AnEmpiricalApproachto AnalysingtheDemographic ConsequencesofSeed Dispersal byFrugivores
H.GODINEZ-ALVAREZANDP. JORDANO
Introduction
Seed dispersalisakeystagein the lifecycleofplantsbecauseitcontributes to the recruitmentofnew individuals(HoweandSmallwood, 1982).Many authors have suggestedthat seed dispersalhasimportantdemographic consequences,and assuch itisnecessarytolink comprehensive analysesof seeddispersal withstudiesofplantdemographyinordertocompletely understandthe populationdynamicsofplants(SchuppandFuentes,1995; GodInez-Alvarezetal.,2002;Jordano and Godoy, 2002; Wangand Smith,
2002;Howeand Miriti,2004).In this chapter,wepresentan approachfor
integrating these two disparate types of study as a way to better understandthe effects ofseed dispersalon the populationgrowthrate of plants.
Dispersalofseedsbyfruit-eatinganimalsisacomplex processwhose consequencesencompassaseriesofconcatenatedstages beyond fruit consumptionandseed removal,such asseeddeposition,germination,and seedlingemergenceand establishment(Jordanoand Herrera,1995;Wang andSmith,2002).Perhapsdueto this complexity,no studyto datehas analysedallthesesequentialstagesanddetermined theirconsequencesby means of an explicitly demographic approach. Since successful seed dispersalbyanimals istheoutcomeofamutuaListicinteractionbetweenthe animalsand the plants,itisexpectedthat ithas net positiveeffects on the populationsof each participatingspecies.The effects of seed dispersal shouldideally beobservedandmeasuredinthe populationgrowthratesof bothanimalsandplants(Addicott,1986),which requiresan understanding offrugivoryandthe behaviouroffrugivoresbeyondfruitremoval.
Theeffects of seed dispersal have traditionally been incorporated into plantdemographicstudiesby meansof simulationswith population dynamics models that are often built withoutconsidering the detailed
©CABInternaUonal2007.SeedDispersal:TheoiyanditsApplicationina391
ChangingWorld(edsA.J.DennisetaL)
392H.GodInez-AivarezandPJordano
natural historyof the particular plant—frugivoresystembeingexamined (Horvitzand Schemske,1995;MartInez-RamosandAlvarez-Buylla,1995; Valverde andSilvertown,1997).On the otherhand,incorporatingthe consequencesofmutualisticinteractionsinto matrixpopulationmodelshas sofar had oniy limitedsuccess (HerreraandJordano,1981; Horvitzand Schemske,1994; GodInez-Alvarezetal., 2002).A particularlimitationto the viable integrationofseeddispersalwith itsdemographicconsequences has beenan inabilityto link the effects of frugivores’actionswith their delayedconsequencesatlaterstagesofrecruitment.
Toaccomplishour goal inthischapter, weFirstreview some existing approachestothe studyofseed dispersal.Wethen presentabrief discussion of the population projectionmatrixmodelswhich are commonlyused in plantdemographicstudies.Finally, we use empiricaldata to illustrateour ideas and then discuss the possible benefitsand disadvantages of our approach.Wearguethat matrixpopulationmodelscan incorporate the necessaryelements of analysis to clearly integratethe effects of frugivore activityon plantpopulationdynamics.Ultimately,wehopetostimulatemore integrated ecologicalstudies,thus contributingtoour understandingofthe keyroleofseed dispersalinplantdemography.
Approachestothe StudyofSeedDispersal
Fromatheoreticalperspective, inordertounderstandthe demographic consequencesofseeddispersal,itisessentialto obtaindata on allcritical stagesandecologicalprocessesfromthe reproductionofthe parentplants throughto reproductivematurityofthe new adults(SchuppandFuentes,
1995).Relevantdata include fruit production, the composition of the frugivoreassemblageandthe quantityof seedsremovedbyeachspecies, frugivoreeffects on seedgermination,the micrositesinto which frugivores deposit seeds, and the effects of those microsites on seed survival, germinationand adultrecruitment.
Becausethe evaluationofsuchbroad-rangingdata types ismethodo logically socomplex,studiesofseed dispersalto datehave only analysed certainstagesof the overallprocess.Somestudieshave concentrated on the pre-disseminationstage,mainlythroughananalysisofthe quantitiesof seedsdispersedbydifferentfrugivorespecies(Schupp,1993;Jordanoand
Schupp, 2000),whereasothers have focusedon the post-dissemination
stage,generallyby estimatingseedlingrecruitmentprobabilities(Jordano and Herrera,1995;Reyand Alcántara,2000;Garcia,2001; Travesetet al.,
2003).
