolated"as sisterto theremainIn theshortest treesobtainedin analysesofdata ingmonocots(Duvall et al., 1993). Phylogenetic re- set 4, the distinctionbetween the monosulcate sults based on 18S rDNA sequences also suggest grade and the eudicot lade is less clear than in thatAcorusis anomalousamongmonocots.Given theshortest treesobtainedfromtheanalysesofdata itslongbranchlengthand unexpectedposition,the sets 1 and 2 (see "LowerEudicots/Monosulcates," 18S rDNA ofAcorusshould be resequenced,and Fig. 4). Platanaceae, Trochodendraceae/Tetracenadditionalmonocotsshould be added to the data traceae,ranunculids(whichare Proparaphyletic), set beforethe affinities ofthisenigmaticgenusare teaceae (Knightia),Buxaceae, Sabiaceae, and a addressedfurther. lade ofChloranthaceae (Hedyosmum)/Winteraceae Because our samplingof.monocots was limited, (Drimys)again appear as sistergroupsto the reto permitmorethoroughtreatment elsewhere,re- mainderof the eudicot"lade. Also in this same lationshipswithinthe monocotswill not be dislowereudicot/monosulcate grade,however, are Calcussed herein anydetail.Nonetheless, severaltraycanthaceae,Annonaceae(Mkilua),and Lauraceae ditionallyrecognizedgroupsofmonocotsappearto (Sassafras),uniaperturate familiesthatappear in a be monophyletic, includingZingiberales,Liliales, lade withsome ranunculids(Fig. 4). and highercommelinoids.Furthermore, the broIn the shortesttreesobtainedin analysesof all meliadsare groupedwiththegrassesand allies, as data sets,theremainderoftheeudicot lade is esexpected(Duvall et al., 1993). The twobest supsentiallycomposedoftwolargesubclades,one conportedclades withinthemonocotsare Zingiberales of Rosidae plus-some Dilleniidae (Maranta,Zingiber,Costus,Canna, Heliconia,and sistinglargely and the other to the Asteridaes.l. corresponding Musa; jackknifevalue of 58%) and a lade com(labeled Rosidae and Asteridae in s.l., respectively, posed of Sparganiaceae,Cyperaceae,Poaceae, and 1, 2, Figs. With a few 4). most exceptions, notably Bromeliaceae(Sparganium,Cyperus,Oryza,Zea, and Glomeropitcairnia; jackknifevalue of 59%). the placementof the monocotswithinthe Rosidae Surprising results,givenrbcLtopologies(Duvall et lade, thesetwolargeclades also are presentin the al., 1993) and morphological features,includethe treesderivedfromsearchesofdata set 3. The RosplacementofOrchidaceae(Oncidium)as thesister, idae and Asteridaes.l. clades werealso recovered or one ofthesisters,to theremaining monocotsand in broadanalysesofrbcLsequences (Chase et al., the placementof Arecaceae (Veitchia)and Alis- 1993; Olmsteadet al., 1992, 1993; Rice et al., submataceae(Sagittaria)withintheAsparagales(Figs. mitted),althoughthe placementof Caryophyllidae in the 18S rDNA and rbcL 1, 2). These unusualplacementsshouldnotbe con- s.l. is verydifferent sidered seriously,however,due to the low repre- topologies(see below).These twolargeclades,Rosidae and Asteridaes.l., reflectthebasic divisionof sentationof the monocots. higherdicotsintotwomajorgroups(Young& Watson, 1970), with(1) polypetalouscorollasand nonEUDICOT CLADE tenuinucellate ovules and (2) sympetalous corollas Below we Analysesof threeoffourdata sets (1, 2, and 4) and tenuinucellateovules,respectively. clearly reveal a eudicot (or triaperturate) lade discuss in moredetailthemajorclades ofeudicots analysesof 18S rDNA se(Figs. 1, 2, 4), withthe following successivesister based on phylogenetic namesin mostinstances groupsat its base (Figs. 1, 2): Proteaceae,Nelum- quences. We use informal bonaceae, Platanaceae, a lade of ranunculids, to referto stronglysupportedclades (e.g., celasand a lade Trochodendraceae/Tetracentraceae, troids,saxifragoids, ranunculids),some of which composed of Winteraceae(Drimys),Aristolochi- differdramatically fromtraditionalviews of rela-

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

22

Annals of the MissouriBotanical Garden

tionship,butformaltaxonomicchangemaybe war- Saxifragaceaes.l. (D. Soltis & Soltis, 1997). The of this lade is supportednot onlyby monophyly rantedformanyofthese. but also by the presenceof an base substitutions, Ranunculids. The searches of the largerdata insertion(see Table 2) located in a portionof the sets (Figs. 1, 2) recovereda lade (labeled "Ran18S rRNAgene thatis highlyconservedin length. unculids")containingLardizabalaceae,Berberidato as rosid III) is reAn identical lade (referred ceae, Ranunculaceae, Menispermaceae,Euptelevealed in the499-taxonanalysisofrbcLsequences aceae, Fumariaceae, Sargentodoxaceae, and in prelim(Chase et al., 1993) and is also retrieved Papaveraceae. This same lade was foundin the inaryanalysesof a 271-taxon18S rDNA data set broad analyses of rbcL sequences (Chase et al., includingmoreHamamelidaceae,as wellas in phy1993); it representsthe core of the Ranunculales logenetic analyses involving matK sequences closelyto (sensu Cronquist,1981) and corresponds & Soltis,unpublished)and prelimi(Hibsch-Jetter the Berberidalesof Thorne(1992) and the Rannary work with atpB sequences (Hoot, unculifloraeof Dahlgren(1980). Analysesof not unpublished).As reviewedin more detail by D. onlyrbcLand 18S rDNA sequences,butalso atpB Soltis and Soltis (1997), this small lade is notesequences, place Eupteleaceae (Hamamelidae) placed in thatit containstaxa traditionally worthy withinthis lade (Hoot & Crane,1995). Also part in threesubclasses:Paeoniaceae (Dilleniidae);Haof this lade in the 18S rDNA analysesis Sargenmamelidaceae, Daphniphyllaceae,Cercidiphyllatodoxaceae,a familytypicallyplaced in Ranuncuceae (Hamamelidae);the remainingtaxa are all lales and allied withLardizabalaceae (e.g., Cronmembersof Rosidae. quist, 1981). In contrast, analyses of rbcL Althoughthis saxifragoidlade is recoveredby withFabaceae sequences placed Sargenfodoxaceae analyses of both 18S rDNA and rbcL sequences, (Chase et al., 1993). This resultis due to themisthissame groupoftaxa has neverbeen recognized of leaf materialin the rbcLanalysis identification WhereasSaxifragaceaes. str., in anyclassification. for (Qiu, pers. comm.).ReanalysisofSargentodoxa Penthorum, Pterostemon, Ribes,Itea, Tetracarpaea, rbcLplaces itas sistertotheLardizabalaceae(Hoot and Crassulaceae are consideredclosely related & Crane,1995; Hoot et al., 1995). membersof Rosidae in virtuallyall recenttreatthetwosmalldata sets(3 and Searchesinvolving ments(e.g., Cronquist,1981; Thorne,1992; TakhofRanunculales. 4) employedfewerrepresentatives tajan, 1987; Dahlgren,1980, 1983), the affinities fromsearchesofdata set 4, these In treesresulting and the dilleniid of the rosidfamilyHaloragacette taxa forma gradeas some of the early-branching familyPaeoniaceae have been consideredenigmateudicots.In treesfromsearches of data set 3, in ic (e.g., Cronquist,1981). The hamamelidfamilies The contrast,the ranunculidsappear polyphyletic. foundin thisclades(Hamamelidaceae,Cercidiphylplacementsof the ranunculidsin analysesof data laceae, and Daphniphyllaceae)typicallyhave not sets 3 and 4 maywell reflecttheirdecreasedrepbeen consideredclose relativesofSaxifragaceaes. resentation (lowertaxondensity)in thesesearches. str. and allied rosids. The relationshipsof these analysesof a 271-taxon18S rDNA In preliminary more anomalousmembersof this lade are disthe ranuncudata set includingmoreranunculids, cussed in moredetailbyD. Soltisand Soltis(1977). group. lids again forma monophyletic Glucosinolatelade. Anotherlade revealedby Saxifragoids. All analyses of 18S rDNA sequences (Figs. 1-4) reveal a lade composedof all analyses (Figs. 1-4) comprisesglucosinolateHeuchera,Boykinia,Saxifraga (Saxifragaceaes. producingtaxa. The familiesthat compose this str.), Crassula, Sedum, Dudleya, and Kalanchoe lade in Figures1-4 are 7 ofthe15 familiesknown Tetracarpaea,Ribes, to produceglucosinolates(mustardoil glucosides): (Crassulaceae), Pterostemon, and Itea (Grossulariaceae),Penthorum(placed in Limnanthaceae,Brassicaceae, Capparaceae, MorSaxifragaceaeby Cronquist,1981), Altingiaand ingaceae, Caricaceae, Bataceae, and TropaeolaLiquidambar (Hamamelidaceae), Haloragaceae, ceae. WhereasBrassicaceaeand Capparaceaehave and Paeoni- long been recognizedas closely related,the reDaphniphyllaceae, Cercidiphyllaceae, to hereas saxifragoids. mainingfamiliesincludedin this study(Limnanaceae; this lade is referred supported thaceae, Moringaceae,Caricaceae, Bataceae, and This lade is one of the moststrongly diverse and findingsof this investigationjackknifevalue of Tropaeolaceae) are morphologically to havebeen placed in distinctorders(e.g.,Cronquist, 68%). The same saxifragoidlade (also referred in an analysisof130 1981; see reviewby Rodmanet al., 1993). The gewas identified as Saxifragales) 18S rDNA sequences aimed at elucidatingthe af- nus Drypetes(Euphorbiaceae)also producesgludiversemembersof cosinolates,but it does not appear to be closely finitiesof the morphologically

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

23

relatedto theglucosinolatelade in anyofthefour Searches involvingthe two largerdata sets (1 searches.Phylogenetic analysesof 18S rDNA se- and 2) also place threefamiliesof Malvales (Malquences involving additionalglucosinolate taxafur- vaceae, Bombacaceae,and Tiliaceae) withinthenitherdemonstrate the monophyly of the glucosino- trogen-fixing lade; thesetaxa werenotpartofthe late-producers, withthe exceptionofDrypetes, and nitrogen-fixing lade in the rbcL-basedtrees. In also clarifyrelationshipsamongthe membersof analysesofdata set 4, however, thesethreefamilies this lade (Rodmanet al., submitted). These results ofMalvalesare notpartofthenitrogen-fixing lade closelyparallelfindings based on thephylogenetic (Fig. 4). No clear nitrogen-fixing lade emergedin analysisof rbcL sequences (Rodmanet al., 1993; analysesof data set 3; instead,thesetaxa are part Chase et al., 1993) and morphology(Rodman, of a grade thatrepresentsthe firstbranchesof a 1991). Thus, bothrbcL and 18S rDNA sequence primarily rosid-dilleniidlade (Fig. 3). data indicatethatthereweretwoindependentoriAsteridaesensu lato. Analysesof all four18S gins of the mustardoil syndrome(see Rodmanet rDNA data sets also revealan expandedAsteridae al., 1993, submitted). lade (Asteridaes.l.) thatagrees closelywiththat recovered byanalysesofrbcLsequences(Olmstead lade. Species of only10 famNitrogen-fixing ilies of angiospermsare knownto formsymbiotic et al., 1992, 1993; Chase et al., 1993). In addition circumscribed Asteridae,this associationswith nitrogen-fixing bacteria in root to theconventionally nodules (Fabaceae, Betulaceae, Casuarinaceae, lade also includesa numberoffamiliesplaced in Coriariaceae,Datiscaceae, Elaeagnaceae, Myrica- Dilleniidae,such as Ericaceae, Clethraceae,Pyroceae, Rhamnaceae, Rosaceae, and Ulmaceae). laceae, Styracaceae, Ebenaceae, Actinidiaceae, These familiesare distributed amongfourofCron- Sarraceniaceae, Fouquieriaceae, Theaceae, and quist's (1981) six subclasses of dicotyledons, im- Primulaceae.Also presentin Asteridaes.l. are NysApiaceae, Araliaceae,and plyingthatmanyofthesefamiliesare onlydistantly saceae, Pittosporaceae, related.Recent phylogenetic analysesof rbcL se- Hydrangeaceae,all membersof Rosidae. In addithatrepresentatives ofall tion,Eucommiaceae,a memberof Hamamelidae, quences reveal,however, ten of these familiesoccur togetherin a single and Byblis,a genus of carnivorousplantsusually lade ("nitrogen-fixing lade"; Soltis et al., 1995). placed in Rosidae,also appearwithinAsteridaes.l. In additionto these ten families,this lade also All analysesalso place an expandedCaryophylliAsteridaes.l. s.l.) within~the containsseveralfamiliesnot knownto formasso- dae (Caryophyllidae ciations with nitrogen-fixing bacteria, including lade, an unexpectedresultthat is discussed in Moraceae,Cannabaceae,Urticaceae,Polygalaceae, moredetailbelow. Within Asteridae s.l., several subclades or Fagaceae, Begoniaceae,and Cucurbitaceae. thatagree,in large part, Analysesof threeof four18S rDNA data sets gradescan be identified (Figs. 1, 2, 4) suggestan alliance oftaxa similarto withsome of the groupsidentifiedin analysesof that revealed by rbcL sequences. This lade in rbcL sequences (Chase et al., 1993; Olmsteadet oftheseis the large partrepresentsa subset of the taxa present al., 1993). Perhapsmostnoteworthy in the rbcL-basednitrogen-fixing lade. The fami- ericalean grade (the asteridIII lade of Chase et lies in the 18S rDNA-basednitrogen-fixing lade al., 1993) observedin all oftheshortest18S rDNA include Betulaceae, Casuarinaceae, Datiscaceae, trees (Figs. 1-4). Otherclades of Olmsteadet al. including Elaeagnaceae, Rhamnaceae, and Ulmaceae, all are also observedto be monophyletic, familiesthatformsymbioticassociationswithni- Dipsacales, Boraginales, Gentianales,Asterales bacteria.Otherfamiliesknowntoform s.l., and Lamiidae. Additionalasteridtaxa should trogen-fixing such associations(i.e., Coriariaceaeand Myrica- be sequenced for18S rDNA to assess morerigorof these groupsand theirinceae) and thatappearedin therbcL-basednitrogen- ouslythe monophyly terrelationships. fixinglade were not analyzedfor18S rDNA se-

