AmericanJournal ofBotany82(5): 596-606. 1995. DIOECY AND ITS CORRELATES IN THE FLOWERING PLANTS SUSANNE S. RENNER2 AND ROBERT E. RICKLEFS Institute ofSystematic Botany,University ofMainz,Bentzel-Weg 2, D-55099Mainz,Germany; and Department ofBiology,University ofPennsylvania, Philadelphia, Pennsylvania 19104-6018 Considerable effort has beenspentdocumenting correlations betweendioecyand variousecologicaland morphological traitsforthepurposeoftesting hypotheses aboutconditions thatfavordioecy.The dataanalyzedin thesestudies,withfew exceptions, comefromlocalfloras, withinwhichitwaspossibleto contrast thesubsetsofdioeciousand nondioecious taxa withregardto thetraitsin question.However,ifthereis a strong to thepresenceor absenceof phylogenetic component dioecy,regionalsampling mayresultin spuriousassociations. Here,wereport resultsofa categorical multivariate analysis ofthestrengths ofvariousassociations ofdioecywithothertraitsoverall flowering plants.Familieswerescoredforpresence ofabsenceofmonoecyor dioecy,systematic position,numbers ofspeciesandgenera,growth forms, modesofpollination and dispersal, and trophicstatus.Sevenpercent ofangiosperm geographic distribution, genera(959 of 13,500)containat least some dioeciousspecies,and ;6% of angiosperm species(14,620 of 240,000)are dioecious.The mostconsistent associationsin thedata setrelatethepresenceofdioecyto monoecy, windor waterpollination, and climbing At growth. boththefamilyand thegenuslevel,insectpollination is underrepresented amongdioeciousplants.At thefamilylevel,a fromtheassociation positivecorrelation betweendioecyandwoodygrowth resultsprimarily betweendioecyandclimbing growth (whether thetreenortheshrubgrowth woodyorherbaceous) becauseneither forms aloneareconsistently correlated witha family'stendency to includedioeciousmembers.Dioecyappearsto have evolvedmostfrequently via monoecy, of floralsex ratiosbetweenindividualplants.Monoecyitselfis relatedto abiotic perhapsthrough divergent adjustments in theless pollinationand climbinggrowth as revealedby multivariate analysis.Dioecyand monoecyare concentrated advancedsuperorders ofThorne(1992) and subclassesofCronquist ofdioecyfoundin a localflora (1988).The frequency therefore reflects thelevelofdioecyin itsparticular pool offamilies as muchas, or morethan,localselectivefactors. The ofdioecywithabioticpollination andmonoecyarerelatedto floraldevelopmental andmorphological positiveassociations as is thenegativeassociationwithbirdand bat pollination; thepositiveassociationof dioecywithclimbing attributes, is tentatively foroptimalresource Ifrapid growth explainedin termsofdifferential selection allocationto sexualfunction. in climbersand iffruitsetat leasttemporarily is at a premium or requirestheproduction inhibits upwardgrowth growth ofthicker, moreslowlygrowing stemsto support itmight be advantageous topostponefemaleness. Iftheeffect heavyfruits, is strong, thismayfavormaleplants. Overthepast 15 yrconsiderable effort seeds (Bawa, 1980; Givhas beenspent fruitsand frugivore-dispersed documenting correlations betweendioecyand various nish,1980;Steiner,1988;Ibarra-Manriquez andOyama, morphological and ecologicaltraitsforthe purposeof 1992;butseeFox, 1985;Muenchow,1987;Thomsonand testinghypothesesabout conditionsthatfavordioecy. Brunet,1990),smallflowers (BawaandOpler,1975;Fox, Ecologicalcorrelates attracted attention in particular 1985;Ibarra-Manriquez and Oyama,1992),whitetoyelbecausedioecyhas provendifficult to explainsimplyas an low or greenishflowers(Bawa and Opler,1975; Muenoutbreeding mechanismin taxa lackingself-incompati- chow,1987),waterpollination (Cox, 1988),andso-called bility(fordiscussionsof this issue, see Charlesworth unspecializedinsectpollinators (Bawa and Opler,1975; and Charlesworth, 1978; Thomsonand Barrett,1981; Bawa, 1994; contraRennerand Feil, 1993). Charlesworth, 1985;ThomsonandBrunet,1990).AtdifWhile older reportstestedthe associationsbetween ferentspatialand hierarchicalscales,dioecyhas been dioecyand particular attributes morerecent separately, associatedwith,amongotherattributes, windpollination oneshaveusedmultivariate analysis,takingintoaccount (Darwin, 1876; Kerner,1895; Sporne,1949; Stebbins, correlations amongsome of the explanatory variables 1951; Kaplanand Mulcahy,1971; Freeman,Harper,and themselves (Fox, 1985;Muenchow,1987;Steiner,1988; andOyama,1992).Someauthorshave Ostler,1980;Muenchow,1987;Steiner,1988),monoecy Ibarra-Manriquez (Yampolskyand Yampolsky,1922; Lewis, 1942; Wes- attemptedto incorporatephylogenetic in information tergaard, 1958; McComb,1966; Lloyd,1982),perennial analysesoftheassociationofdioecywithothertraitsby growth andlignified secondary xylem,orwoodiness(Dar- lookingat thespecies,genus,and family levelsseparately win, 1876; Stebbins,1950; Baker, 1959; Croat, 1979; or by restricting analysesto monophyletic groups(Hart, Conn,Wentworth, and Blum, 1980; Freeman,Harper, 1985; Donoghue,1989; Cox, 1990, 1993; Lahav-Ginott andOstler,1980;Fox, 1985;butseeSteiner,1988),fleshy and Cronk,1993). Thedataanalyzedinthesestudies, witha fewexceptions 1 Manuscript allcome analyses, received19 May 1994;revisionaccepted23 November suchas Cox's(1990, 1993)phylogenetic 1994. fromlocal floras.Such locallycircumscribed data allow The authorsthankD. Charlesworth, P. Cox,P. Endress, C. Freeman, oneto contrast thesubsetsofdioeciousandnondioecious S. Heard,J. Kohn,-J. Thomson,and K. Bawa forcomments on the taxawithregardto particular ecologicaland morphologmanuscript; and P. Berry(Onagraceae), H.-J.Esser(Euphorbiaceae), S. For example,therelationships beKnapp(Solanaceae),J.Kuijt(parasitic angiosperms), C. Ott(Menisper- ical traitsofinterest. tweena particular floralcoloror size and dioecycan be maceae),andJ.Rohwer(Lauraceae)forinformation on dioecyin paranalyzedonlyiffloralcolorsor sizesfortheuniverseof ticulargroups. 2 Authorforcorrespondence. nondioeciousspeciesare known.At present, thisis pos596 May 1995] RENNER AND RICKLEFS-DIOECY IN THE FLOWERING PLANTS 597 of Thorne(1992); total sible onlyon a local basis. No studyhas yetaddressed (1988) and the 28 superorders plantsat large, numberofgenera(in thefamilydata set)and numberof thecorrelatesof dioecyin theflowering and some apparentassociationsamongdioecyand par- dioeciousspecies(in thegenusdata set);growthforms; distribube theresultofregionalef- modesofpollinationand dispersal;geographic ticulartraitsmay,therefore, com- tion; and trophicstatus.Generaacceptedare thoseof ifthereis a strongphylogenetic fects,particularly (1992); familiesarethoseofCronquist(1988), ponentto thepresenceor absenceofdioecy.As longas Brummitt theymay exceptthatin 18 cases morerecent,broaderfamilyconsuchsamplingproblemsremainunrecognized ad hoc hypoth- ceptswereaccepted.To look at theeffects ofdifferences ofunnecessary resultin theintroduction in familialconcepts,thedata setswerealso brokendown thatdid notplaya eses to explainsupposedcorrelations intothemorenarrowly circumscribed familiesofThorne causal rolein theevolutionofdioecy. Here we reportresultsof a studyin whichwe have (1992). Dioecyandmonoecywerescoredas present ina family ofvariousassociations therelativestrengths determined traitsacrossall flowering whenat leastone speciesin thatfamilyis dioeciousor betweendioecyand particular in thecase offamilieswithfew approach,thepos- monoecious.Particularly plants.Becausewe use a multivariate associationsamongthetraitsthem- dioecioustaxagreatcarewastakentosearchtheliterature sibilityoffortuitous confirmation of theirdioeciousstatus. selvesis reduced.Ouranalysis,to somedegree,also takes forindependent to Numeroustaxa(:20 familiesand ;:200 genera)listedas bias becauseit is not restricted care of phylogenetic