Floral Sex Allocation in Sequentially Blooming Plants Author(s): Johanne Brunet and Deborah Charlesworth Reviewed work(s): Source: Evolution, Vol. 49, No. 1 (Feb., 1995), pp. 70-79 Published by: Society for the Study of Evolution Stable URL: http://www.jstor.org/stable/2410293 . Accessed: 02/03/2012 16:40 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to Evolution. http://www.jstor.org Evolution, 49(1), 1995, pp. 70-79 FLORAL SEX ALLOCATION IN SEQUENTIALLY BLOOMING PLANTS JOHANNE BRUNET1 AND DEBORAH CHARLESWORTH2 IDepartment of Botany and Plant Pathology, Oregon State University, Corvallis, Oregon 97331-2902 and Rocky Mountain Biological Laboratory, Crested Butte, Colorado 81224 2Departmentof Ecology and Evolution, Universityof Chicago, 1101 East 57th Street, Chicago, Illinois 60637-1573 flowersmayexhibittemporalvariationin pollen Abstract.-Inplantswhose flowersdevelopin a sequence,different donationand receiptsuch thatthefitnesscontributions throughmale and femalefunctionscan varyamongflowers. withininflorescences, can createsituationswhereflowersin different Dichogamy,or directional pollinatormovements stages in the sequence may differin the numbersof flowersin the femalestage available as potentialmates.We flowersin presentan evolutionarily stable strategy(ESS) analysisof theresourceallocationsexpectedin different hermaphroditic plantswhenthe matingenvironments varyamongflowers.This introducesa modularelementinto of flowerscan select for sex-allocationmodels.Our analysisshows thatsuch variationin thematingenvironments in sex allocationbetweenflowers.Whenmale and femalefertilities are nonlinearfunctionsof the allodifferences cations,variationin resourceavailabilitycan also selectforvariationin sex allocationamongflowers.The influence of dichogamyand pollinatordirectionality on floralsex allocationis discussed,and theempiricalevidencesupporting the predictionsderivedfromthe model is brieflyreviewed.The implicationsof our resultsfor the evolutionof and monoecyare discussed. andromonoecy pollinator movement, reproductive Keywords.-Dichogamy, gaincurves,malefunction, matingenvironment, resources, selfing,sex allocation. ReceivedApril29, 1993. AcceptedMay 3, 1994. flowers(andprotogyny fortheopposite [W]ithhermaphrodite plantsthatare strongly proterandrous,paredto late-opening stable strategy thestamensin theflowerswhichopen firstsometimesabort; pattern).Here, we presentan evolutionarily andthisseemsto followfromtheirbeinguseless,as no pistils (ESS) analysisof the resourceallocationsexpectedin difplantswhentheamountsof ferentflowersin hermaphroditic are thenreadyto be fertilised. -Darwin (1877, p. 283) reproductiveresourcesavailable and the matingenvironthatvarimentsvaryamongflowers.This analysisconfirms In hermaphroditic plants,the male and femalecontribu- ation in the matingenvironments of flowerscan select for tionsto fitnessare each due to thesumof contributions from variationin sex allocationamong flowers.It suggeststhat each of theplant'sflowers.It is thusimportant to incorporate variationin resourceavailabilityamongflowerscan under themodularnatureof plantreproduction intosex-allocation somecircumstances also selectforvariationin sex allocation. theory(StantonandGalloway1990). In plantswhoseflowers The modelmakespredictions thatcan be testedempirically developin a sequence,pollenfromflowersin different stages (see below). As such predictionsinvolve comparisonsof There- flowerson inflorescences, will have different forsiringoffspring. opportunities theyare not confoundedby any male and femalefunc- otherpossibledifferences fore,thefitnesscontributions through betweenparentalplants. tionscan varyamongflowers.In particular, thenumbersof flowersin the femalestage thatare available as potential MODEL matesto each ofthemale flowerscompeting fortheseovules Fitness Expressionsfor the Model may differforflowersat different stages in the sequence. No Allocation to'Attraction Assuming in the These kindsof difference, whichwe termdifferences of flowers, in termsof matingenvironment can be quantified The modelassumesthatthereare severaldifferent stages thenumbersof ovules availableto pollenof flowersthatare offlowers, whichmight,forexample,correspond to theflowpotentialmates of these ovules, for each of the stages of ers thatopen in sequenceon an inflorescence. We shallrefer It seemsprobablethatvariationin thematingen- to these as flowerpositions.In orderto performan ESS flowering. vironments of flowersduringthe flowering sequence could analysisof the consequencesof this situation,we need to in theirsex allocation. selectfordifferences withdifhave an expressionforthefitnessesof phenotypes Factorssuchas dichogamy(temporalseparationof sexual ferent to possibleallocations.Male and femalecontributions functions withinflowers)andpollinator directionality (move- fitness fromflowersateachposition aresumsofcontributions mentof pollinatorsalong inflorescences) mayproducesuch in the sequence. It is assumedthatsome total amountof in the matingenvironment differences of flowers(Darwin resourcesis availableforreproductive functions andthatthis 1877; Pellmyr1987; Brunet1990). In protandrous plants, does not differbetweenindividualsin the population.The wheretheanthersdehiscebeforethestigmasbecomerecep- reproductive resourcescouldbe eitherthoseavailableatflowtive,flowersproducedearlymayhavea low ratioofavailable eringorcouldincluderesourcesavailableforinvestment into ovules to pollencompetingfortheseovules,comparedwith fruits.Let a fixedfractionTj of the totalreproductive relaterflowers(fig. 1). In thiscase, protandry would be ex- sourcesbe availableto thejth flowerposition.We denoteby of thereproductive pectedto selectforfemale-biasedallocationin early-com- Mj theproportion resourcesallocatedto 70 ?) 1995 The Society for the Study of Evolution. All rightsreserved. 