quence variation.Also part of the nitrogen-fixing Caryophyllidae sensulato. All analysesof 18S lade retrievedhere are Begoniaceae,Moraceae, rDNA sequences reveala lade composedof NycUrticaceae,and Cucurbitaceae,familiesalso found taginaceae(Mirabilis),Chenopodiaceae(Spinacia), to be partofthisalliance based on analysesofrbcL Phytolaccaceae(Phytolacca),Aizoaceae (Tetragosequences. However,neitherRosaceae nor Faba- nia), and Molluginaceae(Mollugo).These fivefamceae, twofamiliesinvolvedin nitrogen-fixing sym- ilies representCaryophyllales(e.g., Cronquist, bioses,are includedwithinthe18S rDNAnitrogen- 1981), themonophyly ofwhichis supportedin this fixinglade, althoughbothfamiliesare partofthis studyby a jackknifevalue of 58%, as well as by alliance in therbcL-basedtrees(Soltiset al., 1995). numerouslines of morphologicaland molecular

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

24

Annals of the MissouriBotanical Garden

data (e.g.,Rodmanet al., 1984; Rettiget al., 1992). butagainappearnearGunnera.Thus,whereasboth Sister to this lade of Caryophyllales is another rbcL and 18S rDNA searches occasionallyplace stronglysupportedlade comprisingPlumbagina- santaloidsnear Gunnera,analyses of threeof the ceae and Polygonaceaejackknifevalue of 77%): four18S rDNA data sets place santaloidsclose to this groupcollectivelyrepresentsCaryophyllidae Fabaceae and Polygalaceae. (sensu Cronquist,1981). The monophyly of CaryCelastroids. Anothersmall lade revealedin all ophyllidaeis only weaklysupportedby cladistic analyses consistsof Lepuropetalon and Parnassia analysis of morphological, chemical,anatomical, and palynologicalfeatures(Rodmanet al., 1984). (Saxifragaceaes.l.), Brexia (Grossulariaceae),and Analysesof 18S rDNA sequencesalso suggestthat Euonymus(Celastraceae).This lade, labeled cetwofamiliesofcarnivorous plants,Droseraceaeand lastroids(Figs. 1-4), was also recoveredin analyses Nepenthaceae,are sisterto Caryophyllidae, and we of rbcL sequences (Chase et al., 1993; Morgan& referto this entireassemblageas Caryophyllidae Soltis,1993). Althoughthisinitiallyappearsto be an eclecticassemblage(Brexiais a genusofsmall s.l. (Figs. 1-4). spathulatumis the smallest Phylogeneticanalyses of rbcL sequences simi- trees; Lepuropetalon angiosperm), and morphoembryological larly recovereda Caryophyllidae s.l. lade com- terrestrial posed of Caryophyllales, Polygonaceae,Plumbagi- logicaldata also unitethesetaxa (reviewedin Mornaceae, Droseraceae,and Nepenthaceae(Chase et gan & Soltis,1993). The celastroidlade consists al., 1993). One of the broad analysesof rbcL se- of two pairs of genera,each of whichis strongly quences (searchA, Chase et al., 1993) placed Vi- supported:Lepuropetalon-Parnassia (jackknife= taceae and Dilleniaceae withthisexpandedCary- 100%) and Brexia-Euonymus (jackknife= 67%). ophyllidae lade. In the analyses of 18S rDNA These same twopairsofgeneraalso wererevealed sequences,Vitaceae werenotsampled,and Dille- in analysesofrbcLsequences (Chase et al., 1993; niaceae are well removedfromCaryophyllidae s.l. Morgan& Soltis,1993). The anomalousplacementofDilleniaceae nearthe Cunonioids. Bauera and Ceratopetalum(Cumonocots(Figs. 1, 2) is discussedbelow. noniaceae) and Eucryphia(Eucryphiaceae)forma Santaloids. Analysesofall fourdata setsreveal lade witha jackknifevalue of 53%. A close rea monophyletic santaloidlade orSantalales,which lationshipamongthesegeneraalso was revealedby are representedhere by onlythreefamilies(Opi- a cladisticanalysisofmorphological features(Hufliaceae, Santalaceae,and Viscaceae). However,in ford& Dickison, 1992). Bauera, Ceratopetalum, preliminary analysesin whichSantalalesare rep- and Eucryphiaconstitutethe core of a verywell resentedby seven families(Opiliaceae, Santala- supportedlade (jackknifevalue of 89%) labeled ceae, Viscaceae, Eremolepidaceae,Misodendra- cunonioids(Figs. 1-4) thatalso containsCephaloceae, Loranthaceae,and Olacaceae), santaloids taceae, a familyofcarnivorous plants,and Sloanea againforma lade. These sevenfamiliesare widely (Elaeocarpaceae). A close relationshipof Cephalconsideredto forma naturalgroupbased on mor- otaceae to these same representatives of Cunoniphology(e.g., Cronquist,1981) and have been aceae and Eucryphiaceaealso is suggestedbyanalshownto forma lade in previous,smalleranalyses yses ofrbcLsequences (Chase et al., 1993; Morgan of 18S rDNA sequences (Nickrent& Franchina, & Soltis,1993). Sloanea was notrepresented in the 1990; Nickrent& Soltis,1995). broadanalysesof rbcLsequences. Othertaxa that Althoughsantaloids appear monophyletic, the appear closelyallied withCunoniaceae,Eucryphipositionofthis lade variesamongtheanalyses.In aceae, and Cephalotaceaein rbcLanalysesinclude analysesof data sets 1 and 2, santaloidsare sister Tremandraceaeand Oxalidaceae; these families to Polygala and closely related to the legumes. werenotincluded,however, in the 18S rDNAanalAnalysisofdata set 4 again places santaloidswith yses. Polygala and a legume (Pisum),as well as with Othernoteworthy Gunnera.Analysisof data set 3 resultsin an unrelationships.As recentlyreusual placementof santaloidswithseveralpaleo- viewed(Qiu et al., 1993), the placementof Lactoherbs.These findings parallelthoseof Chase et al. ridaceae has been controversial, withrelationships (1993) based on rbcLsequences in whichthe po- to Magnoliales,Laurales, and Piperales all prosition of santaloidsdifferedgreatlybetweenthe posed. Analyses of rbcL sequences suggesteda 476- and 499-taxonsearches.In the former, san- close relationshipof Lactoridaceaeto Aristolochitaloids and Gunneraformthe asteridV lade; in aceae (Chase et al., 1993), and analyses of 18S thelatter,santaloidsare sisterto thecaryophyllids, rDNA sequences similarlysuggestthatthese two

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

25

familiesare sisters(Figs. 1-4), an inference strong- DIFFERENCES AMONGTHE SHORTEST TREES ly supportedby a jackknifevalue of82%. The shortest treesresulting fromanalysesofdata Additional,smallmonophyletic groupsalso merit sets 1 and 2 are essentiallyidentical(Figs. 1, 2) briefdiscussion.Bombacaceae,Tiliaceae,and Maland in turnare verysimilarto thosederivedfrom vaceae (representedby Bombax,Luhea, and Gossearchesof data set 4 (Fig. 4). The mostunusual sypium,respectively)forma stronglysupported topologyresultsfromsearches of data set 3 (Fig. lade jackknifevalue of 78%) in all 18S rDNA 3). For example,thedistinction betweenthemonoanalyses,in agreementwithbothtraditional treat- sulcategradeand theeudicot lade does notoccur ments(all are membersofMalvales)and topologies in the shortesttrees from this analysis,withthe based on rbcL sequences. However, Sloanea monocotspartofa predominantly rosidassemblage (Elaeocarpaceae-Malvales)does not appear with and saxifragoidsappearingas one of the earlyBombacaceae-Tiliaceae-Malvaceaein any of the branchinglineages of angiosperms.The ranuncu18S rDNA trees(Figs. 1-4). As notedabove, this lids are notmonophyletic in treesfromsearchesof malvoid lade sometimesis embeddedwithinthe data set 3, withtwogenera(Hypecoum and Dicennitrogen-fixing lade (Figs. 1, 2), a placementat tra) appearingas sisterto the monocotsand the odds withanalysesbased on rbcLsequences. This remaining ranunculidsappearingas partofa lade unusual placementcould be the resultof insuffi- thatoccupies the unusualpositionof sisterto Ascienttaxondensityin thatmanyoftheclosestpu- teridaes.l. (see Asteridaes.l. Plus, Fig. 3). Howtativerelativesof Malvaleswerenotincludedhere ever,some of our numeroussearchesof data set 3 (e.g., Anacardiaceae, Simaroubaceae, Leitneri- foundtreesonlyone step longerthanthe shortest aceae, Sterculiaceae,Dipterocarpaceae). treesthathave a topologyessentiallyidenticalto On a broaderscale, all 18S rDNAtopologiessug- thatresultingfromanalysisoftheothersmalldata gestthatHamamelidaecomprisea numberofphy- set (4). logenetically distinctlineages.For example,TrochAlthoughsearchesofdata sets 1, 2, and 4 yieldodendraceae, Tetracentraceae,and Platanaceae ed similartopologies,severalweaklysupporteddifappear near the base of the eudicotsin treesde- ferencesalso existamongthe shortesttreesfound. rivedfromsearchesof data sets 1, 2, and 4 (Figs. For example,in treesderivedfromanalysesofdata 1, 2, 4). Eupteleaceaealso appearnearthebase of sets 1 and 2, one groupofpaleoherbs(Aristolochithe eudicots,but as partof the ranunculidlade. aceae, Lactoridaceae)appears withinthe eudicot Threetraditional familiesof Hamamelidae,Hama- lade, ratherthanwithinthemonosulcategrade,as melidaceae,Cercidiphyllaceae, and Daphniphylla- would be expected.In contrast,in trees derived ceae, are partofa well supportedsaxifragoidlade fromthe smaller-datasets (3 and 4), Aristolochi(Figs. 1-4). Still otherfamiliesof Hamamelidae aceae and Lactoridaceaeappear withinthe mono(i.e., Betulaceae, Urticaceae,Moraceae, and Ul- sulcategrade,close to otherfamiliesofpaleoherbs maceae) are partof the nitrogen-fixing lade, and (e.g., Piperaceae, Saururaceae). In addition,the treesobtainedfromanalysesofdata sets 1 Eucommiaceaeare nestedwithintheAsteridaes.l. shortest The pronouncedpolyphylyof Hamamelidaewas and 2 showa morewell definedbreakbetweenthe similarlyrevealedby analysesof rbcL sequences. monosulcategradeand lowereudicotsthando trees Both 18S rDNA and rbcL sequence data suggest fromdata set 4 (compareFigs. 1, 2, and 4). For similarplacementsforrepresentatives of thissub- example,treesresultingfromanalysesof data set 4 place the monosulcatefamiliesCalycanthaceae, class. Topologiesbased on 18S rDNA sequences also Annonqceae,and Lauraceae withProteaceae(see These and reveal the polyphyly of subclass Dilleniidae.Taxa Fig. 4, Lower eudicots/monosulcates). other differences the be result of insufficient may attributed to Dilleniidae appear in severalphylogeneticallywell separatedclades. Paeoniaceae ap- taxondensityin certainportionsofthetree,incompear in the saxifragoidlade, Nepenthaceaeand pleteanalysis,orlack ofsignal(see Caveatsbelow).