particularfamiliesthathappento dominatelocal floras includingdioeciousspeciesbyYampolskyand Yampolstriking exampleof sky(1922) had to be removedfromthedata set,either (see Steiner,1988,fora particularly dioeciousplantshaveturnedout in theCape floraof SouthAfrica).It is based becausethesupposedly thiseffect ofdioeciousangiosperm polygamous, monoecious,or gynogeneraand to be sex-changing, on a newcompilation and taxonomic families,the onlyprevioussuch compilationby Yam- dioecious,or because of nomenclatural polskyand Yampolsky(1922) beingoutdated.Systema- changes.Yampolskyand Yampolskybased theircomwithinthe pilationonEnglerandPrantl'sNatiirlichePflanzenfamilirelationships tists'viewsof thephylogenetic and systematic en and thevolumesof Engler'sPflanzenreich published angiospermsand the circumscription placementofnumerousgenerahave changedduringthe by 1912. Theyconcludedthatout ofthetotalof 10,113 and angiosperm 972(9.6%)comprised generathenrecognized, past 70 yr,newcases ofdioecyhave beenreported, manyothersmentionedby Yampolskyand Yampolsky dioeciousand/orpolygamodioecious species.However, haveturnedouttobe in error.In anotherpaper,thesame theirlistincludesat least 90 generasincemergedwith ofdioecy othergenera;some,suchas Streblus, arelistedeighttimes data setwillbe used to studythedistribution phy- underdifferent in theframework publishedangiosperm names. There are also 43 unpublished ofrecently names(nominanuda),theidentity ofwhichis obscure. data). logenies(Renner,unpublished viewatthefamily leveldioe- Severalothergenera,froma modernsystematic weanalyzewhether Specifically dis- point,arecompletely misplacedas tofamily. Conversely, geographic cyis associatedwithmonoecy,particular 170 additionaldioeciousgeneraand manyadditional windor watervs. animalpollination,fleshy tributions, dis- familieswereadded eitherbecauseofnewlydiscovered vs. abiotically animal-dispersed fruits and,therefore, ofdioecyorbecauseoftaxonomic changes, such forms. Becausewelacked instances growth persedseeds,orparticular modernfamilyconcepts.Our data setconwe wereunableto addressas- as narrower data forall angiosperms sociationsbetweendioecyand small,greenishflowers tains959 generathatincludedioeciousspecies. Mostdioeciousgeneracompriseonlydioeciousspecies; (Bawa and Opler, 1975; Fox, 1985; Muenchow,1987; Ibarra-Manriquez and Oyama,1992)andbetweendioecy however,325 ofthe959 generahavemorethanonematand pollinationby small, supposedgeneralistinsects ing system,and forthemnumbersof dioeciousspecies fromfigures giveninrecentliterature, maybe a correlateof theas- hadtobe estimated (Bawa, 1994). The former suchas "rarelydioecious"or belowbetweenwindpollination usuallyfloras.Statements sociationdemonstrated as areoften small "occasionallydioecious"werescoredconservatively flowers anddioecy,becausewind-pollinated and lackattractive coloration;thesecondis basedon the at least one dioeciousspecies,while"commonlydioewith cious"was scoredas 50% dioecious.Speciesnumbersare smallnessis correlated erroneous ideathatpollinator and inefficient notused in theanalyses. relationships pollinator-flower unspecific Growthformsweretakenfromflorasand accountsof pollination(Rennerand Feil, (in termsof outcrossing) particularmodes of growth,such as Gentry(1991) for 1993). are fromMabdistributions climbers,whilegeographic work berley(1987),exceptwheremorerecentsystematic MATERIALS AND METHODS was available.Growthformwas tabulatedas presenceor shrubs, herbs,orclimbers (whether woody This studyis based on twodata sets,one a data base absenceoftrees, of dioeciousangiospermgeneracompiledby searching or herbaceous). Pollinationand dispersalmodes wereassessedfrom withspecialists(see and by corresponding theliterature of the flowersand fruitsof each taxonin theothera data base of all families descriptions Acknowledgments), andmonographic sources(fordetails,see offlowering plants.For detailsofthesecondcompilation variousfloristic andRenner(1994). Thegenusandthefamily Ricklefsand Renner,1994). In bothdata sets,thecateseeRicklefs information foreach gorieswere1) abiotic,2) biotic,and 3) variable,thelast data bases containthe following or dispersalpresentin taxon: presence or absence of dioecy, presence or meaningbothtypesofpollination In thegeneric specieswithinthegenusorfamily. absenceofmonoecy(in thefamilydata set);positionin different scoredas 1) intermsofthe11subclassesofCronquist data set,pollinationmodewas additionally theangiosperms 598 AMERICAN JOURNAL OF BoTANY [Vol. 82 wind,2) water,3) insects,4) bats,5) birds,6) variable, repeatedtheanalysesusingonlysmallfamilies(lessthan or7) unknown. Data onthepollination whichheightens ofmanydioecious fivegeneraperfamily), thelikelihoodof taxa had alreadybeen compiledin connectionwithan co-occurrence ofthevariablesin thesamegenusor speearlierreviewof theliterature (Rennerand Feil, 1993). cies; second,we also examinedassociationsat thegenus A fewadditionalautecologicalstudieshave sincebeen levelwherepossible. published(e.g.,Kato, 1993), and thereare severalsysThe statistical significance of each of thevariablesis tematicreviews(Dobat and Peikert-Holle, 1985; Cox, evaluatedby a x2 withits associatedprobability. Each 1988; Kubitzki,Rohwer,and Bittrich,1993; Webster, model has a likelihoodratio(LR) which,if significant, 1994a,b). For manyoftheremaining genera,pollination indicatesadditionalstructure inthedatasetnotaccounted by eitherwindor insectscould be assessedfromfloral forbythemodel.Initialmodelsincludedno interactions. structure, withparticular weightbeinggivento thepres- Nonsignificant variablesweredeletedfromthemodeland ence or absenceofnectaries.All fleshy whenLR x2 values were significant, as wellas fruits, interactions were indehiscent largedryfruits, werescoredas beingadapted addedbytrialand errorto arriveat a modelwitha nonforanimaldispersal.Some smalldryfruits LR x2. Becausemonoecyis a matingsystem scoredas abi- significant oticallydispersedmayhave seedsinitially dispersedbal- thatmay facilitate the evolutionof dioecy(see below), listically and thenbiotically (byants);itis unlikely, how- thesemodelswererunbothwith(+M) andwithout (-M) ever,thatthisgroupintroduces a largeerrorintothedata. monoecyas a variableto isolatetheeffects ofecological Wheredescriptions offloralstructures on dioecy. and/orfruits were interactions insufficient to assessreproductive modes,thecondition wasleftunscoredinthegenericdataset;all families could RESULTS be scored. For the family-level analysis,geographicdistribution Out ofthetotalof 13,479genera(Brummitt, 1992),at was coded as presentor absentin tropicallatitudesand least 959 (7.1%) containdioeciousspecies;if Thorne's is presentor absentin temperatelatitudes.The category (1992) estimateof 12,650generafortheangiosperms is 7.6%.A conservative estimate in- accepted,thatpercentage "tropical"includesall familieswhosedistributions cludeeitherlow or highelevationsin thetropics;"tem- ofthenumberofdioeciousspeciesis 14,620or 6% ofan plants.Of perate"includesall familieswhosedistributions include assumedtotalof240,000speciesofflowering circumscribed 365 familiesof angioorborealzones-generally, temnperate zoneswithfrost. The a conservatively thefamiliesof Cronquist,1988; see distributions ofthedioeciousgenerawerescoredin con- sperms(essentially siderabledetailforfuture analysisbecausemanygenera Materialsand Methods),157 (43%) containdioecious OfThorne's(1992) morenarrowly defined withvariablematingsystemsalso have theirdioecious members. 437 speciesconfinedto a particular region.For example,in families,167 (38%) have dioeciousmembers. In thefollowing we first resultsofanalysesperpresent Asperula(Rubiaceae),the 16 Australianspeciesare didata set(Tables 1-12); this oecious whilethe Mediterranean speciesare hermaph- formedusingthefamily-level is followedby resultsbased on genus-level data (Tables roditic. Each variablewas scoredas 0 (absent)or 1 (present). 