71 FLORAL SEX ALLOCATION 40 30 1st position flowers 30:0 30 2nd Malephase floweU Female phase 3rd flower position 20 - 2010 10 40 8- 2nd flower position 4th flower position 30 -6cmCY't LO O - co 0) 0 CCO 'o LC)cs r- no 0) 0 o\ CO 'It 0 1. LO (o C\j V) O coo CM C4 ,I Un co fl 00 a) o c\1m ql 20 -4- 10 0 2- .... .. ~0 Day . . . . . . . . . . . . . Day FiG. 1. Sequentialbloomingin thefirst fourflowerpositionsin inflorescences of Aquilegiacaerulea,a plantwithprotandry creatinga timedelay betweenthemale and femalephases of flowers. male functionsby flowersof thejth position,withthe re- to fitnessfromall stagesof flowersis givenby thisexpresmainderbeingallocatedto femalefunctions (i.e., we ignore sion. Withresourcelimitationof seed output,the quantity allocationto attraction, so thatallocationto femalefunctions underthesummation wouldbecomea function ofthenumber of fertilizedovules. We have used a power function,with is Fj = 1 -Mj). thepossibilityof resourcelimitaThe ESS value of themale allocationparameters forflow- exponentkl, to represent ers at thejth position,Mj, was foundby themethodof Char- tion, as in previousmodels (Charnov 1982; Lloyd 1984; and Charlesworth 1987). The exponentsof the lesworth(1990), givenin theAppendixforthegeneralcase Charlesworth in whichreproductive resourcescan be allocatedto male or gain curvesare notallowedto evolve in theallocationmodfemalefunctionsor to pollinatorattraction. Here, we give els. The model assumesa pool of pollen thatwill in partbe thefitnessexpressions,ignoringallocationto attraction. in this pool Assumingcompleteoutcrossingand a constantcost per depositedon stigmas,and thatrepresentation male fertility. Thus,themale contribution to fitovule,thenumberoffertilized ovulesfromall stagesofflow- determines nesscomesfromthecontributions ofpollenfromeach flower ers is proportional to position,to the outcrossingpollen pools of flowersof difWfX (1) ferent Tj(1 -M ). positionsthatareopenduringpollenproduction ofthe flowersunderconsideration. Let Kij be a matrixof relative Assumingthatall fertilizedovules matureinto seeds (i.e., probabilitiesthatpollen producedin flowersof thejth poresourcesdo notlimitseed output),thefemalecontribution sitionwill becomepartof thepollen pool of flowersat the 72 J. BRUNET AND D. CHARLESWORTH ith position.This matrixallows us to separatethe pollentransfer portionof thereproductive processfromtheeffects on pollenand ovule production. of theallocationdifferences Pollen transfer presumably dependschieflyon thedegreeof temporaloverlapbetweenthemale and femaleflowerstages of flowersat different positions,or on pollinatorbehaviors betweenflowerpositions(as well as thatinfluencetransfer to pollinators,whichis not discussed on the attractiveness oftheKij valuesfromfielddata further here).The estimation will be describedbelow. fora to fitnessthrough The male contribution outcrossing particularphenotypeis Wm j Wmj = Ii wmji whereSi is theselfingrateforflowersof thejth position,and 8 is the inbreedingdepression(i.e., the loweringof fitness relativeto thatof outcrossedprogeny). in progenyof selfing, In addition,thenumberofovulesin flowersofthejthposition by thepollenpool forthat thatare availableforoutcrossing positionwas replacedby TjFj(1 - Sj). The malecontribution to fitnessfromflowersof thejth positionbecomes: Win = TM j2j E T1Fli(I -Si TM K i (5) Apartfromthese changesin the fitnessformulae,and the alterationsin the derivativesof the fitness corresponding themodelwas as describedpreviously. functions, Afterwritingthe expressionsfor the derivativesof the fitness equations,we solvedfortheESS allocationsusingthe using iterativemethoddescribedin theappendix,arbitrarily 0.1 forall VAvalues. The calculationsweremadeon a MacusingMicrosoftQuickBasicprogram,and intoshcomputer, on a PC usinga Fortranprogram.In bothcases, theprogram includedthe derivativesof fitness,based on the fitnessexpressionsexplainedabove. Calculationof thesecondderivatives of the fitnessfunctionat the ESS was done to test whetherthevalues foundwere stable. wherewmjiis the numberof ovules in flowersof the ith to their position,multipliedby theratioof thecontribution pollenpool fromflowersat thejth positionto thetotalpollen pool thatis competingforthe ovules in question(i.e., the fractional contribution of pollenfromthejth flowerposition to thetotalpollenpool forflowersof the ithposition).The in flowersof the numberof ovules availableforoutcrossing ithpositionis givenby Tj(1 - Mi). In thesimplestpossible of flowersof each positionto the model,the contributions are proportional to theallocation pollenpool foroutcrossing parameters, Mj; thatis, pollenoutputfromflowersat thejth Estimationof the Matrix of positionis givenby Tj X Mj. Withno pollen limitationof Pollen-TransferProbabilities this norresourcelimitation of seed production, fertilization ratiofora givenallocationphenotype designatedby thesubOurmodelseparatesthepollen-transfer portion(Kijvalues) at fre- of thereproductive script1, in a populationwithsome set of phenotypes in processfromtheeffectsof differences quenciesPr, is then To makethe allocationto totalpollenand ovule production. (2) modelconcrete,it is helpfulto show how the matrixof Kij IMljKii . valuescan be estimatedfromfielddata.The rawobservations EEPrTMrjKij consistof recordsof the numbersof flowersopen at each r j withthe starting flowerposition,foreach day of flowering, Withthemorerealisticassumption thatpollenoutputis pro- day of openingof the firstflowerof the earliest-flowering to thepollen individual.Ideally,thesedata coLuldbe collectedseparately portionalto theMj values,butthatcontributions withpollenout- forflowersopen as each sex. These data are thencondensed pool showa diminishing-returns relationship put,we can replacetheproductTj X Mj by a powerfunction, into tables containingthe numbersof flowersin the popusay,(Tj X Mj)k2.Notethattherelative,ratherthanabsolute, lationthatare open and have dehiscedantherson each day period,classifiedby theirpositions,and a is ofinterest allocationto malefunction here,as reproductive of the flowering similartable forall flowerswithreceptivepistils,tabulated resourcesdifferbetweenpositions. To performan ESS analysis,we need the fitnesscontri- by day and by position.These data are denotedby Nmnjand wherej refers, respectively, fora new phenotype(type2) Nf,j,formaleandfemaleflowers, butionvia male reproduction introducedinto a populationconsistingalmostentirelyof as before,to flowerposition,and c denotestheday. In reality,theflowersmaynotbe classifiedas to theirsex type1. For thesimplestcase discussedabove, we have whentheyare scored,butthedaythateach flower functions, E T-(1 - M,j)TjM2jKU (3) opens may be known.In thatcase, the raw data fora proWin1 potandrousplantconsistof recordsof flowersat different sitionsopen on each day of observation.Knowingthetime andassuminga fixedaverageamount to the case wheneitheror bothof the courseoftheprotandry, The generalizations thesedata in beforepistilmaturity, of time the male stage show diminishing returnswithinmale or femalefertilities of a