Gender Variation and Transitions between Sexual Systems in Mercurialis annua (Euphorbiaceae)

GENDER VARIATION AND TRANSITIONS BETWEEN SEXUAL SYSTEMS IN MERCURIALIS ANNUA (EUPHORBIACEAE) John R. Pannell,1Ma cel E. Do ken, Benoi Pujol, and Regina Be jano2 Depa men o Plan Sciences, Uni e si y o Ox o d, Sou h Pa ks Road, Ox o d OX1 3RB, Uni ed Kingdom E olu iona y ansi ions be ween he maph odi ism and dioecy ha e occu ed nume ous imes in he land plan s. We b ie ly e iew he ac o s hough o be esponsible o hese ansi ions, and we p o ide a syn hesis o wha has been lea ned om ecen s udies o he annual he b Me cu ialis annua, in which dioecy (males and emales), monoecy ( unc ional he maph odi es), and and odioecy (males and he maph odi es) occu in di e en pa s o i s geog aphic ange. P e ious esea ch on M. annua has e ealed he impo ance o genome duplica ion and hyb idiza ion in he o igin o much o he obse ed a ia ion. He e we show, howe e , ha spa ial ansi ions in he sexual sys em also occu wi hin he same ploidy le el. In pa icula , we p esen an analysis, using low cy ome y da a, o ploidy a ia ion ac oss a p e iously uns udied ansi ion be ween he maph odi ism and and odioecy, in which we ind ha he sexual-sys em ansi ion is uncoupled om he shi in ploidy le els. We e iew ecen esea ch ha shows ha such ansi ions be ween sexual sys ems in M. annua a e consis en wi h di e en ial selec ion a he egional le el o ep oduc i e assu ance du ing coloniza ion. We also p esen new expe imen al da a ha highligh bo h he impo ance o he esou ce s a us o plan s and ha o hei local ma ing con ex in egula ing gende s a egies and sex a ios. The s udies e iewed and he new esul s p esen ed emphasize he ole ha shi s in he ecological and gene ic con ex o plan popula ions may play in causing ansi ions be ween sexual sys ems. Keywo ds: and odioecy, dioecy, gynodioecy, he maph odi ism, ma ing sys em, monoecy. In oduc ion By a , he majo i y o lowe ing plan s a e he maph odi ic (Yampolsky and Yampolsky 1922; Sakai and Welle 1999), bu ansi ions be ween he maph odi ism and dioecy ha e been equen (Renne and Rickle s 1995; Weiblen e al. 2000; Vamosi e al. 2003; Case e al. 2008). Mos esea ch has been di ec ed owa d unde s anding shi s om he maph o- di ism owa d dioecy, wi h a ange o hypo hesized e olu- iona y pa hs in oked ( e iewed by Cha leswo h [1999]; Webb [1999]). Indeed, Da win (1877) se he s age o much o his wo k by no ing he subs an ial ad an ages o he - maph odi ism and asking why ‘‘he maph odi e plan s should e e ha e been ende ed dioecious’’ (p. 279). In con as , we migh simila ly ask why males and emales, which may enjoy ad an ages o gende specializa ion, should e e be eplaced by he maph odi es. Al hough p obably less equen han ansi ions om he maph odi ism o dioecy, he b eakdown o dioecy owa d he maph odi ism is known o ha e oc- cu ed in bo h plan s and animals (Des eux e al. 1996; Wol e al. 2001; Kion ke e al. 2004; Weeks e al. 2006), and a g owing body o ideas exis s on when such shi s migh occu (Cha no e al. 1976; Cha no 1982; Mau ice and Fleming 1995; Wol and Takebayashi 2004). Ne e heless, he empi i- cal ounda ion o unde s anding ansi ions om dioecy o he maph odi ism emains weak. F om a heo e ical poin o iew, he b eakdown o dioecy equi es he in asion and sp ead in a popula ion o ei he pollen-p oducing emales o seed-p oducing males. Ini ially, he sp ead o he maph odi es in a dioecious popula ions will yield ‘‘ ioecy,’’ a sexual sys em in which all h ee gende classes a e main ained oge he , bu he condi ions unde which such a gende imo phism can be main ained e olu iona ily ap- pea o be a he limi ed (Mau ice and Fleming 1995; Wol and Takebayashi 2004). Mo e likely is he apid displace- men o ei he he emales o he males om he popula ion, wi h he in ading he maph odi es main ained in an and odioe- cious o a gynodioecious popula ion, espec i ely. Al hough gynodioecy is unde s ood o be an impo an s ep in he e o- lu ion o dioecy om he maph odi ism, i is no known o ha e played a majo ole in he b eakdown o dioecy. In con- as , while and odioecy is exceedingly a e in absolu e e ms (Da win 1877; Cha leswo h 1984), almos all he known and odioecious species, bo h in animals and in plan s, appea o ha e e ol ed om a dioecious a he han a he maph o- di ic ances o (Pannell 2002; Weeks e al. 2006). T ansi ions om dioecy o he maph odi ism by way o an- d odioecy ha e been s udied in only wo plan species in much de ail. In he wind-pollina ed No h Ame ican pe en- nial he b Da isca glome a a, he maph odi es appea o be modi ied (pollen-p oducing) emales wi h he same sex de e - mina ion as emales in i s dioecious sis e species Da isca cannabina (Wol e al. 2001). In he Eu opean he b Me cu i- alis annua ( ig. 1), which is also wind-pollina ed, and odioecy 1Au ho o co espondence; e-mail: john.pannell@plan s.ox.ac.uk. 2Cu en add ess: Depa amen o de Biologı ´a Vege al y Ecologı ´a, Uni e sidad de Se illa, Apa ado 1095, 41080 Se illa, Spain. Manusc ip ecei ed Feb ua y 2007; e ised manusc ip ecei ed July 2007. 129 In . J. Plan Sci. 169(1):129–139. 