Droseraceaeappearin Caryophyllidae s.l., Capparales, Batales, and Violales appear in the glucosinolate lade, and several orders (e.g., Violales, Ebenales, Ericales, Diapensiales,Primulales,and Theales) appear in Asteridaes.l. Otherrepresentativesof Dilleniidae (e.g., Turneraceae,ElaeocarthelargeRosidae paceae) are scatteredthroughout lade.

ANOMALOUSPLACEMENTS

Perhapsthe mostunusual consistentfeatureof the 18S rDNAtreesinvolvestheplacementofCaryophyllidaes.l. withinAsteridaes.l. AlthoughCaryophyllidaes.l. forma well supportedlade, the positionof this lade withinAsteridaes.l. is not strongly supported.Someofthemanysearchescon-

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

26

Annals of the MissouriBotanical Garden

The threesubfamiliesof Fabaceae, Papilionoiducted retrievedtrees onlytwo steps longerthan treesobtainedin whichCaryophyllidae deae, Mimosoideae,and Caesalpinioideae(repretheshortest s.l. are notpartofAsteridaes.l., butappearinstead sented by Pisum and Glycine;Albizia; Bauhinia, althoughpresentin the same small analysesof respectively), withinthe Rosidae lade. Furthermore, 130 dicot18S rDNA sequencesaimedat resolving lade withseveralotherfamiliesin Figures1 and groupin any ofour ofSaxifragaceaes.l. did notplace 2, do notforma monophyletic therelationships a truecase of s.l. withinAsteridaes.l., butin- searches.Ratherthanrepresenting theCaryophyllidae to be embedded discordancebetween18S rDNAand rbcLtrees,this stead showed the caryophyllids withina rosid lade (D. Soltis& Soltis,1997). Al- likelyrepresentseitherthe lowerlimitsof resolus.l. varies tionof18S rDNAsequences(see below)orretrieval thoughthe placementof Caryophyllidae in the broadanalysesofrbcL sequences (Chase et of onlya small sample of all equally mostparsial., 1993), this lade does notappearcloselyrelat- monioustrees(i.e., thestrictconsensusofall shorted to the asteridsin any of the shortesttreesob- est trees,had theybeen found,wouldhave led to tained.The 476-taxonanalysisplaces Caryophyl- the collapse of thispartof the tree).In supportof thatmore conclusionis the observation lidae s.l. withina lade of rosids,whereas the theformer studiesof18S sequencesrep499-taxonanalysisplaces themnear the splitbe- focusedphylogenetic tweenthe clades of highereudicots(i.e., Rosidae resentingonlyRosidae, some ofwhichswappedto Fabasimilarlysuggesta polyphyletic completion, and Asteridaes.l.). Otheranomalousplacementsinclude the posi- ceae; bootstrapanalyses indicatethatthese relationin some analyses(Figs. 1, 2) of one groupof tionshipsare poorlysupported,however(D. Soltis Lac- & Soltis,1997, unpublished). Aristolochiaceae, paleoherbs(Chloranthaceae, The positionof the monocotgenusAcorus(Aratoridaceae)plus Winteraceaeof Magnolialesnear the base ofthe eudicot lade. These taxa oftenare ceae) (Figs. 1, 2) also is unusual.Ratherthanapor primi- pearingwiththe monocots,Acorusappears as an consideredto representearly-branching as it did in a previous angiosperm, in bothanalysesofrbcL sequenc- early-branching tiveangiosperms es (Chase et al., 1993; Qiu et al., 1993) and recent analysis of 64 18S rDNA and rRNA sequences classification schemes (e.g., Cronquist, 1981; (Nickrent& Soltis,1995). OtheranomalousplaceThorne,1992; Takhtajan,1987). Based on phylo- mentsincludethepositionofDilleniaceaenearthe geneticanalysesof rbcL sequences (e.g., Chase et monocots(Figs. 1 and 2) and the unexpectedpoal., 1993; Qiu et al., 1993), forexample,Chloran- sitionofOncidium(Orchidaceae)as a first-branchthaceae, Aristolochiaceae,and Lactoridaceaeare ing monocot. not only because Several taxa are noteworthy part of the monosulcatelade. As noted above, positionsare unusual,but also however,in some of our searches(see Figs. 3 and theirphylogenetic and Lactori- because their phylogeneticpositionvaries from Aristolochiaceae, 4), Chloranthaceae, daceae do appear closer to the base of the angio- search to search.For example,the close relationship of Gunnerato the monocots(Figs. 1 and 2) is spermswithothermonosulcatetaxa. Analysesof data set 4 recovereda lade con- unexpected,butit is notseen in thetreesresulting sistingof Sagittaria (Alismataceae)and Cuscuta fromanalysisof data sets 3 and 4 whereGunnera (Cuscutaceae), placed in the Rosidae lade. In appears in a lade withSantalales,Polygalaceae, ofGunnera fromanalysesof all otherdata sets, and Pisum(Fabaceae). The relationship treesresulting thesegeneraappearwiththe monocotsand Aster- also is uncertainin rbcL topologies,in whichits withtraditional placementvariesfrombeingembeddedwithinAsin agreement idae s.l., respectively, views and with trees based on rbcL sequences teridaes.l. (the476-taxonsearch)to sistergroupof sug- the higherdicots(the499-taxonsearch). (Chase et al., 1993). The unusualrelationship gestedby searchesofdata set 4 likelyresultsfrom the morelimitedtaxon samplingof this data set COMPARISONWITH HAMBY AND ZIMMER (1992) (fewermonocotsare included,forexample,comHambyand Zimmer(1992) used partial18S and attracparedto data sets 1 and 2) and long-branch tion.Sagittaria and Cuscuta haveverylongbranch- 26S rRNA sequences to examine relationships in Fig. 2) in amongland plants.Because theiranalysesinvolved es (e.g.,39 and 65 steps,respectively, taxonsamplingclearlydiffers treesobtained.In analysesofdata only46 angiosperms, all oftheshortest sets 1 and 2, thelongbranchofCuscuta also seems betweenthatand the presentstudy.Nonetheless, to affectthe placementof Ipomoea (Convolvula- briefcomparisonof the topologiesresultingfrom ceae), withbothappearingin Lamiales insteadof the twostudiesis instructive. In mostoftheshortest treesobtainedhere(Figs. Solanales.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

27

1, 2, 4), as well as in the studyof Hambyand sity,the presenceof "older,"erroneous18S rDNA and theoveralllower is allied withthe sequencesin thedata matrix, Zimmer(1992), Ceratophyllum monocots.Both studies also concurin suggesting rate of evolutionof 18S rDNA comparedto rbcL. thatNymphaeaceaeappearnearthebase ofthean- We discuss these potentialfactorsin moredetail giospermradiation.Nymphaeaceaeare the sister below. cannotbe expectAn analysisofthismagnitude in Hambyand Zimgroupto all otherangiosperms in a reasonable parsimony mer's (1992) shortesttrees; however,Amborella- ed to achievemaximum and Schisandraceaewere amountoftime.It is likelythatwe did notfindall ceae, Austrobaileyaceae, not included in thatstudy.In all of our shortest classes of most-parsimonioustrees, despite a (cf.Maddisonet al., 1992) designed trees,Nymphaeaceaefollowthelatterthreefamilies searchstrategy and Illiciaceae as the sistergroupto all remaining to identifymultipleislands (Maddison,1991) of treesexistthat trees,and thateven shorter shortest flowering plants. althoughour betweenthe shortesttreesin were not recovered.Furthermore, Anothersimilarity well over two yearsofcominvolved strategy search Drimys (Winteraof is the placement bothstudies there ceae). Drimysoccupies an unusual phylogenetic putertime,no searchswappedto completion; thatthese treesreprepositionin trees presentedby both Hamby and is no assurance,therefore, optimum.Althoughit Zimmer(1992) and Nickrentand Soltis(1995), ap- sent even a local parsimony pearingas sisterto Glycineand Pisum (Fabaceae), is, ofcourse,impossibleto knowhowfarfromcomDri- pletion any search is when it is truncated,the angiosperm. ratherthanas an early-branching basis mysoccupies an unusualpositionin treesderived search designused here offersan insightful analyses as well,appearingamong forcomparison.Data sets 1 and 2, and 3 and 4 are fromthecurrent fromtheanal- identicalexceptforthe inclusionof twogap charthelowereudicots.In treesresulting ysisofdata set3, DrimysagainappearswithPisum. acters(indels)in data sets 2 and 4, each ofwhich The 18S rDNAsequence ofDrimysexhibitsa num- apparentlyaccounts for only four steps on the notfoundin othermagnoliids. shortesttrees obtained.Thus, the fact that the ber of substitutions treesobtainedin searchesofdata set 2 are ofWin- shortest In an attempt to ascertaintherelationships teraceae,we sequenced twospecies ofDrimys,D. seven stepslongerthanthoseobtainedin searches winteriand D. aromatica,and theyhave identical of data set 1 indicatesthatthe shortesttreesobanothermemberofWin- tainedin oursearchesofdata set 2 are threesteps sequences. Morerecently, thantreesderivedfromsearches has been sequenced for less parsimonious teraceae (Pserudowintera) ofthesearches 18S rDNA (Hoot, unpublished);this sequence is ofdata set 1. A similarcomparison nearlyidenticalto-thesequencesforDrimys.Add- of data sets 3 and 4 revealsthatthe shortesttrees to theanalysisdoes notalterthe fromsearchesofdata set 4 are twostepsless paringPseudowintera unusualpositionofWinteraceae(treesnotshown). simoniousthanthoseobtainedfromdata set 3. We also sampledamongthe largeset ofequally thatexist relationships The unusualphylogenetic Sandersonand Doyle treesfollowing amongthe eudicotsin the shortesttreesofHamby parsimonious and Zimmer(1992) probablyderivefrominsuffi- (1993b). Using treesobtainedin searchesof data cient samplingin thatstudy.The presentanalysis set 1, we examinedthe numberof distinctcompoofthesize ofthesample of eudicotsreveals nents(clades) as a function withits greaterrepresentation muchmorein accordwithrecentclas- oftrees(numberoftrees).We wantedto determine relationships sifications (e.g., Cronquist,1981; Takhtajan,1987) whetherincreasingthe set of treesuncoversnew of parand/orthe rbcLtopologiesof Chase et al. (1993). componentsthatbear on the relationships subThus, the presentstudysuggeststhatmanyof the ticulartaxa or,in contrast,includesdifferent highlyunusual relationshipsseen in Hambyand sets of the componentsthatare essentiallyvarialowtaxondensityrather tions on the same theme (Sanderson& Doyle, Zimmerare likelyto reflect thanan inherentinabilityof 18S rDNA sequences 1993b). We foundthata plotofthenumberofdistinctclades versus the numberof trees sampled to resolverelationships. fora small numberof trees, reachesan asymptote has been thatmostofthe lade diversity suggesting CAVEATS found,despitethe factthatall mostparsimonious of The development in anylarge treeshave notbeen retrieved. are inherent A numberoflimitations analysisoflarge studysuchas this.Severalfactorsmay improvedmethodsofphylogenetic phylogenetic be one ofthecentralissues contributeto the anomalouspositionsof certain data setswillultimately duringthenextseveral reconstruction taxa, including uncertaintyregardingmaximum ofphylogeny insufficient taxonsamplingand/orden- years(see discussionsin Chase et al., 1993; Doyle parsimony,