table the numbers could used to construct be N1ci and creasingallocations,can be made simplyby replacingthe flowers on each of male and female day,and open stage NfCi termsin thisequation,and thatforthe femalecontribution the methodoutlinedherecould stillbe used. The same apto fitness, mentionedabove. by thepowerfunctions plant.Figure1 To includeselfing,thefitnessequationsabove weremod- proachcould also be used fora protogynous this data forAqan of of illustrates type example (above) to fitnessresulting ifiedas follows.The femalecontribution uilegiacaerulea.These data werecollectedas thefirstdays fromflowersat positionj was replacedby: on whichflowerswere open (in the male phase), and the (4) femalephase was assumedto begin5 d laterthanthe male TjFj[(l - Sj) + 2Sj(1 - 6)], 73 FLORAL SEX ALLOCATION 1. Estimatesof therelativepollen-transfer probability (Ki1) allocationamongpositions.The parametervaried,and the values forAquilegiacaeruleain Coloradoin 1987,calculatedfrom resultsof thecalculationsare summarized in table 2. theraw data shownin figure1. TABLE Position of female flower Obligate Outcrossing Flower position as male (j) 1 3 2 4 5 Case 1. No Effectof Position on Pollen Transfer(All Kij thecase ValuesEqual).-It is helpfulto startby considering of are independent of pollentransfer whentheprobabilities thepositionsof theflowersproducingthepollen(i.e., pollen randomly,and ovules of all flowerpositions is transferred sharethesamepollenpool, see table2, firstkindof matrix). The ESS allocationsthendo notdifferamongpositions,unbetween resources(Tj) differ less theamountsofreproductive (kl and k2) are unpositions,and thegain-curveparameters phase, based on observations of the average time delay be- equal. Whentheseconditionsare satisfied, in sex differences tween the sex stages. allocationare expectedbetweenpositions.Figure2 shows To obtain the Kij values, one firstneeds to know what somenumerical exampleswhenthevalue of Tj decreasesfor fractionof the ovules available on a given day, c, will be laterflowerpositions(the resultsforthe oppositedirection fertilizedby pollen from position j flowers. Assuming that of change in Tj values were exactlythe oppositeof those the transferprobabilitiesare determinedby the extentof tem- shown).Whentheexponentforthemalegaincurveis larger poral overlap between the male and female stages of flowers thanthatforthefemalecurve(k2 > kl), positionswithgreatat differentpositions, this will depend on the fraction of er amountsof reproductive resourcesare moremale-biased flowers open in the male phase on day c whose position is thanpositionswithfewerresources,and the reverseholds j. These fractionsare easily found fromtables of flowersin whenk2 < kl (fig.2). 1 2 3 4 5 Totals 0.506 0.077 0.022 :0 :0 0.611 0.422 0.284 0.087 0.024 :0 0.821 0.299 0.296 0.095 0.041 :0 0.734 0.414 0.630 0.721 0.172 0.072 2.101 0.524 0.920 1.541 0.359 0.169 3.511 the male and female phases. We denote these fractionsby fcj given by Case 2. All Flower Positions Equally Successful As Pollen Donors, But Transfer to Ovules of DifferentPositions (K11 Values) Varies withBoth Donor and Recipient Position.-In is necessarily of pollentransfer case 1, thetotalprobability to examine thesame forall flowerpositions.It is interesting flowerpositions the effectof the situationswhen different where success as donorsof pollen. Denotingby Kj have different thatpollenof flowersat a givenposition the total probability ENm ji in thepollen pool, it is is successfulin being incorporated possible(thoughnotbiologicallyrealistic)thattheKj values is the total numberof flowersopen on day c. The numberof could be the same forall flowerpositions,even if thereis position i flowersthatwill be pollinated by positionj donors variationin the Kij values. In this case also, the ESS sex is thereforethe sum, over all days, of these fractionsmul- allocationsof different positionsdo not vary unless the tiplied by the number of ith position flowers whose pistils amountofreproductive resourcesat each position(Tj values) are receptive on day c, that is, also varies,even thoughnow the patternof pollen transfer differs forovules in different positions(table2, secondkind E fmcjNfCi. c of matrix). If theKij values differ, however,thenunlikecase 1, variTo put this on a per-flowerbasis, and obtain relative K11 in sex alfordifferences ationin the T7values is sufficient values, these sums are divided by the total numbersof flowers locationbetweenflowersto evolve. The extentof the alloat the relevant donor positions, betweenthe cationdifferences dependson the relationship probabilitiesand theamountof reproductive pollen-transfer E Nmcj: c resourcesavailable at each position.Whendonorflowersat flowersfrom ofpollinating positionj have a highprobability f mCjNf ci resources(high c positionswithlargeamountsof reproductive success as T1 values), theywill have greaterreproductive E Nmcj~ c males; thus,even if the male and femalegain curves are identical,one expectsa greaterbias in sex allocationthan The results for the Aquilegia caerulea data of figure I are whentheseflowerspollinatestigmasat positionsreceiving given in table 1. lesser resourceamounts.Figure 3 shows some examples in thesecondtypeof K whenthevalues of theparameters m = Nm Cj Nm C RESULTS As qualitatively similar results were obtained for three-, four-,or five-floweredinflorescences,the resultsas presented for four-floweredinflorescences.We examine several cases i;o clarifywhich factorsmost stronglycause variation in sex matrixof table 2 were a = 0.6, b = 0.2, c = 0.1, d = 0.1. To obtainmuchvariationin sex allocationamongpositions, For werenecessaryin theKijprobabilities. strongdifferences thesame Tj, kl, and k2 values as in figure2, theresultsare butmalebias is lowerforflowerslateinthesequence, similar, 74 J.BRUNET AND D. CHARLESWORTH Examples of the typesof pollen-transfer matricesused in the calculations,and trendsin the ESS sex-allocationpatterns amongpositionsin obligateoutcrossers.The examplesare forfour-flowered inflorescences. TABLE 2. Position of ovules 1 2 3 4 Total Resources (Tj) Position of pollen donor (j) (i) 1 a a a a 4a No. Effectof Positionon Pollen Transfer 3 4 a a constant a a a a variable a a 4a 4a 2 a a a a 4a Gain-curve parameters kl = k2 k2 =Ak2 kl = k2 kl =Ak2 Allocation patterns all Mj= 0.5 all Mj = k2/(kl+ k2) all Mj = 0.5 M MI= M2 =A.. All FlowerPositionsEqually Successfulas Pollen Donors,butDifferential Transfer to Ovules of Different Positions* 1 2 3 4 b a c kl = k2 d all Mj = 0.5 constant b d a c kl?Ak2 allMj=k2/(kl +k2) c d b a variable kl = k2 M1I ? M2 ?A....A Mi 1 2 3 4 d c 1 2 1 a 3 a 2 b kl =Ak2 a M1I #M2 ?. ... .A Mj Transfer Different forPollen fromFlowersof Different Positionst 3 4 b c d constant d variable a b c d 4 Total a 4a b 4b c 4c d 4d 1 2 3 4 Total 1 c d 0 0 c+ d 2 a c d 0 a+ c+ d b c kl = k2 kl ?k2 kl = k2 kl =Ak2 Pollen Best Transferred to Flowersof thePreviousPositiont 3 4 b kl = k2 d constant b kl =Ak2 a kl = k2 c a variable d kl =Ak2 c a+ b+ c+ d a+ b+ c+ d M1I #M2 . . . =? Mj M1I $ M2 $A...