2008. Ó2008 by The Uni e si y o Chicago. All igh s ese ed. 1058-5893/2008/16901-0011$15.00 DOI: 10.1086/523360 is ound only in polyploid popula ions o a complex in which dioecy is clea ly he ances al ai (Du and 1963; Du and and Du and 1992; Pannell 1997d; Pannell e al. 2004). Me - cu ialis annua is unusual in he ex ao dina y di e si y o sexual sys ems i displays ac oss i s geog aphic ange; his a ia ion makes i a use ul s udy sys em in which o add ess ques ions conce ning e olu iona y ansi ions be ween sexual sys ems and he o igins and main enance o and odioecy. In his a icle, we p esen new da a ha ad ance ou unde - s anding o ansi ions in he sexual sys em o M. annua in wo ways. Fi s , we ask how closely he spa ial ansi ion be- ween monoecy and and odioecy in a hi he o poo ly s udied pa o he species’ ange on he A lan ic coas o Mo occo co esponds o a shi be ween e aploidy and hexaploidy. This ques ion is impo an because elsewhe e in he dis ibu- ional ange o he species complex, spa ial (and pe haps e o- lu iona y) ansi ions in he sexual sys em and ploidy le els a e con ounded; ins ances in which sexual-sys em ansi ions a e uncoupled om hose in ploidy p o ide pa icula ly e ile g ound o in oking he main enance o di e en sexual sys- ems by na u al selec ion ( a he han as a esul o he his o i- cal dis ibu ion o ploidy le els). Second, we p esen esul s o an expe imen ha asks how he ma ing sys em is a ec ed by plas ic esponses o esou ce a ailabili y in he ela i e alloca- ion o male and emale unc ions. We add ess bo h hese issues wi hin he con ex o a syn hesis o ecen wo k on M. annua. We begin by conside ing he phylogene ic and phy- logeog aphic his o y o he species complex in Eu ope. We hen e iew empi ical es s o a hypo hesis ha in okes di e - en ial ex inc ion-coloniza ion dynamics in a me apopula ion o explain he main enance o di e en sexual sys ems. Fi- nally, we conside he po en ial impo ance o pheno ypic plas ici y in sex alloca ion in egula ing he selec ion o com- bined e sus sepa a e sexes. O e all, we highligh he impo - ance o bo h shi s in he gene ic sys em associa ed wi h polyploidiza ion and hyb idiza ion and shi s in he ecological and demog aphic con ex o selec ion o ansi ions be ween sexual sys ems. T ansi ions in he Sexual Sys em: The Role o Polyploidy and Hyb idiza ion Me cu ialis annua belongs o a small Eu opean genus in he Eupho biaceae ha comp ises mainly dioecious woody o he baceous pe ennials. Phylogene ic econs uc ion o he Fig. 1 Male (A), emale (B), and monoecious (C) indi iduals o Me cu ialis annua. The male lowe s on he male and he monoecious indi iduals ha e he same mo phology, bu male lowe s o male plan s a e held on e ec peduncles. Also, he emale lowe s o bo h emale and monoecious plan s ha e he same mo phology and placemen in he lea axils (in he pho og aphs, emale lowe s ha e al eady se ui ). Males o diploid and hexaploid popula ions ha e he same mo phology. Monoecious plan s a e e ec i ely emales ha p oduce s amina e lowe s a ound hei pis illa e lowe s. 130 INTERNATIONAL JOURNAL OF PLANT SCIENCES genus by K ahenbuhl e al. (2002), based on ITS sequence analysis, indica ed ha dioecy and pe enniali y a e ances al in he genus and ha monoecy has e ol ed on a leas wo independen occasions, one o which was in he annual poly- ploid complex M. annua. This complex comp ises ploidy le els anging om diploid h ough a leas 12-ploid, wi h di- oecy con ined o he diploids and monoecy ound only in he polyploid lineages. He e, he e aploids, oc aploids, and highe ploidy le els a e exclusi ely monoecious, bu he hexaploids show ema kable a ia ion in hei sexual sys ems, wi h pop- ula ions anging om monoecy h ough and odioecy o sub- dioecy in a ious pa s o hei ange (Du and 1963; Du and and Du and 1992). Ea ly s udies o mo phology and meio ic pai ing beha io sugges ed ha M. annua was an au opolyploid se ies and ha polyploidiza ion had p ecipi a ed he e olu ion o mono- ecy (Du and 1963). Howe e , mo e ecen analysis iden i ied a leas wo di e gen ITS pa alogues in hexaploid popula- ions o M. annua, only one o which occu ed in diploid and e aploid indi iduals (Obba d e al. 2006b). The o he ITS sequence, absen in diploids and e aploids, was ound in Me cu ialis hue ii, a diploid sis e species o M. annua, which has he same numbe o ch omosomes and is also annual and dioecious. Obba d e al. (2006b) in e p e ed hese esul s as e idence o hyb idiza ion be ween e aploid M. annua and M. hue ii, yielding iploids and ollowed by polyploidiza ion o p oduce he hexaploid lineage ( ig. 2). This hypo hesis e- qui es u he es ing wi h mo e exhaus i e sampling o geno- ypes and loci, bu p elimina y esul s om mic osa elli e loci, some o which ampli y only in hexaploid M. annua and M. hue ii, a e consis en wi h he ITS sequence analysis (P. Ryme , H. S one, G. Ko becka, and J. R. Pannell, unpublished da a). Polyploidy ep esen s a d ama ic shi in he gene ic sys em o a lineage. No only may i al e he exp ession o inb eeding dep ession, po en ially allowing he sp ead o sel - e iliza ion in a popula ion (Ron o 1999; Husband e al. 2008), bu i may also al e a b oad ange o pheno ypic ai s ha may al- low a lineage o occupy new and di e en habi a s; i.e., i can cause ecological shi s ha may also a o a ansi ion in he sexual sys em o a popula ion (S ebbins 1950; Pannell e al. 2004). Pannell e al. (2004) e iewed he in e ac ions ha a e expec ed o occu be ween ploidy and sexual-sys em e olu ion. Thei main conclusion was ha he complexi