Phylogenetic Relationships in the Commelinaceae

Phylogenetic Relationships in the Commelinaceae: I. A. Cladistic Analysis of Morphological Data Author(s): Timothy M. Evans, Robert B. Faden, Michael G. Simpson, Kenneth J. Sytsma Source: Systematic Botany, Vol. 25, No. 4 (Oct. - Dec., 2000), pp. 668-691 Published by: American Society of Plant Taxonomists Stable URL: http://www.jstor.org/stable/2666727 Accessed: 10/09/2010 13:35 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aspt. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

American Society of Plant Taxonomists is collaborating with JSTOR to digitize, preserve and extend access to Systematic Botany.

http://www.jstor.org

Systemnatic Botany(2000), 25(4): pp. 668-691 ? Copyright2000 by the AmericanSocietyof Plant Taxonomists

PhylogeneticRelationships in the Commelinaceae: I. A Cladistic Analysis of Morphological Data TIMOTHY M. EVANS1

DepartmentofBotany,University ofWisconsin,430 LincolnDrive,Madison,Wisconsin53706 Presentaddress,authorforcorrespondence: Departmentof Biology,Hope College,35 East 12thStreet, Holland,Michigan49423-9000 ROBERTB. FADEN DepartmentofBotany,NHB 166,NationalMuseumof NaturalHistory,SmithsonianInstitution, Washington, DC 20560 MICHAEL G. SIMPSON

DepartmentofBiology,San Diego StateUniversity, San Diego, California92182

KENNETHJ.SYTSMA DepartmentofBotany,University ofWisconsin,430 LincolnDrive,Madison,Wisconsin53706 Editor:RichardJensen Communicating ABSTRACT. The plantfamilyCommelinaceaedisplaysa wide rangeofvariationin vegetative, floral,and inflorescence morphology.This high degree of variation,particularlyamong charactersoperatingunder has made theassessmentofhomologyamongmorphostrongand similarselectivepressures(i.e.,flowers), and has resultedin severaldiscordantclassification schemesforthefamily.Phylogicalcharacters difficult, logeneticrelationships among40 of the 41 generain the familywere evaluatedusing cladisticanalysesof data.The resulting morphological phylogeny showssomesimilarity to themostrecentclassification, butwith some notabledifferences. was placedbasal to therestofthefamily. Cartonemna (subfamily Cartonematoideae) of Triceratella (subfamily Cartonematoideae), however,was placed amonggenerawithintribeTradescantieae were subfamilyCommelinoideae. Likewise,thecircumscriptions of tribesCommelineaeand Tradescantieae in disagreement withthe mostrecentclassification. The discordancebetweenthephylogenyand themost recentclassification is attributed to a high degreeof convergence in variousmorphological characters, particularlythoserelatingto the androeciumand theinflorescence. Anatomicalcharacters (i.e.,stomatalstrucwithintheCommelinaceae, ture),on theotherhand,show promiseforresolvingphylogenetic relationships based upon theiragreement withthemostrecentclassification.

The Commelinaceae, a well definedfamilyof41 neae is foundmainlyin Africaand Asia. Only six Floscopa, generaand about 650 species,is characterized Buforrestia, Commelina, by genera (Aneilema, several featuresincludinga distinctclosed leaf Murdannia, and Pollia)have indigenousspecies in sheath,a succulentleaf blade, and three-merous bothhemispheres. flowerswith distinctpetals and sepals (Cronquist oftheCommelinaceaerelied Olderclassifications 1981;Faden 1985;Faden and Hunt 1991).The gen- heavilyon floralfeatures(see Fadenand Hunt1991, era are largelytropicaland subtropical, butseveral fora review).Woodson (1942) firstused infloresextendintotemperateregions.The greatestdiver- cence charactersforhigher-level but classification, sity is in Africa,where,along with Madagascar, he was largelyunfamiliar withtheOld Worldgennearlyhalfthegeneraand about40% ofthespecies era. Pichon(1946) employedanatomicalcharacters are found(Faden1983a). as thefamilyCarto separatethegenus Cartonema Thereis a naturaldivisionof taxa betweenthe tonemataceae, but he returnedto moretraditional Old and the New World.The seven subtribesof charactersto definehis ten tribesof CommelinatribeTradescantieae(sensu Faden and Hunt 1991) ceae. Brenan(1966)used a varietyofmorphological are each whollyconfinedto eitherthe Easternor charactersto define15 informal"groups" of genWesternhemisphere,whereas the tribeCommeli- era. These have served as the basis of surveysof 668

2000]