Theoverallconsequence ofseed dispersalcan perhapsbe viewed best within the frameworkof‘disperser effectiveness’.Dispersereffectivenessis defined as the relative contribution of a frugivore species to plant reproduction,which dependson the quantityof seeds removed andthe qualityof treatmentgiven to the seedsduring the processof dispersal (Schupp,1993).Atpresent,no single studyhas thoroughlyanalysedboth
AnalysingtheDemographic ConsequencesofSeedDispersal393
the quantitativeand qualitativecomponentsfor all the frugivoresof a particular plantspecies,but very goodbeginningshavebeenmadefor somesystems(Howe,1977; Herreraand Jordano,1981; Murray,1988; Reid,1989;Jordano andSchupp,2000).Thefrequencyof visits and the
number of seeds removed per visit are among the best-documented
quantitativeaspects,while fruithandling,post-feedingmovementpatterns offrugivores,andthe effects ofhandlingon seed germinationare some of the best-studiedqualitativeaspects(seeTable 17.1 forexamples).
Studiesaddressingrecruitmentprobabilitiesare generallybasedon the
analysisofseed andseedlingsurvivalindifferentmicrohabitatswhereseed rain may or may not have beenadequately quantified.Important microhabitat-specificdataincludethe numberof seedsdispersed to that site by the frugivore community as a whole, the proportion of seeds sufferingpredation, the proportion of seedsgerminating, and at least someminimaldataon survivalof the seedlings.Actual dataon seedling survival has been extremely variablein duration, rangingfromseveral monthsup to 4years post-germination(Table 17.1).Based on such data, recruitment probabilitiesfor particularplantspecieshave beencalculated as the productof the transition probabilitiesamong the early life cycle stages (e.g. ovule to seedling, Traveset et at., 2003; seed to seedling, Herreraet at., 1994; Jordanoand Herrera, 1995; Rey and Alcántara,
2000).In manycases at least,the processesactingduring thesestagesof
the life cycle are independentandthe eventualoutcomesare site-specific (Jordanoand Herrera,1995; Schupp,1995) andoftencontext-dependent inotherways(seeSchupp,Chapter20,thisvolume).
In order to determine the role of frugivores in plant population dynamics,webelievethat itisnecessarytoimproveourknowledgeofthe pre-andpost-disseminationstagesof seed dispersalandthento identify the methodological limitationsfor integratingfrugivoreactivity with the success of seedsand seedlingsfollowinghandling and dissemination.A review ofthe relevantliteratureshowedthat for somespecies the post- disseminationstagehas beenintensivelystudiedwhile data regardingthe pre-disseminationstage arescarceor non-existent,andvice versa (Table
17.1). Thus,an imbalanceisevidentbetweenthe amountof datafor the
pre-andpost-disseminationstagesofagiven species.
Inaddition,evenifwehave adequatedata weare limitedinourability to effectivelylink thesestages.Effectivelylinkingsuch dataisessentialin ordertointegrateseed dispersalwith informationon plantdemography. Wesuggestthis integrationmay beachievedthroughthe useofprojection matrices,which incorporatedataon survival, growthand fecundityof individualsatdifferentstagesofthe lifecycle, andsummarizeitin relevant populationparameterssuch as the populationgrowthrate,orlambda(K; Caswell,2001).In particular,wesuggestthatone waytobetterunderstand the impactoffrugivoreson the demographyofplantpopulationsistouse projectionmatricestolink the pre-and post-disseminationstagesfor each individualdisperserspecies.
394
H.GodInez-AlvarezandP.Jordano
Table17.1. Asampleofstudiesthathaveassessedaspectsofthequantitative and qualitativecomponentsofdispersereffectiveness, sensuSchupp(1993).Numbersindicate distinctaspectsofthequantitativeorqualitative componentsofdispersereffectiveness:
1=Relativeabundance,expressedmainlyasnumber ofindividualsperhour ofobservation;
2=Frequencyofvisitstotheplant,estimatedasthenumberofvisitsperhour;3=Time spentfeedingontheplant;4=Numberorproportionofseedsand/orfruitsremovedper dropping,cacheorvisit;5=Post-foragingbehaviour,expressedasdirectionofmovements, timebetween movements,movementdistance,and/ornumber ofmovementstodistinct landscapeperches, unitsormicrohabitats;6=Numberorproportionofseedsdefecatedper minute;7=Numberorproportionofseedsgerminatingaftergutpassageand/orgermination rate;8=Numberorproportionofseedsand/orseedlingsindifferentmicrohabitats;
9=Numberorproportionofseedlingssurvivingfor1—4years; —= Notevaluated.