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

28

Annals of the MissouriBotanical Garden

et al., 1994; Mishler,1994; P. Soltis & Soltis, tythatsurrounds someangiosperm relationships in1997). ferredfromanalysesof 18S rDNA sequences.FurAlthoughthe anomalousrelationships described thermore,because relativelyfew character-state forsome taxa may be unsettling, extremely short changes occur on manyof the branches,a small branchescharacterizemostof the majorclades in amountof homoplasyor errorin the data set may the 18S rDNA trees.The internalsupportformany be sufficient to distortsomerelationships. branchesis verylow,as indicatedbytheparsimony Additionally, someofthe anomalousplacements jackknifeanalysis (Farriset al., 1997). Although could reflectinsufficient and/oruneventaxonsamthemonophyly oftheangiosperms is wellsupported pling.The somewhat uneventaxonomic distribution (jackknifevalue of 100%), fewmajorclades within of the sequences presentlyavailable means that the angiospermshave highjackknifevalues. For some groups,such as Asteridae,and muchofRosexample,largeclades suchas eudicotsand Rosidae idae and Hamamelidae,are relativelywell repredo nothave jackknifevalues above 50%; the sax- sented here, whereasMagnoliidae,the monocots, ifragoidsrepresentthe largest lade havinga high Dilleniidae,Caryophyllidae, and severalordersof jackknifevalue (jackknifevalue of68%). The other Rosidae are under-represented. monophyletic groupswithhighjackknifevaluesare The importance ofsufficient taxondensityis rerelativelysmall,such as cunonioids,Zingiberales, vealed hereby some ofthe differences in topology Malvales, Caryophyllales, Lactoridaceae-Aristolo-observedbetweentreesresultingfromanalysesof chiaceae, and Schisandraceae-Illiciaceae-Austro-the smallerand largerdata sets. Manyof the taxa baileyaceae.Significantly, a numberofmajorclades notpresentin the twosmallerdata sets (3 and 4) seen in all shortesttrees,as well as in treesmany representmonosulcatesand lower eudicots.It is steps longerthanthe mostparsimonioustrees,do this portionof the overalltopologythatshowsthe not have jackknifevalues above 50%, including mostspuriousrelationshipsin trees derivedfrom monocots,glucosinolates,Caryophyllidae s.l., and analysesof these twosmall data sets (the distincAsteridaes.l. The majority ofhighjackknifevalues tion betweenthe monosulcategrade and eudicots correspondto pairs of sistertaxa representing ter- largelybreaksdownin Fig. 3, forexample).In conminalnodes (e.g., Calycanthus-Sassafras, Brexia- trast,themuchmorethoroughly represented AsterEuonymus,Lepuropetalon-Parnassia, Plumbago- idae s.l. and Rosidae clades are littleaffectedby Cocoloba, Helwingia-Phyllonoma,Tragopogon- slightly decreasedrepresentation in data sets3 and Tagetes,Francoa-Greyia,Trochodendron-Tetracen4. These findingslend further supportto the imtron,Menispermum-Tinospora). portanceof sufficient and equal taxondensityin Examination oftreesobtainedfromsearchesthat attempts to inferangiospermphylogeny (e.g., Sytsfoundtreesone ora fewstepslongerthantheshort- ma & Baum, 1996). est treesalso suggestslowinternalsupportforsome One ofthemajorlessonsofthisstudyis thatthe branches.The phylogenetic positionof the mono- 18S rRNAgeneis difficult to sequence,apparently cots appears weaklysupported.In some searches due in largepartto the secondarystructure inherofdata set2, forexample,treesonlyone steplonger ent in the rRNA.As a result,manypublishedsethan the shortesttreesplace the monocotswithin quences are erroneous, somehighlyso, and theexthe eudicots,as part of Rosidae, a positionalso tent of insertionand deletion events has been observedin theshortest treesobtainedfromsearch- greatlyoverestimated. We reiterate thatwhereasthe es ofdata set 3 (Fig. 3). Althoughall ofthestarting totallengthof the aligned 18S rDNA data matrix treesand shortesttreesshowedAmborellaceae,Il- of64 taxa used by Nickrentand Soltis(1995) was liciaceae, Schisandraceae,and Austrobaileyaceae 1853 bp, the lengthof our 228-taxondata matrix to be at the base ofthe angiosperms, one searchof actuallyis shorter,1850 bp. Afterresequencing data set 2 resultedin treestwo steps longerthan over 20 dubious 18S rDNA sequences, we were the shortesttrees and placed these fourfamilies able to removenumerous"false"indelsand reduce near the monocots,withAcorusand Oncidiumas thelengthofthealignedsequences.The greatmathe first-branching angiosperms.Trees two steps jority(70%) ofthe 18S rDNA sequences used here longerthan the shortesttrees showthe Asteridae were generatedvia cycle sequencingfollowedby s.l. embeddedwithinRosidae, ratherthansisterto automatedsequencing,an approachthatprovides thislarge lade. In treestwostepslongerthanthe morereliablerDNA sequences. Additional"older" shortesttreesfoundfordata set 3, Caryophyllidae 18S rDNA sequences shouldbe replacedwithses.l. are not part of Asteridaes.l. but instead are quences generatedvia thisapproach. partofthe largeRosidae lade. The overallslowerrateofevolutionof18S rDNA These fewexamplesillustrate welltheuncertain- comparedto rbcL (see Nickrent& Soltis, 1995)

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1

Soltis et al.

in part,to the widespreadbeliefthat contributed, 18S rDNA sequences wouldnotcontribute greatly in angiosperms. to phylogenetic inference Although thisstudyand otherrecentpapersemploying entire 18S rDNA sequences (e.g., Nickrent& Soltis, 1995; Kron,1996; D. Soltis & Soltis,1997; Rodman et al., submitted; Johnsonet al., unpublished) have dispelled this notion,18S rDNA sequences will, in mostcases, not elucidaterelationships to the degreepossiblewiththemorerapidlyevolving rbcL. In some groupssuch as Orchidaceae,however, 18S rDNA has been foundto evolve faster thanrbcL(Cameronand Chase, unpublished).

CONCLUSIONS

1997

18S RibosomalDNAPhylogeny

29

This study providesgeneral insightsinto the structure and evolutionof the 18S rRNA gene in angiosperms and dispels certain"myths"aboutits evolution.Indels are neitheras commonnor as problematicforalignmentas previouslybelieved. Instead,theyare largelyconfinedto a few,small, specificregionsthatcorrespondto the terminiof certainhelices presentin the proposedsecondary structure for18S rRNA.Whenthesefew,shortareas are eliminatedfromconsideration, alignment of 18S rDNA sequences is straightforward and easily Conaccomplishedby eye across all angiosperms. versely,indelsare rarethroughout mostofthe 18S FUTURE CONSIDERATIONS rRNA gene; whenpresent,theytypicallyinvolvea indels presentin single base pair. Furthermore, These exploratory analysesclearlyillustratethe highlyconservedregionsof the gene may,in fact, phylogenetic potentialof 18S rDNA sequences for be phylogenetically informative, such as the inserat highertax- tion thatunitessaxifragoids elucidatingangiosperm relationships and the deletionthat onomiclevels. Futureattemptsto conductbroad uniteshighereudicots. phylogeneticanalyses of 18S rDNA sequences Initial attemptsto evaluate the impactof secshouldnotonlyadd moretaxa,butshouldalso in- ondarystructureof the 18S rRNA transcripton volve the resequencingof the 18S rRNA gene for phylogenyreconstruction in angiospermssuggest some of thosetaxa forwhicherroneoussequences that both stem and loop regions appear to be are suspected. sourcesofphylogenetic witha slightly information, This studysuggeststhata broad,nuclear-based greaterproportion (58% vs. 42%) of informative phylogenetichypothesisfor the angiospermsis sitesfoundin stemratherthanloop regions.Ofthe achievablevia sequence analysisofthe 18S rRNA stemchangeswe analyzed,only27% destroyeda of 18S sequence data base-pairingcouplet;73% restoredor maintained gene. One of the strengths appears to be the abilityto recognizea suite of stembase pairingand hence are consideredcomgroupsthatappear in all shortesttrees(e.g., glu- pensatory. The mostfrequenttypeof stemchange cosinolateclade, saxifragoids, Caryophyllidae s.l., observed involvedsingle base substitutions that Asteridaes.l., celastroids).This mayreflectsubsti- changedone base-pairingcoupletto another(e.g., tutionsthatoccurredin highlyconservedportions U-G to C-G; U-A to U-G). The highfrequencyof of the 18S rRNA gene duringthe earlydiversifi- compensatory change indicatesthat some downcation of a lineage, resultingin a well-supported weighting of stemcharactersrelativeto loop bases clade. Such substitutions are rare,however, and the maybe warranted in futurebroadanalysesof 18S in someareas ofthe18S rDNA sequences. resultis limitedresolution rDNA topologies.Thus,ourresultsalso clearlyreThe phylogenetic treesobtainedin theseexplorveal that18S rDNA topologieswill,in mostcases, atory,broad analysesof 18S rDNA sequences are not exhibitthe degree of resolutionand internal largelyconcordantwiththoseresultingfromanalsupportpossible withrbcL sequences. Increased yses of rbcL sequences. Areas of generalconcorsamplingof angiospermsfor18S rDNA sequence dance include the presenceof a tricolpateor euanalysisis desirable.However,to achievea nucle- dicot lade, whichin turnincludestwolargeclades ar-based estimateof angiospermphylogenycom- corresponding mostlyto Rosidae and Asteridaes.l., parableto thatrealizedwithrbcL,it probablywill respectively. However,thelatterlade also includes be necessaryto includeall, or portionsof,the26S Caryophyllidae s.l. in 18S rDNA trees,but notin rRNA gene as well. The utilityof portionsof the treesretrieved fromanalysesofrbcLsequences.In 26S geneforinferring has addition, the monocotyledons family-level relationships are monophyletic been demonstratedfor angiosperms(Hamby & (withthe possible exceptionofAcorus)and generZimmer,1992), as well as forothergroupsof or- allyappearwithothertaxahavingmonosulcate polganisms(e.g., Buchheim& Chapman,1991; Chap- len. One of the most noteworthy differences beman & Buchheim,1991; Chapela et al., 1994; Wa- tweenthis studyand thatof Chase et al. (1993) terset al., 1992). concerns the first-branching angiosperms.The

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Annals of the MissouriBotanical Garden

30

plastids:A reviewbased on comparisonsofsmall-subwoody magnoliids Amborellaceae, Illiciaceae, J.Phycol.31: 489consistent- unitribosomalRNA codingregions. Schisandraceae,and Austrobaileyaceae 498. angiospermsand are Bremer,K. 1988. The limitsofaminoacid sequencedata ly appear as first-branching Evolution alwaysfollowedby the paleoherbNymphaeaceae. in angiospermphylogeneticreconstruction. is closelyallied withthe monocots 42: 795-803. Ceratophyllum of and does not appear as sisterto all otherangio- Buchheim,M. A. & R. L. Chapman.1991. Phylogeny the colonialgreenflagellates:A studyof 18S and 26S sperms,as in analysesofrbcLsequences (Chase et rRNAsequence data. BioSystems25: 85-100. al., 1993). Monophyleticgroupsapparentin all , M. Turmel,E. A. Zimmer& R. L. Chapman. 1990. Phylogenyof Chlamydomonas(Chlorophyta) s.l., Asteridaes.l., analysesincludeCaryophyllidae based on cladisticanalysisof nuclear 18S rRNA setaxa, santasaxifragoids,glucosinolate-producing quence data. J. Phycol.26: 689-699. loids, and cunonioids.Otherclades apparentin Bult,C., M. Kallersjd& Y. Suh. 1992. Amplification and and nitrogen-fix- sequencing of 16/18S rDNA fromgel-purifiedtotal mostanalysesincluderanunculids majorclades ingtaxa.Thus,thisanalysisidentifies plantDNA. PI. Molec. Biol. Reporter10: 273-284. ofbryophytes of angiospermsthat are largely consistentwith Capesius, I. 1995. A molecularphylogeny based on the nuclearencoded 18S rRNAgenes. J. PI. fromrbcLanalyses. thoseinferred Physiol.146: 59-63. that18S rDNA Chapela,I. H., S. A. Rehner,T. R. Schultz& U. G. Mueldemonstrates This studyfurther to conduct information sequencescontainsufficient historyof the symbiosisbeler. 1994. Evolutionary antsand theirfungi.Science266: tweenfungus-growing studiesat highertaxonomiclevels in phylogenetic 1691-1694. analyses Additionalphylogenetic the angiosperms. Chapman,R. L. & M. Buchheim.1991. RibosomalRNA shouldbe conductedusinga larger ofangiosperms in the phygene sequences: Analysisand significance 18S rDNA data set thatUnprovestaxonsampling ofgreenalgae. C. R. C. Crit.Rev. logenyand taxonomy for Magnoliidaeand DilTeniidaein particular.In PI. Sci. 10: 343-368. this larger data set, some taxa for Chase, M. W., D. E. Soltis,R. G. Olmstead,D. Morgan, constructing D. H. Les, B. D. Mishler,M. R. Duvall, R. A. Price, which published sequences are available should H. G. Hills, Y.-L. Qiu, K. A. Kron,J. H. Rettig,E. firstbe resequenced. Conti,J. D. Palmer,J. R. Manhart,K. J. Sytsma,H. J. Althoughcomparativesequencingof the entire Michaels,W. J. Kress, K. G. Karol, W. D. Clark,M. 18S rRNA gene holds greatpromiseforretrieving Hedren,B. S. Gaut, R. K. Jansen,K.-J. Kim, C. F. S. H. Strauss,Q.-Y. Wimpee,J.F. Smith,G. R. Furnier, at thefamilylevel and above in theanphylogeny Xiang,G. M. Plunkett,P. S. Soltis,'S. M. Swensen,S. giosperms,this nucleargene will rarelyelucidate E. Williams,P. A. Gadek,C. J. Quinn,L. E. Eguiarte, totheextentposfamilialand genericrelationships E. Golenberg,G. H. Learn,Jr.,S'W. Graham,S. C. H. sible withrbcL(see also Nickrent& Soltis,1995). Barrett,S. Dayanandan& V. A. Albert. 1993. Phylogeneticsof seed plants:An analysisof nucleotideseDue to the slowerrate of evolutionof 18S rDNA quences fromtheplastidgenerbcL.Ann.MissouriBot. comparedto rbcL,it likelywill be necessaryto seGard.80: 528-580. quence the 26S rDNA as well to obtaina nuclear- Chaw,S.-M., H. Long,B.-S. Wang,A. Zharkikh& W.-H. based estimateof phylogenycomparableto that positionofTaxaceae based Li. 1993. The phylogenetic on 18S rRNAsequences.J. Molec. Evol. 37: 624-630. achievedwithrbcL.Lastly,because of the general , H.-M. Sung,H. Long,A. Zharhikh& W.-H.Li. congruenceof 18S rDNA and rbcL topologiesfor positionsoftheconifergenera 1995. The phylogenetic suggeststhat thisstudyconcomitantly angiosperms, from Phyllocladus,and Nageia inferred Amentotaxus, 18S rDNAand rbcLsequencesshouldbe combined 18S rRNAsequences.J. Molec. Evol. 41: 224-230. to providea moreaccurateestimateofangiosperm , A. Zharkikh,H.-M. Sung,T.-C. Lau & W.-H.Li. and seed of gymnosperms 1997. Molecularphylogeny One can anticipatethatothersequences phylogeny. plant evolution:Analysisof 18S rRNA sequences. J. also be (e.g., atpB and 26S rDNA) will ultimately Molec. Evol.: in press. combinedwithrbcL and 18S rDNA sequences to Crane,P. R. 1985. Phylogenetic analysisof seed plants providea largerdata setfromwhichto infera more Ann. MissouriBot. and the originof the angiosperms. phylogeny. completepictureofangiosperm Gard.72: 716-793. Cited Literature of and limitations Bailey,I. W. 1957. The potentialities and claswoodanatomyin the studyof the phylogeny sificationof angiosperms.J. ArnoldArbor.38: 243254. Bakker,F. T.,J. L. Olsen,W.T. Stam& C. Vanden Hoek. New 1994. The Cladophoracomplex (Chlorophyta): viewsbased on 18S rRNAgenesequences.Molec.Phylogeneticsand Evol. 3: 365-382. of D. & L. Medlin. 1995. The phylogeny Bhattacharya,