=. Mj M1I $M2 ?. ... .? Mj M1 I ?M2 ?A....A M M1I $M2 $A...=. M1I $ M2 ?A...=. M1 I ?M2 ?A....A M1 I #M2 ?A....A Mj Mi Mj Mj * Sums of each row and column equal a + b + c + d. t Sums of each row, but not of the columns, are equal. t In the protandrouscase, a > b > c > d. Protogynycan be modeled by the corresponding matrix inverted around the top rightto bottom left diagonal. which have the lowest T1 values. Differences in the gain curves affectthe allocations only when small differencesin the Kij exist or when the gain curves are very different. Case 3. Kj Values Differ between Flower Positions.-In biologically realistic cases, variationin Kij values is unlikely to occur withoutvariation in the total probabilities of pollen transferfromflowersof differentpositions, Kj, and such variation alone can stronglyselect forvariation in sex allocation among positions, even if the proportionsof resources allo- Matrix oftransfer probabilities 0.25 0.25 0.25 0.25 1.0 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.8 0.25 0.25 0.25 0.25 Matrix oftransfer probabilities 0.6 0.2 0.1 0.1 1.0 0.2 0.6 0.1 0.1 00.1 0.1 0.6 0.2 0.1 0.1 0.2 0.6 0.8| 0.8 *0j ,6 Lr Gain curveparameters kl, k2 -U 0.3, 0.8 0.8 -0-0.6, i 0.4 -0--0.6, 0.6 -k- 0.2 0.8, 0.6 --A--- 0.8, 0.3 _ * - T values 0.6 0.42K 0.2 - ^ ^ ~~~~~~~~~~~~~~~~~~~0.2 Gain curveparameters kl, k2 0.3, 0.8 1--00.6, 0.8 V --- 0.0 1 2 3 4 Position (j) FIG. 2. ESS allocationsto male functionin flowersof foursuccessivepositionswhenpollentransfer is random(firsttypeofmatrix of table2 withall KU1 values equal to 0.25), buttotalallocationof resourcesis highestforthe earliestflowersand declinesforlater flowers(T1 = 0.4, T2 = 0.3, T3 = 0.2, T4 = 0.1). Variousgaincurveparametervalues are shown. 6,0.6 -A-0.8, 0.6 0.0 1 2 3 0.8, 0.3 T values 4 Position (j) FIG. 3. ESS allocationsto male functionin flowersof foursuccessive positionswhenpollen transfer fromall flowerpositionsis equallysuccessful(equal Kj values,secondtypeof matrixof table 2). The same set of Tj values,and thesame gain-curveparameters, as in figure2 were used. 75 FLORAL SEX ALLOCATION oftransfer Matrix probabilities 0.4 0.5 0.1 0 0 0.4 0.5 0.1 1.0- 0 0 0.8 ,// 0.6 0.4 ~ ~~ ~ ~ ~ 0.1 0.4 0.5 0.1 0.3 0.6 curveparameters _-Gain kl, k2 U 0.3, 0.8 ~~~~---0.6, 0.8 and 3, theESS allocationsto malenesswere0.29 forposition 1 and 0.89 forposition4, whenbothgain-curveexponents were 0.6. Highervalues of the exponentsmade the latest flowerseven moremale,and less female,in theirallocation. SelfingPopulations SelfingRate Constant at All Positions.-When the selfing rateis constantamongpositions,it influencesthe ESS sex O~0.6, 0.6 allocationbut notthepatternof sex allocationamongposi0.2 0.8, 0.6 * tions (resultsnotshown).Like previoussex-allocationmod-A-S 0.8, 0.3 and betweenflowers(Charlesworth differences els without Tvalues 0.0 3 4 1 ~~~2 ~ 1981; Lloyd 1987), our model foundan inCharlesworth Position (j) crease in relativeallocationto femalefunction(decreasein in the in flowersof foursuc- sex allocation)withincreasedselfingand an increase FIG. 4. ESS allocationsto male function of with levels increased function male to allocation relative (witha slightlymodifiedformof cessivepositionsunderprotandry resources matrixof table 2, adjustedso that inbreedingdepression.Variationin reproductive the last typeof pollen-transfer to each ovule positionwere equal). The T. amongpositions,differences the sums of transfers in theexponentsto thefemale values were all 0.25, and the same gain-curveparametersas in and male gain curves,and variationin the probabilitiesof figure2 wereused. positionshad thesame forpollenat different pollentransfer influenceon thepatternof variationin sex allocationamong cated to reproduction(T1) are the same for all positions. This positionsas forthe case of obligatelyoutcrossedhermaphremains true even if the exponents of the male and female rodites. gain curves are equal, and when donor flowerposition does positions not affecttransferof pollen to recipientsat different (table 2, thirdkind of matrix). Flower positions with higherprobabilities of pollen-transfer Kj allocate proportionatelymore reproductiveresources to male function relative to positions with lower transfer probabilities (fig. 4). Variationin Tj, and differencesbetween kl and k2, do not modifythe qualitative patternof variation in sex allocation among positions. Figure 4 shows an example of the increase in ESS allocation to maleness with successive flowerpositions, for one hypotheticalexample of a. protandrous situation with Kj values 0.4, 1.1, 1.3, and 1.2. This example is not intended to be particularlyrealistic. It does demonstrate,however, that under protandry(which implies that the firstflowers have low pollen success), the ESS allocations may differbetween the earliest and later positions, withoutlarge differencesbetween the allocations of flowers of the differentlater positions. Moreover, the last flowerposition has a lower total pollen success thanthe precedingone, and thatthe allocation to maleness is correspondinglylower than of the previous position (resulting from the lower Kj value for the final flowerposition). In other words, the allocation patternneed not be monotonic. When the differentfactorspredict differenttrends,strong variationin the probabilityof pollen transferamong positions (the Kj values) determines the pattern of variation in sex allocation among positions. Only when the values of Kj are uniformor differlittle from one another will resource distribution among positions and the exponents of the gain curves control the trends in sex allocation among positions (fig. 2). In some cases, differencesbetween expected allocations of differentflower positions were extreme, amounting essentially to unisexualityof flowersat one or more positions. For example, using the transferprobabilities estimated fromthe A. caerulea data given in figure 1 and table 1 (firstfour positions only), and with the same T7values as in figures2 SelfingRate Variable among Positions.