Plants/Animals
Caseariacoi’ymbosa
(Flacourtiaceae)/Birds(6spp.)
Prunusmahaleb(Rosaceae)/ Birds(>lospp.)
Plants(3spp.)/Birds(6spp.)
Amyemaquandang
(Loranthaceae)/Birds(2spp.)
Quantity Quality Reference
1,2,47,8Howe,1977
1,2,3,45,8HerreraandJordano,1981; JordanoandSchupp,2000
16,7Murray,1988
47,8Reid,1989
Pinusjeffreyi(Pinaceae)/Rodents(2spp.) Juniperusashei(Cupressaceae)/Birds(2spp.) Maesalanceolata(Maecaceae)/Birds(2spp.) Phillyrealatifolia(Oleaceae)/Birds
Plants(6spp.)/Frugivores
Phoradendroncalifornicum
(Viscaceae)/Birds(3spp.) Plants(2spp.)/Rodents(2spp.)
4
3,4
2,4
4
4
1,2
8
5,8
6,7
7,8,9
7,9
8,9
VanderWall,1993
Chávez-RaniIrezandSlack,1994
Grahameta!.,1995
JordanoandHerrera,1995
ChapmanandChapman,1996
Larson,1996
Forget,1997
Plants(6spp.)/Birds(3spp.)
Ocoteaendresiana(Lauraceae)/Birds(5spp.) Plants(4spp.)/Birds(6spp.)
Plants(9spp.)/Birds(2spp.)
Oleaeuropaea(Oleaceae)/Birds
Plants(18spp.)ILagothrixlagothricha(Atelidae)
Bei!schmiediapendula(Lauraceae)/ Birds(4spp.)
3,45,6
5,8
48
5,6
47,8,9
45,7
7,8,9
Suneta!.,1997
WennyandLevey,1998
LoiselleandBlake,1999
HolbrookandSmith,2000
ReyandAlcántara,2000
Stevenson,2000
Wenny,2000a
OcoteaendresianalBirds(5spp.)
Juniperuscommuriis
(Cupressaceae)/Birds(2spp.)
Coremaalbum(Lmpetraceae)/
7,8,9Wenny,2000b
47,8,9GarcIa,2001
Birds(2spp.),Rodents(1spp.)
Ruppiamaritima(Ruppiaceae)/Birds(9spp.)
Neobuxbaumiatetetzo(Cactaceae)
/Bats(1spp.),Birds(4spp.)
7,8,9
47
1,25,7
Calvino-Cancela,2002,2004
Figuerolaetat,2002
Godinez-Alvarezeta!.,2002
Plants(12spp.)/Monkeys(3spp.)
Leptonychiausambarensis
(Sterculiaceae)IBirds (11spp.)
Rhamnusludovici-salvatoris
(Rhamnaceae)/Birds(2spp.)
Pinusmonophylla(Pinaceae)/Rodents(6spp.)
2,48KaplinandLambert,2002
49CordeiroandHowe, 2003
48,9Traveseteta!.,2003
48HollanderandVanderWall,2004
Analysing theDemographicConsequencesofSeedDispersal395
PopulationProjectionMatricesandSeedDispersal
Projection matrix models are commonly used in the study of plant demographybecause they providebasic populationparameterssuchas populationgrowthrates,stable size distributions,and reproductive values (Caswell,2001).Projection matricesused in the studyof plantsgenerally use size or life cydestagefor categorizingthe populationstructure (Fig.
17.la)becausetheseare generallybetterpredictorsthanage ofplantfate.
Thestructurecan be depictedwith lifecycle diagramsin which different stagesof the life cycle are representedby nL,wheren is the number of individualsinstagei,alongwith the probabilitythat an individualinstage
iat time tcontributesthroughsurvival,growth,or reproductiontostagej
at timet+l. Based on these diagrams, projection matrices are easily derivedby consideringthe links betweenthe different stagesas well as theirassociatedtransitionprobabilities(Fig. 17.ib).