ies in ol ed make gene al p edic ions di icul . Fo example, on he one hand, polyploidiza ion may allow sel - e ile he - maph odi ism o eplace dioecy by causing a educ ion in in- b eeding dep ession (Lande and Schemske 1985; Pannell e al. 2004; Husband e al. 2008). On he o he hand, genome du- plica ion can cause he b eakdown o sel -incompa ibili y in he maph odi ic popula ions, allowing dioecy o e ol e as an al e na i e ou c ossing mechanism (Mille and Venable 2000; bu see Mable 2004; Pannell e al. 2004). Thus, i can be emp ing o in oke polyploidiza ion as a cause o ansi ions bo h om he maph odi ism o dioecy and ice e sa. In he case o M. annua, i would seem no only ha polyploidiza ion ini ia ed he shi om dioecy o he maph odism bu also ha i has no p ecluded he main enance o males in hexaploid popula ions o M. annua o he exis ence o dioecy in he newly disco e ed e aploid species Me cu ialis cana iensis (Obba d e al. 2005). Phylogeog aphy and Regional T ansi ions in he Sexual Sys em Spa ial ansi ions in he sexual sys em o Me cu ialis annua in Eu ope ( ig. 3) co espond, o an impo an ex en ( hough no exclusi ely; see below), wi h ansi ions in he ploidy le el. Fully dioecious popula ions a e exclusi ely diploid; hese a e widesp ead h oughou Eu ope. In con as , popula- ions con aining he maph odi es (wi h o wi hou males) a e polyploids; hese a e la gely es ic ed o he wes e n Medi e - anean Basin and no hwes e n A ica. Te aploids occu sou h o Raba on he A lan ic coas o Mo occo, and hese mee wi h hexaploid popula ions o he no h. Te aploids a e he maph odi ic, whe eas he hexaploids, which a e e y wide- sp ead in no hwes e n Mo occo and a ound he coas o he Ibe ian Peninsula, a e a iously he maph odi ic o and odioe- cious. Hexaploid popula ions mee he diploids a wo con ac zones in no heas e n and no hwes e n Spain (Du and 1963; Obba d e al. 2006b). Pa e ns o allelic ichness and gene ic di e si y a se e al isozyme loci sugges ha he Spanish diploid-hexaploid an- si ions a e seconda y con ac zones, wi h diploids ha ing expanded ac oss Eu ope om an eas e n Medi e anean e u- gium and he hexaploids ha ing mo ed no h along he coas s o he Ibe ian Peninsula om sou he n Spain o No h A ica (Obba d e al. 2006b). This expansion o he geog aphic ange o diploid M. annua appea s o be con inuing apace, wi h su - eys sugges ing ha he diploids ha e displaced he hexaploids by some 80 and 200 km wi hin abou ou decades in no h- eas e n and no hwes e n Spain, espec i ely, appa en ly bo h as a esul o diploid supe io i y in hei g ow h (Buggs and Pannell 2007) and h ough he abili y o diploids o swamp he monoecious hexaploids wi h he la ge amoun s o pollen ha diploid males p oduce (Buggs and Pannell 2006). The highe pollen p oduc ion o diploid males o e he - maph odi es p o ides a plausible explana ion o he apid displacemen o he maph odi ic popula ions by he dioecious lineage. Howe e , wha should we p edic o a con ac zone be ween dioecy and and odioecy, whe e males co-occu wi h he he maph odi es? This is a pe inen ques ion because, should he dioecious lineage con inue i s ad ance down he coas s o he Ibe ian Peninsula, i will soon encoun e and o- dioecious popula ions a he sou h. Do ken and Pannell (2007) add essed his ques ion, using spa ially explici compu e sim- ula ions. They expec ed ha he p esence o males in and o- dioecious hexaploid popula ions migh slow he diploid ad ance by compe ing wi h he diploid males o si e ou - c ossed p ogeny. Howe e , because ange expansion equi es ecu en success ul coloniza ion (Bake 1955; Pannell and Ba e 1998), he occu ence o males wi h monoecious indi- iduals ac ually dilu ed one o he bene i s ha monoecy has o e dioecy: he ad an age o ep oduc i e assu ance du ing coloniza ion. Do ken and Pannell’s (2007) simula ions he e- o e p edic ha he diploid ad ance migh accele a e when and odioecious egions a e encoun e ed, and hey ecall he complexi ies ha can esul when selec ion ac s on plan e- p oduc i e s a egies a bo h he popula ion and me apopula- ion le els (Ba e and Pannell 1999; also see nex sec ion). The ansi ions be ween dioecy and he maph odi ism in no he n Spain, desc ibed abo e, a e ully con ounded by 131 PANNELL ET AL.—SEXUAL-SYSTEM TRANSITIONS IN MERCURIALIS ANNUA shi s be ween diploidy and hexaploidy. Howe e , ansi ions also occu be ween and odioecy and he maph odi ism else- whe e in he Ibe ian Peninsula and in Mo occo. In he Ibe- ian Peninsula, hese ansi ions occu wi hin he hexaploid lineage, bu in Mo occo, he ansi ion occu s along he A - lan ic coas in a egion b oadly coinciden wi h a ansi ion be ween hexaploidy in he no h and e aploidy in he sou h (Du and 1963). To de e mine whe he hese wo ansi ions coincide p ecisely, we used low cy ome y o assay he ploidy le el o 22 popula ions a in e als along he A lan ic coas o Mo occo ( ig. 4). Fo he low cy ome y measu emen s, we used lea ma e ial om seeds g own in he g eenhouse. Ma e- ial was p epa ed using he ‘‘LB01’’ me hod o Dolezel e al. (1989) and Lycope sicon esculen um c . Ga dene ’s Deligh Fig. 2 Hypo heses o he ela ionships be ween he annual lineages o Me cu ialis. Hea y lines indica e he phylogene ic ela ionships be ween diploid species, and hin a ows show polyploidiza ion o hyb idiza ion e en s; Mindica es p oposed ma e nal