EVANS ET AL.: COMMELINACEAE

669

studiesare similarto anatomical(Tomlinson1966,1969) and cytological molecularand morphological (Jonesand Jopling1972) charactersin the family, the molecularphylogenies(e.g.,Chase et al. 1995; but theyhave littletaxonomicsignificance. Linderand Kellogg1995).Recentworkby Givnish, The mostrecentclassification, whichwillbe used Evans,Pires,and Sytsma(Givnishet al. 1995,1999) in thisstudy,was put forwardby Faden and Hunt generallysupportsthesefindings.It also placesthe (1991).In theirclassification, theCommelinaceaeis Hanguanaceae as the sisterfamilyof the Commedivided into two unequal subfamilies(Table 1). linaceae and the Xyridaceaeand Eriocaulaceaein SubfamilyCartonematoideaecontains the tribes thegrass/sedgeclade. and Triceratelleae, Cartonemateae each comprising Certainmorphologicalcharactersalso lend supa single genus. SubfamilyCommelinoideaecon- portto the separationof the Commelinaceaefrom tainstheremaining38 generathatare arrangedin otherfamiliesof the originalCommelinales(sensu two tribes,Tradescantieaeand Commelineae,the Cronquist1981,or Dahlgrenet al. 1985). As disformer ofwhichis dividedintosevensubtribes. cussed above,Dahlgrenet al. (1985) separatedthe Evans (1995)conducteda cladisticanalysisofthe Commelinaceaefromthe othercommelinoidfamiand mo- lies in partby thepresenceof raphides,whichare Commelinaceaeusingbothmorphological This paper representsa modifi- absentin theMayacaceae,Rapateaceae,Xyridaceae, lecularcharacters. cation and expansionof the morphologicalcom- and Eriocaulaceae(Dahlgren and Clifford1982). Raphidesare widespread,however,in thePontedponentsofthatstudy. Relationshipsof Commelinaceae and Haemodoraceae(Dahlto otherMono- eriaceae,Philydraceae, whereasthe Comcot Families. The familyCommelinaceaeis the grenet al. 1985). Additionally, namesakeforthe orderCommelinales.Cronquist melinaceaewere separatedfromotherfamiliesin (1981) definedtheorderby thepresenceofperfect the Commelinalesby thepresenceof an amoeboid flowersthatare adapted forgeneralinsectpolli- tapetum,the genus Pontederia and all investigated nation,havingshowypetals thatare differentiatedmembersoftheHaemodoraceaealso havean amoefromthesepals. He includedthreeotherfamiliesin boid tapetum(Dahlgrenand Clifford1982; Simpthis order:Xyridaceae,Rapateaceae,and Mayaca- son 1988, 1990). Finally,flowersin the Commeliceae. He separatedthe Commelinaceaefromthe naceae show a strongtendencytowardzygomorin other three families by their well-definedand phy,whereastheyare generallyactinomorphic closed leafsheath,and thesucculentblade. the other membersof the Commelinalessensu Dahlgrenet al. (1985)includedthesame families Cronquist(1981) or Dahlgrenet al. (1985). Zygoin Commelinalesas Cronquist,but withthe addi- morphicflowersare common,however,in thePonand Haemodoraceae. tionof theEriocaulaceae.Eriocaulaceaewas placed tederiaceae,Philydraceae, in itsown orderbyCronquist,who maintainedthat have Althoughnumeroustaxonomictreatments it was mostlikelyderivedwithinXyridaceae. Dahl- been produced forthe Commelinaceae,therehas should theCommelinaceae been littleconsensusas to whichcharacters grenet al. (1985)distinguished fromthe othercommelinoidfamiliesby the pres- be used to definerelationships amongthegenera. ence of raphides,an amoeboid tapetum,and the The earliersystemsreliedheavilyupon featuresof leaf charactersused by Cronquist (closed leaf the highlyvariableandroecium,but theyignored sheathand succulentblade). charactersof the inflorescence. The classification Although cladistic analyses of morphological producedby Fadenand Hunt (1991),whichwillbe data supporta monophyletic this study,incorporateda broad Commelinalessensu used throughout Dahlgrenet al. (1985;Stevensonand Loconte1995), rangeof information, such as characters frommorrecentlypublishedmoleculardata suggestother- phology,anatomy,and palynology.None of these wise.Nucleotidesequencedatafromthechloroplast classifications, is rootedin an evolutionary however, The purpose of this studywas to exgene rbcL(Chase et al. 1993;Clarket al. 1993;Du- framework. vall et al. 1993) place the Commelinaceaenear the amine the intergeneric relationshipsin the Comand Haemodoraceae, melinaceae using morphologicaland anatomical Pontederiaceae, Philydraceae, all of whichare part of a largerclade containing characters in a cladisticanalysis. thefamiliesoftheorderZingiberales.The Rapateaceae (theonlyotherrepresentative ofCommelinales MATERIALS AND METHODS included in the previous molecular studies) is clade that contains,among placed in a different OperationalTaxonomicUnits (OTU's). A critiotherfamilies,Poaceae and Cyperaceae.Combined cal step in any cladisticanalysisis theselectionof

670

SYSTEMATICBOTANY

[Volume 25

TABLE 1. Classification oftheCommelinaceae(as presentedbyFadenand Hunt1991).*indicates Old Worldsubtribe, **indicates New Worldsubtribe.