13 and 14). analysesrelatingeach variableindividuWithineach of thecategoriesof variables(distribution, Contingency in Table 1. Variablesthatare matingsystem,pollination, dispersal,growthform,tro- allyto dioecyarepresented and strongly associatedwithdioecyare,in phic status)each familyhad to have a scoreof 1 forat significantly oftheircorrelation, climbleastone ofthevariablesbutcouldalso have a scoreof orderofthestrength monoecy, form, bioticdispersal, abioticpollination, shrub 1 forall ofthem.None couldhave scoresof0 forall of inggrowth them.For some of the analyses,the familieswerealso growthform,and tropicaldistribution. Categoricalmodels are presentedin Table 2. The dividedintotwogroupsbasedon numberofgenera:size in an initialmodel(no interactions) effects were = 0, fewerthan 5 genera; size = 1, 5 or more genera. strongest In orderto examinethe relationship growth form,bioticdispersal,abiotic of dioecyto all monoecy,climbing The individualasand tropicaldistribution. variablessimultaneously, categoricalmultivariate con- pollination, sociationof shrubbygrowthwithdioecyevidently was tingency analyses(SAS Institute, Inc., 1988,CATMOD withother correlation in whichthepresenceor ab- subsumedin thismodelthrough procedure)wereperformed senceofdioecywas relatedto presenceor absenceofall variables.In thisanalysis,thelikelihoodratiowas sigin thedata setin additionalstructure indicating the othervariablesin the data set. Categoricalmodels nificant, A seriesofmodelswithvarious take into accountassociationsamongthe independent theformofinteractions. termswerethenrun.The onlysigvariablesand revealtheuniquestatistical re- two-wayinteraction (predictor) interaction was abioticpollinationx monoecy. betweendioecyandothervariablesinthedata nificant lationships was includedand nonsignificant set.Whenappliedat thetaxonomicleveloffamily, this Whenthisinteraction weredeletedfromthemodel,thelikelihoodratio theassociationof traitswithintaxa effects approachidentifies (Table 2, model+M). Thus,sevbut does not implythata pair of associatedtraitsnec- becamenonsignificant to thepresenceof independently essarilyoccursinthesamegenusorspecies.Indeeddioecy eral factorscontribute indescending orderofimportance: andmonoecycannotco-occurina singlespeciesandhence dioecywithina family, abioticpollination, shrub theirassociationcouldonlybe identified bya higher-level monoecy,tropicaldistribution, howandclimbing form, growth form.In addition, withthisapproachis thattwotraits growth analysis.A difficulty withabioticpollinationin conmaybe associatedbecauseeach is morelikelyto be rep- ever,monoecyinteracts to dioecy(Table 3). When monoecywas exresentedin familieswithlargenumbersof species.We tributing in two ways:first, analysis(Table2,model-M), triedto deal withthis difficulty we cludedfromthecategorical May 1995] TABLE 1. Family-level relationshipof presence (in 157 families) or absence (in 208 families)of dioecy to each of the variables. and of abioticpollination tableof theeffects 3. Contingency areincludedin bioticdispersalon presenceofdioecy.All families withdioecyovertotal offamilies theanalysis.Ratiosarenumbers numberoffamilies.a TABLE Number of families Variable absent Categoryand variable Monoecy Dioecy absent Dioecy present 179 54 Variable present Monoecy Dioecy present x2a 29 103 107.6 0.000 Dioecy absent 41 111 12 80 167 97 145 77 11.2 0.2 0.001 0.648 177 24 104 33 31 184 53 124 17.9 6.0 0.000 0.014 Dispersal Abiotic Biotic 55 117 59 46 153 91 98 111 5.1 26.8 0.023 0.000 Growthform Herb Shrub Tree Climber 103 119 118 157 87 59 72 72 105 89 90 51 70 98 85 85 1.2 13.9 4.2 33.8 0.264 0.000 0.039 0.000 Parasite Size of family 203 130 149 53 5 78 8 104 1.9 30.0 0.172 0.000 X2 is the likelihood ratio x2 value. abioticpollinationand climbinggrowthformremained ofdioecy. strongcorrelates Because,on average,largefamiliescontainmoredioecioustaxa thando smallfamilies(Table 1), thedata weresplitintotwosetshavingnearlyequal numbersof familiesbased on numberofgeneraperfamily(lessthan thestronfive,and fiveor more).Amongsmallfamilies, are monoecyand thepollination gestindividualeffects and dispersalvariables,withdioecyless oftenpresentin and abioticdispersaland familieswithbioticpollination moreoftenpresentin familieswithabioticpollination TABLE Absent p Distribution Tropical Temperate Pollination Abiotic Biotic a Abioticpollination 33/198 Absent (17%) 21/35 Present (60%o) 54/233 Total (23%) Total Present 71/83 (86%) 32/49 (65%) 103/132 104/281 (37%) 53/84 (63%) 157/365 (43%) (78%) betweenmonoecyand dioecyin theabsenceof The relationship had LR X2= 123.4,P < 0.001,andin thepresence abioticpollination LR X2 = 0.2,P = 0.62. ofabioticpollination, a and bioticdispersal(Table 4, butsee below).In addition, herbsandmore dioecyis lesslikelyinfamiliescontaining shrubsand trees.A categorlikelyin familiescontaining and including effects nonsignificant ical modelexcluding of monoecywithabioticpollinationhad theinteraction likelihood-ratio andrevealedabioticpola nonsignificant growth lination,bioticdispersal,monoecy,and climbing effects (Table 5, Model +M). to be the onlysignificant contributed bymonoecyis conMuchoftheinformation tainedin themonoecyx abioticpollinationinteraction (Table 6). Whenmonoecywasexcludedfromtheanalysis (Table 5, Model -M), we couldnotfinda whollysatismodel (LR, P = 0.034), but the same variables factory rebioticdispersal, climbing growth) (abioticpollination, mainedsignificant. effects Amonglargefamilies(Table 7), the strongest andshrubgrowth bioticdispersal, climbing weremonoecy, withabioticdispersal forms,and tropicaldistribution, A categorical modelshowed effect. a weaknegative exerting to be theonlysignifmonoecyand tropicaldistribution at thefamily levelbetweendioecyand theindependent 2. Categorical modelsoftherelationship variablesin thisanalysis. Initial model Variable Monoecy Tropicaldistribution Temperate distribution Abioticpollination Bioticpolliantion Abioticdispersal Bioticdispersal Herbgrowth form Shrubgrowth form Treegrowth form form Climbinggrowth Abioticpollinationx X2 52.1 4.8 0.5 5.7 0.1 0.0 6.1 0.4 2.0 0.0 10.1 Model +M P 0.000 0.029 0.500 0.017 0.710 0.912 0.014 0.539 0.157 0.874 0.002 X2 192.2 (143)b Variablenotincludedin analysis. b Degreesoffreedom forlikelihoodratio. 0.004 Model -M P X2 P 27.6 11.7 0.000 0.001 5.2 0.023 8.5 0.004 39.9 0.000 10.7 0.001 8.3 0.004 36.8 0.000 26.1 monoecy interaction Abioticpollinationx bioticdispersal Tropicaldistribution x bioticdispersal Likelihoodratio a 599 DIOECY IN THE FLOWERING PLANTS RENNER AND RiCKLEFS- 19.5 (20) 0.000 0.493 -a - - 6.3 0.012 6.9 11.4 0.008 0.249 (10) 600 [Vol. 82 AMERICAN JOURNAL OF BOTANY 4. Relationship ofpresence(in 53 families) orabsence(in 130 families)ofdioecyto each ofthevariablesindependently among smallfamilies(<5 genera). TABLE 6. Contingency tableof theeffects of abioticpollination and bioticdispersalon presenceof dioecy.Onlysmallfamilies(<5 genera)areincludedintheanalysis.Ratiosarenumber ofdioecious familiesovertotalnumberoffamilies.a TABLE Number of families Variable absent Categoryand variable Dioecy absent Monoecy Variable present Dioecy present Dioecy absent 113 31 Monoecy Distribution Tropical 30 9 74 28 Temperate Pollination Abiotic 20 107 20 Biotic 23 Dispersal 46 Abiotic 30 Biotic 76 21 Growthform Herb 40 75 Shrub 80 25 Tree 71 20 117 Climber 43 a x2 is thelikelihood ratiox2. Absent Dioecy present P x2 17 22 16.8 0.000 100 56 44 25 0.9 0.3 0.35 0.61 23 110 33 30 33.8 15.4 0.000 0.000 84 54 23 32 7.0 5.4 0.009 0.020 55 50 59 13 13 28 33 