pa en age, and Pindica es p oposed pa e nal pa en age. A, Diploid M. annua p oposed as he pa en o polyploid M. annua. The he e ogeneous ITS ypes p esen in hexaploid M. annua show ha i has an allopolyploid o igin h ough hyb idiza ion be ween M. annua and M. hue ii, and he hexaploid ch omosome complemen is consis en wi h hyb idiza ion be ween a e aploid and a diploid lineage, ollowed by ch omosome doubling. ITS da a also show M. cana iensis o be allopolyploid in o igin, p obably a hyb id be ween M. annua s.l. and an unknown axon (dashed box). Chlo oplas sequence simila i y o M. annua s.l. sugges s ha M. annua s.l. was he ma e nal pa en , and ch omosome numbe s a e consis en wi h bo h pa en s being diploid. B, Gi en he g ea e simila i y be ween ITS sequences om M. cana iensis and e aploid and hexaploid M. annua, he diploid p ogeni o o hese axa may ha e be an ea lie wes e n lineage o M. annua, di e gen om he lineage ha has ecen ly expanded om he eas . F om Obba d e al. (2006a). 132 INTERNATIONAL JOURNAL OF PLANT SCIENCES lea ma e ial as a s anda d. Each sample was assessed on a Bec on Dickinson FACScan low cy ome e and analyzed us- ing Bec on Dickinson CellQues so wa e (BD BioSciences, F anklin Lakes, NJ). Ou esul s indica e ha he ansi ion be ween e aploidy and hexaploidy in Mo occo occu s some 100 km sou h o he monoecy-and odioecy ansi ion, which is loca ed a Ra- ba ; males a e ound only in popula ions no h o his poin ( ig. 4). (No e ha wo e aploid popula ions we e ound no h o he majo ploidy ansi ion wi hin he hexaploid zone. Such popula ions a e likely o be he esul o ea h- mo ing oadwo ks, al hough hey migh also be emnan popula ions ollowing a hexaploid ad ance sou h; see Buggs and Pannell 2006.) The geog aphic shi in he sexual sys em is hus uncoupled om he shi in ploidy in Mo occo, and i occu s wi hin he hexaploid lineage jus as i does in he o he monoecy-and odioecy ansi ions in he Ibe ian Penin- sula. This poses he ques ion o why he sexual sys em should change o e geog aphical space, gi en ha popula ions on bo h sides o hese ansi ions ha e he same gene ic sys em and appea o exchange genes (Obba d e al. 2006b). Mechanisms o T ansi ions among Sexual Sys ems: The Role o Me apopula ion Dynamics To explain he main enance o he maph odi ism e sus di- oecy o and odioecy in di e en egions o he ange o hexa- ploid Me cu ialis annua, Pannell (1997c) hypo hesized a me apopula ion model ha in okes di e en ial selec ion o ep oduc i e assu ance du ing coloniza ion. Because only sel - e ile he maph odi es, and no males, can colonize un- occupied habi a on hei own, we should expec ex inc ion- coloniza ion dynamics o educe he equency o males a he me apopula ion (i.e., egional) le el. A he same ime, he maph odi es wi h emale-biased sex alloca ion should be a o ed by selec ion du ing coloniza ion o e hose wi h mo e equal sex alloca ion because hei popula ions will g ow mo e quickly (no e ha selec ion o emale-biased sex alloca- ion a he me apopula ion le el can be seen equi alen ly in e ms o local ma e compe i ion du ing sel ing o in e ms o deme-le el selec ion; see F ank 1986). As popula ions g ow dense a e coloniza ion, he selec i e ad an age o maleness inc eases, bo h because o he a ailabili y o la ge numbe s o o ules o e ilize and because o he low numbe s o pol- len g ains p oduced by he maph odi es wi h which hey mus compe e. Thus, whe eas males a e dis a o ed du ing coloni- za ion, hey may be a o ed as mig an s (o , p obably much mo e a ely, as new mu an s) in o es ablished demes. Unde his model, whe he males a e ound a a egional le el de- pends la gely on he balance be ween hese wo opposing se- lec i e o ces (Pannell 1997a; see also Pannell 2001). The me apopula ion model makes a numbe o p edic ions ha ha e ecen ly been es ed. One demog aphic p edic ion is ha , i he egional absence o males is he esul o apid Fig. 3 Dis ibu ion o Me cu ialis annua in Eu ope and a ound he Medi e anean Basin. In he no h and eas o his ange, M. annua is dioecious and diploid, whe eas in Ibe ia and No h A ica, i is monoecious (and and odioecious) and polyploid. Ci cles indica e he loca ions o seed collec ions used o es ima e pa e ns o di e si y by Obba d e al. (2006b). Regions ma ked ‘‘Dioecy,’’ ‘‘Monoecy,’’ and ‘‘And odioecy’’ deno e zones occupied by he co esponding sexual sys ems (see ‘‘Phylogeog aphy and Regional T ansi ions in he Sexual Sys em’’). Modi ied om Obba d e al. (2006b). 133 PANNELL ET AL.—SEXUAL-SYSTEM TRANSITIONS IN MERCURIALIS ANNUA popula ion u no e compa ed wi h egions in which males occu , hen popula ions in monoecious egions ough o be smalle han hose in and odioecious egions (because hey a e, on a e age, younge ). Ano he p edic ion is ha a es o habi a occupancy should be lowe in monoecious han in and odioecious egions i ex inc ion-coloniza ion a es a e highe in he o me (Gas on e al. 2000; F eckle on e al. 2005). Pa e ns o occupancy and abundance obse ed ac oss se e al clines in he sexual sys em o M. annua in Spain we e consis en wi h bo h hese p edic ions (Eppley and Pannell 2007b). F om a popula ion gene ics pe spec i e, he me apopula- ion model p edic s ha monoecious popula ions should ha e lowe gene ic a ia ion and should be mo e s ongly di - e en ia ed om one ano he han and odioecious