-With selfing,the fitnesscontribution throughfemale,but not throughmale, depression(eqs. function dependson thelevel of inbreeding 4, 5). Wheninbreedingdepressionis 0.5, thefitnesscontriof the level butionthroughfemalefunctionis independent of selfing,but whenit is less than0.5, flowerswithhigher selfingratespropagatemoregenes throughfemalefunction relativeto flowerswithlowerselfingrates;thereverseis true wheninbreeding depressionis greaterthan0.5 (eq. 4). forpollenat of pollentransfer Withconstantprobabilities different positions(equal Kj), theaveragenumberof ovules is the same forpollen at all posiavailable forfertilization one expectsa value of intions.Underthesecircumstances, breedingdepressionof 0.5 to lead to allocationof thesame resourcesto male functionin all of reproductive proportion of theirselfingrates.Withinbreeding flowers,irrespective depressiongreaterthan0.5, plantsshould allocate proporresourcesto male functionin tionatelymore reproductive flowerswith largerselfingrate; and, with inbreedingdepressionless than0.5, plantsshouldallocateproportionately moreresourcesto femalefunctionin flowerswithgreater selfingrates.These patternswereobservedwithTi constant or ki k2. resourcesallocatedto eachposition Whenthereproductive also vary and the exponentsto the femaleand male gain thesefactorsaffectthepattern curvesdiffer fromone another, ofvariationin sex allocationamongpositions(case 1,above), and thatwas foundin the case of selfingpopulationsalso. Similarly,whenthe probabilityof pollen transfer (Kj) and theselfingratevaryfordifferent flowers,thematrixof polthe trendsin sex alloprobabilitiesdetermines len-transfer cationexpectedamongpositions.As forobligateoutcrossers, amongpositions ofpollentransfer variationintheprobability of variationin sex is thestrongest of thepattern determinant amongthe values of allocation,and only whendifferences in resources do variation reproductive are small or absent Kj and different gain curvesinfluencethepattern. 76 J. BRUNET AND D. CHARLESWORTH DISCUSSION 0 Pollen-TransferProbabilities We have shown that sex allocationamong positionsis stronglyinfluencedby the matrixof pollen-transfer probabilitiessuchthatpositionswithmoresuccessfulpollentransferallocate proportionately morereproductive resourcesto male functionthando flowersof otherpositions.The probabilitiesof pollentransfer in ourmodeldependon theratios of available ovules (flowersin thefemalestageon potential mates) to the numbersof potentialcompetitors (flowersin themale stagecompetingfortheovules in question). Why mighttheseratiosdifferbetweenflowerpositions? Dichogamy(timedelaybetweenexpressionof male and female functions withinflowers)and pollinatormovements on inflorescences can createdifferent transfer probabilitiesfor pollenat different positionsandthusinfluence sex allocation. We discusstheinfluence ofthesetwofactorson pollentransferand sex allocation. Dichogamy 6- 0.4 0 C 0 0 b b 02 1W 0 - 2 0.1 0.0 Flower number FIG. 5. Estimatesof allocationsto male functionin flowersof successivepositionsin Aquilegiacaerulea, in inflorescences with different numbersof flowers(filledcirclesshowdataforall flowers combined;inflorescences withdifferent numbersof flowersare denotedas follows:open triangles, two-flowered inflorescences; open circles,threeflowers;filledsquares,fourflowers;filledtriangles, fiveflowers;open squares,six flowers).Allocationis expressedas theproportion of totaldryweightsof flowersand fruitsthatwere due to anthers.For thecombineddata,errorbarsequal to 2 SE are shown,and lettersindicatesignificant differences by Tukey'stest. Dichogamyshiftsthe flowering phenologyof flowersin thefemalestagerelativeto flowersin themale stage(see fig. Thomson 1989; Brunet 1990; Spalik and Woodell in prep.), 1 forprotandry). It therefore modifiesthenumberofpotential and the patternof variation is in some cases associated with mates (flowersin the female stage) at different positions dichogamy (Brunet 1990). available to pollen froma givenpositionwithoutchanging Few allocation data exist for species with hermaphroditic the numberof male flowerscompetingforaccess to these flowers,butAquilegia caerulea, a protandrousspecies, shows ovules.Withprotandry, theratioof potentialmatesto pollen the predicted pattern(Brunet 1990). In inflorescences with competitors (flowersin themale stagecompetingforaccess 2-6 flowers, fruitdry weights showed a highly significant to femaleflowers)is lowestin first-position flowersand in- decrease fromfirstto last flowers.Stamen weights tended to creases in later-position flowers.As a result,fewerstigmas decrease very slightlywith position, but the differencewas are available when first-position flowersshed pollen, but significant only in two-flowered inflorescences. Figure 5 plentyof pollenis aroundwhentheseflowershave receptive shows the allocation values calculated fromthese data. Simthustendsto reducethe probabilityof ilar patternswere found when sex allocation was measured stigmas.Protandry fromfirst-position pollen transfer flowersbut tendsto in- before pollination. Ovule number and fruitand seed set decrease it forlast-positionflowers(comparedto an adicho- creased fromearly to late flowers(Brunet 1990, in prep.). In gamousspecies).Therefore, thereproductive successoffirst- another population, first-positionflowersproduced seeds in positionflowersis expectedto be greaterthroughfemale 42.2% of carpels, compared with only 0.003% of flowersin (relativeto male) function,whereasthe reversepatternis the last position on the inflorescence(usually fourthto sixth expectedin later-opening flowers.Our analysisconfirms that position), a highly significantdifference(44 [SE ? 1.0] and this selectsforlower (relative)allocationto male function 41 [SE ? 0.001] flowers,respectively). The mean numbers inearly-opening flowersandtolowerfemalefunction inlater- of seeds in follicles with seeds were also much lower forlast openingflowers.Protogyny createstheoppositepattern. than for firstflowers (firstflowers: 26.3 ? 6.9; last flowers The femalebias expectedwithprotandry, for example, 0.098 ? 