10 5.3 3.2 4.3 2.5 0.021 0.076 0.038 0.112 Present Abioticpollination Absent 13/114 (11%) Present 18/30 (60%) Total 31/144 (22%) Total 7/13 (54%) 15/26 (58%) 22/39 (56%) 20/127 (16%) 33/56 (59%) 53/183 (29%) a The relationship betweenmonoecyand dioecyin theabsenceof abioticpollination had LR x2 = 11.8,P < 0.001,and in thepresence of abiotic pollination,LR x2 = 0.0, P = 0.86. sistentassociationsin thedata setrelatethepresenceof tomonoecy, abidioecywithina family climbing growth, oticpollination, andbioticdispersal;tropicaldistribution is associatedwithdioecyonlyin largefamilies. Becausemonoecywas suchan important factorin asof sociationwithdioecy,we examinedtherelationship monoecyitselfto thedistribution ofothervariables(excludingdioecy)withinfamilies (Table 11).Severalfactors icanteffects, but therewas a stronginteraction between had significant associationswith monoecy:climbing bioticdispersaland abioticpollination(Table 8, Model growth (+), abioticpollination (+), bioticpollination (-), +Ma). Climbinggrowthformwas notsignificant. bioticdispersal(+), and temperate distribution (+). A The interaction effect ofabioticpollinationand biotic categorical modelofthedependence ofmonoecyon other dispersalin largefamiliescanbe visualizedbyexamining variablesproducedthe following results:abioticpollithecontingency tablerelating presenceorabsenceofdioe- nation(x2 = 46.8, P = 0.000), climbinggrowth(36.9, cy to combinationsof presenceand absenceof each of 0.000), temperate distribution (3.9, 0.049), abioticpolthesevariables(Table 9). Bioticdispersalandabioticpol- linationx bioticdispersalinteraction (11.2,0.001),likelinationwerenot individuallysignificant effects. How- lihoodratio(12 df,23.4,0.025).Amongfamilies including ever,familieswithneitherof theseattributes included taxawithabioticpollination, presenceorabsenceofbiotic fewerdioecioustaxathanfamilieswitheitheroneorboth dispersalhad no influence on monoecy(60.9%vs. 55.3%, of them.The same is trueof the interaction between abioticpollinationand monoecy(Table 10),whichalso amonglargefamilies(?5 genera)ofpresence makesa suitablecategorical model(Table8,model+ Mb). TABLE 7. Relationship (in 104 families)or absence(in 78 families)of dioecyto each of Whenmonoecywas deletedfromthecategorical model thevariables. forlargefamilies,abioticpollination, tropicaldistribution,and climbinggrowthformweresignificant effects, Number of families alongwiththeabioticpollinationx bioticdispersalinVariable present Variable absent teraction (Table 8, model -M). Categoryand Dioecy Dioecy Dioecy Dioecy To summarizethecategorical variable absent present absent analyses,themostconpresent P x2 TABLE 5. Categorical modelsforsmallfamilies(<5 genera)of the betweendioecyand theindependent relationship variablesin this analysis. Model +M Variable Monoecy Abioticpollination Bioticdispersal form Climbing growth Abioticpollinationx monoecy interaction Abioticpollinationx form climbing growth Likelihoodratio (df) Model -M X2 P 6.2 0.013 13.4 7.8 4.3 0.000 0.005 0.039 4.8 0.028 15.6 (10) 0.111 x2 - 12.3 9.8 9.5 - 4.0 10.4 (4) P - 0.000 0.002 0.002 - 0.046 0.034 66 23 Monoecy Distribution 11 Tropical 3 37 52 Temperate Pollination 84 Abiotic 70 Biotic 4 10 Dispersal 29 Abiotic 9 Biotic 41 25 Growthform 47 Herb 28 34 Shrub 39 47 52 Tree Climber 40 29 a x2 iS thelikelihood ratiox2. 12 81 75.3 0.000 67 41 101 52 8.1 0.1 0.005 0.73 8 74 20 94 2.9 1.3 0.09 0.25 69 37 75 79 7.6 15.7 0.006 0.000 50 39 31 38 57 70 52 75 1.6 5.6 1.9 10.4 0.21 0.018 0.17 0.001 May 1995] RENNER AND RICKLEFS-DIOECY bemodelsforlargefamiliesoftherelationship 8. Categorical variablesin thisanalysis. tweendioecyand theindependent TABLE Variable Monoecy Tropicaldistribution Abioticpollination form Climbinggrowth Abioticpollinationx Bioticdispersal Abioticpollinationx Monoecy interaction Likelihoodratio (df) Model +Ma Model +Mb Model -M x2 x2 x2 p - - P 48.9 0.000 4.2 0.040 p 20.9 0.000 12.4 0.000 14.7 0.000 13.3 0.360 (12) 7.6 0.006 9.7 0.002 5.8 0.017 19.9 0.000 7.1 0.008 3.5 0.360 (5) - - 9.8 0.455 (10) x2 = 0.3, P = 0.60). Amongfamilieslackingabioticpollination,presenceof bioticdispersalincreasedthelikelihood of monoecy(39.0% vs. 16.2%,x2 = 17.9, P < 0.001). herbaceousor woody,also whether Climbinggrowth, factorin association and unexpected, was an important, ofthis withdioecy,and so we examinedtherelationship ofothervariableswithin growthformto thedistribution families(Table 12). Significant associationswere:dioecy (+), monoecy(+), bioticdis(+), tropicaldistribution (+), form(+), bioticpollination persal(+), shrubgrowth and abioticpollination(-). A categoricalmodelof the dependenceof climbinggrowthon othervariablesprogrowth(X2= 26.5, results:shrubby ducedthefollowing P = 0.000), herbaceousgrowth(21.0, 0.000),bioticdispersal(8.3, 0.004), monoecy(8.1, 0.004), abioticpolli(22 df,24.9, 0.30). nation(5.2, 0.022), likelihood-ratio herb x In addition,thereweretwo stronginteractions: shrubgrowthform(19.0, 0.000) and abioticpollination x monoecy(6.8, 0.009). sectionspresentresultsbased on thegeThe following nus-leveldioecydata set. Detailed global comparisons of the betweenthedioeciousgeneraand theremainder angiosperm generawerenotfeasable.However,because thisis thefirst compilationofdioecyin theangiosperms since Yampolskyand Yampolsky(1922) and Charlesofseveral worth(1985)andbecausethedataaresuggestive trendsamongdioecioustaxa,suchas thelikelyoverre(see below),we hopethat ofheterotrophism presentation thesedataherewillstimulate further datacolpresenting analyseson thegeneric lectionand eventualmultivariate level. and of abioticpollination tableof theeffects 9. Contingency bioticdispersalon presenceof dioecy.Onlylargefamilies(25 ofdioecious Ratiosarenumbers genera)areincludedintheanalysis. familiesovertotalnumberoffamilies. and ofabioticpollination tableoftheeffects 10. Contingency monoecyon presenceof dioecy.Onlylargefamilies(25 genera) ofdioeciousfamareincludedin theanalysis.Ratiosarenumbers iliesovertotalnumberoffamilies.a TABLE Monoecy Absent Biotic dispersal Abioticpollination 14/53 Absent (26%) 11/13 Present (85%) 25/66 Total (38%) Present 70/101 (69%) 9/15 (60%) 79/116 (68%) Total 84/154 (55%) 20/28 (7 1%) 104/182 (57%) Total Present Abioticpollination 84/154 64/70 20/84 Absent (55%) (91%) (24%) 20/28 17/23 3/5 Present (7 1%) (74%) (60%) 104/182 81/93 23/89 Total (57%) (87%) (26%) a The relationship betweenmonecyanddioecyin theabsenceofabihad LR x2= 79.1,P < 0.001, and in thepresenceof oticpollination LR x2= 0.4,P = 0.54. abioticpollination, Dioecy and phylogeny-Dioecy is slightlymore common among dicot genera than among monocot genera (8.2% vs. 5.1%; x2 = 30, P < 0.001). It occurs in 24 of the 28 superordersof Thorne (Table 13) and in each of the six dicot and five monocot subclasses of Cronquist (1988). Because some ofCronquist'ssubclassesare poorly the followinganalysis definedand probablypolyphyletic, concentrateson Thorne's superorders,which because of theirnarrowercircumscriptionstand a betterchance of being monophyletic. Among the 14 dicot superordershaving > 100 genera the distributionof dioecious generais extremelyheterogeneous (X213 = 909, P < 0.0001). In absolute numbers, most dioecious genera are found among Thorne's Malvanae (208), his Magnolianae (134), Violanae (90), Rutanae (71), Commelinanae (62), Theanae (58), and Gentiananae (49). In relativeterms,dioecy is concentratedin monocotyledonous Pandananae (100%), Alismatanae (47%), and Triuridanae (25%) and the dicotyledonous Rafflesianae(80%), Magnolianae (29%), and Malvanae (28%). The Magnolianae and Malvanae are thus among the most importantdioecious clades, in both relativeand absolute terms.This is mainly due to a few dioecy-rich