popula- ions. This is because monoecious popula ions a e expec ed o display gene ic signa u es o mo e ecen coloniza ion bo - lenecks ha subsequen mig a ion has no had ime o e ase (e.g., Sla kin 1977; Wade and McCauley 1988; Pannell and Cha leswo h 2000). To es hese p edic ions, Obba d e al. (2006b) measu ed gene ic di e si y in hexaploid monoecious and and odioecious popula ions ac oss hei ange in he Ibe- ian Peninsula and No h A ica and in dioecious popula ions ac oss hei Eu opean ange. They ound ha , e en a e he e ec s o he hypo hesized ange expansion had been ac- coun ed o (see ‘‘Phylogeog aphy and Regional T ansi ions in he Sexual Sys em’’), hexaploid and odioecious popula- ions we e s ikingly mo e di e se han hei monoecious coun e pa s, as expec ed. Mo eo e , whe eas pai wise ge- ne ic di e en ia ion among hexaploid monoecious popula ions was highly a iable and o en la ge, pai s o and odioecious popula ions we e in a iably e y simila , as we e dioecious popula ions. These pa e ns sugges a his o y o homogenizing gene low among and odioecious popula ions and among di- oecious popula ions ( hough no be ween and odioecious and dioecious popula ions, owing o hei di e en ploidy le els; Obba d e al. 2006b). Al hough popula ions o M. annua wi h sepa a e sexes migh be linked by mo e gene low because hey con ain males ha dispe se mo e pollen han he maph odi es, such an explana ion would no accoun o he low di e si y o monoecious popula ions occu ing wi hin and odioecious egions (Obba d e al. 2006b). A key assump ion o he me apopula ion model is ha sel - ing a es a e densi y dependen : he maph odi es sel - e ilize hei p ogeny when ma es a e absen , bu oppo uni ies o ou c ossing inc ease when popula ions g ow. Eppley and Pannell (2007a) es ed his assump ion by es ima ing sel ing a es in popula ions g owing a di e en densi ies. Sel ing a es we e high when indi iduals we e mo e han ca. 30 cm apa bu quickly d opped in dense s ands, as p edic ed by he model. Eppley and Pannell (2007a) used hei esul s o p edic he h eshold densi y below which males should be excluded om a popula- ion and abo e which hei main enance was assu ed. This h eshold densi y was much lowe han commonly measu ed in pu ely monoecious popula ions, indica ing ha he p esence o males in a popula ion is indeed mig a ion limi ed (Eppley and Pannell 2007a). E olu ion o And odioecy: Sex Alloca ion and In lo escence S uc u e We ha e seen ha dense popula ions o Me cu ialis annua a e la gely ou c ossing, easing he po en ial in asion o males in o he maph odi ic popula ions. Bu how do males su pass he wo old h eshold in pollen p oduc ion equi ed o hei main enance a equencies g ea e han ze o (Lloyd 1975; Cha leswo h and Cha leswo h 1978; Cha leswo h 1984)? The me apopula ion hypo hesis sugges s one mechanism by in oking selec ion o emale-biased sex alloca ion in he he - maph odi es. Thus, i he maph odi es alloca e a p opo ion x<0:5 o hei ep oduc i e esou ces o hei male unc- ion, hen emale-s e ile indi iduals ha di e all hei e- sou ces o he wise in es ed in seeds o pollen p oduc ion will be able o dispe se 1=x>2 uni s o pollen ela i e o he pol- len p oduced by he maph odi es. Full ‘‘compensa ion’’ in e- sou ce alloca ion be ween male and emale unc ions would he e o e be su icien o male in asion as long as he he - maph odi es a e emale biased in hei sex alloca ion. Ano he way in which males migh c oss he wo old in a- sion h eshold is by dispe sing he pollen hey p oduce be e han he maph odi es. Assume again ha he maph odi es al- loca e a p opo ion xo hei esou ces o pollen p oduc ion and ha males alloca e a p opo ion 1.0. Assume u he ha pollen g ains dispe sed by males a e g imes mo e likely o ind a ecep i e s igma han hose dispe sed by he maph- odi es. Then males should enjoy a ela i e si ing success o g=x. Clea ly, i g>1:0, hen males migh in ade a popula- ion o he maph odi es e en i x¼0:5, i.e., he popula ion is unbiased in i s sex alloca ion. Because we expec he sex allo- ca ion o ou c ossing he maph odi es o be ca. 0.5, i he male and emale i ness gain cu es a e no e y di e en Fig. 4 Map showing he loca ions o 22 popula ions o Me cu alis annua o which he ploidy le el was de e mined along he A lan ic coas o Mo occo. Open and illed ci cles ep esen e aploid and hexaploid popula ions, espec i ely. The solid line nea Raba indi- ca es he poin o ansi ion be ween monoecious popula ions ( o he sou h) and and odioecious popula ions ( o he no h). Two e aploid popula ions we e ound wi hin he a ea occupied p incipally by hexa- ploid popula ions. 134 INTERNATIONAL JOURNAL OF PLANT SCIENCES (Lloyd and Bawa 1984), his easoning shows ha he condi- ion g>1:0 ough o be su icien o he e olu ion o maleness ai ly gene ally. Eppley and Pannell (2007a) es ed his idea and es ima ed a alue o g¼1:6. Tha males a e so much be e han he maph odi es a dis- pe sing hei pollen is almos ce ainly due o he di e ences be ween he wo mo phs in hei in lo escence a chi ec u es. Al hough monoecious plan s ha e a emale mo phology, wi h male lowe s held a ound a subsessile emale lowe in he lea axils, males dispe se hei pollen om lowe s held abo e he plan on long, e ec peduncles. In lo escences such as hose o M. annua males a e widely ound in wind-pollina ed he bs and a e expec ed o inc ease