The entries in a projection matrix can vary depending on the
complexityof the life cycle (Silvertownetal., 1993;Caswell,2001).For instance,projectionmatricesfor specieswith relativelysimplelifecycles in
which individualsat onestagecan only eithergrow to the nextstageor
p3p4
(b)
Fig.17.1. Anexampleofalifecyclediagram(a)anditscorresponding projectionmatrix(b) foraplantwitharelativelysimplelifecyclecategorized infourstages(n1).Individuals instage icansurviveandgrow(G1)intothenextstage,orsurviveandremaininthesamestage(F1), duringagiventimeinterval.Individualsinthelastthreestagescontributetothefirststage throughsexualreproduction (F1).
surviveat the samestageduring a specifiedtimeinterval,andin which adultscan only reproduce sexually,willonly have possiblevaluesalongthe main diagonal, the first lower subdiagonal,and the first row (see Fig.
17.lb).Values alongthe main diagonalrefertothe survivalprobabilitiesof individuals that do not grow to the next stage, those in the first subdiagonalrepresentthe transitionorgrowthprobabilities,andthosein the first rowindicatefecundities.
The fecundityvalueof a given adultstagecan be calculatedas the product of the probability of reproduction for that stage, the mean numberofseedsproducedbyareproductiveindividualofthat stage,and the probabilityofaseed becomingaseedlingduringthe time-step, which isconsideredconstantforalladultstages(GodInez-Alvarezetal., 1999; Box
17.1). Frugivoreeffects may be incorporatedbyconsideringdataon quantitative(i.e. frequencyof visits to the plantsandseed removal)and qualitative(i.e.seed germinationaftergut passageand numberofflightsto differentlandscapepatches)aspectsofeach speciesoffrugivore.Thefirst stepistocalculate:
1. Theprobabilityofseedremoval;
2. Theprobabilityofseed germinationaftergut passage;
3. Theprobabilitythat seeds would bedeliveredtoeach distinctpatchtype;
4. Theprobabilityof makingthe transitionfromseed toseedlingin each distinctpatchtype.
Usingtheseprobabilities,new fecundityvaluescan be estimated,which in turn may be incorporated in the projection matrix to calculatethe populationgrowthrate duetofrugivoreeffects (see Box 17.1 for adetailed descriptionabouthow to calculateandintegratethesefourprobabilities intofecundityvalues).
With this approachwecan begin toevaluatethe effectsoffrugivoreson plant demography. Depending on the system, we can do this for each individualspecies offrugivoreseparatelyor for disperserfunctionalgroups composedofspecies sharingsome taxonomicalorecologicalaffinities(see DennisandWestcott,Chapter 9,this volume).Eitherway,wecan use fairly easily collecteddata toestimate, respectively,the contributionof particular speciesor offunctionalgroupsto populationgrowthrate (seeBox 17.2). In addition,we can use this approachto evaluate,amongotherthings,the consequencesfor plantsofan avian seed predatororofan invasive disperser species,as well as the populationcollapseoreven extinction of a given species offrugivore.Byestimatingthe populationgrowthrate undervarious scenarios (e.g.withandwithoutaparticularfrugivore)wecan determinethe relativeimportanceofgiven species tothe maintenanceofplantpopulations.
Dispersa’ofNeobuxbaumiatetetzoSeeds:anExample
In this section we present a brief overview of seed dispersal of the columnarcactusNeobuxbaumzateteizo(Cactaceae)asanexampleofthe ideas
AnalysingtheDemographicConsequencesofSeedDispersal397
Box1?1Incorporatingtrugivoreeffectsintothefecundlt9rvaluesofmatrixmodels
Frugivereeffectscanbeincorporatedintothefecundityvaluesofmatrixmodelsthrough
theestimationoffourprobabilities..
ITheprobabilityofseedremoval,
2 Theprobability01seedgerminationaftergutpassage,
3 Thepmbabflityofdeliveringseedstoparticularpatches,
4 TheprobabilityofmakingthetransitionfromseedtoseedlingIneachpatch
ToIllustratetheestimationsofthese probabilities consideranexamplewiththree speciesoftrugivores(ABandC)forwhichwehavedataonthefrequency ofvtstttothe plantsandthenumberofseedsremovedpervisiL -
Toestimatetheprobabilityofseedremoval,thesedataarefittmultipliedtogether andthnthisproductissummedacrossallspeciesEromthistotaltherelativeproportign
oftotLseedsremovedbyeach speciesoffrugivore iScaIculatadthisrepresentsthe probabilityfseedremovalperspecies
:
tc0.22’
B7535005
C5100500073
Total6851OQ
Otherspectaofseeddispersalsuchasfruithandlingsuccess(egproportionffruits ingestedaridsuccessfullyremovedawayfromtheparenttreereItivetothenumberof fruitshandled)-caneasilybeincrpratedintothisapproach
Theproliabllity’ofseedgermlnatiønaftergutpassagecanbeestimatedwith lebrnaforyexpertrnentsusingseedsdefecatedbyeachspeciesot’frugivoraTheneari
proportionof seedsgerminatingfoteachspeciesoffrugivoremaybeinterpretedasthe probabilityofseedgerminationaftergutpassage
Thestimate theprobabilityofdeliveringseedstoparticularpatches,onecat
determinethenumberofpost-feedingflightstodifferent landscapepatchesforeach
speciesofIrugivore aridthencalculatetherelativeproportionoftotalflightstoa
microhabitattypemataremadebyeachspecies Ahypotheticalexamplewiththree
speciesoffrugivoresandtwopatchesis
•.:;:‘4’
AB
ofRetstiveNoof Ret1iN.ofe1atl flights proportion fiihts proportion tliprs proportion<
Patchl10220905G
Patch236075301013050
Total4810030100<100.