families,such as the Lauraceae, Menispermaceae, Myristicaceae,and Monimiaceae in the firstsuperorderand the Euphorbiaceae, Moraceae, and Urticaceae in the secofthepresence 11. Family-level (in 132families) relationships or absence(in 233 families)ofmonoecyto eachoftheothervariables. TABLE TABLE Absent 601 IN THE FLOWERING PLANTS Variable X2 is Variable present Mon- Monoecy oecy absent present Mon- Monoecy oecy absent present 40 13 132 59 198 83 31 26 44 70 116 47 130 60 119 59 123 67 168 61 226 126 thelikelihoodratiox2value. Tropicaldistribution distribution Temperate Abioticpollination Bioticpollination Abioticdispersal Bioticdispersal form Herbgrowth form Shrubgrowth form Treegrowth form Climbinggrowth Parasitehabit a Variable absent 193 101 35 207 163 117 103 114 110 65 7 x2. P 119 3.8 0.057 4.8 0.028 73 49 22.5 0.000 101 9.4 0.002 0.4 0.516 88 6.9 0.009 85 72 3.6 0.057 1.4 0.241 73 0.1 0.709 65 71 23.9 0.000 0.6 0.452 6 602 12. Family-level ofthepresence relationships (in 136families) ofclimbing orabsence(in 229 families) formtoeachofthe growth othervariables. TABLE Variable absent Variable Tropicaldistribution Temperatedistribution Monoecy Dioecy Abioticpollination Bioticpollination Abioticdispersal Bioticdispersal form Herbgrowth form Shrubgrowth Treegrowth form Parasitehabit a [Vol. 82 AMERICAN JOURNAL OF BOTANY X2 5 76 65 51 118 10 39 39 62 45 72 132 Number of families Variable present Climb- Climb- Climb- Climbers ers ers ers absent present absent present 48 115 168 157 163 47 75 124 128 133 118 220 Total ofdioecyinThorne's(1992)superorders. TABLE13. Distribution ofgenerainthistablearefromThorne,whilethefamilies numbers as in Cronquist(1988; see Methods). arecircumscribed 181 114 61 72 66 182 154 105 101 96 111 9 131 60 71 85 18 126 97 97 74 91 64 4 x2. 24.4 1.1 23.9 33.8 12.4 12.4 0.7 23.0 3.6 21.7 0.1 0.2 P 0.000 0.294 0.000 0.000 0.000 0.000 0.415 0.000 0.057 0.000 0.794 0.617 is the likelihood ratio x2value. Superorder Dicots 1 Magnolianae 2 Nymphaeanae 3 Rafflesianae 4 Caryophyllanae 5 Theanae 6 Celastranae 7 Malvanae 8 Violanae 9 Santalanae 10 Geranianae 11 Rutanae 12 Proteanae 13 Rosanae 14 Comnanae 15 Asteranae 16 Solananae 17 Loasanae 18 Myrtanae 19 Gentiananae Dio- MonoTotal ecious ecious 34 2 3 12 40 2 20 26 9 16 23 1 39 23 6 8 1 12 0 1 9 14 1 17 13 8 4 8 1 15 16 3 2 0 14 0 1 7 10 1 12 11 6 3 8 1 14 11 2 1 0 Total Dioecious genera number of NumPercent ber genera 466 134 0 81 8 10 563 30 540 58 7 60 769 208 725 90 162 37 7 189 1,129 71 2 75 381 36 650 44 1,251 28 4 296 0 13 29 0 80 5 19 12 27 12 22 4 6 2 9 7 2 1 0 ond superorder. Familieswiththehighest concentrations of dioeciousgenerain bothabsoluteand relativeterms are the Menispermaceae (100% of thegeneraare dioe1 2 3 463 3 13b cious),Myristicaceae (78%),Moraceae(62%),Urticaceae 2 6 2,135 49 5 23 (52%), Anacardiaceae(50%), Monimiaceae(47%), Eu8 305c 132 110 9,958 817c Total dicots phorbiaceae(39%), and Cucurbitaceae (32%). 1 2 1,184 16 8 15 Lilianae Amongthesubclasses,Cronquist(1988, pp. 263-265 20 1 0 0 2 0 0 21 Hydatellanae and pp. 451-457) considerstheMagnoliidae,Hamamel- 22 Triuridanae 2 25 1 8 1 1 idae, and Caryophyllidae as less advancedthantheDil- 23 Alismatanae 11 57 27 47 7 6 1 1 2 2 1 108 leniidae,Rosidae,and Asteridaeamongdicots,and the 24 Aranae 11 0 1 1 0 0 Alismatidaeand Arecidaeas relatively moreprimitive 25 Cyclanthanae 3a 0 3 100 1 1 thattheCommelinidae, andLiliidaeamong 26 Pandananae Zingiberidae, 1 1 1 200 31 15 27 Arecanae in thelessad- 28 Commelinanae monocots.Dioecyis strongly concentrated 5 9 1,192 62 26 6 vancedsubclasses.Also Thorne(1992,p. 244, fig.1) tries 5 60 25 22 2,765 142 Totalmonocots to arrangehis superorders so that"The positionofthese 7.5 365 157 132 12,723 959 plants Totalflowering of the phyleticshrub stem,cross-sections superordinal a SomePandanaceaeare monecioussex-changers thanconsisrather indicatesas closelyas possiblemyinterpretation oftheir tent dioecists(Cox, 1982). relationships and theirrelativedegreeof specialization b This numberis inflatedby Cronquist'srecognition of Punicaceae fromtheirmoreprimitive, archaicancestors."We have and Sonneratiaceae fromLythraceae. as distinct c Including in hissequence(Table 13) and numbered thesuperorders families thatareunassigned byThorne: fourofCronquist's andPandaceae;ofthese, Corynocarpaceae, Barbeyaceae, dividedthe dicotsuperorders roughlyintotwogroups, Aextoxicaceae, and Barbeyaaredioecious. one relativelyless advanced (numbers1-9), the other Aextoxicon relatively moreadvanced(numbers10-19).As withCronin theless adquist'ssubclasses,dioecyis concentrated Old World tropicsthan in the New as suggestedby Thox2 = 519,P < 0.0001; vancedsuperorders ofdicots(genera: families:x2 = 5.4, P < 0.025). Monoeciousfamilieshave mas and LaFrankie (1993), due primarilyto the highOld a paralleldistribution todioeciousfamilies (Table 13)and World diversityof such dioecy-richfamiliesas the Euin less advanceddicot phorbiaceae, Menispermaceae,and Restionaceae. exhibita similarconcentration superorders (X2= 4.4,P < 0.05). Possiblereasonsforthe Growth genera-We have angiosperm formsofdioecious concentration of dioecyin theless advancedclades are informationon thegrowthformsof 818 ofthe959 genera givenin thediscussion. Geographicdistribution of dioeciousangiosperm genera-Of the959 dioeciousgenera,217 occurin theneo86 are pantropical, and tropics,402 in thepaleotropics, in thetemperate zone. Thirty 149 are foundexclusively generahave temperateand tropicalmembers.The-reto Hawaii,New maining75 generaare mostlyrestricted NewCaledonia,orNewSouthWales. Zealand,Australia, inthelargegeographic Becausethetotalnumbers ofgenera regionshave notbeentabulated,it is notpossibleto calculatethedifferent relativefrequencies of dioecy,butit in the seemslikelythatdioecyis indeedmorefrequent with dioecy. Of the 818, 198 (24%) comprise only tree species, 157 (19%) only herbs, 128 (16%) only climbers, and 94 (11%) only shrubs.Tree and shrubspecies occur togetherin 195 (24%) generawith dioecy. Other combinations of growthforms,such as trees,shrubs,and herbs (seven genera),shrubsand herbs(19 genera),shrubsand climbers(ten genera),and so on are rare. Of the genera with dioecy, :z551 (67%) are woody (namely those with trees and shrubs or both growthforms,plus half those with only climbers).In the angiospermsas a whole, 234 familiesincludingtreeand shrubgrowthformshave 7,929 genera,or 59% of the total 13,500 genera; 182 families May 1995] RENNER AND RICKLEFS-DIOECY 14. Genus-level analysisoftheassociationbetween dioecyand pollinators. TABLE Number of genera Pollinated by Insects Windor water Batsor birds With dioecy 550 (5.1%) 232 (21.0%) 4 (0.4%) Lacking dioecy 10,317a 874 750 IN THE FLOWERING PLANTS 603 species(Porsch,1931;and personal withbird-pollinated plants(Table 14) it estimate)in therestoftheflowering pollinationare both appearsthatinsectand vertebrate amongdioeciousgenera.Thiswas also underrepresented notedby,forexample,Richards(1986);belowwediscuss possiblereasonsforthenearabsenceof bat pollination amongdioecists. genera-Ofthe angiosperm Dispersalmodesofdioecious data 814 dioeciousgeneraforwhichthereare sufficient orpredomon modeofdispersal,526 (65%) areentirely bywind, 254(31%)aredispersed animal-dispersed, inantly and sixhavebothbioticand abi28 arewater-dispersed, oticmodesofdispersal.We haveno dataon thedispersal data set,1,782 modesof 145 genera.In thefamily-level withwhollybioticdispersal generabelongto 114families and 5,362generabelongto 88 familieswithbothabiotic havingonlyshrubsortreeshave 3,185genera,or 24% of andbioticdispersal.Takinghalfthelatterplustheformer thetotalangiosperm genera.Thus,between24% and 59% givesan estimateof 4,463 animal-dispersed genera,or