he si ing success o pollen by allowing i o a el a he a e elease (Le in and Ke s e 1974; Niklas 1985; Bu d and Allen 1988). Such adap a ions almos ce ainly gi e ise o an accele a ing male i ness gain cu e, and hey should hus s abilize he main enance o sep- a a e sexes (Cha no e al. 1976). I is poo ly unde s ood whe he he e olu ion o such in lo escences is mo e likely o p ecede, and hus o p ecipi a e, a ansi ion om he maph- odi ism o dioecy o o ollow he e olu ion o dioecy and wind pollina ion. Howe e , ansi ions be ween wind pollina- ion and dioecy o en go hand-in-hand (Renne and Rickle s 1995; Wallande 2001; F iedman and Ba e 2008), and hey may coincide wi h seconda y sexual adap a ions such as hose ound in M. annua (see Welle e al. 1998, 2006; Ka - enbe g e al. 2002; F iedman and Ha de 2004; Golonka e al. 2005). Main enance and B eakdown o And odioecy: The Role o Pheno ypic Plas ici y Sex exp ession is no o iously labile in plan s, wi h males and, less o en, emales o dioecious popula ions equen ly p oducing lowe s o he opposi e sex (Lloyd and Bawa 1984; Ko pelainen 1998). In gynodioecious and and odioe- cious species, he he maph odi es oo may a y in hei sex alloca ion in esponse o en i onmen al cues ( e iewed by Delph and Wol [2005]). Whe eas he implica ions and unc- ional signi icance o comple e gende swi ches emain poo ly unde s ood, mo e sub le exp essions o sexual labili y may ha e impo an consequences o sexual-sys em e olu ion, e.g., by con e ing upon a lineage an abili y o sel - e ilize a e long-dis ance dispe sal (Bake and Cox 1984). Plas ici y in he sex alloca ion o he maph odi es is p obably d i en by a ia ion in he ma ginal cos o esou ces needed o ep o- duc ion o in changes o he shapes o i ness gain cu es (Lloyd and Bawa 1984; Klinkhame e al. 1997), and i has impo an implica ions o sexual-sys em e olu ion h ough i s e ec on he main enance o males o emales. The pheno ypic plas ici y in sex alloca ion o Me cu ialis annua he maph odi es should ha e impo an implica ions o he equency o males ha can be main ained in and o- dioecious popula ions. Indeed, along wi h me apopula ion dynamics, i migh be a u he ac o unde lying he la ge among-popula ion a ia ion obse ed in male equencies o he species. As a gued by Delph (2003), plas ici y in sex allo- ca ion by he maph odi es can change he ela i e e ili y o unisexual and he maph odi e plan s (see also Delph and Lloyd 1991). Because he ela i e pollen o seed e ili y o uni- sexuals and he maph odi es egula es hei equilib ium e- quencies (Lloyd 1976), such plas ici y can cause a ia ion in unisexual equencies ac oss ecological g adien s (Delph 1990; Delph and Lloyd 1991; Asikainen and Mu ikainen 2003; Ba 2004). Fo example, males always alloca e all o hei ep oduc i e esou ces o pollen p oduc ion (i.e., xm¼1). I he maph odi es change he p opo ion o esou ces alloca ed o pollen in esponse o an ecological g adien (i.e., x h a ies ac oss he g adien ), hen his also changes he ela i e pollen e ili ies o males and he maph odi es (i.e., ¼xm=xh). Hal o he p ogeny si ed by males a e, on a e age, male, ia he seg- ega ion o dominan male-de e mining alleles (Pannell 1997b). Thus, di e ences in ac oss en i onmen al g adien s will a - ec he ela i e equency o male-de e mining pollen in he popula ion and he equency o males a equilib ium. In e - es ingly, such a scena io would concu wi h Da win’s (1877) easoning ha un a o able en i onmen al condi ions can lead o an inc eased sepa a ion o he sexes and ollows simila pa - e ns shown by o he species ( e iewed by Delph and Wol [2005]). In a p e ious expe imen , M. E. Do ken and J. R. Pannell (unpublished manusc ip ) ound ha plan densi y go e ns e olu iona y ajec o ies in male equencies by a ec ing he magni ude o . Unde high plan densi ies, was mo e han h ee imes highe han unde low densi ies (and see Pannell 1997c). In he nex gene a ion, male equencies we e 38% highe among he p ogeny o plan s g own unde high densi- ies (M. E. Do ken and J. R. Pannell, unpublished manu- sc ip ). Densi y can a ec male si ing success in wo ways. Fi s , densi y may a ec he local a ailabili y o esou ces by inc easing compe i ion among plan s. Second, because pollen is dispe sed locally (Eppley and Pannell 2007a), densi y di- ec ly a ec s male si ing abili y. We a emp ed o dissec hese con ounded e ec s by con olling plan densi y and ma- nipula ing esou ce a ailabili y. Does esou ce a ailabili y a ec he magni ude o ? I so, does a ia ion in esou ce a ailabili y egula e male equencies in M. annua? To add ess hese ques ions, we mixed seeds om i e nea by si es in Mo occo, each wi h high equencies o males (i.e., be ween 25% and 47%). We g ew plan s om seed in 12 3 33-m aised beds in s anda dized, low-nu ien soil mix- u es a he Wy ham Field Lab be ween July and Oc obe 2004. We added slow- elease nu ien e ilize pelle s (Os- moco e, Sco s, Ma ys ille, OH) a he ecommended dosage o hal o he beds, using a andomized block design wi h h ee blocks. On a e age, he e we e 566 697 (SD) plan s in he high- esou ce plo s and 566 642 (SD) plan s in he low- esou ce plo s. We main ained high male equencies in each plo by ansplan ing males om addi ional simila plo s no included in his analysis he e (a e age male equency in high- esou ce plo s ¼52:8% 61:4% [SE], in low- esou ce plo s ¼52:5% 68:2% [SE]). A he end o he expe imen , we ha es ed a s anda dized sample o 20 he maph odi e and 10 male plan s om each plo by sampling indi iduals a egula in e als along ou e enly spaced ansec s. Follow- ing Pannell (1997c), we calcula ed he maph odi e alloca ion o male unc ion ( ) as he abo eg ound p opo ion o bio- mass alloca ed o pollen o he maph odi es (p h ) ela i e o 135 PANNELL ET AL.