0.001). The differenceswere notcaused by depletion could manifest itselfby hermaphroditic flowersopeningear- of resources by seed maturationin early flowers. An experlier than male flowersin andromonoecious plants,female imentdone in 1987 provides the evidence forthis. Pollination flowersopeningearlierthanhermaphroditic flowersin gy- in all but the last flowersof several inflorescenceswas prenomonoeciousplants(Pellmyr1987), or early-opening her- vented by covering the stigmas with straws. Fruit and seed flowersallocatingmostof theirresourcesto fe- set were zero in the 34 last-positionflowers.When pollination maphroditic male functions,and later-opening flowersallocatingmore was preventedforall but first-positionflowers,fruitand seed resourcesto male functions(Brunet1990). The empirical set were increased, compared with control flowers at that evidenceaccumulatedso farsuggeststhatin andro-and gy- position (fruitset, 80.8 + 6.8; seed set, 119.4 ? 17.7; control nomonoeciousplants,earlyflowersare morefemale-biased values given above). Thus, although the results for first-pothanlate flowersin protandrous species,whereasthereverse sition flowers stronglysuggest that maturationof seeds in is truewithprotogyny (reviewedin Pellmyr1987), although otherflowersof the same inflorescencediminishes resources some exceptionsexist (see Andersonand Symon 1989). It for first-positionfruitsand seeds and lowers their numbers, appearsthatspecieswithhermaphroditic flowersdo notnec- it is neverthelessnotpossible forflowerslate in the sequences essarilyallocatetheirreproductive resourcesequallyamong to use these resources to mature seeds. In other words, late maleandfemalefunctions (Lord 1980;Bawa andWebb1983; flowersdo not participatemuch in female reproduction. FLORAL SEX ALLOCATION 77 To determinewhetherdifferences in the probabilitiesof Heinrich(1979) demonstrated thatbees can learnto modify pollentransfer are responsiblefortheobservedvariationin theirpattern ofmovement dependingon nectarrewards.This sex allocationamongpositions,one needs to quantifythe suggeststhatpollinatordirectionality maybe quitecommon influence of dichogamyon pollen-transfer probabilities. Data in hermaphroditic species. Althoughthe influenceof pollion flowering phenologysimilarto thoseof figure1 can pro- natordirectionality on probability of pollentransfer aiid sex vide information on pollen-transfer probabilitiesforflowers allocationamongpositionsremainsspeculativeat thispoint, at different positions.Species withthe strongestdegreeof it represents a plausiblefactorthatdeservessome empirical dichogamyshouldhave thegreatestdifferences in sex allo- attention. To testwhether inpollinator directionality behavior cationamongpositions.For anylevel of dichogamy, thein- does indeedinfluence pollen-transfer one could probabilities, fluenceon pollen-transfer probabilities shouldalso be greater use color dyes to label pollen fromflowersof different poin species wherethenumberof days betweenexpressionof sitions.At presentwe are awareof no suchdata,noraredata male and femalefunction represents a significant fractionof on allocationsavailable. thelengthof theflowering periodforanyposition.Although themodelrequiresdifferences in theflowering of phenology. Resource Variation flowersat different positions,it does notrequiresynchrony Besidesdifferences inpollen-transfer probabilities, flowers of flowering of different individuals.The magnitudeof the positionsalso mayvaryin thetotalamountsof differences in contributions to the pollen pools forstigmas at different Such variationcould reof different flowerpositionswill, however,dependon the resourcesallocatedto reproduction. appropriate degreeofsynchrony, whichwilltherefore influence thequan- sult if flowersat the base of the inflorescence moreofthereproductive resourcesthando moredistantflowtitativepatternsof allocationto be expected. If all flowerson a plantopen at the same time(simulta- ers(Lee 1988),orfromtemporalprocessesifflowersopening startproducingovules and deneousblooming)and ifindividualplantsdiffer in theirtimes earlieron the inflorescence of flowering, theprobabilities of pollentransfer betweenin- velopingseeds earlierand acquiremoreof thereproductive dividualsin a populationwill be affectedlike thosebetween resourcesvia source-sinkrelationships(Lee 1988). In the here,reproductive resourceswereassumed flowersat different positionsdiscussedabove. Withprotan- modelspresented dry,pollenof early-flowering individualshas fewstigmasto to varyamongpositions,but the relativeamountof reprobutwhentheseindividualsare in thefemalestage, ductiveresourcesallocatedto each positionwas fixed(was fertilize, pollen will be plentiful.The reverseis trueforindividuals notallowedto evolve). We thusexaminedonlytheeffectof bloominglate in the flowering season. If plantscan adjust resourcevariationon sex allocationamongpositions,notthe whichfactorsinflutheirsex allocationin responseto the matingenvironment,evolutionof thisvariation.Determining resourceallocationis an one mightexpectearly-flowering individualsto allocaterel- ence the patternof reproductive relatedquestionthathas notbeen previouslyadativelymorereproductive resourcesto femalefunctionand interesting later-opening individualsto allocatemoreto male function, dressed,butit can be examinedseparatelyand is beyondthe study, as suggestedby Pellmyr(1987). However,Pellmyr'sdiscus- scope ofthepresentwork.This factordeservesfurther To date, sion does notstrictly applyto different allocationsof flowers as it is likelyto be of real biological importance. modelsthatexaminedthefactorsinfluencing in thesame inflorescences. Protogyny wouldcreateopposite othertheoretical floweror inflorescence numberon a plant(Cohen and Dukas patterns. 