ofall angiosperm generamaybe woody,andso woodiness ;33% oftheangiosperm total.As in thecase ofwoody wouldappeartobe overrepresented amongdioeciousgen- growth, animaldispersalappearsto be overrepresented era.In thefamily-level analysis, woodinessperse wasnot amongdioeciousgenera,eventhoughthiswas nota para factorbecausethetreegrowthformhad no influence ticularly analysis. strongfactorin thefamily-level on thepresenceor absenceofdioecywithina familyand theshrubgrowthformhad an effect onlyin someofthe DISCUSSION analyses(butsee Materialsand Methodsfora discussion ofthelimitations ofa multivariate approachatthefamily systematic As has oftenbeenpointedout,thescattered level). of dioecyin the angiosperms suggeststhat distribution manytimes thismatingsystem hasevolvedindependently Dioecy and trophiccondition-Ofthe 959 dioecious and possiblyfordifferent reasons(Lewis,1942; van der bythefact genera,at least43 (4%) areholoparasites, hemiparasites, Pijl, 1978;Lloyd,1982).Thisis also suggested and or saprophytes. Dioecious parasitesoccurin the Bala- thatin some lineages,suchas theCaryophyllaceae variesalmostcontinnophoraceae, Eremolepidaceae, Loranthaceae, Misoden- Chenopodiaceae, sexualexpression ofmaleor femalefunction draceae,Opiliaceae,Rafflesiaceae, Santalaceae,and Vis- ually,withreadysuppression conditions(Heslop-Harrioftheangiosperms, caceae. Amongtheremainder para- in responseto environmental intheConvolvulaceae(Cus- son,1957;Freeman,McArthur, and Harper,1984;Freesiticgeneraareconcentrated cutoideae),Cynomoriaceae, Hydnoraceae, Krameriaceae, man et al., 1993), whilein otherlineages,such as the andCucurbitaceae, dicliny Lennoaceae, Loranthaceae, Olacaceae, Santalaceae, Monimiaceae,Euphorbiaceae, as angiosperms and is basicallyfixed.In thedicotyledonous Scrophulariaceae (Rhinanthoideae, Orobanchoideae), intheless Rafflesiaceae (Mitrastemonoideae), whichtogether may a whole,monoecyand dioecyareconcentrated ofThorne(1992). A consequence comprisesome 135 heterotrophic genera(1.1% of all advancedsuperorders ofdioeis thatfrequencies distribution 13,500angiosperm genera).Dioecytherefore maybe ov- ofthisnonrandom floras(made muchofby manyauthors) errepresented amongheterotrophic plants.Thiswas also cy in particular pool of the level of dioecyin a particular suggested bya species-level analysisofdioecyin theflora likelyreflect ofthesoutheastern UnitedStates(Conn,Wentworth, and familiesas muchas, 'ormorethan,localselectivefactors. whichof the various the questionregarding are Therefore, Blum,1980); t 17% ofall theCarolinaheterotrophs features associatedwithdioecyplaya dioecious. ecomorphological willhaveto causalrole,and underwhichcircumstances, Pollinationmodesofdioeciousangiosperm genera-Of be answeredfora numberof individuallineagesbefore the799 generawithdioeciousspeciesforwhichthereare we can extrapolatefromcommonfeaturesto common 232 (30%) are causes. sufficient data on themodeofpollination, resemblancesbetween wind-or water-pollinated, 550 (68%) arepredominantly To controlfor phylogenetic analyzing and 14 (2%) havewindand insectpol- membersofthesamelineage,we arecurrently insect-pollinated, ofrecently linationin closelyrelatedspeciesand sometimeswithin ourdioecydatasetwithintheframework pubdata). a species.Threegenera,theLiliaceaeCollospermum and lishedmolecularphylogenies (Renner,unpublished (letterof 16 Astelia(Dobat and Peikert-Holle, 1985) and the Bala- However,as suggested by D. Charlesworth betweenand similarities nophoraceaeDactylanthus (Webband Kelly,1993),are March 1994) the differences buttrioecious withinlineagesshouldtendto obscuregeneralpatterns; pollinatedbybats.Anotherbat-pollinated apparentwe speciesisPachycereuspringlei (Cactaceae;Flemming, per- so, wheregeneralpatternsare nevertheless sonalcommunication). Freycinetia (Pandanaceae)is pol- can accordthemsomeweight. betweendioecyand linatedbybirdsand bats(Cox, 1982).Whenthesefigures Atthefamilylevel,thecorrelation but does not are comparedto the ,:250 generawithbat-pollinated fleshyfruits(bioticdispersal)is significant species(Dobat and Peikert-Holle, 1985) and 500 genera accountfora largeproportionof the variationamong a This number ofnondioecious generapollinated by insectswas arfromthetotalnumberof angiosperm rivedat by subtracting genera (13,500) all thedioeciousgenera(959), 750 bird-and bat-pollinated genera,874 generain familieswithsolelywind-or water-dispersed withvariousdispersal members, andan estimated 600generainfamilies modes.Thereare;250 bat-pollinated genera(DobatandPeikert-Holle, 1985) and 500 bird-pollinated genera(Porsch,1931 and our ownestimate). 604 AMERICAN JOURNAL OF BOTANY [Vol. 82 familiesin presenceor absenceofdioecy;at thegeneric ceae, Campanulaceae,Goodeniaceae,Lobeliaceae,and level,bioticdispersal appearstobe overrepresented among otherfamiliesin the Asteridaeexhibita complexsyndioecists. Theassociation ofdioecywithfleshy fruits found organization ofmaleandfemaleorgans, fQrexample,with in particularfloras(see the introduction) likelyreflects pollenfirst beingdepositedon and thenpresented bythe thecombinedeffects of local samplingand multipleas- style(secondarypollenpresentation) or withcongenital sociationsamongthecorrelates themselves. Thus,forex- and postgenital fusionofcarpels,petals,and stamens.A ample,acrossall angiosperms fleshyfruitsare strongly similarsynorganization of male and femalestructures correlated withthetreegrowth form(Ricklefs andRenner, characterizes manyFabaceae (forfurther examples,see in press),whichin turnpredominates in certaintropical van der Pijl, 1978). All thisshouldmaketheontogenic families. suppressionof male or femaleorgans,withoutradical Treegrowthaloneis notassociatedwithdioecyat the alterationsin the structural organization of the flower, in themoreadvancedgroupsthanin the familylevel,whilethe shrubgrowthformhas a weak moredifficult positiveeffect onlyin some analyses;climbinggrowth, less advancedones. whetherwoodyor herbaceous,by contrasthas a strong Modem familiesthathave managedto escape from effect onthepresenceofdioecy(possiblymech- suchconstraints enhancing are,e.g.,theApiaceae,Araliaceae,and anismsforthisassociationaresuggested below).The sup- Asteraceae(the latterin spiteof secondarypollenpreall ofwhichhavesmallflowers posed correlation betweenwoodinessper se and dioecy sentation), into aggregated has led variousauthors(e.g.,Lloyd,1982)to hypothesize inflorescences and all ofwhicharecapableofsuppressing in some flowers. abouta particular needforobligatory inlong- sexualfunction outbreeding Theyhave andromonlived,woodyplantsdue to geneticloads. Anothersug- oeciousand gynomonoecious species,yetrelatively little gestionrelateswoodinessanddioecyvia themoreintense dioecy(sixof403 generaand :z34 of3,100speciesin the in theclimaxhabitatswherethese Apiaceae,fourof 47 generaand ;:69 of 800 speciesin selectionforheterosis plantsoccur(Bell, 1982). The presentanalysisprovides the Araliaceae,and 26 of 1,509 genera,and ;:526 of littleempiricaljustification 20,000 speciesin theAsteraceae). forthesehypotheses. A secondcorrelateof dioecy(and monoecy)is polliDioecy is strongly associatedwithmonoecy,abiotic pollination,and climbinggrowth.We willnow address nationbywindorwater.Severalcausativefactors forthe possiblemechanisms thesethreeassociations. associationbetweendioecyand abioticpollinationhave underlying The singlemostimportant predictor ofa group'stendency beensuggested (Darwin,1876;Kerner,1895;vanderPijl, to acquiredioecyis thepresenceofmonoecyin thegroup 1978; Freeman,Harper,and Ostler,1980).In wind-pol(as also foundbyYampolskyand Yampolsky,1922,and linatedplants,pollen release may requirependulous, Lewis,1942).As pointedoutbyWestergaard (1958),the