...-‘Cor#inued
I Box17.1Continued
Theseprobabiiftiescanbecombinedwiththefrequencyofvisitstotheplantsandthe numberofseedsremovedpervisittoestimatethenumberofseedsinapatchtype contributedbyagivenspecies.
The•probability ofmaking thetransition fromseedtoseedling ineachpatch canbeestimatedthroughfieldexperimentsinthosepatchtyposconsideredrelevant.In eachofthesepatchesseedsdirectlyobtainedfromfruitscanbesowninthefieldto determinethenumberofemergingseedlingsaswellasthenumber.of surviving seedlingsafter1year.Themeanproportionofsurvivingseedlingswithrespecttothe• totalnumberofseedssownmaybeconsideredastheprobabilitythatseedsgerminate andsurviveineachpatch.
Toincorporatetheseprobabilities,intothefecundityvalues,firstconsiderthatthe fecunditybasedontherecruitmentcit1-year oldseedlingscanbecalculatedas:
F.=R1S,C...(1)
whereFisthefecundityforindividualsinstage Ristheprobabilityofreproduction,Sis themeannumberofseedsproducedperindividual,andCistheprobabilityofpassing fromseed‘toseedlinginthefieldafter1year,whichisconstant’foralladultstages. Basedonthisdefinition,Ccanbereplacedin(1)bythefourestimatedprobabilities
.
above.That is:
=R;SaPG>).(D.,.iT,.7)(2)
where.F;.:’ R,,andS,arethesameasin’(i),Pis.theprobabilityof seedremoval by the trugivores.Gis.theprobabilityofseedgerminationaftergutpassage.Ps,,, isthe probabilitythatseedswouldbedeliveredinthepatchmbys,andTmistheprobabilityof makingthetransitionfromseedtoseedlinginm.Thesimplestcaseiswhen‘plantsare visitedonlybyonefrugivorespecies(s=1)andtheirseeds.andseedlingssuccessfully germinateandsurviveonlyinonepatchtype(m=1).Inthiscase,P.G,and are theprobabilitiesestimatedforthe‘onlyspeciesoffrugivoreandT,1,isconstant.When thereismorethan onetrugivore‘species(S =x).andpatchtype.m=y.Pcanbe calculatedasthesumoftheremovalprobabilitiesestimatedforeachspecies,andGas themeanweightedbytheseedremovalprobability.
‘
presentedabove.Detailedinformationon this plantanditsseed dispersers can befoundinGodInez-Alvarezetal.(2002).
Neobuxbaumia tetetzoisacommonplantin the TehuacánValley,a dry zone in south-central Mexico. Fleshy fruits produced on branch tips dehisceat night,exposingawhitish, sugarypulpwith hundredsof small black seeds. Fruits are consumed at night by the long-nosed bat Leptonycteris curasoae (Phyllostornatidae)and during the day by several species ofbirdssuchascactuswren(Campylorhynchusbrunneicapillus; Troglodytidae),curved-billedthrasher(Toxostomacurvirostre;Mimidae)and grey-breastedwoodpecker(Melanerpes hypopolius;Picidae).Sincethe traits of N.tetetzofruitsagreewith thosesuggestedfor fruitsconsumedby bats (van der Pijl, 1982),it is reasonableto assumethat bats have a greater contributiontocactuspopulationgrowthrate thanbirdsdo.