1990; SchoenandDubuc 1990 ) haveassumedconstantcosts per flower. Pollinator Movement For variationin reproductive resources(withoutdifferIf pollinatorsshow directionality in theirmovementson ences in pollentransfer) to influencefloralsex allocationin thismayinfluence inflorescences, theprobability thatpollen sequentiallybloomingplants,the exponentsof the female fromgivenpositionsreachesstigmasat different positions. and male gaincurvesmustdiffer fromone another. The evoPollinatorsthattendto move frombottom-to top-position lutionof variationamongflowersin sex allocationis then flowerswithininflorescences will go fromtop- to bottom- similarin principleto the evolutionof variationamonginpositionflowerswhenmovingbetweeninflorescences. Ifpol- dividualswhenthedistribution of resourcesvary.Different linatorspick up some freshpollenfromeach flowervisited exponentsof themale and femalegain curvesimplythatthe while displacingsome of the pollen previouslycollected, shapeofthereturn function-that between is,therelationship such thatmost pollen carriedcomes fromthe last flower investment in a sexual functionand thereproductive return visited,directionality of movements will influencetheprob- fromthatsexual function-differs formale and femaleinabilityof pollentransfer amongflowers.These assumptions vestments. FollowingFrank(1987), positionswithmorereseembiologicallyjustified(Robertson1992). Thiseffectmay productive resourcesshouldbe biasedtowardthesexualfuncbe greatestin self-incompatible species, wherepollen de- tion withthe largestexponentto its gain curve (given expositedon a plant'sown stigmaswill notlead to fertilization ponentssmallerthanor equal to one). This is indeedwhat of ovules. we found.Accordingto theseprinciples, whentransfer probBumblebeestendto move up inflorescences (Pyke 1979; abilitiesforflowersat different positionsare equal and when and Heinrich1979; Corbetet al. 1981; Best and theexponentsto themale and femalegain curvesare equal, Waddington Bierzychudek1982; Haynesand Mesler 1984). Most of the no variationis expectedin sex allocationamongpositions, plantsin thesestudiesweresequentialbloomerswithflowers even whenresourcesvaryamongpositions. at thebottomofinflorescences openingfirst. and It is notknownwhether Waddington variationin reproductive resources 78 J. BRUNET AND D. CHARLESWORTH variationin sex allocationamongindividualsand oecy, as was alreadyclear empiricallybecause some moninfluences (LloydandWebb1986). among flowersin naturalpopulationsof hermaphroditicoeciousspeciesareselfincompatible this.As difplants.Empiricaldata are neededto determine poprobabilitiesfromdifferent ferencesin pollen-transfer Modular Approach to Sex Allocation sitionshave a stronginfluenceon sex allocationamongpoStantonand Galloway (1990) have previouslyexamined to eliminatethisfactorfromconsidsitions,it is important how floralsex allocationcan influencean individual'srewhether variation eration.A firststepwouldbe to determine productive success.In theirmodel,floralsex allocationsare in sex allocationamongflowersoccursin specieswherethis factordoes not apply,thatis, in adichogamoushermaphro- randomlyassignedto individuals,and each floweron a plant theirworkdoes not in whichreproductive is identical.Owing to thisassumption, diteswithoutpollinatordirectionality sex allocation, althoughthey floral examine the evolution of resourcesdifferamongpositions.If variationin sex allocaof structure of considering the modular importance stress the tionamongflowersis commonin suchpopulations,one can when dealing with sex allocation.The thenask whetherthemechanismproposedherecan explain plant reproduction introduces themodularnature hereexplicitly it. This wouldrequiremeasuringtheshapesof themale and modelpresented into sex-allocation theory. of inflorescences femalegain curvesin naturalpopulations-a verydifficult Althoughdiscussedmostlyin termsof flowerswithinintask. themodel analyzedherecan of coursebe apflorescences, otherthanflowers,such as umbels,infloresplied to units Selfing-RateDifferences cences, or tillers.As long as theyopen sequentiallyon a Withsynchronous floweringand dichogamy(Lloyd and plant,and flowering withintheseunits, is fairlysynchronous Webb 1986) or temporaldioecism (Crudenand Hermann- one can assignpositionsto units,accordingto the orderin and whichtheirflowersopen, and thenexaminesex-allocation Parker1977), thereis no overlapof pollenpresentation amongflowerswithina plant,and hence patterns stigmareceptivity on plantswithmanyflowers, amongthem.Similarly, dichogamy sex allocationcan be examinedamonggroupsof flowerson no geitonogamousselfing.Strongwithin-flower selfing.In sequentialbloomers,certain theinflorescence, reduceswithin-flower synchrowhereflowerswithinsufficiently combinationsof floweringphenologyand dichogamycan nous groupscan be identifiedas different "positions" in cause variationin selfingrateamongflowers.Withprotandry, termsof themodelpresentedhere. a few if first-position flowerson a plantopen synchronously flowerswill flowers,first-position days beforelater-position Conclusion flowers generallyreachthefemalestagewhenlater-position The chiefconclusionof thesecalculationsis thattheevoare in the male stage. Selfingratesmay thenbe greaterin lutionof allocationsto male and femalefunctionscan be earlyflowersthanin laterones. probabilitiesbetweendifferent Whethervariationin selfingrate among hermaphroditicaffectedby pollen-transfer in differences thetotalresourcesplantsdevote flowers and by flowersexists in naturalpopulationsof dichogamoussepositionswhenthe gain curvesrelatingreproHowever,variation to different quentialbloomersawaits investigation. forthetwosex functions. in selfingratesexpectedin the situationsjust describedis ductivesuccessto allocationdiffer betweenflowersare biologically unlikelyto have a majorinfluenceon sex allocationamong Both kindsof differences differences plausible,and thereis some empiricalevidencefortheirocflowers.Our resultsindicatethat selfing-rate amongflowerscan influencesex allocationto a majorextent currence.We have also reviewedthe small amountof eviin malenessactuallyoccur thatdifferences Di- dencesuggesting do notdiffer. probabilities onlywhenthepollen-transfer that seem to satisfytheconditions between flowers of plants chogamyin sequentialbloomerswould most likelycreate is sequential).This that of our model blooming (principally, probabilitiesamong positions, variationin pollen-transfer in a situationwheredifferent flowerson inwhichwould generallybe the dominantfactorcontrolling correspondence dividualplantsare considered,and thereare no confounding allocationpatternsof flowers. between or geneticdifferences factorssuchas environmental approach. validates the allocation individuals, The Evolution of Andromonoecyand Monoecy in the The role of dichogamyand pollinatordirectionality evolutionof plantbreedingsystemshas receivedsome attention(Webb1981; LloydandWebb1986),buttheseeffects intoallocationmodels. havenotpreviously beenincorporated Ourresultssuggestthattemporalvariationin pollendonation andreceiptmayplaya rolein theevolutionofthesebreeding in the transfer probabilitiesfor systems.Large differences flowerpositionscan lead to the evopollen fromdifferent lutionofandromonoecy andmonoecyin sequentialbloomers. These modelssupporttheview thattheevolutionof andromonoecyis based on allocationof resources,in particular wheremale flowersenhancemale fitnessthrough pollendonation(Bertin1982). They also show thatavoidanceof inbreedingneed notbe a conditionforthe evolutionof mon- ACKNOWLEDGMENTS and mathJ.B.thanksJ.Coopersteinforhis programming ematicaladvice. The developmentof theseideas benefitted fromdiscussionswithJ.Thomson.SuggestionsbyT. Meagher and two anonymousreviewershelpedimprovethemanuscript.The RockyMountainBiological Laboratory(RMBL) whileJ.B.collecteddatain provideda congenialatmosphere thefield.The fieldworkonAquilegiacaeruleawas supported bygrantsto J.B.fromSigmaXi, andbytheLee R. G. Snyder Botany MemorialFundfromtheRMBL, and theHutchinson Fundof theUniversity of Chicago.Duringtheearlystageof of thismanuscript, J.B.was supportedby a docpreparation toralfellowshipfromtheQuebec Government (FCAR). FLORAL SEX ALLOCATION 79 mentbetweenflowerswithininflorescences. AnimalBehaviour 27:1167-1181. Anderson,G. J.,and D. E. Symon. 