. 1988. Major clades and relationshipsin the Pp. 218-272 in C. B. Beck (ed"higher"gymnosperms. Columbia itors),Originand EvolutionofGymnosperms. Univ.Press,New York. , E. M. Friis& K. R. Pedersen. 1995. The origin of angiosperms.Nature374: and earlydiversification 27-33. of Cronquist,A. 1968. The Evolutionand Classification Boston. Mifflin, HigherPlants.Houghton, of SystemofClassification . 1981. An Integrated FloweringPlants.ColumbiaUniv.Press,New York. of Curtiss,W. C. & J. N. Vournakis.1984. Quantitation

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

31

base substitutions in eukaryotic 5S rRNA:Selectionfor Field, K. G., G. J. Olsen, D. J. Lane, S. J. Giovannoni, themaintenance ofRNA secondarystructure. M. T Ghiselin,E. C. Raff,N. R. Pace & R. A. Raff. J.Molec. Evol. 20: 351-361. 1988. Molecularphylogeny oftheanimalkingdom. SciDahlgren,R. T. 1980. A revisedsystemofclassification ence 239: 748-753. of the angiosperms. Bot.J. Linn.Soc. 80: 91-124. Forster, H., M. D. Coffey, H. Elwood& M. L. Sogin. 1990. . 1983. Generalaspects of angiospermevolution Sequence analysisofthesmallsubunitribosomalRNAs and macrosystematics. NordicJ. Bot. 3: 119-149. ofthreezoosporicfungiand implications forfungalevoDahlgren,R. T., H. T Clifford & P. F. Yeo. 1985. The lution.Mycologia82: 306-312. Families of the Monocotyledons: Structure, Evolution, Gargas,A., P. T. De Priest& J.W.Taylor.1995. Positions and Taxonomy. Springer-Verlag, New York. of multipleinsertionsin SSU rDNA of lichen-forming Dixon,M. T. & D. M. Hillis. 1993. RibosomalRNA secfungi.Molec. Biol. Evol. 12: 208-218. ondarystructure-Compensatory mutations and impli- Goremykin, V.,V. Bobrova,J.Pahnke,A. Troitsky, A. Ancationsforphylogenetic analysis.Molec.Biol. Evol. 10: tonov& W. Martin.1996. Noncodingsequences from 256-267. the slowlyevolvingchloroplastinvertedrepeatin adDonoghue,M. J. & J. A. Doyle. 1989a. Phylogenetic ditionto rbcLdata do notsupportGnetaleanaffinities analysisof angiosperms and the relationships of Hamof angiosperms. Molec. Biol. Evol. 13: 383-396. amelidae. Pp. 17-45 in P. R. Crane & S. Blackmore Hamby,K. R. & E. A. Zimmer.1992. RibosomalRNA (editors),Evolution,Systematics and Fossil Historyof as a phylogenetic toolin plantsystematics. Pp. 50-91 the Hamamelidae.ClarendonPress,Oxford. in P. S. Soltis,D. E. Soltis& J. J. Doyle (editors),Mo& . 1989b. Phylogenetic studiesof seed lecularSystematics of Plants.Chapmanand Hall, New plantsand angiospermsbased on morphological charYork. acters.Pp. 181-193 in B. Fernholm,K. Bremer& H. Hendricks,L., R. De Baere,Y. Van de Peer,J. Neefs,A. Jornvall (editors),The Hierarchy ofLife:Moleculesand Goris& R. De Wachter.1991. The evolutionary posiin Phylogenetic Morphology Analysis.ElsevierScience tionoftherhodophyte Porphyraumbilicalisand thebaPublishers,Amsterdam. scottiiamongothereukarysidiomycete Leucosporidium otes as deduced fromcompletesequences of small , R. G. Olmstead,J. F. Smith& J. D. Palmer. 1992. Phylogenetic of Dipsacales based relationships ribosomalsubunitRNA. J. Molec. Evol. 32: 167-177. on rbcLsequences.Ann.MissouriBot. Gard.79: 333Hillis, D. M. & M. T Dixon. 1991. RibosomalDNA: 345. Molecularevolutionand phylogenetic inference. Quart. Doyle,J. A. & M. J. Donoghue. 1986. Seed plantphyRev. Biol. 66: 411-453. logenyand the originof the angiosperms:An experi- Hinkle,G., J. K. Wetterer, T. R. Schultz& M. L. Sogin. mentalcladisticapproach.Bot. Rev. 52: 321-431. 1994. Phylogeny oftheattineantfungibased on anal& . 1992. Fossils and seed plantphylogysis of small subunitribosomalRNA gene sequences. 44: 89-106. enyreanalyzed.Brittonia Science 266: 1695-1697. & C. L. Hotton. 1991. Diversification of early Hoot,S. B. & P. R. Crane. 1995. Inter-familial relationangiospermpollenin a cladisticcontext.Pp. 169-195 ships in the Ranunculidaebased Qnmolecularsystemin S. Blackmore& S. H. Barnes (editors),Pollen and atics. P1. Syst.Evol. 9: 119-131. Spores.ClarendonPress,Oxford. , A. Culham& P. R. Crane. 1995. Phylogenetic , M. J. Donoghue& E. A. Zimmer.1994. Interelationships of the Lardizabalaceaeand Sargentodoxgration ,of morphological and ribosomalRNA data on aceae: Chloroplastand nuclear DNA sequence evitheoriginofangiosperms. Ann.MissouriBot.Gard.81: dence. P1. Syst.Evol. 9: 195-199. 419-450. Hufford,L. & W. C. Dickison. 1992. A phylogenetic analysisofCunoniaceae.Syst.Bot. 17: 181-200. , C. L. Hotton& J. V. Ward. 1990. EarlyCretaceous tetrads,zonasulculatepollen,and Winteraceae. Huss, V. A. R. & M. L. Sogin. 1990. Phylogenetic posiII. Cladisticanalysisand implications. Amer.J.Bot.77: tion of some Chlorellaspecies withinthe Chlorococ1558-1568. cales based upon completesmall-subunitribosomal Duvall,M. R., M. T. Clegg,M. W. Chase, W. D. Clark,W. RNA sequences.J. Molec. Evol. 31: 432-442. J. Kress,H. G. Hills, L. E. Eguiarte,J. F. Smith,B. S. Kantz,T. S., E. C. Theriot,E. A. Zimmer& R. L. ChapGaut,E. A. Zimmer& G. H. Learn,Jr. 1993. Phyloman. 1990. The Pleurastrophyceae and Micromonaforthe monocotyledons genetichypotheses constructed dophyceae: Cladistic analysis of nuclear rRNA sefromrbcLsequence data. Ann.MissouriBot. Gard.80: quence data. J. Phycol.26: 711-721. 607-619. Kranz,H. D., D. Miks,M.-L. Siegler,I. Capesius,Ch. W. Eckenrode,V. K., J. Arnold& R. B. Meagher. 1985. Sensen& V. A. Huss. 1995. The originofland plants: Comparisonofthenucleotidesequenceofsoybean18S Phylogenetic relationships betweenCharophytes, BryorDNA with the sequences of other small-subunit phytes,and vascular plants inferredfromcomplete rRNAs.J. Molec. Evol. 21: 259-269. small subunitribosomalRNA gene sequences.J. MoEmbley,M. T., R. P. Hirt& D. M. Williams. 1994. Biolec. Evol. 41: 74-84. at themolecularlevel:The domains,kingdoms Kress,W. J.,W. J. Hahn, T. M. Evans & E. A. Zimmer. diversity and phylaof life.Philos.Trans.,Ser. B 345: 21-31: 1995. Unraveling theevolutionary radiationofthefamEndress,P. K. 1994. Evolutionary ilies of the Zingiberanaeusingmorphological, aspects of the floral molecuin Ceratophyllum. P1. Syst.Evol. 8: 175-183. structure lar,and fossilcharacters. Amer.J.Bot.82 (suppl.):142. & R. Honegger.1980. The pollenoftheAustro- Kron,K. A. 1996. Phylogenetic of Emperelationships baileyaceae and its phylogenetic significance.Grana traceae,Epacridaceae,and Ericaceae: Evidence from 19: 177-182. nuclearribosomal18S sequence data. Ann. Bot. (LonFarris,J. S., V. A. Albert,M. Kallersjo,D. Lipscomb& don) 77: 293-303. A. G. Kluge. 1997. Parsimony jackknifing & M. W. Chase. 1993. Systematics outperforms ofEricaceae, neighbor-joining. Cladistics11: in press. Empetraceae,Epacridaceae and related taxa based