Family:CommelinaceaeR. Br. Subfamily: Cartonematoideae (Pichon)Faden& D. Hunt Tribe: Cartonemateae (Pichon)Faden & D. Hunt Cartonema R. Brown Tribe: Triceratelleae Faden & D. Hunt Triceratella Brenan Faden& D. Hunt Subfamily: Commelinoideae (Bruckner) Tribe: Tradescantieae(Meisner)Faden & D. Hunt Subtribe: *Palisotinae Faden & D. Hunt PalisotaReichb. Subtribe: *Streptoliriinae Faden & D. Hunt Streptolirion Edgew. Spatholirion Ridley Aetheolirion Forman Subtribe: *Cyanotinae (Pichon)Faden & D. Hunt D. Don (Including Cyanotis R. Rao & Kamm.) Amischophacelus Belosynapsis Hassk. Faden& D. Hunt Subtribe: *Coleotrypinae Coleotrype C. B. Clarke Porandra D. H. Hong Amischotolype Hassk. Subtribe: **Dichorisandrinae (Pichon)Faden & D. Hunt Mikan Dichorisandra Siderasis Raf. Geogenanthus Ule Lem. Cochliosteina Undescribedgenus Subtribe: **Thyrsantheminae Faden & D. Hunt Thyrsanthemumn Pichon D. Hunt Gibasoides TinantiaScheidw. ElasisD. Hunt D. Hunt Matudauithus Weldenia Schult.f. Subtribe: **Tradescantiinae Rohw. GibasisRaf. Tradescantia L. (IncludingSetcreasea Schumann& Sydow,Separotheca Waterf., Cymnbispatha Pichon,CampeliaRich.,RhoeoHance,ZebrinaSchnizl.) H. Moore,Cuthbertia CallisiaLoefl.(IncludingHadrodemnas Small, C. B. Clarkeex Hemsley) AploleiaRaf.,LeiandraRaf.,Phyodina Raf.,Leptorhoeo Raf. Tripogandra SatIvalleaWright Tribe: Commelineae(Meisner)Faden& D. Hunt Brenan Stanfieldiella FloscopaLour. Buforrestia C. B. Clarke Murdannia Royle Anthericopsis Engl. Tricarpelema J.K. Morton H. Perrier Pseudoparis Benth. Polyspatha Dictyospermum Wight PolliaThunb. R. Brown Aneilemna Hassk. Rhopalephora Commelina L. (IncludingPhaeosphaerion Hassk.,Commelinopsis Pichon)

2000]


671

evolutionary units,or operationaltaxonomicunits son 1990; see also Bininda-Emondset al. 1998). (OTU's). It is assumed thatthe unitsbeing exam- Some of the difficulties of addressingthefirstopined have all descendedfroma mostrecentcom- tion(monophyly) are discussedabove.The theoretmon ancestor,and representdistinctevolutionary ical and practicaldifficulties in dividingall polyentities.If the OTU's are polyphyletic, theninac- morphicOTU's into monomorphicunits is discurate("misleading")reconstructions of theirevo- cussed in some detailin Donoghue (1994). Therefore, in this analysis,a single species was lutionary historymaybe produced(Nixonand Davis 1991;Donoghue1994). scored for each genus where possible (Table 2). Of the 41 generain the Commelinaceae, ten are Data forindividualspeciesoffourgenera(Porandra, monotypic(Aetheolirion, Anthericopsis, Elasis,Giba- Pseudoparis, Rhopalephora, and an undescribedgesoides, Matudanthus, Sauvallea, Streptolirion, Tapheocar- nus) were unavailable,so theywere leftas polyand Weldenia), and thereis morphicto reflectvariationwithineach genus.An pa Conran,Triceratella, in treatingeach as an OTU. For each attemptwas made,wheneverpossible,to selectthe no difficulty moof the remaining31 genera,however,claims of same speciesbeing used in a complementary monophylymightbe misleading.The genus Calli- lecular cladistic analysis of the Commelinaceae sia, for example,as circumscribed by Faden and (Evans et al., in prep.)so thatthe resultsfromthe Hunt (1991),contains20 species in seven sections two data sets would be moredirectlycomparable. (Hunt 1986b). There is no single morphological On the basis of the recentmolecularstudieson characterthatdefinesthe genus,and its question- monocots(Chase et al. 1993;Clarket al. 1993;Duablymonophyletic status(Hunt 1986b)mightpro- vall et al. 1993;Givnishet al. 1999),representatives in a cladisticanaly- of the familiesPontederiaceae(Heteranthera) duce misleadingrelationships and were used as outSiS. Haemodoraceae (Haemodorum) Likewise,thegenusTradescantia sensuFadenand groups.These two taxa were selecteddue to their Hunt (1991) is large (containingabout 70 species) putatively basal positionin theirrespectivefamilies and morphologically diverse.Somemembersofthis (Simpson1990;Grahamet al. 1998)and availability data. Fortyingroupand two outhavebeen placed intoas many of morphological genusuntilrecently as six othergenera(Setcreasea, Separotheca, Cymbis- group generaor exemplarswere includedin this patha,Campelia,Rhoeo,and Zebrina).Hunt (1975, study(Appendix1, 2). The recentlydescribedge1980,1986a)has graduallylumpedall ofthesegen- nus Tapheocarpa Conran (1994) was not included data. era into Tradescantia, but themonophyly of thege- due to insufficient nus is yetto be demonstrated. Character Selection. Characterswere scored Althoughit is clearthatlargegenerasuchas Cal- frombothlivingand pressedplantshoused at the lisiaand Tradescantia mustbe treatedcautiouslyin US NationalHerbarium(US). Observationswere a cladisticanalysis,the same difficulties may also made on living plant materialgrowingin the arise forsmaller,more "clearlydefined"genera. SmithsonianInstitutionBotany Research Greenfor example,is distinguishedfrom houses and in the field.Whereliving or pressed Rhopalephora, Aneilema by a combinationof characters:inflores- plants were not available for examination,inforcence axis veryshort,stamenfilamentsfusedba- mationwas takenfrompublishedliterature (Table werescored. sally,ovaryand capsule denselycoveredby hook 2). Fortysevencharacters hairs,and dorsal capsule valve deciduous(Faden Phylogenetic Analysis. Data were enteredinto 1977).All ofthesecharacters occurwithinAneilema, a matrixusing the computerprogramMacClade and no singlemorpho- version3.05 (Maddison and Maddison 1992),and butnotin thiscombination, is unique to Rhopalephora. The im- the phylogeneticanalyses were performedwith logicalcharacter plied hierarchicalrelationshipsbetween the two PAUP*version4.Ob2a(Swofford1999). A total of generasuggestsone or the otherof themmay be 3.7 percentofthecellsofthematrixwerescoredas paraphyletic. "missing,"eitherbecausethedatawereunavailable werenotapplicable To avoid the problemsdiscussed above, each orbecauseparticularcharacters OTU shouldbe treatedin one of threeways: 1) it to certaintaxa. weretreated shouldbe clearlydemonstrated tobe monophyletic; In an initialanalysis,all characters it should be as unordered.A multiple-islandsapproach was 2) if polymorphicforany character, split into smaller groups that are monomorphic used to findthemostparsimonioustrees(modified (Nixon and Davis 1991); or 3) it shouldbe repre- fromOlmsteadet al. 1993;see also Maddison1991; sentedin the analysisby exemplarspecies (Simp- Olmstead and Palmer1994). To evaluatethe sup-