shakingstamenswhereaspollencapturemayfavorlarge, hydrophilous plantshave generaldirectionof evolutionary changeis likelyto be erect,freestigmas.Similarly, frommonoecyto dioecy;clearly,however,phylogenetic floralmechanisms forthecaptureofwater-borne pollen interfere withmechanisms forthedispersal analysesof individualexamplesare neededto support thatphysically thisassumption. The routefrommonoecyto dioecymay of pollen(Cox, 1988, 1991). Also,theimprecisemovebe especiallycommonbecause once a lineagehas the mentofpollenbywindorwateris likelytocausefrequent in an her(in theabsenceofself-incompatibility) physiological and morphological abilityto suppressmale selfing in someflowers, abioticpollinationshouldfavor or femalefunction Therefore, thesubsequentstep maphrodite. to dioecymay occurthroughdivergent thelatterthenapparently of unisexualflowers, facilitating adjustments floralsex ratiosbetweenindividualplants(Lloyd,1972; theevolutionofdioecyfrommonoecy.Supportforsuch McArthur et al., 1992).It maybe easierforstrictly comes fromrelativelyfewstudiesof mon- an interpretation oecious populationsto evolve dioecy(since mutations changesin breedingsystemin connectionwithchanges mode.InAcer(Hesse,1979)andThalictrum affecting pollenand ovuleproduction mustalreadyhave inpollination occurredin theirunisexualflowers) thanforpopulations (Kaplan and Mulcahy,1971) a switchin some species thatproducebisexualflowers, be theygynomonoecious, frominsecttowindpollination accompaniedtheacquisiand dioecy;whileHesse andromonoecious,polygamo-monoecious,gynodio- tionofdicliny, polygamodioecy, to have precededdiclinyin Acer, or trioecious.Indeed,how fre- interprets anemophily ecious,androdioecious, in quentlygynodioecyhas led towarddioecy is unclear Kaplanand Mulcahyfavortheoppositeinterpretation studiesmayeventually andCharlesworth, Phylogenetic (Charlesworth permit 1978;GouyonandCou- Thalictrum. us toestablishtheevolutionary vet,1987; Wellerand Sakai, 1991; Barrett, sequenceofeventsinboth 1992). in theless genera.In tropical,entomophilous Whyare dioecyand monoecyconcentrated Fagaceae,diclinyand to a lateracquiadvanceddicotsin spiteof thefactthatveryfewofthe monoovulymayhave beenpreadaptive mostprimitiveextantMagnoliidaeare wind-pollinated sitionofanemophily as suggested byvan derPijl (1978). frombisexualto uni(Endress,1990a)? We believethatthereasonhas to do In stillothergroups,thetransition was notassociatedwitha switchfrominwithfloralmorphological theirab- sexualflowers constraints, or,rather, sence.In primitive floralphyllotaxis is rel- sectsto wind,and in thesetaxa,insectpollinationmay angiosperms, to do withtheevolutionof dioecy,albetweenmale have something ativelyplastic,andthereis littleintegration forsuchan effect and femalefloralparts(Endress,1987, 1990b). In the thoughthemechanism (increasedselfing more advancedgroups,by contrast,numbersof floral ratesin partsof the rangewithdifferent pollinatorbeandstructures haveevolved havior?)is unclear. partsoftenarelowand fixed, to directpollinatormovementor to concealnectar(usuThirdly,dioecyis strongly associatedwithclimbing orepigyny, andzygomorphy). growth, ofwhether irrespective allyvia perigyny theplantis a woodyliana sympetaly, Apocynaceae,Asclepiadaceae,Brunoniaceae,Calycera- oran herbaceousvine.Familieswithclimbers, as a group, May 1995] RENNER AND RICKLEFS-DIOECY IN THE FLOWERING PLANTS 605 seemtohavea combination offeatures predisposing them LITERATURE CITED todioecy,namelyhighincidence ofmonoecy, fleshy fruits, and tropicaldistribution. However,thepositiveeffect of theclimbing habiton a family's likelihoodofcomprising BAKER, H. 1959. Reproductive methods as factorsin speciation in floweringplants. Cold Spring Harbor Symposiumin Quantitative dioeciousspeciesis independent oftheseothervariables Biology 24: 177-19 1. (Table 2). We suspectdifferential selectionforoptimal BARR=r, S. C. H. 1992. Gendervariationand theevolutionofdioecy resourceallocationto sexualfunction to be responsible in Wurmbeadioica (Liliaceae). JournalofEvolutionaryBiology 5: An increasein femalefloralallocationwith forthiseffect. 423-444. increasedplantsizehasbeenfoundinseveralmonoecious BAWA,K. S. 1980. Evolution of dioecy in floweringplants. Annual Review ofEcology and Systematics11: 15-39. climbers(Bickeland Freeman,1993). This impliesre1994. Pollination of tropicaldioecious angiosperms:a reassourcelimitations to fruitsetbelowa certainthreshold, sessment?No, not yet.AmericanJournalof Botany 81: 456-460. andin Cucurbitapepo is indeedstrongsubsequent growth , AND P. A. OPLER. 1975. Dioecism in tropical foresttrees. lyinhibitedafterthefirst fruitis produced(Nitschet al., Evolution 29: 167-179. 1952).Ifrapidupwardgrowth is ata premium inclimbers BELL, G. 1982. The masterpieceof nature.Croom Helm, London. and if fruitset at least temporarily inhibitsgrowthor BICKEL,A. M., AND D. C. FREEMAN. 1993. Effectsofpollen vectorand plant geometryon floralsex ratio in monoecious plants.American requirestheproductionof a thick,slowlygrowing stem Midland Naturalist 130: 239-247. to supportheavyfruits, itmightbe advantageous topostR. K. 1992. Vascular plant families and genera. Royal If the effect pone femaleness. is strong,thismay favor BRummITr, Botanic Gardens, Kew. male plants.Structural necessitiesmayalso favorsepa- CHARLESWORTH, B., AND D. CHARLESWORTH.1978. A model forthe rationofthesexes,withflowers bearingfruit better placed evolutionofdioecy and gynodioecy.AmericanNaturalist112: 975997. onthicker stemsandpollen-dispensing flowers onthinner, ofdioecyand self-incompatCHARLEsWORTH, D. 1985. Distribution moreexposedmodules. ibilityin angiosperms.In P. J. Greenwood and M. Slatkin [eds.], The nearabsenceof birdand bat pollinationamong Evolution: essays in honour of John Maynard Smith, 237-268. dioeciousplants(Table 14) maybe explainedbythefact Cambridge UniversityPress, Cambridge. thatvertebrate pollinationrequiresadaptationsoffloral CoNN,J. S., T. R. WENTWORTH, AND U. BLUM. 1980. Patternsof whichseemsto almostprecludethe evomorphology, dioecism in thefloraoftheCarolinas.AmericanMidland Naturalist 103: 310-315. lutionof dicliny.First,pollinationby large-bodied animals,suchas birdsand bats,requireslargequantitiesof Cox, P. A. 1982. Vertebratepollinationand themaintenanceofdioein Freycinetia.AmericanNaturalist 120: 65-80. nectar(Opler,1983),whichcanbe offered adequately only - cism1988. Hydrophilous pollination.Annual Review of Ecology . in cups,tubes,orspurstypically formed byconcavefloral and Systematics19: 216-280. receptacles (oftenwithinferior ovaries)and/orfusedpet1990. Pollination and the evolution of breedingsystemsin als (sympetaly). Secondly,pollinationby thesevectors Pandanaceae. Annals of the Missouri Botanical Garden 77: 816840. requiresrelatively largequantitiesof pollen,whichare 1991. Abiotic pollination:an evolutionaryescape foranimaloftenproducedby numerousstamens(secondarypolypollinated angiosperms.Philosophical Transactions of the Royal and secondary andry).Inferior ovaries,sympetaly, polySociety of London B 333: 217-224. oftherelatively advancedsubclasses andryareattributes evolution 1993. Hydrophilouspollinationandbreedingsystem butnearlyabsentin Dilleniidae,Rosidae,and Asteridae, in seagrasses:a phylogeneticapproach to the evolutionaryecology theMagnoliidaeandHamamelidae(Cronquist, 1988;Enof Cymodoceaceae. Botanical Journalof theLinnean Society 113: con217-226. dress,1990b).Thusthereis a strongmorphological straint on vertebrate pollination. Indeed,ofthecirca250 CROAT, T. B. 1979. The sexualityof the Barro Colorado Island flora (Panama). Phytologia42: 319-348. generawithbat-pollinated speciesknownin the angioCRoNQuIsT, A. 1988. The evolution and classificationof flowering none and to sperms(Dobat Peikert-Holle, 1985), belong plants, 2d ed. Columbia UniversityPress, New York, NY. theMagnoliidaeand Hamamelidae,while28% arein the DARWIN,C. 1876. The effectsof cross and self fertilisationin the Rosidae,26% in theAsteridae,25% in theDilleniidae, vegetablekingdom.JohnMurray,London. 