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

32

Annals of the MissouriBotanical Garden

uponrbcLsequencedata. Ann.MissouriBot.Gard.80: 735-741. of the CeraLes, D. H. 1988. The originsand affinities tophyllaceae.Taxon37: 326-345. , D. K. Garvin& C. F. Wimpee. 1991. Molecular historyof ancient aquatic angiosperms. evolutionary Proc. Natl.Acad. U.S.A. 88: 10119-10123. Loconte,H. & D. W. Stevenson.1991. Cladisticsof the Magnoliidae.Cladistics7: 267-296. ofmulMaddison,D. R. 1991. Discoveryand importance trees.Syst.Zool. 40: tipleislandsofmost-parsimonious 315-328. 1992. Geographic M. Ruvolo& D. L. Swofford. evDNA: Phylogenetic originsof humanmitochondrial idence fromcontrolregionsequences. Syst.Biol. 41: 111-124. Martin,P. G. & J. M. Dowd. 1991. Studiesofangiosperm usingproteinsequences.Ann.MissouriBot. phylogeny Gard.78: 296-337. Mindell,D. P. & R. L. Honeycutt.1990. RibosomalRNA applications. Evolutionand phylogenetic in vertebrates: Ann. Rev. Ecol. Syst.21: 541-566. Mishler,B. D. 1994. Cladisticanalysisofmolecularand 94: 143data.Amer.J.Phys.Anthropol. morphological 156. , K. Bremer,C. J. Humphries& S. P. Churchill. 1988. The use of nucleic acid sequence data in phyTaxon37: 391-395. logeneticreconstruction. , L. A. Lewis,M. A. Buchheim,K. S. Renzaglia, D. J. Garbary,C. F. Delwiche,F. W. Zechman,T S. relationKantz& R. L. Chapman. 1994. Phylogenetic Ann.Misshipsofthe"greenalgae" and "bryophytes." souriBot. Gard.81: 451-483. relaMorgan,D. R. & D. E. Soltis. 1993. Phylogenetic tionshipsamongmembersof Saxifragaceaesensu lato based on rbcLsequence data. Ann.MissouriBot.Gard. 80: 631-660. Nairn,C. J. & R. J.Ferl. 1988. The completenucleotide ribosomalRNA coding sequence of the small-subunit imregionforthe cycad Zamia pumila: Phylogenetic plications.J. Molec. Evol. 27: 133-141. D. L. 1994. Fromfieldto film:Rapid sequencNickrent, ing methodsfor field collected plant species. BioTechniques16: 470-474. relation& C. R. Franchina.1990. Phylogenetic shipsoftheSantalalesand relatives.J.Molec.Evol. 31: 294-301. & D. E. Soltis. 1995. A comparisonof angiospermphylogeniesfromnuclear 18S rDNA and rbcL sequences. Ann. MissouriBot. Gard.82: 208-234. & E. M. Starr. 1994. High ratesof nucleotide (18S) rDNA from in nuclearsmall-subunit substitution plants.J. Molec. Evol. 39: 62holoparasiticflowering 70. Nixon,K. C., W. L. Crepet,D. Stevenson& E. M. Friis. Ann. 1994. A reevaluationof seed plant phylogeny. MissouriBot. Gard.81: 484-533. of ribosomalRNA. Annual Noller,H. F. 1984. Structure Rev. Biochem.53: 119-162. Olmstead,R. G., H. J. Michaels, K. M. Scott& J. D. of the Asteridaeand idenPalmer. 1992. Monophyly fromDNA seoftheirmajorlineagesinferred tification quences of rbcL. Ann. MissouriBot. Gard. 79: 249265. , B. Bremer,K. M. Scott& J.D. Palmer.1993. A analysisoftheAsteridaesensulatobased on parsimony rbcLsequences.Ann.MissouriBot.Gard.80: 700-722.

Olsen, G. J. 1987. Earliest phylogeneticbranchings: with treesinferred rRNA-basedevolutionary Comparing varioustechniques.Cold SpringHarborSymp.Quant. Biol. 52: 825-837. Olsen, J. L., W. T Stam,S. Berger& D. Menzel. 1994. 18S rDNA and evolutionin the Dasycladales (ChloModernlivingfossils.J.Phycol.30: 729-744. rophyta): Qiu, Y.-L., M. W. Chase, D. H. Les & C. R. Parks. 1993. of the Magnoliidae:Cladistic Molecularphylogenetics analyses of nucleotidesequences of the plastidgene rbcL.Ann. MissouriBot. Gard.80: 587-606. Ragan,M. A., C. J. Bird,E. L. Rice, R. R. Gutell,C. A. Murphy& R. K. Singh. 1994. A molecularphylogeny based on nuclear ofthe marinered algae (Rhodophyta) rRNAgene.Proc.Natl.Acad. U.S.A. 91: small-subunit 7276-7280. J. & I. Capesius. 1990. Nucleotidesequence Rathgeber, of the intergenicspacer and the 18S ribosomalRNA genefrommustard(Sinapisalba). Nucl. Acids Res. 18: 1288. Rettig,J. H., H. D. Wilson& J. R. Manhart.1992. Physequencedata.Taxon logenyoftheCaryophyllales-gene 41: 201-209. Rice, K. A., M. J. Donoghue& R. G. Olmstead.1997. A reanalysisof the large rbcL data set: Implicationsfor studies.Syst.Biol. in press. futurephylogenetic Rodman,J. E. 1991. A taxonomicanalysisof glucosiplants,Part 2: Cladistics.Syst. Bot. nolate-producing 16: 619-629. , P. S. Soltis,D. E. Soltis & K. J. Sytsma.Dual fromcongruinferred originofmustardoil biosynthesis ent nuclearand plastidgene phylogenies.Proc. Natl. Acad. Sci. U.S.A., submitted. J.U. McClammer, , M. K. Oliver,R. R. Nakamura, Jr.& A. H. Bledsoe. 1984. A taxonomicanalysisand Syst. Bot. 9: revisedclassificationof Centros-permae. 297-323. , R. A. Price,K. Karol,E. Conti,K. J. Sytsma& J. D. Palmer. 1993. Nucleotidesequencesof therbcL oil plants.Ann. ofmustard gene indicatethemonophyly MissouriBot. Gard.80: 686-699. of Sanderson,M. J.& J.J. Doyle. 1993a. Reconstruction organismalgene phylogeniesfromdata on multigene families:Concertedevolution,homoplasy,and confidence. Syst.Biol. 42: 4-17. in . 1993b. Phylogenetic relationships & NorthAmericanAstragalus(Fabaceae) based on chlosite variation.Syst. Bot. 18: roplastDNA restriction 395-408. & M. J.Donoghue.1994. Shiftsin diversification Science264: 1590ratewiththeoriginofangiosperms. 1593. Savard,L. & M. Lalonde. 1991. Sequence of 18S rDNA Al/us glutinosa(Betulaceae).P1. Molec. ofactinorhizal Biol. 16: 725-728. Schlegel,M., H. J. Elwood& M. L. Sogin. 1991. Molecciliates:Sequence ofthe ular evolutionin hypotrichous small subunitribosomalRNA genes fromOnychodro(Oxytriand Oxytricha granulifera musquadricornutus Ciliophora).J. Molec. Evol. 32: chidae, Hypotrichida, 64-69. Senecoff, J. F. & R. B. Meagher.1992. In vivoanalysis of plant RNA structure:Soybean 18S ribosomaland carboxylasesmall subunit ribulose-1,5-bisphosphate RNAs. P1. Molec. Biol. 18: 219-234. Smith,A. B. 1989. RNA sequence data in phylogenetic

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

33

of flowering Cla. 1980. Outlineof the classification reconstruction: Testingthe limitsofits resolution. Bot. Rev. 46: 225-359. plants(Magnoliophyta). distics5: 321-344. Academyof . 1987. Systemof Magnoliophyta. Sogin, M. L., H. J. Elwood & J. H. Gunderson.1986. Sciences U.S.S.R., Leningrad. Evolutionarydiversityof eukaryoticsmall-subunit oftheflowrRNAgenes.Proc. Natl.Acad. U.S.A. 83: 1383-1387. Thorne,R. F. 1992. An updatedclassification eringplants.Aliso 13: 365-389. relationSoltis,D. E. & P. S. Soltis. 1997. Phylogenetic I., U. Stocker& P. Gruendler.1989. Nucleotide of Unfried, shipsamongSaxifragaceaesensu lato:A comparison thalsequence ofthe 18S rRNAgene fromArabidopsis topologiesbased in 18S rDNA and rbcL sequences. iana ColO. Nucl. Acids. Res. 17: 7513. Amer.J. Bot. in press. relationships , D. R. Morgan,S. M. Swensen,B. C. Wada, H. & N. Satoh. 1994. Phylogenetic I as inferred from amongextantclasses of echinoderms, Mullin,J. M. Dowd & P. G. Martin.1995. Chloroplast desequencesof18S rDNA,coincidewithrelationships gene sequence data suggesta singleoriginof the preduced fromthe fossilrecord.J. Molec. Evol. 38: 41dispositionfor symbioticnitrogenfixationin angio49. sperms.Proc. Natl.Acad. U.S.A. 92: 2647-2651. Wainright, P. O., 0. Hinkle,M. L. Sogin& S. K. Stickel. analysis Soltis,P. S. & D. E. Soltis. 1997. Phylogenetic originsof the Metazoa:An evolu1993. Monophyletic of largemoleculardata sets. Bol. Soc. Bot. Mexico.In tionarylinkwithfungi.Science 260: 340-342. press. Waters,D. A., M. A. Buchheim,R. A. Dewey & R. L. Plants:EvolutionAbove Stebbins,G. L. 1974. Flowering inferences ofthephylogChapman. 1992. Preliminary the Species Level. HarvardUniv.Press,New York. fromnuclear-encoded ribosomalRNA enyofbryophytes Swann,E. C. & J. W.Taylor.1993. Highertaxaofbasidsequences.Amer.J. Bot. 79: 459-466. iomycetes:An 18S rRNA gene perspective.Mycologia Wheeler,W. C. & R. L. Honeycutt.1988. Paired se85: 923-936. and in ribosomalRNAs: Evolutionary quence difference Molec. Biol. Evol. 5: 90-96. implications. phylogenetic D. L. 1993. PAUP-Phylogenetic AnalysisUsSwofford, Il- Woese,C. R. 1987. Bacterialevolution.Microbiol.Rev. version3.1. CenterforBiodiversity, ing Parsimony, 51: 221-271. linoisNaturalHistorySurvey,Champaign,Illinois. J. & V. A. Erdmann.1986. Cladisticanalysisof & G. J. Olsen. 1990. Phylogeny reconstruction. Wolters, struc5S rRNAand 16S rRNAsecondaryand primary Pp. 411-501 in D. Hillis & C. Moritz(editors),Molecand theirrelationto ture-The evolutionof eukaryotes ular Systematics. Sinauer,Sunderland,Massachusetts. J. Molec. Evol. 24: 152-166. Archaebacteria. Sytsma,K. J. & D. A. Baum. 1996. MolecularphylogevesD. A. 1981. Are theangiosperms primitively of the angiosperms.Pp. Young, nies and the diversification selless? Syst.Bot. 6: 313-330. 314-340 in D. W.Taylor& L. J.Hickey(editors),Flow- Young,D. J. & L. Watson. 1970. The classification of Chapman eringPlantOrigin,Evolution,and Phylogeny. A studyof the upperlevels ofhierarchy. dicotyledons: and Hall, New York. Austral.J. Bot. 18: 387-433. Takaiwa,F., K. Oono & M. Sugiura. 1984. The complete Zimmer, E. A., R. K. Hamby,M. L. Arnold,D. A. Leblanc nucleotidesequence of a rice 17S rRNA gene. Nucl. & E. C. Theriot.1989. Ribos6malRNA phylogenies Acids. Res. 12: 5441-5448. and flowering plantevolution.Pp. 205-214 in B. FernTakhtajan,A. 1969. FloweringPlants: Originand Dis(editors),The Hierarchy holm,K. Bremer& J.Jornvall of Life.ElsevierScience Publishers,Amsterdam. persal.Oliver& Boyd,Edinburgh.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Annals of the MissouriBotanical Garden

34

1A

EUDICOTS -

100

I

Calycanthus Sassafras Tacca Bowiea

Chiorophytum Allium Sagittaria Hippeastrum

Eucharis Gladiolus

Isophysis Xanthorrhoea Veitchia Cyanella Helmholtzia Maranta

Zingiber Costus Canna Heliconia

61

z 0

X

0

Z

Musa Glomeropitcairnia Cyperus

Sparganium ______________

Elasis>

m

Coichicum

Calla Oncidium

Ceratophyllum

Dillenia Gunnera LIsolona 100

Acorus

-

Houttuynia

a

~~~~~Nymphaea

Peperomia ;r Saururus C Amborella Schisandra 94 Austrobaileya Iificium 0 Gnetum gnemon C nodiflorum Gnetum ___________________________________Gnetum urens Ephedratorreyanac U Ephedrasinica analysisof223 species phylogenetic fromtheexploratory treesresulting Figure1. Strictconsensusof5294 shortest jackknife ofangiosperms. Each oftheshortest treeshas a lengthof3923 steps,CI = 0.235, and RI = 0.535. Parsimony values (Farriset al., 1997) of 50 or above (based on 1000 replicates)are givenabove nodes (runtime= 949 sec.). Because of its size, the treehas been brokenintofourparts(1A, 1B, IC, and ID).

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

35

1B

ASTERIDAES. L. ROSIDAE l ||

75

82

96

>

Hedyosmum Aristolochia

a

~4(LAsarum

M m

Saruma L Lactorisw

;

Drimys Sabia Trochodendron

Cl)

Sargentodoxa Ranunculus

x

>

m c

Xanthorhiza

z

)

Tinospora aulophyllum

5

C,

E Tetracentron Akebia-Coptis

7 90

Menispermum C

Podophyllum

____________________ 97

I 77_C

0 m

Euptelea

|,

-i

Dicentra Hypecoum -P~atanus Nelumbo Knightia Placospermum

MONOSULCATE GRADE FigurelB.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

36

Annals of the MissouriBotanical Garden

Tetracarpaea Haloragis Penthorum Paeonia Itea Pterostemon Crassula

66 92

10

C > X -i

Sedum>

74

C)

Dudleya Kalanchoe

0

Heuchera Boykinia

68

____

____

____

___

C

Saxifraga

Ribes

Daphniphyllum

79

79

c

Cercidiphyllum

~~Liquidambar

Altingia Zelkova Celtis

Ceanothus

Morus Symbegonia

10 I

1C

S.L. ASTERIDAE

54,

h