672

[Volume 25

SYSTEMATICBOTANY

TABLE 2. Primary literature sourcesforinformation at thespecieslevelused in a cladisticanalysisofmorphological characters in theCommelinaceae. Species

Forman Aetheolirion stenolobiutn Ainischotolype hispida(Lesson & A. Rich.)D. Y. Hong Anithericopsis sepalosa(C. B. Clarke)Engl. Brenan Aneilemna calceolus kewensis Hassk. Belosyinapsis Buforrestia mauniiC. B. Clarke CallisiaelegansAlexanderex. H. E. Moore E Muell. Cartonema philydroides Lemaire odoratissirntm Cochliosterna C. B. Clarke Coleotrype natalensis Commnelinia benghalensis L. barbatus D. Don Cyanotis Dichorisandra hexandra (Aubl.)Standl. rontacmtm Dictyospermum Wight Hunt Elasishirstita Floscopaafricania (P. Beauv.)C. B. Clarke Geogenanthus poeppigii (Miq.) Faden Gibasisgeniculata (Jacq.)Rohweder Gibasoides (C. B. Clarke)Hunt laxiflora Mattudanthus nanius(Martens& Gal.) Hunt Murdannia edutlis (Stokes)Faden Palisotaambigua(P. Beauv.)C. B. Clarke Pollicmannii C. B. Clarke Polyspatha paniculcta Benth. Porandra Pseudoparis Rhopalephora Sauvallea Siderasis (Lodd.) H. E. Moore ftuscata Spatholirion longifoliumn (Gagnep.)Dunn Stanfieldiella imperforata (C. B. Clarke)Brenan voltubile Streptolirion Edgew. Thyrsanthentinu floribundum (Martens& Galeotti)Pichon Tinantiti erecta(Jacq.)Schltdl. Tradescantia L. virgiiiacma Tricarpeleina gigantea(Hassk.) Hara Triceratella drurnniondii Brenan Tripogandra amplexicatulis (Klotzsch)Woodson Weldenia Schultess. catndida UndescribedGenus(Faden,in prep.)

port foreach node, 100 bootstrapreplicateswere conducted.Due to the high numberof treesproduced,a maximumof 5000 treesper replicatewas saved. As a measureofwhichcharacters are morephylogenetically informative and whichare morehomoplasious (with respect to the topologiesof the treesproduced),as well as to arriveat a morestable topology,the characterswere re-weighted a posteriori(usingthe"REWEIGHT CHARACTERS" command

Source

Forman1962 Clarke1881 Brenan1966;Faden,unpubl.data Faden 1991 Clarke1881;Faden,unpubl.data Brenan1960;Faden,unpubl.data Moore1958 Brenan1966;Faden,unpubl.data Read 1965;Faden,unpubl.data and Faden 1985 Obermeyer and Faden1985 Obermeyer Brenan1966 Matuda 1955 Morton1966;Faden,unpubl.data Hunt 1978;Faden,unpubl.data Berhaut1988 Moore1954 Hunt 1985 Hunt 1978;Faden,unpubl.data Hunt1978;Faden,unpubl.data Faden 1980 Clarke1881 Brenan1966 Clarke1881;Faden,unpubl.data Hong 1974;Faden,unpubl.data Faden,unpubl.data Faden 1975,1977 Faden,unpubl.data Clarke1881 (as Pyrrhema); Faden,unpubl.data Faden 1985 Brenan1960 Forman1962 Hunt 1993 Hunt1993 Andersonand Woodson1935 Morton1966;Faden,unpubl.data Brenan1961 Hunt1993 Hunt 1993;Faden,unpubl.data Faden,unpubl.data

in PAUP) based on theirfitto the treesproduced in the unorderedanalysis.The multi-step analysis describedabove was thenrepeatedusing the reThisprocesswas repeatedunweightedcharacters. til the same set of most parsimonioustreeswas producedin twoconsecutiveruns.PAUPallowsthe characterto be re-weighted based on consistency index (CI), retention index(RI), or rescaledconsistencyindex (RC). The maximalvalue of each of theseindiceswas used in threeseparateanalyses.