AnalysingtheDemographicConsequencesofSeedDispersal399
Box17,2 Disperser functiOnalgroupscombiningtheprobablhtiesoffrugivore
Specieswithecoloica1ortaxonomicataffinities
The probabilitiescafcuiatedforeachspeciesoffruivorecartbeuSedindividuallyorin combinationWithothetspeciesthatsbare>SomeecpIogioatdrtexonomicalaffinitiesinorder toevaluatethespeciesspoiJooç4unchnl groupeffectsoffrugivors onpopulation gowfhrate
5ConstderforsimpkcityfbstprobabilitiesofseedrinOv5l seederminatioxafter gutpassag,e{G6) andseedpositron(D)intwoparticularlandscapepatcheshave bee4calculatedfortwospeciesftnrds(Aand aridonespeciesofbat(C The prthaihtIeofmakingthetransitionfromseedtcseedltng (7)inéathpafohwere cIciulatedilIdeperittentbrthØfugivorespecianc1temjnconatanASartexample
ffugwores
4si m2 flI1 m—2
A ø22 090 O25 075 1‘ 001
B 605 060 O90- 010 01 0&t
c :o.m 0.80 oo
6ased-nthesedata5andEqn(2,itispossIL1odstimate, forexample5theeffectsof J.irds A÷B).bats(,andallfngrc’orespeciesA+±CInthecasesotbirdsaidall fruvorespeciestheprobabilityofseedremovat canbecalculatedasthesumofthe
11bbd1tjesestimated-foreachspecies-biconsidered(A÷arAB+C)While the
prdbabihtyofseedgerminationcanbeestimatedasthemeanweightedbytheseed
• NflàvaIprobabilit.•. •.: •..
Fuivoresj(Ds,Tm)
specres
-
02209000325
B / &05 / 060 / 0091Ai+B / 027 / J84 / 01235
C / 073 / 0B0
A-iS÷C / 100 / G81 / 01765
TheseprobabiIitiemaybsubstitutedirtthecalcu1ationofthe-fecundies toobtainnew vaiuestliatintearat&theeffêotcifbate;ofbirds!orofllfrvnrespecies..
5S4’
.
-
‘-I
‘
To test this assumption,informationon seed dispersalfor different frugivoreswasincorporatedinto apopulationprojectionmatrixofN.tetetzo (GodInez-Alvarezetal.,1999).Thisprojectionmatrixwasbuilt bygrouping individualsin size categoriesbased on the heightof the principaltrunk andthenby consideringsurvival,reproductionand growth probabilities
ifindividualsin each categoryover a1-year interval.Size categorieswere:
-dling1 (<2cm), seedling 2 (2—8cm), seedling3 (8—15cm), sapling
(15—45cm), juvenile (45—100cm), immature (100—150cm), mature 1 (150—250cm),mature 2(250—350cm),mature 3 (350—450cm),mature4 (450—550cm), mature5(550—650cm)andmature6(> 650cm), yieldinga total of 12 categories.The projectionmatrixonly has valuesin the main diagonal(survival withoutgrowthprobabilities),the first lower subdiagonal (growthprobabilities),andthe first row (fecundityvalues).Fecundityvalues were estimatedasthe productof:(i)the probabilityofreproduction;and (ii) the meannumberofseeds producedbyareproducingindividualfor each reproductive category,as well as (iii)the constantestimated probabilityof passingfrom seed toseedlingover a1-yeartime-step(Fig.17.2).
Theseed dispersal dataincludedquantitativeand qualitativeaspectsof disperser effectiveness(Schupp, 1993). The quantitative aspectswere frequencyof visits to plants(visits/h)and seed removalby each species (number of seeds removed/fruit). This latter aspect was estimated as numberof seedsremovedperfruit,insteadof per visit, becauseof the largenumberofseedsandtheirsmall size.Thequalitativeaspectswere the proportionofseed germinationaftergutpassageandthe numberofpost- feedingflightsto treesandshrubsfor each speciesof frugivore,andthe numberofsurvivingseedlingsbeneathtrees andshrubsafter1year. The numberofpost-feedingflightsand the numberofsurvivingseedlingswere estimatedonly for treesandshrubs,since theseplantsprovidethe only microsites suitable for seed germination and seedling establishment (GodInez-Alvarezetcii.,1999).Using thesedata,an estimateof: (i)the probabilityofseedremoval;(ii)the probabilityof seed germinationafter gut passage; (iii) the probabilityof delivering seeds to tree andshrub microsites;and(iv)the probabilityofmakingthe transitionfromaseed to a seedlingbeneathtreesandshrubs,were calculatedfor eachspeciesof
Fig.17.2. LifecyclediagramofthecolumnarcactusNeobuxbaumiatetetzo.Valuesinside boxesrefertosurvivalprobabilities (P1valuesinFig.17.la)andarrowsrepresenttransitionor growthprobabilities (G1valuesinFig.17.la).Boxesabovematurecategories2—6indicate fecundityvalues(F1valuesinFig.17.la).
frugivore,andwere used to modifythe fecundityvalues of the N. tetetzo matrixmodel.Thecalculationof theseprobabilitiesand the modification of the fecundityvalues were madefollowingthe proceduresdescribedin Box17.1.