1989. Functionaldioecy and Robertson,A. W. 1992. The relationshipbetweenfloraldisplay in Solanum.Evolution43;204-219. andromonoecy size, pollen carryoverand geitonogamyin Myosotiscolensoi Bawa, K. S., and C. J. Webb. 1983. Floral variationand sexual (Kirk)Macbride(Boraginaceae).Biological Journalof theLinin Muntingiacalabura.Evolution37:1272-1282. differentiation nean Society46:333-349. Bertin,R. I. 1982. The evolutionand maintenanceof andromon- Schoen,D. J.,and M. Dubuc. 1990. The evolutionof infloresence oecy. Evolutionary Theory6:25-32. size and number:a gamete-packaging in plants.Amerstrategy Best, L. S., and P. Bierzychudek.1982. Pollinatorforagingon ican Naturalist 135:841-857. foxgloveDigitalispurpurea:a testof a new model. Evolution Stanton,M. L., and L. F Galloway. 1990. Naturalselectionand in flowering 36:70-79. allocationto reproduction plants.Pp. 1-50 in M. Mangel, ed. Some mathematical Brunet,J. 1990. Genderspecializationof flowerswithininfloresquestionsin biology.Sex allocation and sex change: experimentsand models. American cences of hermaphroditic plants.Ph.D. diss. StateUniversity of MathematicalAssociation,Providence,R.I. New York,StonyBrook. Bull, J.J. 1981. Sex ratioevolutionwhenfitnessvaries.Heredity Thomson,J. D. 1989. Deploymentof ovules and pollen among flowerswithininflorescences. Evolutionary Trendsin Plants3: 46:9-26. 65-68. , ed. 1983. Evolution of sex determiningmechaWaddington, D., and B. Heinrich.1979. The foragingmovements nisms.Benjamin/Cummings, Menlo Park,Calif. of bumblebeeson vertical"inflorescences":an experimental B. 1990. Optimization Charlesworth, models,quantitative genetics, analysis.Journalof ComparativePhysiology134:113-117. and mutation.Evolution44:520-538. and sexual selecprotandry Charlesworth, D., and B. Charlesworth.1981. Allocationof re- Webb,C. J. 1981. Andromonoecism, tionin Umbelliferae.New Zealand Journalof Botany19:335sourcesto male and femalefunctionsin hermaphrodites. Bio338. logical Journalof theLinneanSociety 15:57-74. . 1987. The effectof investment in attractive structures on Corresponding Editor: T. Meagher allocationto male and femalefunctions in plants.Evolution41: 948-968. Charnov,E. L., ed. 1982. The theoryof sex allocation.Princeton APPENDIX N.J. University Press,Princeton, Cohen, D., and R. Dukas. 1990. The optimalnumberof female Methodof Analysis flowersand thefruits-to-flowers ratioin plantsunderpollination The methodis based on a quantitative-genetic approachthatassumes and resourceslimitation. AmericanNaturalist135:218-241. Corbet,S. A., I. Cuthill,M. Fallows, T. Harrison,and G. Hartley. thatthe evolutionof a characteris determinedby the vectorof partial 1981. Why do nectar-foraging bees and wasps workupwards derivativesof fitnesswithrespectto theparametersof interest,together with a matrix of additive variances and covariances (Charlesworth on inflorescences? Oecologia 51:79-83. Cruden,R. W., and S. M. Hermann-Parker. 1977. Temporaldioe- 1990). We describe the analysis for the most general case studied,in which the reproductiveresources may be allocated to the threefunccism: an alternative to dioecism?Evolution31:863-866. Darwin,C. R. 1877. The different formsof flowerson plantsof tions,male and female primaryreproductivestructures,and structures for pollinatorattraction.The ESS values of the male, female,and atthesame species. Murray,London. parametersforflowersat thejthpositionare denoted Frank,S. A. 1987. Individualand populationsex allocationpat- traction-allocation by Mj, Fj, and Aj, whereFj = 1 - Mj - Aj. For thisexample, the vector terns.TheoreticalPopulationBiology 31:47-74. Haynes,J. R., and M. Mesler. 1984. Pollen foragingby bumble- of derivativesof fitnesswithrespectto theallocation parametersunder bees: foragingpatternsand efficiency of Lupinuspolyphyllus. selection can be writtenas Oecologia 61:249-253. = A AF) (Ala) Vo Lee, T. D. 1988. Patternsof fruitand seed production.Pp. 179202 in J. L. Doust and L. L. Doust, eds. Plant reproductive ( ak'A? |aF] a] ecology.OxfordUniversity Ab Press,New York. Lloyd, D. G. 1984. Genderallocationsin outcrossingcosexual plants.Pp. 277-300 in R. Dirzo and J. Sarukhan,eds. Perspectiveson plantpopulationecology.Sinauer,Sunderland,Mass. wherew is the fitnessexpression.To obtainthematrixof variancesand 1987. Allocationsto pollen,seeds and pollinationmech- covariances, we use the relation:6F = -(8M = 6A), which gives: anismsin self-fertilizing plants.FunctionalEcology 1:83-89. VA(F) VA(M) + VA(A), (A2a) Lloyd, D. G., and K. S. Bawa. 1984. Modificationof the gender of seed plantsin varyingconditions.EvolutionaryBiology 17: CA(F, M) -VA(M), (A2b) LITERATURE 255-338. CITED Lloyd,D. G., and C. J.Webb. 1986. The avoidanceof interference CA(F, A) -VA(A). (A2c) betweenthepresentation of pollenand stigmasin angiosperms. The matrix of additive variances and is genetic covariances therefore: I. Dichogamy.New Zealand Journalof Botany24:135-162. Lord, E. M. 1980. Intra-inflorescence variabilityin pollen/ovule 0 [VA(M) -VA(M) ratiosin thecleistogamousspeciesLamiumamplexicaule.AmerG= O VA(A) -VA(A) . (A3) ican Journalof Botany67:529-533. L-VA(M) -VA(A) J -VA(F) Pellmyr,0. 1987. Multiplesex expressionsin Cimifugasimplex: dichogamydestabilizeshermaphroditism. Biological Journalof These expressionswere used to computethevectorof allocation values theLinneanSociety31:161-174. in thenextgeneration,and thecalculationswere iterateduntilthevector Pyke,G. H. 1979. Optimalforagingin bumblebees:ruleof move- of allocations converged.
© Copyright 2026 Paperzz