2000]


673

In additionto theunorderedanalysis,an analysis deredanalysis.The CI- and RC-basedanalysesprowas performedin whichthreecharactertransfor- duced identicalsets of 15 equallymostparsimonimationswere ordered(fruitlocule number[Char- ous trees(notshown)thatdiffered fromtheunoracter35] and seed numberper locule [Characters dered topologyin the lineages sisterto the zygo37 and 38]). The same proceduredescribedabove morphicclade. In the unorderedtree (Fig. 1), the was used in thisanalysisto findmultipleislands zygomorphic clade has a sistergroupconsistingof The Pollia,Siderasis, and Dichorisandra. Stanfieldiella, of mostparsimonioustrees. Forpurposesof characterstatemapping,one of CI- and RI-basedanalysesproduceda setofnested select- relationshipsfromthe morebasal Siderasisto, sethemostparsimonioustreeswas arbitrarily Dischored forillustration. However,due to thelack ofres- quentially, Anthericopsis, Pollia,Stanfieldiella, and thezygomorphic clade. olutionofseveralcladesin thestrictconsensustree, isandra, wereexaminedon all ofthemostparOrdered CharacterAnalysis. When assumpdistributions steps in tions about the sequence of evolutionary simonioustrees. 115 characters35, 37, and 38 were incorporated, were produced most trees equally parsimonious RESULTS (notshown).Thistopologyis nearlyidenticalto the Unordered Analysis. In theunorderedanalysis, unorderedtopology,exceptthatit showed greater clade.The undescri154 equally most parsimonioustrees were pro- resolutionin thezygomorphic were united in one of bed genus and Cochliostema duced,witha length(excludingautapomorphies) Aneilema,and Commelina 239 steps and a consistencyindex (CI) of 0.43 clade and Polyspatha, (length= 272 and CI = 0.44whenautapomorphies formedan unresolvedclade. are included).Figure 1 illustratesthe tree chosen forcharacterstatereconstructions. DISCUSSION Cartonema is sisterto therestofCommelinaceae. thathavebeenproThe remaininggeneraformfourlineages (Fig. 1): The discordantclassifications Callisia,which is sisterto the remaininggenera; posed forthe Commelinaceaereflectthe high dein homologyamong morphoand twolargelyunresolvedclades.Within gree of uncertainty Weldenia; 15 genera logical charactersin the family.Likewise,the low the clade thatis sisterto Anthericopsis, Polys- CI values formanycharactersin thisstudy(Table (Buforrestia, Floscopa, Pseudoparis, Tricarpelemna, a highamountofhomoplasyin the Dictyosper-3) demonstrate patha,Aneilema, Commelina, Rhopalephora, Aetheolirion,family.The cladisticanalysis presentedhere,almum,Palisota,Tinantia,Cochliostema, and an undescribedgenus) forma thoughsupportingparticularelementsofsomepreGeogenanthus, antese- vious classifications, does not agree closelywith clade thatis supportedby two characters, palous staminalfilamentlengths(Character26), any one of them.The zygomorphicclade (Fig. 1) and the presenceof a zygomorphicandroecium includes all but threegenera (Pollia,Stanfieldiella, (Character31; note:Floscopaproducesan asymmet- and Murdannia)of Faden and Hunt's (1991) tribe ric androecium).For ease of reference, this clade Commelineae.However,severalmembersof Trawill be referred clade."The descantieaeare also includedin this clade. Liketo as the"zygomorphic remaininggenera(Murdannia, Gibasis,Elasis,Thyr- wise, fewelementsof Brenan's(1966) classification Por- are supported.Ofthe15 informal Coleotrype, "groups"putforsanthemumn, Gibasoides, Matudanthus, Tricera- ward by Brenan,one,"Group VI" (= subtribeCyandra,Amischotolype, Cyanotis, Belosynapsis, here.Relativelynarrow Tradescantia,anotinae),is monophyletic tella,Streptolirion, Spatholirion, Sauvallea, and Tripogandra) forma large clade that is sup- circumscriptions of severalofBrenan'sgroups(i.e., ported by the presenceof bearded antesepalous theinclusionofonlyone or two generaper group) staminalfilaments(Characters22, 23). For ease of make a directcomparisonof relationships among Pichon's(1946) classification this clade will be referred to as "Clade thosegeneradifficult. reference, 1." includedten tribes,five of which containonly a A posterioriRe-weighting.Each of the re- singlegenus.Of theremainingfivetribes,not one weightedanalyses(CI-,RI-,and RC-based;Table3) is supportedby thisanalysis.The largenumberof producedtopologiesthatwere similarto the un- taxonomicgroupswith only a singlegenus in Piorderedtree.The RI-based analysisproduced 45 chon'sand Brenan'ssystems,as well as lack ofstaequally mostparsimonioustrees(not shown) that tisticalsupportformostcladesin thisanalysis,sugin were a subset of the 154 treesfoundin the unor- gest thatthereis a high degree of uncertainty

674

[Volume 25

SYSTEMATICBOTANY Anthericopsis Buforrestia Floscopa Pseudoparis 4

||0j

C Tri carpelema

60*%O

r"

[j

g||1

mu.

| |Commelina

| Commelineae

Aneilema

_Rhopalephora

Dictyospermum Palisota

I Thyrsantheminae

Tinantia _..