Therole of different frugivores on the population dynamicsof N. tetetzowas determined through matrixsimulationsassessingthe effects of individualspeciesaswellasthe effect ofalldispersersactingtogether.The effect of a species of frugivore was simulated by incorporating the fecundity valuescalculated for that species into the matrix model. To simulatethe effect ofall dispersersactingtogether,the probabilityofseed removalwas calculatedas the sum of all the probabilitiesestimatedfor each species,whereasthe probabilityofseed germinationwasestimatedas the meanweightedby the seed removalprobability(see Box 17.2).The populationgrowthrate (lambda,X)wascalculatedfor each simulation by multiplyingthe projectionmatrixbyavectorrepresenting the numberof individualsin each size category.Thisprocedurewas repeatedwith each resultingvectoruntilitsproportionsremainedconstant,atwhich pointthe populationgrowsatarate equaltolambda.
Resultsshowedthat speciesoffrugivoredifferedin theireffects on the
population dynamicsofN. tetetzo(Table 17.2).Asexpected,the estimated populationgrowthrate dueto the long-nosedbat, L.curasoae,washigher thanthose estimatedtobe duetothe birdsC.brunneicapill’us,M. hypopolius andT curvirostre.ThehighercontributionofL.curasoaewas mostly dueto itshigherprobabilitiesofseed removalandofdeliveringseedstotreeand shrub microsites.The matrix simulationsconducted for all species of frugivores acting together also showed a high contribution to the populationgrowthrateofthiscolumnarcactus(Table 17.2).
ConclusionsandPerspectives
Many authorshave pointedout that inordertoadvanceourunderstanding of seed dispersalby frugivoresitisessentialtoevaluatethe demographic consequencesofdispersalfor plantpopulations(SchuppandFuentes,1995; GodInez-Alvarezetal.,2002;Jordanoand Godoy, 2002; WangandSmith,
2002;Howe and Miriti, 2004).To accomplishthis goal, it is necessaryto
obtaindetailedinformationon the outcomeoffrugivoreactivity and their potentialeffects on critical stages of the life cycle of plants,such as seed germinationand seedling,juvenileand adultsurvivaland growth— the delayed consequencesof dispersal.The studiesconducted up until now have providedsomeinformationon the dispersaleffectivenessoffrugivores andthe demographicfate ofseeds andseedlings.However,thisinformation has not allowedan evaluationof the demographic consequencesof seed dispersal because it is incomplete for most plants and because the methodologicalapproachesneededtointegratethis informationinto plant demography have not been clearlydescribed.The importance of this chapter is the demonstration that an empirical approach based on
-
Table17.2. Frugivoreeffectsemployedtomodifytheprojectionmatrix ofNeobuxbaumia tetetzo.Effectswereestimatedaccording toEqn(2)inBox17.1,whereP=probabilityof seedremovalbyspeciess,G5=probabilityofseedgerminationafter gutpassage,Dc,,,= probabilityofdeliveringseedsinthepatchmbys,andTm=probabilityofmakingthe transitionfrom seed toseedlinginm.
.,,(DsmTm)
FrugivorespeciesGDsmTms=1 m=1 / Estimated rateof increase
Leptonycteriscurasoae(bat) / 0.987 / 0.86 / 0.72 / 0.00139 / 0.001 / 1.003
Campylorhynchus / 0.002 / 0.98 / 0.43 / 0.00139 / 6.0x10 / 0,97
brunneicapillus(bird)
Melanerpeshypopolius(bird) / 0.008 / 0.96 / 0.03 / 0.00139 / 4.2x10’ / 0.97
Toxostomacuivirostre(bird) / 0.003 / 0.98 / 0.12 / 0.00139 / 1.7X1O / 0.97
Allfrugivores / 1.000 / 0.86 / 1.8x1O / 1.009
Pisestimatedastheproductofthefrequencyofvisitsandnumberofseedsremoved perfruitfor eachfrugivore;thisproductwas summedacrossallfrugivores,andtherelative proportionofthetotalwascalculatedforeachspecies.Forallfrugivores,itwascalculatedasthesumoftheremovalprobabilities estimatedforeachspecies.