13%inthemonocots,and 8% in theCaryophyllidae. The DOBAT, K., ANDT. PEIKERT-HoLLE. 1985. BlutenundFledermaiuse. Verlag von Waldemar Kramer,Frankfurt. distributions ofbat pollinationand dioecyin theangioDONOGHUE,M. J. 1989. Phylogeniesand the analysisof evolutionary exclusive. spermsare thusalmostmutually sequences, with examples fromseed plants. Evolution 43: 1137Whiletheenvironmental conditions underwhichpop1156. ulationschangebreedingsystemswill have to be ad- ENDRESS, P. K. 1987. Floral phyllotaxisand floralevolution. Botanin allianceswithrelatively dressedexperimentally labile ische Jahrbiicherflir Systematik108: 417-438. sex expression,such as Begonia, Atriplex,or Wurmbea etal.,,1992),theglobalanalysis (Barrett, 1992;McArthur famthatacrossall angiosperm heresuggests performed conandclimbing iliesmonoecy, windpollination, growth favortheevolutionofdioecy.Othertraits,such sistently as heavy,fleshy fruits, mayplaya rolein certainlineages. Wherethe ontogenetic floraladjustmentsrequiredfor unisexualflowersare possiblewithout jeopardizingpollination,specializationforone sexualfunctionappears favoredbyselectionundera widerangeofenvironmental conditions as is apparentfromourfinding thatgeographic distribution has no strongeffect on presenceor absence ofdioecy. 1990a. Evolution of reproductivestructuresand functionsin primitiveangiosperms(Magnoliidae). Memoirs of the New York Botanical Garden 55: 5-34. 1990b. Patternsof floralconstructionin ontogenyand phylogeny.Biological Journalof theLinnean Society39: 153-175. Fox, J. F. 1985. Incidence of dioecy in relationto growthform,pollination and dispersal. Oecologia 67: 244-249. FREEMAN, D. C., K. T. HARPER, AND W. K. OSTLER. 1980. Ecologyof plantdioecyin theintermountain regionofWesternNorthAmerica and California.Oecologia 44: 410-417. , E. D. McARTHuR, AND K. T. HARPER. 1984. The adaptive significanceof sexual labilityin plants usingAtriplexcanescens as a principalexample. Annals of the Missouri Botanical Garden 71: 265-277. S. C. SANDERSON, AND A. R. TIEDEMANN. 1993. The 606 AMERIcAN JOURNAL OF BOTANY influenceof topographyon male and femalefitnesscomponentsof Atriplex canescens.Oecologia93: 538-547. GENmy, A. 1991. The distributionand evolution of climbingplants. In F. E. Putz and H. A. Mooney [eds.], The biology of vines, 349. Cambridge UniversityPress, Cambridge. GIVNISH,T. J. 1980. Ecological constraintson theevolutionofbreeding systemsin seed plants:dioecy and dispersalin gymnosperms.Evolution34: 959-972. GOUYON,P.-H., AND D. COUVET. 1987. A conflictbetweentwo sexes, femalesand hermaphrodites.In S. C. Stearns[ed.], The evolution of sex and its consequences. BirkhauserVerlag,Basel. HART, J. A. 1985. Evolution of dioecism in Lepechinia Willd. sect. Parviflorae(Lamiaceae). SystematicBotany 10: 147-154. HEsLoP-HARRsoN, J. 1957. The experimentalmodificationof sex plants.Biological ReviewsoftheCambridge expressionin flowering Society32: 38-90. Philosophical und Verteilungdes Pollenkittsin der HESSE,H. 1979. Ultrastruktur GattungAcer (Aceraceae). Plant Sysinsekten-und windbliutigen tematics and Evolution131: 277-289. G., ANDK. OYAMA. 1992. Ecological correlates IBARRA-MANRiQuEz, ofreproductivetraitsofMexican rainforesttrees.AmericanJournal of Botany 79: 383-394. KAPLAN, S. M., AND D. L. MuLcAHY. 1971. Mode of pollinationand floralsexualityin Thalictrum. Evolution 25: 659-668. gracilis(Nepenthaceae) KATO,M. 1993. Floral biology of Nepenthes in Sumatra.AmericanJournalof Botany 80: 924-927. KERNER VONMARIAUN, A. 1895. Pflanzenleben, vol. 2. Bibliographisches Institut,Leipzig. KUBIrzKI, K., J. G. ROHWER, AND V. BrTRIucH[EDS.]. 1993. The Berfamiliesand generaof vascular plants,vol. 2. Springer-Verlag, lin. LAHAv-GINoTT,S., AND Q. C. B. CRoNK. 1993. The matingsystem ofElatostema(Urticaceae) in relationto morphology:a comparative andEvolution186: 135-145. study.PlantSystematics LEWIS,D. 1942. The evolution of sex in floweringplants. Biological Philosophical Society17: 46-67. ReviewoftheCambridge [Vol. 82 NrrscH,J. P., E. B. KURz, J. L. LIVERMAN, AND F. W. WENT. 1952. The developmentof sex expressionin cucurbitflowers.American JournalofBotany 39: 32-43. OPLER,P. A. 1983. Nectar productionin a tropicaLecosystem.In B. Bentleyand T. Elias [eds.], The biology of nectaries,30-79. Columbia UniversityPress, New York, NY. PoRscH, 0. 1931. Grellrotals Vogelblumenfarbe.Biologia Generalis 7: 647-674. RENNER,S. S., AND J. P. FEIL. 1993. Pollinatorsof tropicaldioecious angiosperms.AmericanJournalof Botany 80: 1100-1107. RIcHARDs,A. J. 1986. Plant breeding systems.Allen and Unwin, Winchester,MA. RICKLEFS,R. E., AND S. S. RENNER. 1994. Species richnesswithin familiesof floweringplants. Evolution 48, in press. INc. 1988. SAS/STAT (TM) user'sguide,Release 6.03 SAS INSTITUrE, ed. SAS Institute,Inc., Cary,NC. K. R. 1949. A new approach to the problemof the primitive SpoRNiE, flower.New Phytolologist48: 259-276. STEBBINS, G. L. 1950. Variation and evolution in plants. Columbia UniversityPress, New York, NY. of angiosperm 1951. Natural selectionand the differentiation families.Evolution 5: 299-324. K. E. 1988. Dioecism and its correlatesin the Cape floraof STENmER, South Africa.AmericanJournalof Botany 75: 1742-1754. THOMAS, S. C., AND J.V. LAFRANKIE. 1993. Sex,size,and interyear variation in floweringamong dioecious treesof the Malayan rain forest.Ecology 74: 1529-1537. THOMSON, J.D., ANDS. C. H. BARRErr. 1981. Selectionforoutcrossing, sexual selection,and the evolution of dioecy in plants. American Naturalist 118: 443-449. fortheevolutionofdioecy , AND J.BRUNET. 1990. Hypotheses in seed plants. Trendsin Ecology and Evolution 5: 11-16. THORNE,R. F. 1992. Classificationand geographyof the flowering plants. Botanical Review 58: 225-348. VAN DER PuL, L. 1978. Reproductiveintegrationand sexual disharmony in floralfunctions.In A. J. Richards [ed.], The pollination of flowersby insects,79-88. Academic Press, London. LIOYD, D. G. 1972. Breedingsystemsin Cotula L. (Compositae, Anbiologyof the 1993. The reproductive themideae). New Phytologist71: 1181-1194. WEBB, C. J.,AND D. KELLY. New Zealand flora.Trendsin Ecology and Evolution 8: 442-447. 1982. Selection of combined versus separate sexes in seed plants. The AmericanNaturalist 120: 571-585. WEBSTER, G. L. 1994a. Classificationof the Euphorbiaceae.Annals of theMissouri Botanical Garden 81: 3-32. MABBERLEY, D. J. 1987. The plant-book.CambridgeUniversityPress, 1994b. Synopsis of the genera and supragenerictaxa of EuCambridge. S. C. SANDERSON, phorbiaceae.Annals oftheMissouri Botanical Garden 81: 33-144. E. D., D. C. FREEMAN,L. S. LUCKINBILL, McARTHuR, ANDG. L. NOLLER. 1992. Are trioecyand sexuallabilityinAtriplex WELLER, S. G., AND A. K. SAKII. 1991. The geneticbasis of male in Schiedea (Caryophyllaceae),an endemicHawaiian genus. canescens geneticallybased?: evidence fromclonal studies. Evosterility Heredity67: 265-273. lution46: 1708-1721. in M. 1958. The mechanismof sex determination WESTERGAARD, McCoMB, J. A. 1966. The sex formsof species in the flora of the dioecious floweringplants.Advances in Genetics9: 217-281. south-westof westernAustralia.AustralianJournalof Botany 14: 303-316. YAMPOLSKY, C., AND H. YAMPOLSKY.1922. Distributionof sex forms in the phanerogamicflora.BibliothecaGenetica 3: 1-62. MuENcHow, G. E. 1987. Is dioecy associated withfleshyfruit?Amer- icanJournalofBotany74: 287-293.
© Copyright 2026 Paperzz