| ||||||||||||||||||||

I Palisotinae

Undescribed genus Dichorisandrinae

|_Cochlostema |||||| IHU

Tradescantieae

S Streptoliriinae

Aetheolirion Geogenanthus Dichorisandra

Dichorisandrinae

Siderasis

I

Pollia Stanfieldiella

I Thyrsantheminae

Weldenia Murdannia Gibasis

j

I Tradescantieae I Commelineae

nae

ITradescant

F/asis I Thyrsanthemum |

1

Gibasoides

~~~~Thyrsantheminae

Matudanthus Coleotrype

*Tradescantieae

Porandra

Coleotrypinae

Amischotolype Cyanotis

Cn Caoiu

Be/os ynapsis

a

Triceratella

80%

Commelineae

I (lade

Streptolirion Spatholirion I

Streptoliriinae

Sauvallea ~~~~Sauvallea

I

Tradescantia Tripogandra

5

I Triceratelleae

||

Tradescantieae

Tradescantiinae

~~~~~~~ ia| ~ ~~~ ~~~~~~~~~~~~~~~~~~~~Ca//lis

E

Cartonema

| Cartonemateae

Haemodorum Heteranthera

FIG. 1. A representative of the154 equallymostparsimonioustreesfoundin theunorderedanalysisof47 morphoin CommelinaceaeusingHaeniodorurin logicalcharacters and Heteranzthera as outgroups.Dashed linesrepresent branches thatcollapse in thestrictconsensusof mostparsimonioustrees.Bootstrapvalues greaterthan50% are shownabove branches.Subtribaland tribalaffinities are shownto rightof genericnames."Clade 1" and the"zygomorphic clade" discussedin thetextare indicated.

assessing homology ofstructures amonggeneraof Cartonema.Callisia is a memberof subtribeTradescantiinae,a well definedgroup based on feaCommelinaceae. Basal Lineages in the Commelinaceae. Each

tures of the inflorescence(cincinnicontracted,

analysisdividedtheCommelinaceae intoseveral fusedin bifacialpairs; Faden and Hunt 1991).The of thisentiresubtribe,as mainlineages,withCartonema sisterto therestof New Worlddistribution thefamily and Callisiasisterto everything except well as its putativelyrecent origin (Faden and

2000]

675


fromeach analysis.CI = consistencyindex; RI = retentionindex;RC = rescaled TABLE 3. Characterstatistics index;Wc, = characterweightafterre-scalingaccordingto theCI; WRI = character weightafterrescaling consistency accordingto theRI; WRC = character weightafterre-scalingaccordingto theRC. Weight Character

States

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

01 01234 01 012 01 012 01 01 012 01 01 01 01 01 01 01

17

18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

01

01 01 01 01 01 01 01 01 012 012 012 012 01 012 012 01 012 012 01 0123 0123 0123 01 01 01 0123 01 01 012 01

Steps

3 13 9 6 2 7 5 4 7 3 4 5 3 2 2 3 1

1 2 3 3 3 3 1 3 2 2 7 9 2 5 3 7 3 4 6 7 8 11 6 2 5 9 3 2 2 1

CI

RI

RC

WCI

WRI

WRC

.333 .308 .111 .333 .500 .286 .200 .250 .286 .333 .250 .200 .333 .500 .500 .333

.500 .308 .429 .556 0.00 .444 .636 .400 .375 .600 0.00 .429 .333 .500 .500 .500

.167 .095 .048 .185 0.00 .127 .127 .100 .107 .200 0.00 .086 .111 .250 .250 .167

333 307 100 286 500 286 200 250 286 333 250 200 333 500 500 333

286 357 400 500 0 500 583 286 375 800 0 375 667 333 333 500

167 94 36 127 0 127 127 100 107 200 0 86 111 250 250 167

1000

1000

1000

1.00

1.00 .500 .333 .333 .333 .333 1.00 .333 1.00 1.00 .286 .222 .500 .400 .667 .143 .333 .500 .167 .286 .250 .273 .167 .500 .200 .333 .333 .500 1.00 1.00

-

-

1.00 .500 .333 .778 .882 .882 1.00 .889 1.00 1.00 .167 .222 0.00 .812 0.00 0.00 .667 .333 .444 .737 .667 .500 .500 .500 .750 .667 .600 .667 1.00 1.00

1.00 .250 .111 .259 .294 .294 1.00 .296 1.00 1.00 .048 .049 0.00 .325 0.00 0.00 .222 .167 .074 .211 .167 .136 .083 .250 .150 .222 .200 .333 1.00 1.00

500 1000 333 250 500 500 1000 250 1000 1000 333 222 500 500 667 143 333 500 125 286 250 273 200 500 167 375 333 500 1000 1000

1000 1000 333 778 889 889 1000 842 1000 1000 333 333 0 813 0 0 667 333 444 609 545 421 455 500 750 722 800 667 1000 1000

417 1000 111 167 471 471 1000 208 1000 1000 111 49 0 438 0 0 222 167 28 211 167 136 120 250 115 271 200 333 1000 1000