Gisestimatedasthemeanproportionofgerminatedseedsinlaboratoryexperimentsusing
seedsdefecatedbyeachfrugivorespecies.Forallfrugivore species,itwasestimatedasthe meanweightedbytheseedremovalprobability.
Dsmisestimatedastheproportionofpost-feedingflightstotreesandshrubs outofthetotal numberofflightsrecorded.
Tmiscalculatedasthemean proportionofseedlingssurvivingundertreesandshrubsafter
1year.
projectionmatricesallowsustoexplorethe ecologicalconsequencesofseed dispersalin an explicitlydemographiccontext.Thisapproachisbased on the incorporationof frugivoreeffects intothe matrixelementsconcerned with plantreproductioninordertosimulatefrugivoreimpacts on the populationgrowthrateof plants.Thisallows acomprehensive analysisof the effectiveness of different frugivore species or guilds with the best measurepossible:theireffects on plantpopulationdynamics.
Since thisapproachmay enhanceourunderstanding ofseeddispersal, it is essential to conduct more studies using these ideas in order to determine theirvalidityandapplicability.In thiscontext,itisimportantto define those aspects of seed dispersal that provide the minimum informationneededtoconductthesedemographicevaluations.Generally, the basicdataneededare:
1.Thefrequencyofvisitsbyaspeciesoffrugivoretoplants;
2. Thenumberofseedsdispersedpervisit;
3. Thespecies-specificeffects on seed germination;
4. The species-specificcontribution to seedrainin distinctand relevant microhabitats;
5. Data onseed andseedlingsurvivalinthesemicrohabitats.
Item 4requiresasimultaneousassessmentofseed rain,frugivoreforaging andhabitatpreferencesatthe landscapelevel(e.g.WennyandLevey, 1998; Jordanoand Schupp,2000).With this information,itispossible toestimate the probabilitiestobeincorporatedinto thefirst rowofaprojectionmatrix.
Otheraspectssuch asthe reliabilityoffrugivores,digestivephysiology,and post-foragingbehaviour,amongothers,may also be incorporatedintothe calculation of these probabilities,but muchof these data may be most importantfor dissectingwhyspecies differ in effectiveness,rather thanfor actuallyquantifyingdifferences.
The likelihoodof seeds being dispersed to sites where seedling
recruitmentcan successfullyoccur in the field may besignificantlyaltered byspatio-temporal variability,such aslandscapeheterogeneityand annual changesinthe frugivorecommunity(JordanoandHerrera,1995;Schupp,
1995;Schuppand Fuentes,1995;Herrera, 1998;HoweandMiriti, 2004).
In addition,the ‘quality’ofagiven type ofdispersalsite (e.g. microhabitat) may change temporally and spatially(see Schupp, Chapter 20, this volume). Thus, in order to completely understand the demographic consequencesof seed dispersal,it would be importantto also begin to quantifyvariabilityin criticalfactorsand to evaluatethe effects of such variabilityon the populationgrowthrateof plants.Theseevaluationsare especiallyimportant, since spatialand temporal variability can affect not only the estimatesof the frugivoreeffects, but also the estimatesof the otherniatrixelements,influencingthe rateof populationgrowthandthe contributionof differentlife stagesto this rate(i.e. elasticities;see Pfister,
1998).Theapproachdescribedheremay beused asapreliminarybasis to incorporatethesefactors, throughmonitoringannualchangesin plant demographyas well as in some selected stagesof seed dispersal.Such monitoringalso mightincludedifferent habitatswhereit is knownthat somedemographicprocessessuch as recruitmentor survivalprobabilities may changethroughoutthe life cycle of plants(Howeand Miriti,2004). Basedon this information,itispossibletobuild multi-annualmatricesfor differenthabitats,which integratefrugivoreeffects and plantdemography. Comparingthe populationgrowthratesand elasticitiescalculatedfor each of thesematriceswould permitan evaluationof the extentto which the relativeimportance of particular speciesof frugivoresand different life stagestoplantpopulationdynamicschangesamonghabitatsand/oryears.
Theapproachdiscussedheredependson the availabilityofprojection
matricestoestimatethe growthrate of populations.Thesematrixmodels haveassumptionsthat mightlimit the useofthe proposedapproachunder someconditions(Caswell,2001).
•Projection matrices are time-invariant in that survival and growth probabilitiesand fecundityvalues ofindividualsremainconstantthrough time.
•They also assume that populations grow at a constant rate until reaching a stable distribution; thus they do not consider density
dependent effects.
404H.God/nez-AlvarezandPJordano