~~~~~~ ~~~~~~Begonia

X

Bombax Gossypium Luhea Elaeagnus

Z

g -

I

)

m Z

4

Tetrameles DatiscaZ

Z

C

Abobra Francoa

98

Greyia

Alnus

Casuarina Crossosoma Geranium

Muntingia Sloanea> Bauera z 0 ~~~~Ceratopetalum Z Eucryphia Ce~phalotus Koe/reuteria

55 89

Guaiacum,

99

Pisum

Crarkia

Poly~gala

66 {

T~apilia

o Osyris Punicau

=

100 70

> Z

a

AAlbizia | O Bauhinia r Lepuropetalon >

Parnassia Brexia

67

Euonymus

00

C 0

Acer Citrus U Malpighia Turnera Linum Chrysobalanus Licania

~~~~~Euphorbia

70

Spiraea

_____________________________

Drypetes Floerkea Koeberfinia Arabidopsis Brassica Cleome Batis Moig

62 _________________________~__

0

_-

zA o

Momaar

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

)

m

37

Soltiset al. 18S RibosomalDNAPhylogeny

Volume84, Number1 1997

>i1D

Drosera

Nepenthes

E<

0

Plumbago

Cocoloba Mollugo

58

r

Tetragonia Phytolacca

Spinacia Mirabilis Manilkara Fouquieria

77

DiapDensiam

> =: 90

9gcon

~Galax Impatiens Prmula

m ;

Gilia Sarracenia

m

Pogifia

r

Clethra Vaccinium

Z

Arctostaphylos

88

5 m

)

Styrax

,

lonotropa Pyrola

Actinidia DioispyrTos Camellia ESycos Philadelphus

>

Hydrangea

~~Pittosporum

71

71 77

PHedera

Lomatium> Escallonia Eremosyne

rOymphoricarpos 60 58 P

0~~~~590 P _ _ _ _ _ _ __

_ _ _ _ _ _ _ _

_______

______

_ _ _ _ _ _ _ _ ______

V

~~Dipsacus Lonicera Roussea | Campanula achysacy Lobelia

Tragopogon

Tagetes Eucommia

Aucuba

Heiwingia

Phyilonoma

Olea Phytachys

/

m

Berbeia Pedicutari

Aecyanthus motec La mpu Phacelia Bourreria Tabernaemontana Mitchella ROSIDAE Brunfelsia Montinia

57

____

____

____

____

____

____

Vahfia

Berzelia Camptotheca

ROSIDAF

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Annals of the MissouriBotanical Garden

38

A 3

A

~~4

r'-

~~~~~9 Calycanthus E19Sassafras

6

26

Tacca

7 a 2

4 4

3

212

~Gladiolus -

6

5

17

1

40

25 2913

t__

5

6

and R

Gn23

0

Aw

i

9

n

40 29

n

-|

4 Oncidium Houttuynia e 17? 9Ceratophyllum 22 Dillenia

E Gunnera 5_Mlu

32 ~~~~~Acorus

r

-

Nymphaea r~~~~~~~12 ~ ~~ Houttuynia0 Peperomia M Saururus w

4

2

0

m

GColchicum

31 1

oMusa Glomeropitcairnia

~~~~Cyperus 17 Sparganium Elasis

21

69

a

Zingiber

7

11

6

0

Helmholtzia

8 Maranta

6

9

I~~~sophysis

~~Xanthorrhoea

18 Veitchia F18 ~~~~~~~~~~Cyanella

4

4

31

Sagittaria Hippeastrum

9

_

Chlorophytum

7

2

7

1Heiconia

Bowiea

85

7

4

L

2A

EUDICOTS

4

t

h

i

o

3

tts fAmborella - Schisandra

Austrobaileya

16 -fcium

Gnetumgnemon a ~~~~~1 nodifiorum L9 Gneturn c sinica 2_-E~ohedra torreyana Ephedra

fromtheexploratory analysisof223 species ofangiophylogenetic Figure2. One of2508 shortesttreesresulting sperms;twoindelswereincludedin theanalyses.Each of theshortesttreeshas a lengthof 3930 steps,CI = 0.235, and RI = 0.540. Arrowsindicatenodes notpresentin the strictconsensusof all shortesttrees.The lettersA and B indicatetheoccurrenceoftheindelsdescribedin Table 2. Because ofits size, thetreehas been brokenintofourparts (2A, 2B, 2C, and 2D).

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

A

3

Soltis et al. 18S Ribosomal DNA Phylogeny

2

1

ROSIDAE

17

4

2

i1 4~~

3

Sabia

1

15 4 Sau a

4 6

2

Hedyosmum 10 Aristoloch i

15

>

0

5 4erceo

10

2uCoptis

to

Akebia u

-3~~~~~~ Sargentodoxam 62Ranunculus

~

l

2

>

C Xanthorhiza Dr L ~ O~~~~~~~ frodendronu ~~~~~~~ imyns 36 cMenispermum ~~ ~~~21 4 -C _~~~~~~~~~ Tinospora 5 2Caulophyllum 5 1

~~~~~~~~~~~0Euptelea

3 7

2B

ASTERIDAE S. L.

3

5

39

GA 00

-Hi

:Podophyllum

11

Dicentra C13~~Hypecoum Platanus r3~~~~~~Knightia 2,Placospermum

MONOSULCATEGRADE Figure2B.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Annals of the MissouriBotanical Garden

40

18

Tetracarpaea

~

2

Haloragis

Penthorum Paeonia Itea

6

6

Kalanchoe

C) O

Dudleya

0

Daphniphyllum Cercidiphyllum

2

1

v

Sedumn

1 Heuchera 0 13Boykinia Saxifraga 14 Ribes

1

2

B

><

-n,

Crassula

~2{2

1

2

~~

1

3

t' >

-~Pterostemon

6 13

7

2C

ASTERIDAE S. L.

2

Liquidambar

0

E -

13 1,

Altingia Zelkova

Z

Ceanothus Elaeagnus

15

2

0

Francoa

1

C)

Greyia Tetrameles

23

3

m

z 3

f~~~~~~~~~~~3

~

s

16Bgoi Symb<

~

1

7

4

L

_3

7

CIarkia Pun/ca ~Polygala Opilia

19

Osynis

3

B

8

Bauhinia e Lepuropetalon Parnassia O 7Brexia

0

O

~~~~~~~~5 19Malpighia

4

266Chrysobalanus

Li~~~~~~~~~~cania

7 9

13

0

mus Citrus

3

3

9 16 3 2

3

6

36

0

4u11

43r

zi

~~~~~~~~~~Abibzia ~

2

2

0

C

Glycine

3 30

~~~z

Pisum

4

3

Ceratopetalum |

Eucryphia 10 Cephalotus Guaiacum 4 14Koelreuteria

12

110 3

>

Muntingia

l

9

0

(I

Geranium

7

2

2

X

~~~Casuarina SLoaeam Crossosoma

11 1

S5

1

1

fegonia

is Morlus ~ ~~~~ Luhiea

~ ~ ~ ~ ~ ~~~~~0

4~~~~~~~3

10Euphorbia

7

Prunus

~~~~~~~Photinia ~~~~~~Spiraea

~~~~~~~9 ~ Drypetes Troaeoum 4

10

15 9 8

18

Floerkea Koeberlinia Arab/dopsis Brassica

Cleome ii Bat/s Moringa Carnca 26

0

(A z -

(

LOWEREUDICOTS Figure2C.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

3

5 1010

6

6

7

.

7

2

DroSera

Nepenthes 14 -Plumbago 1' ' Cocoloba 5 - Mollugo Tetragonia 3 17Phytolacca

10 10

r

2

6

1212 12

2 2

<

Primula Cobaea>

Acanthogilia

6 -

4 510

~~~~~2> 8 l(

2 2

2

8

21

2

fl2 15

2 1

12 13

5

:

~~~~~Gilia ~ ~~~~~ Sarracenia Clethra

~~~Arctostaphylos Vaccinium Styrax

Cyrilla onotropa

r >m Z

C) ; m

~~~~~~~~~~~Actinidia Diospyros Camellia Symplocos Philadelphus

2

1

1

Corokia 10Symphoricarpos Lonicera Roussea

10d 1r10

Campanula

2

Tagetes

12 8

13

4

4

6 4

18 2

2

2

41

2 3

3

12 3

~~~s L

73

> (I

Helwingia Phyllonoma

~~~~~~~~~lea> ~ O~~~ Pachystachys

m

Lamium

15

2

2

Lobelia Tragopogon Eucommia

20 4

13

0

Eremosyner

'-L

18

m

Hedera Lomatium Escallonia

4

1

1

>

14Pittosporum

7'3

9

Buddleja

Parmentiera Pediculanst Orthocarpus

h ~~~Cuschuta 8Aeschynanthus Phacelia Bourreria Tabemaemontana Mitchella 8 Brunfelsia Montinia 8 8

Berzelia

Camptotheca

ROSIDAE

LOWEREUDICOTS Figure2D.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

0 m

klHydrangea

79Garrya

Z-

i 2D

Impatiensm

31

2 4k

4

.

Mirabilis Manilkara Fouquieria Diapensia Galax

3

3

2 2

c

0 > E5 0

~~~~~~~~SpinaciaF

_

2

41

Soltis et al. 18S Ribosomal DNA Phylogeny

Volume 84, Number1 1997

42

Annals of the MissouriBotanical Garden

3A 3

ASTERIDAES. L.

1

PLUS

ROSIDAE PLUS 22

Hedyosmum

16

4k

42

Drimys

4

Polygala

S

>

isu Nelumbo 3

|

Malpighia <

0

7

12Aristolochia

Asarum Lactoris -~~~26

12Houttuynia

1 6

_

~1 6

4

6

11

13

6Sedum 67 67

0Ribes -- - Itea Pterostemon 5 6

Schisandra Austroballeya

13

Illicium

20

16

1 9

38

2 2

Crassula

~~~~~~~~~~~Dudleya Nymphaea

31 31

37

rn

I Saururus A Peperomia Tetracarpaea Haloragis Penthorum Paeonia Liquidambar (A Altingia1 x Daphniphyllum > 1 Cercidiphyllum C) Heuchera 1 Boykinia 4 Saxifraga

10 18 12 7 7 2 1

1-2

1

-

o

Amborella Gnetum gnemon o Gnetum nodiflorum urens Gnetum sinica Ephedra c Ephedratorreyana C

analysisof 194 species ofangiophylogenetic Figure3. One of8897 shortesttreesresultingfromthe exploratory sperms.Each of the shortesttreeshas a lengthof 3501 steps,CI = 0.249, and RI = 0.531. Arrowsindicatenodes notpresentin the strictconsensusof all shortesttrees.Because of its size, the treehas been brokenintothreeparts (3A, 3B, and 3C).