676

SYSTEMATICBOTANY

[Volume 25

Hunt 1991) shed doubt on the basal positionof used anatomicalcharactersto suggestthatthe geCallisia. nus may forma linkbetweenCartonema, whichhe The positionof Cartonema sisterto therestofthe placed in a separatefamily,and the Commelinafamilyis consistentwith otherdata, particularly ceae. He indicatedthatit mightbe treatedas a seprbcLsequences(Evans 1995;Evans et al., in prep.). aratesubfamily ofCommelinaceae. Fadenand Hunt On the basis of various morphologicalcharacters (1991) placed Triceratella and Cartonema in a sub(e.g.,actinomorphic flowerswithsixfertile stamens, familywithintheCommelinaceae, witheach genus the absenceof raphidecanals),Cartonema has been in a separatetribe. The morphological data presentedhereplace Tripostulatedto be one of the moreprimitivemembers of the family(Pichon1946; Hutchinson1959; ceratellaas a sistergroup of subtribeCyanotinae Brenan1966;Faden and Hunt 1991). (Cyanotisand Belosynapsis), nestedwell withinthe Relationshipsin the Tradescantieae.With the Tradescantieae.The clade containingBelosynapsis, exceptionsof Murdannia(tribeCommelineae)and Cyanotis, and TriSauvallea, Spatholirion, Streptolirion, Triceratella (subfamily is supportedby a singlecharacter(sessile Cartonemateae), Clade 1 con- ceratella ofmembersofTradescantieae(Fig. flowers).The presenceof sessile flowersis labile sistsexclusively 1). Several membersof the tribe,however,are withinCommelinaceae, havingbeen lost once and placed in otherparts of the tree. As discussed gained at least threetimes(Fig. 2). Althougha reabove,Callisia,forexample,is basal to all othergen- lationshipbetweenTriceratella and the Streptoliriera except Cartonema. study.Until Likewise,eight genera are inae has somesupport,itneedsfurther placed eithersisterto the zygomorphicclade (Di- less homoplasiouscharactersare foundto support and Siderasis)or withinit (Palisota,Tin- such a relationship, chorisandra it will remainsuspect. antia,undescribedgenus,Cochliostema, Aetheolirion, Of the seven subtribescircumscribed by Faden and Geogenanthus). The position of Dichorisandraand Hunt (1991),onlytwo,Coleotrypinae and Cyand Siderasis as thesistergroupofPollia(tribeCom- anotinae,are supportedas monophyletic in this melineae)is supportedby thepresenceofbiseriate study. The remainingsubtribesare polyphyletic seeds (Character36). This character,however,is and largelyunresolved.Two genera in subtribe and Elasis) conhighlyvariablewithinthe family.Withinthe zy- Thyrsantheminae (Thyrsanthemum gomorphicclade and its sisterclade,biseriateseed tributeto a polytomyin Clade 1. Thyrsantheminae, has arisena minimumof threetimes as definedby Faden and Hunt (1991), is based arrangement (Fig.2). This character is evenmorevariablein oth- largelyupon the absenceof characters or thepreser most parsimonioustrees,some of whichforce ence ofputatively and maybe primitive characters, and Geogenanthus to gainbiseriateseeds an unnaturalassemblage(R. Faden,unpubl.data). Aetheolirion independently. Likewise,relativefilamentlength Recentmoleculardata also suggestthatthis subwithinthe antepetalousstaminalwhorl(Character tribeis polyphyletic (Evanset al.,in prep.),and the 27) is the singlecharacterthatsupportsthe clade lackofresolutionand weak supportin thisanalysis in findingtaxonomiccharcontainingAetheolirion, Cochliostema, Geogenanthus,illustratethe difficulty Tinantia,and the undescribedgenus as the sister acters that will unambiguouslyunite these taxa groupofthezygomorphic too withothergenera. clade.This character is variable within the zygomorphicclade (not Relationshipsin the Commelineae.Ten of the thirteengenerain tribeCommelineaewere placed shown). The taxonomicplacementofthegenusTriceratellawithinthezygomorphic clade (Fig. 1). Of thethree has been problematic. When Triceratella was origi- remainingCommelineaegenera,two (Pollia and nally described,a loose affinity with StanfieldiellaStanfieldiella) are partof thesistergroupof thezywas suggested (Brenan 1961). Tomlinson(1964) gomorphicclade;Murdannia is nestedwithinClade

of characters FIG. 2. Distribution on therepresentative tree(see Fig. 1) fromtheunorderedanalysisof47 morphologicalcharacters in Commelinaceae.Left:Seed arrangement. Biseriateseed arrangement has arisenat leastthreetimes withinthe zygomorphicclade and its sisterclade. One additionalstep is requiredin some otherequallymostparsimonioustreesdue to shiftsin positionofAetheolirion and Geogenanthus. Right:Pedicel.Thepresenceofpedicellateflowers is homoplasiouswithinCommelinaceae. The distribution ofthischaracter is notalteredin anyothermostparsimonious trees.

2000]

U)

U


677

it-o -5

U.~~~~~~ ~~~-~~~~

Q) QE

~

U~~~~~?~~

~ ~ ~ ~~

U

E

E

EE

U-~~~~~~~~~~~~~~~~~~~~~~~~~~~j.U ~ ~ ~ ~ ~ ~ ~~ ~ ~ ~ ~ ~ U

-

US

Z

-,CZ

E

-8

Q

CU

E

E

EE

)

EE

E

E

)E

E

E

E

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~Q

:

678

SYSTEMATICBOTANY

[Volume 25

measures 1. The only synapomorphy thatsupportsthe zy- weightedon the basis of threedifferent (CI, RI, and RC). Each gomorphicclade is thepresenceof a zygomorphic of theirinternalconsistency androecium(Character31), a characterthatis var- of thesere-weighting schemesstartswiththe uniable withinthe familyas a whole. However,all orderedtopology,and thenre-evaluatesthe fitof membersofthisclade,exceptPalisota, sharea com- the charactersto thattopology.The CI, a measure mon stomataltype consistingof six cells, with of thefitof the characters to thetree,is derivedas small terminalsubsidiarycells (Figs. 3 and 4). Al- theratioofthenumberof statechangesfora charto generabe- acterin thedata to thenumberofstatechangesfor thoughthisstomataltypeis restricted longingto thetribeCommelineae,it is nota syna- the characteron thehypothesizedtree(Kluge and pomorphyforthezygomorphic group,as it is also Farris1969).A potentialproblemwiththe consisand Murdannia.This tencyindex,however,is thatit onlymeasureshow foundin Pollia,Anthericopsis, anatomicalcharacter is in agreementwitha molec- well a characterfitsonto a particulartopology;it ular phylogenyfor Commelinaceae(Evans 1995; does not evaluatehow well thatcharacteractually thetopology(i.e.,ifitis synapomorphic for Evans et al. 2000),and thedistribution of stomatal supports intypeshereraisesquestionsregardingtheplacement anyclade). Farris(1989)developedtheretention index(RC) in an of MurdanniawithinClade 1, and calls fora re- dex (RI) and rescaledconsistency evaluationofthehomologyofthefilament bearding attemptto incorporatethe internalconsistencyof betweenMurdanniaand the othergenerain that each characterwithrespectto the worstand best clade. possiblefitto a tree.The RI and RC bothincorpoAlthoughthe unorderedanalysis produced a ratetheminimumnumberofstepsa character must highly unresolved zygomorphicclade, the re- undergoon any tree,as well as the actual amount on thehypothesizedtree. weightinganalyses each yielded a smallerclade of synapomorphies withinthe zygomorphicgroupcontainingPolyspa- By using thesevalues as bases forre-weighting propertiesof thechartha,Aneilema,Rhopalephora, Commelina, Dictyosper-characters, slightlydifferent This clade is supportedby actersare beingselected.When CI is used, forexmum,and Tricarpelema. a reductionin seed numberin each locule (Tricar- ample,each characteris re-weightedaccordingto pelemais polymorphicforthis character).As rela- the numberof steps it undergoeswhen mapped tionshipsin the tribeCommelineaehave been dif- onto the tree.However,thisschemedoes not take ficultto evaluatedue to a lack of informative char- intoaccountwhetherthecharacter supportsanyof RI and RC, acters(Fadenand Hunt1991),theformation ofdis- the clades in the tree.Incorporating tinct clades based on fruitcharacteristics may however,does take this into account.A good exin thesemeasurescan be providesome insightas to where such characters ample of the differences seen withCharacter#5(numberofflowersper cinmaybe found. A ComparisonofDifferent MethodsofParsimony cinnus).This characterhas a CI of 0.5 (Table 3). Analysis. Severaldifferent methodsof character However,it does notuniteanyclades (Fig.4), so it weightingand orderingwere exploredwith this lends no supportto the topology.Because it does data set,and thequestionarisesas to whichone(s) not supporta clade,it has an RI and rC value of are the best approximation of the truehistoryof zero (Table 3). Therefore,the CI re-weighting the family.The unorderedanalysis containsthe schemeassigned a relativelyhigh weightto that fewesta prioriassumptionsabout characterevolu- characterwhereas the RI and RC decreased its fewassumptionsaboutchar- weight. tion,as it incorporates acterweightor directionoftransformation. While each of the re-weightingschemes proThe absenceof additionalassumptions, however,imposes duced phylogeniessimilarto the unorderedphya penaltyin termsof the potentialloss of phylo- logeny(RC-and Cl-basedanalysesproduceda subFor example,the numberof set oftheunorderedtrees),theyall showedgreater geneticinformation. locules in thefruit(Character35) may representa resolutionthan the unorderedanalysis.Whereas series,in whichfruitswith threeunequal locules the unweightedanalysisproduced a nearlycombe- pletelyunresolvedzygomorphic clade,each of the (e.g., one locule is reduced) are intermediate tweenfruitswiththreeloculesand fruitswithtwo re-weighting analysesproduceda highlyresolved locules.By coding such charactersas ordered,hy- zygomorphic clade. The re-weighting analysesdiftrendsmay be incor- feredfromeach othermainlyin the relationships pothesesabout evolutionary poratedin estimatesofphylogenetic relationships. of the sistertaxa to the zygomorphicclade. These In additionto ordering,characters characwere also re- differences reflectbiases towarddifferent

2000]

679


t.m.c.

)O )Q

07

~~g.m.c.

0

not Xinfrequent

_la

I~

2-celled (e.g., Cuthbertia) 0

\ o0

(e

R

0

%~~~~~ % %

~~~a

0~~~~~~~~~~~~~~~~ 0~~~~~~ I-e led

le 6-cel~~~~~~~ (Geogenanthus)~

FIG. 3.

~ '4 ~

~

~~~~~~L.s.c. (2)

(e.g.,Comm elin ) 2.

stoforthefourstomatal typesfoundintheCommelinaceae: Diagramsdepicting developmental pathways

mata withtwo subsidiarycells,foursubsidiarycells,and two typeswithsix subsidiarycells. Abbreviations: t.m.c., terminalsubsidiarycellmothercell;g.m.c.,guardcellmothercell;l.m.c.,lateralsubsidiarycellmothercell;t.s.c.,terminal lateralsubsidiarycell; ls.c. (2), outermostlateralsubsidiarycell (fromTomlinson subsidiarycell; ls.c. (1), innermost 1966).

680

[Volume 25

SYSTEMATICBOTANY

-s-S

t

c

4-

S

~

0

.

~

? U

>