—SEXUAL-SYSTEM TRANSITIONS IN MERCURIALIS ANNUA ha o males (p m ; i.e., ¼pm=ph). Thus, lowe alues o indica e highe he maph odi e alloca ion o pollen. We ound ha he maph odi es g own unde high- esou ce condi ions (a e age ¼11:260:9 [SE]) had nea ly double he alloca ion o pollen o plan s g own unde low- esou ce condi ions (a e age ¼21:662:3 [SE]; ig. 5a; wo- ailed - es wi h equal a iances: ¼4:28, d ¼10, P<0:005). A he end o Sep embe , we ha es ed all he maph odi es in a plo and collec ed hei seed in bulk, and we sowed an a - e age o 1500 seeds (es ima ed by weighing ou s anda d amoun s o seeds) back in o he plo in June 2005. Because he maph odi es had highe alloca ion o pollen in high- esou ce condi ions, he p opo ion o p ogeny si ed by males should ha e been lowe han ha unde low- esou ce condi- ions. Mo eo e , because maleness is go e ned by he seg e- ga ion o a dominan Mendelian allele (Pannell 1997b), we p edic ed ha he equency o males among he p ogeny o plan s g own unde high- esou ce condi ions should be lowe han ha among hose om low- esou ce condi ions. Ou esul s a e consis en wi h his p edic ion ( ig. 5). The a e age equency o males om high- esou ce plo s was 40:9% 6 1:1% (SE; a e age numbe o plan s pe plo ¼720 674 [SE]), compa ed wi h 45:8% 61:6% (SE) o low- esou ce plo s (a e age numbe o plan s pe plo ¼501 683 [SE]; one- ailed - es wi h equal a iances: ¼1:92, d ¼10, P<0:05). Ou esul s indica e ha plas ici y in sex alloca ion egu- la es he equency o unisexual plan s in gende -dimo phic popula ions, suppo ing he p edic ion made by Delph (2003). E en hough ou expe imen was conduc ed ac oss a single gene a ion, we ound conside able changes in male equen- cies in esponse o en i onmen al condi ions. Speci ically, we ound ha male equencies no only can espond o densi y, which egula es male si ing abili y (as p edic ed by Eppley and Pannell 2007a), bu also can espond di ec ly o a ia ion in imposed by a ia ion in he esou ce s a us o he maph- odi es. These esul s demons a e he impo ance o pheno- ypic plas ici y in he maph odi e sex alloca ion o egula ing male equencies in and odioecious M. annua. Gene ics o Sex De e mina ion and Sex Alloca ion in Me cu ialis annua The gene ic basis o sex exp ession in dioecious Me cu ia- lis annua was s udied in ensi ely by Du and and co-wo ke s (Louis and Du and 1978; Dauphin-Gue in e al. 1980; Du- and and Du and 1991a, 1991b). On he basis o c ossing expe imen s using a small sample o o iginal geno ypes, hey concluded ha sex was de e mined by epis a ic in e ac ions be ween alleles seg ega ing a h ee independen ly seg ega ing loci. The ac ha amily sex a ios a e o en s ongly male o emale biased, despi e a 1 : 1 sex a io a he popula ion le el (J. R. Pannell, pe sonal obse a ions), would seem o con i m ha mo e han one locus is in ol ed, bu esul s o ecen gene ic analysis based on sequence-cha ac e ized ampli ied- egion (SCAR) ma ke s we e consis en wi h a sin- gle locus (Khadka e al. 2002). Pannell (1997b) concluded ha maleness in an and odioecious popula ion was de e - mined by he p esence o a single dominan allele, bu his analysis also poin ed o he impo ance o a plas ic compo- nen o sex exp ession. Clea ly, much emains o be lea ned abou sex de e mina ion in M. annua and abou he e olu- ion o sex de e mina ion in he complex h ough i s sexual- sys em ansi ions. I is clea om he s udies e iewed abo e ha sex alloca- ion in he maph odi es o M. annua is also highly a iable, albei as a con inuous ai . Much o his a ia ion is e i- den ly due o pheno ypic plas ici y, bu compa isons be ween Fig. 5 Pheno ypic plas ici y in he maph odi e sex alloca ion a - ec s e olu iona y ajec o ies in male equencies. a, The ela i e pollen p oduc ion ( ) o males, compa ed wi h ha o he maph odi es, di e s be ween low- and high- esou ce condi ions. b, Because he pollen e ili y o males ela i e o ha o he maph odi es egula es he equency o males, di e en alues o be ween low- (ci cles) and high- esou ce condi ions (squa es) lead o di e gence in male e- quencies in he nex gene a ion. In b, e o ba s ep esen s anda d e o s. 136 INTERNATIONAL JOURNAL OF PLANT SCIENCES Mo occan and Spanish popula ions sugges an impo an ge- ne ic componen , oo (J. R. Pannell, R. Be jano, and S. M. Eppley, unpublished manusc ip ); howe e , he a chi ec u e o his gene ic a ia ion in M. annua is unknown. I a mo e comple e pic u e o pas e olu iona y ansi ions be ween sexual sys ems in he species complex is o eme ge, we need u he in o ma ion abou he quan i a i e gene ics o sex al- loca ion. Two speci ic ques ions a e cu en ly he ocus o a en ion o ou esea ch, and bo h ela e o he so-called b eede ’s equa ion, R¼h2S, which ela es he po en ial esponse o a popula ion o selec ion, R, o he p oduc o he selec ion co- e icien , S, and he na ow-sense he i abili y o he ai , h 2 . Because h2¼VA=VP, i.e., he ac ion o