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Volume 84, Number1 1997

Soltis et al. 18S Ribosomal DNA Phylogeny

4

6 2 5

9

4

20 Hippeastrum - Gladiolus

Xanthorrhoea 4Veitchia

3

1

r

_42

k2

_

5 5

610 6L~

26

-

12

~~~9 2626 1

4

O

CMaranta

z

10 Zingiber ECanna Heficonia

0

~~18 Musa

3Glomeropitcairnia z4Oncidum

0

|1

22

Cyanella Helmholtzia

4 5

9

21 24

1

3B

30 Tacca Colchicum 7Bowiea Chiorophytum

2

3~~~~~

43

Zea Oryza

Cyperus

EucyElasis ~~~~~~Oncidium

Calla Sagittana - Dicentra 1 Hypecoum

E

14

Tum

20

12

=

LiBauhnia

6~~~411

3

5

4~~

Khoblinia

~~~~~8 2 Mhyoraaus

PArnasia Gauariacu Brexiia AcFlerke ErLBopCaelum

25 2

Aj

s~

39

~~~~~~~~~~~~~~~~~Crpaohsssm 10 _~~~~~~~~~~~~~~~ A2Zlbkzva ~~~~~~~~~~1 Cer4$ _~~~~~~~~ 44 2

raPiea

i

O

atopetaiu 8 Gossyium Geranium

C~~~~~~~~~~~~~~~~~rosseosoma

12

3

Elaeagnus

Ak. T~~~~~~~~~~~~~~~~Cetrmeles D~~~~~~~~~~~~~ STaERIAESL.PU atisca

3

442

TAXA

KoeeCasurfinia

29 8

3

0

15 AFranopa

10

144

LepralonB

c

Barassc

2

?LoLU L~~~~~~~SepRIDeaE

Figure3B.~~~~~~~~~~~~~~~~~~~Bei

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Annalsofthe MissouriBotanicalGarden

44

ROSIDAEPLUS 30 Ceratophyllum i

S

4 4

3

19

6

2

Gunnera

CalyCanthUS Sassafras Mkilua Knightia

Akebia

3

Ranunculus

Xanthorhiza

2

Menispermum Caulophyllum

3

2

4 Euptelea

7

Z z

c

C U

Trochodendron

Tetracentron

Platanus Sabia Buxus

17 9

33 Drosera> Nepenthes

5

3

W5w 10

r|6

1Plumbago Cocoloba

7

Mollugo

2

Manilkara

Fouquieria 3Diapensia

3

2

06

29 LomGalax

3

m

Impatiens

Primila 8 69

>

GiiahlCobaea

Acanthogilia

12 Sarracenia

-

4 44

3

3

m

Spinacia Mirabilis

10

441

t;

CTetragonia Phytolacca

4

1

<

o

5

Actinidia

4

Arctostaphylos Vaccinium Symplocos

11

Z

13 Camellia ~~Philadeiphus> Hydrangea(I 7 Gan~ya 44 6 4 Pittosporum Hedera 1 7 3_ 440 12 ~~~~~Lomatium> ~~~~~Corokia 44 Symphoricarpos 5

2

Escalionia Eremosyne Roussea 3~ Campanula 2 Lobelni>

-- n 1 1

20 Tagetes r= ~Aucuba

4

4

13 4k

2

-

m

E12 Hwngia

Phyllonoma Olea

m

14 ~~~~~~~Pachystachys

Byblis Lamium

~~~~~Parmentiera

13 2

2

2 3

10

12

8 Buddleja 2 8PFhacelia T57

emaemontana Mitchella Cuscuta Brunfelsia Montinia Vahlia

19Aeschynanthus

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

m

i-r

45

Soltis et al. 18S Ribosomal DNA Phylogeny

Volume 84, Number1 1997

4A

EUDICOTS -Tacca Bowiea Chlorophytum

Hippeastrum

Gladiolus Xanthorrhoea Veitchia Cyanella Helmholtzia

Maranta Zingiber Costus

0

c0

Canna

Z 0

Heficonia Musa

Glomeropitcairnia

Zea Oryza Cyperus Elasis Coichicum __

_

__

_

_

__

_

_

j

Calla

Lt |

I

;

Oncidium Ceratophyllum Nelumbo | Aristolochia

Asarum> {

m G

Lactoris Houttuynia

Peperomia Saururus

> m

o 0 c

Nymphaea Schisandra Austrobaileya Iflicium Amborella 0 Gnetumgnemon < G Gnetumnodiflorum 0 Gnetumurens Ephedrasinica Ephedratorreyana

of194species analysis phylogenetic theexploratory from treesresulting of2582shortest consensus 4. Strict Figure of3507 steps,CI = treeshasa length Eachoftheshortest in theanalyses. twoindelswereincluded ofangiosperms; in Table2. Becauseof oftheindelsdescribed A andB indicate theoccurrence 0.249,andRI = 0.536.Theletters intothreeparts(4A,4B, and4C). itssize,thetreehasbeenbroken

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

Annals of the MissouriBotanical Garden

46

A

4B

ASTERIDAES. L.

|

ROSIDAE Hedyosmum ; m

Buxus Calycanthus Sassafras

Mkilua

Knightia Akebia Ranunculus Xanthorhiza Menispermum Caulophyllum

l

Dicentra L Hypecoum

______________________________________ _ l LEuptelea E

o

Trochodendron Tetracentron Platanus

TE MONOSCULCA GRADE Figure4B.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

H

> z c z C

0 Z (|)

n -H

on

47

Soltis et al. 18S Ribosomal DNA Phylogeny

Volume 84, Number1 1997

ASTERIDAES. L.

Sagittaria Cuscuta Chrysobalanus Licania Linum Geranium Muntingia Bombax

4C

Luhea

Gossypium Malpighia Drypetes Turnera Euphorbia

r

Prunus

Photinia Spiraea Zelkova Cetris

z

Alnus Casuarina Mowus _______

c

0

m z

MFrancoa

G

_

-

Plea Elaeagnus

Greyi'a Ce-anothus Tetrameles

__

Symbegonia Z C Begonia Abobra Datisca Crossosoma

Guaiacum Sloanea

C:

o Bauera z Ceratopetalum

r

EZAlbizia Bauhinia

:D Lepuropetalon > ParnassiaO

____

____

____

____

____

____

___

Brexia us E uonym Acer Floerkea

o Cn

Koeberlinia0 Arabidopsis0 z Brassica o Cleome Moninga

B

.

m Canica CDaphnipylum Tropaeolum I/ Tetracafpaea Haloragis Penthorum Paeonia> ltea Pterostemon Crassula Sedum> a Dudley Heuchera F-Boykinia

Cf

C)

C

~~~~~~~~~~Sakifraga

_____

____

_____

____

Cercidiphyllum

Daphniphyllum Uiquidambar AItingia Polygaa z > Opilia Osynis rViscum Gunnera vh

LOWEREUDICOTS/ TES MONOSULCA Figure 4C.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

m

Annals of the MissouriBotanical Garden

48

>

4D

0 f

E

O ~~~~~~~Drosera

Nepenthes

0

Mollugo Tetragonia

5 >

Plumbago Cocoloba

-

I

, Phytolacca r Spinacia Mirabilis Manilkara Sarracenia Clethra Arctostaphylos Vaccinium Actinidia Camellia Symplocos Fouquieria

m

5 >

m

>

Diapensia Galax

Impatiens

Primula Cobaea

> m

Acanthogilia Gilia

Philadelphus Hydrangea Garrya Pittosporum Hedera Lomatium Escallonia Eremosyne Roussea Campanula Lobelia Symphoricarpos Dipsacus Corokia Tragopogon Tagetes Aucuba Helwingia Phyllonoma Olea

Pachystachys

> V -)

m >

m

U)

>

t)

m

5

>

Byblis

Parmentiera Buddleja Lamium Phacelia Bourreria Tabemaemontana Mitchella

Ipomoea Brunfelsia Montinia Vahil

Aeschynanthus BerzeMniaa

Camptotheca

ROSIDAE LOWEREUDICOTS! MONOSULCATES Figure4D.

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

49

Soltis et al. 18S Ribosomal DNA Phylogeny

Volume 84, Number1 1997

forthe 18S ribosomalRNA of Glycinemax (modifiedfromNickrent& APPENDIX. Proposedsecondarystructure sequence of Glycine(Eckenrodeet al., 1995) and follows modelis based on the primary Soltis,1995). This structural in general.Tertiary interactions are indicatedbythicklines.The positions thegeneralmodelsproposedforeukaryotes sequence and length(positions,230proneto variationin primary indicatedby arrowsare thoseregionsparticularly to align overa broadtaxonomicscale and werenot 237, 496-501; 666-672, 1363-1369); these regionsare difficult analyses(see text). includedin our phylogenetic 25

A AU aAUA I ACU'Ua UCUU UUUAUUAAAUAC .AAC AUC AU U A26 ,bil UAUACUUUAUUC A CU"AA aG U AU au 'U AUU..U C-goU USC UUCUU-CU-A U C U U UC

-A A

UAUUCA: C-U U'BA UA

UU UC A CA ' e .UA

~

A

C-.QC-~

UA

CUU

U

UA

UC 40IC

A UaU

CUA CA U A10 U-CC-

aUU

A

37

2

AUA

U' C-U A-UU.

42 39 4 is130s CCU CCUCUUA AUU' ' UGGG~a U% UUUA AA A ACU CCUAUA AU OUG AC' ' I'' I 0 U UUUUUAA U C UU CCU CC UUUU U GA IU U'S UA U;

AU

41A UUUUUC

C 50 UU C AUSC A'CU

~

A

V

U-C

U C-U C-U -U'U UU CU'

~

V'

GUUU A-U C-U C-U Ul

UU C E223 U. AUUA. 24CCU--UC"CUU C A-U A-U U9_UC GA 4 UU 38UAaUCC U aAUUU UIUA U CUU" C U.U -U'U AU U UU U UA AMU- AC A 0 U-A-A C A A C GUUU:U AaC U ,UAUU AU CC IA-U UUU U-CUa' A U" U 29U C Ca 'U AA A 4UGAUUAU'U9 A A 1450 I' II I CUU A' -C U CUCUUUUU UUaCCUU UA' UA ICCAUUA U UC U-A U C ii 23b %-A ' U" UU UU AU'CC A'U, UUUUUAa CU CUAAAC AA A AGC U A AA36(A C C U Ca G%\ CUCUUA-U UaC6AAA A E23-7 U23 -C IUCU P C/ CUC: a G~CaAACU G AU UU' ICU A, GA CUAU'AU AUUUI CCU C.A' U U UU'UUUC AA UJC7UU U a\G AU A A .\U C C'U U Uc'UC CE23-1A GU AU28 U UA CU.U AUAUs a U A% A U'UUa Cu AUU, AA UA 28CCaU. aC UU U. AU I C 'N 'U '' UC'UC UC U 'U UA CAC CU'UC A U-U aC/UUC GCC' 'ACCAC UUACAA U A E23-6 L~~:CUJU70UGUUUCUU ,UUC E23-6 lUffA a A a0CA' ,UCC ' C-UU'G UAA C U Cu GE23-2A4~U C'AA 3CU (U:c aC AU UC AU A UC U UU U'UAU' U CC UA 18 U U UU AUA C UAA -CU-C A JU U AAAVCY" A-U U UC-(' 'U UUUAA a G'C a 223UAUAUUUCU ak G C aC U.UU.aCU A U -AA UUa U

~~~~~~~-U ~~~C ~ ~~~

E23-8 G E23

~~~~A-U

~~~~~~~~~TA

~~~~~

6~~~.Ua

~~~CC

~

U~

C-~~3-

~

~

~

AV

I-C A

U~~~U

U

U-U

E2-UU

AA UACCU~Ua U.. UUU1gA-U U U"U AU U C UCC" -A CA A CC aA AU AUAA A .-U C2'U3 U G-C CA U' U"U C UA USC ''A'AUa';

~

3C

'''''~~~~ CAU

C AA UUC 5

'

U~~~~"C

CU. USC

15G CUG

U U

CA AAUAA

CG1A

A.-A UU U

AA a

SC

AU

47,U

UA

U

A

3501a

UA

''

a' UA

A

UAU U

a

UCA

UUC

A7

a

U

U-

Aa A U 4

C

C-

UUA

C

U U a

-80

0-15

A_ UAU UUUAU

U

UC U

a

A A

C

AGCUAU U A US U C AUUU UUUCA AU UAA

CUCU

C UaUUAUIAA a aGUC . AlA U a C C-A UUU

CUUCA5'

UISCa

UCG -U i1''''i UG- U -ow7SC a

U U

GAGC

CC ACUUACU

U UA aUUUUCUCOUAAAA 'U AU:A UU~C E1O

U \ GAUcuU A' A A .UIS Ca U U'U U A'A

C UU

A

,A

AAUU

UaUCCAUAGUUUUU 'U U 'UU

14

AAC UG A A AUU

UUCCUUAA CA

U

U

AU71 U

3U

C

A-

CUA

~~~~~~~~~~~~6 AA~~~~~~~Aa U~~~~~

IAAUC U

AV 2

AU UU.UC5

A

C UC

AG-U5a CACU CAA C-U A U UAUC AC' A 'Uc AACU ' UU 1A U UCjGG GGUAC U C I.II. III 'A ACAUU CUa

12

C U

UUUU '(j(jUG A CCUAUA U48GUUAA U

CUAA

V317Uua

C

~~~~~~~~U

ACU~~~~~~~~~~~~~~~~~~~~~UUUC ~~~~~~~~~~~~~~~~AAQUUCCU C

A

ACUUUC UUUUUUUACUAUA' 3 C 1 - AU' C A A C-U1: UaA U. A IAA CA AU A 2 U2CAU P U4 UC5 AGG

u Glyinema 4

A(UC U'c UCG 10

U

This content downloaded from 128.104.46.196 on Fri, 13 Feb 2015 09:43:00 AM All use subject to JSTOR Terms and Conditions

A

U

46