o al pheno ypic a iance ha is due o addi i e gene ic a iance, he po en ial o esponse o selec ion on sex alloca ion will depend on bo h he absolu e amoun o addi i e gene ic a iance, V A , and he size o V P , which will be in la ed by pheno ypic plas- ici y (Falcone and Mackay 1996). The i s ques ion hus conce ns how pheno ypic plas ici y in sex alloca ion in M. annua will a ec esponses o selec ion on pollen p oduc ion. This ques ion is o gene al impo ance o ou unde s anding o ansi ions om gynodioecy and and odioecy o dioecy (Delph and Wol 2005). The second ques ion conce ns he size o V A : how much addi i e gene ic a iance in sex alloca ion is p esen in popu- la ions o M. annua wi h di e en sexual sys ems, and how migh his a y geog aphically? The geog aphic pe spec i e on his ques ion is impo an because ange expansions, which a e known o ha e occu ed in M. annua (see ‘‘Phylo- geog aphy and Regional T ansi ions in he Sexual Sys em’’), ypically in ol e epea ed popula ion bo lenecks ha a e ex- pec ed o educe quan i a i e gene ic a ia ion (Lande 1992). Indeed, ange expansion in M. annua was in e ed on he ba- sis o educed gene ic a ia ion a isozyme loci wi h dis ance om pu a i e e ugia (Obba d e al. 2006b). Wha we do no know is whe he he hypo hesized ange expansions a - ec ed gene ic a ia ion in sex alloca ion, which is unde equency-dependen selec ion, di e en ly. Cu en wo k is add essing his ques ion h ough an assessmen o he e- sponse o selec ion on pollen p oduc ion in he maph odi es sampled om popula ions a di e en la i udes in Spain and No h A ica. Conclusions Me cu ialis annua displays unusual a ia ion in i s sex ex- p ession a se e al le els in he genealogical hie a chy, in- cluding wi hin and among geno ypes, among popula ions in pu a i e me apopula ions, among egions in di e en pa s o Eu ope, and be ween ela ed species in a clade ep esen ed by lineages wi h e icula e phylogene ic ela ionships. This a ia- ion e lec s a complex his o y o di e gence, mig a ion, and gene low, which ha e all con ibu ed o ansi ions be ween sexual sys ems. Resea ch on M. annua has h own ligh on o a numbe o hemes in sexual-sys em e olu ion. These include he impo ance o ploidy, hyb idiza ion, and ange expan- sions in egula ing sexual-sys em a ia ion and hus se ing he s age on which na u al selec ion hen ac s; he impo ance o pheno ypic plas ici y unde lying much o he a ia ion ob- se ed be ween indi iduals and i s esponsibili y in egula ing de ails o sex a io e olu ion; he ole o con ex -dependen ma ing in species cha ac e ized by demog aphic luc ua ions and possible me apopula ion dynamics; he adap i e alue o sexual specializa ion o pollen dispe sal unde wind pollina- ion; and he po en ial ole o and odioecy in e olu iona y pa hs be ween dioecy and he maph odi ism, a pa hway abou which e y li le was hi he o known. Fu u e wo k aims o cha ac e ize he demog aphic p ocesses hypo hesized o ha e played a ole in d i ing he obse ed sexual-sys em ansi- ions a egional and local-pa ch scales as well as o desc ibe he quan i a i e gene ic a chi ec u e o sex de e mina ion and sex alloca ion ac oss he phylogene ic and geog aphic ange o he species complex. Acknowledgmen s We hank Spence Ba e o in i ing us o con ibu e o his special issue o he jou nal, Anne Sakai and wo anony- mous e iewe s o use ul commen s on he manusc ip , and he Wellcome T us and F. M. Pla o low cy ome y acili- ies. J. R. Pannell and B. Pujol we e suppo ed by g an s om he Na u al En i onmen Resea ch Council (NERC) o he Uni ed Kingdom; M. E. Do ken was suppo ed join ly by NERC and he Na u al Sciences and Enginee ing Re- sea ch Council o Canada; and R. Be jano was suppo ed by a Spanish g an o he Minis e io de Educacio ´n y Ciencia. Li e a u e Ci ed Asikainen E, P Mu ikainen 2003 Female equency and ela i e i ness o emales and he maph odi es in gynodioecious Ge anium syl a icum (Ge aniaceae). Am J Bo 90:226–234. Bake HG 1955 Sel -compa ibili y and es ablishmen a e ‘‘long- dis ance’’ dispe sal. E olu ion 9:347–348. Bake HG, PA Cox 1984 Fu he hough s on dioecism and islands. Ann Mo Bo Ga d 71:244–253. Ba CM2004Soilmois u eandsex a ioinaplan wi h nuclea -cy oplasmic sex inhe i ance. P oc R Soc B 271:1935–1939. Ba e SCH, JR Pannell 1999 Me apopula ion dynamics and ma ing- sys em e olu ion in plan s. Pages 74–100 in P Hollingswo h, R Ba eman, R Go nall, eds. Molecula sys ema ics and plan e olu- ion. Chapman & Hall, London. Buggs RJA, JR Pannell 2006 Rapid displacemen o a monoecious plan lineage is due o pollen swamping by a dioecious ela i e. Cu Biol 16:996–1000. ——— 2007 Ecological di e en ia ion and diploid supe io i y ac oss a mo ing ploidy con ac zone. E olu ion 61:125–140. Bu d M, TFH Allen 1988 Sexual alloca ion s a egy in wind-pollina ed plan s. E olu ion 42:403–407. Case AL, SW G aham, TD Mac a lane, SCH Ba e 2008 A phylo- gene ic s udy o e olu iona y ansi ions in sexual sys ems in Aus al- asian Wu mbea (Colchicaceae). In J Plan Sci 169:141–156. Cha leswo h D 1984 And odioecy and he e olu ion o dioecy. Biol J Linn Soc 23:333–348. ——— 1999 Theo ies o he e olu ion o dioecy. Pages 33–60 in MA Gebe , TE Dawson, LF Delph, eds. Gende and sexual dimo phism in lowe ing plan s. Sp inge , Heidelbe g. 137 PANNELL ET AL.—SEXUAL-SYSTEM TRANSITIONS IN MERCURIALIS ANNUA