Models of Facultative Mutualism: Density Effects Author(s): Carole L. Wolin and Lawrence R. Lawlor Reviewed work(s): Source: The American Naturalist, Vol. 124, No. 6 (Dec., 1984), pp. 843-862 Published by: The University of Chicago Press for The American Society of Naturalists Stable URL: http://www.jstor.org/stable/2461304 . Accessed: 11/06/2012 15:02 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]. The University of Chicago Press and The American Society of Naturalists are collaborating with JSTOR to digitize, preserve and extend access to The American Naturalist. http://www.jstor.org December 1984 The AmericanNaturalist Vol. 124, No. 6 MODELS OF FACULTATIVE CAROLE L. MUTUALISM: DENSITY EFFECTS WOLIN* AND LAWRENCE R. LAWLORt Departmentof Zoology, Universityof Texas, Austin,Texas 78712 SubmittedMay 10, 1982; Accepted May 4, 1984 Mutualism is a commonoccurrence innature.Notableamongsuchinteractions and plants(Faegriand van der Pijl 1979; are associationsbetweenpollinators Gilbert1975;Richards1978); seed dispersersand plants(Handel 1978;Howe and Orians1982);mycor1977;Pudloet al. 1980;van derPijl 1972;Wheelwright bacteria rhizaeand plants(Janos1980;Meyer1967;Tinker1975);nitrogen-fixing and plants(Burnsand Hardy1975;Lange 1967);insectsand fungi(Batra 1979; Cooke 1977);algaeandprotozoa(Karakshian1970,1975;Taylor1975);coelenter(Dustan1979;Goreauet al. 1979;KinzeeandChee 1979); atesand zooxanthellae nectaries(Bentley1977;Janzen and Hymenoptera and plantsbearingextrafloral to interactions resultin directcontributions 1966). Some of these mutualistic suchas forexample,gameteor zygotedispersal.Indirectfitness benefits fitness, or herbivory occurin others. againstpredation nutritional rewardsor protection The relativeimportof such effectsvariesnotonlyfromspeciesto speciesand but also withpopulationparameterssuch as frominteraction to interaction, forexample,intensity ofcomCommunity parameters, densityofthemutualists. withor predation on mutualistically associatedspecies,mayalso modify petition theimpactof mutualistic interaction. In addition,mutualismmay be obligateor facultative.Models of obligate mutualism havequalitatively thanthoseoffacultative different stability properties and Boucher1978).In thisstudywe consideronlyfacultative ones (Vandermeer themmutualism. arethosewhichcan existandperpetuate Facultativemutualists selveswithor withoutthepresenceofa mutualist partner. A formulation Lotka-Volterra equaoffacultative mutualism basedon modified (May 1974).This tionshas been used to predictthatmutualism is destabilizing is to increasethe ofmutualism modelincorporates theassumption thattheeffect to exceedits (i.e., to allowthepopulation equilibrium densityoftheparticipants ariseforspeciespairs,each describedbysuch carrying capacity).Two situations is greater a model.Whentheproductoftheinterspecific interaction coefficients thanor equal to one, populations bound.However,thisseemingly growwithout andBoucher1978), unstablebehaviorallowspopulation persistence (Vandermeer *Presentaddress: Departmentof Zoology, Universityof California,Davis, California95616. tPresentaddress: Box 119, Gualala, California95445. Am. Nat. 1984. Vol. 124, pp. 843-862. All rightsreserved. C 1984 by The Universityof Chicago. 0003-0147/84/2406-0007$02.00. 843 844 THE AMERICAN NATURALIST and the addition of a predatorcan lead to a stable equilibrium(Heithaus et al. 1980). Smaller interactioncoefficients,on the other hand, result in a stable equilibriumin which both species coexist at higherequilibriumdensities than whenalone. In thislattercase, however,Lyapunov stabilityis reducedrelativeto a systemwithnoninteracting species, i.e., thereturntimeto equilibriumfollowing perturbation is increased. Goh (1979) has demonstratedthatin two-speciesLotkaVolterra models of mutualismthat local stabilityimplies global stability.The importanceto stabilityof highlevels of intraspecificcompetitionrelativeto the intensityof mutualisticbenefithas been noted by Dean (1983), Gause and Witt (1935), May (1981), and Travis and Post (1979). Not all models of mutualismincrease the participants'equilibriumdensities. Rather than simplyincreasingeach other's equilibriumdensity,mutualistscan increase r, the intrinsicrate of increase, or both r and the equilibriumdensity. Simulationsdemonstratethatthese lattertwo typesofmutualism,and particularly mutualismwhich increases the intrinsicrate of increase alone, enhance stability (Addicott 1981). Population models of facultativemutualismhave not yet been based on the mechanism of action of mutualism. Here we develop mechanisticmodels of mutualismratherthan simplyexpandingor modifying termsof a Lotka-Volterra type equation. Mutualism is an interspecificinteractionwhich increases births and/ordecreases deaths of individuals.We base our models on this simple assumption.The impactof mutualisticinteractionon an individualis a function,of among other things,recipient density. In certain natural systems mutualistic benefitincreases withincreasingrecipientdensity(Beattie 1976; Plattet al. 1974; Silander 1978) while in others it decreases with increasingrecipient density (Addicott 1979; Antonovicsand Levin 1980; Gilbert1982). We presentsix possible relationsbetween recipientdensityand the effectof mutualismon birthsand deaths of individualsof the recipientpopulationand use these to generatecorresponding population growthequations. We then analyze two-species systems based on these models and characterizethe dynamicalbehaviorof these systems of mutualistswithrespect to stabilityand persistence. This approach has several advantages. The impactof mutualismon thedynamics of populationgrowth(i.e., the effecton r-actual)is clearlyevidentin each of the models presented. This facilitatesrelatingmodels to the "real world" and testingpredictionsgeneratedby them in real systems. By constructinga set of models, each withdiffering responses to recipientdensity,we can compare the dynamicalbehavior of a varietyof two-species systems. In so doing we reach some generalizationsabout the dynamicsof mutualisticsystemsand the conditions under which mutualismenhances systemstability. MODELS The startingpointforeach of the models is one in whichper capita birthsand deaths exhibitlinearresponses to densitywhen no mutualistis present,per capita birthsdecreasingand per capita deaths increasingas a functionof density,thatis MODELS OF FACULTATIVE MUTUALISM 4= bo- 845 bN, and (1) do + dN1. = 4 and d are per capita birthand death rates, respectively,bo the birthrate at zero density, do the death rate at zero density, and b and d per capita densitydependentregulatoryfactors.From these assumptionsWilson and Bossert (1971) derive the logisticequation, N1 = rNj(K- N- K by definingr = bo - do and K = rl(b + d). Mutualisticeffectsare superimposedon this basic model. The mutualistincreases births and/or decreases deaths in a density-dependentor a densityindependentmanner.Since we consideronly facultativemutualistshere, we use the situation in which species exist alone, described by equations (1), as a referencepoint forcomparingthe effectsof different models of mutualism. Density-Independent Model Mutualisticeffectson birthsand/ordeaths may act in a density-independent manner.In such cases additionofa givennumberofmutualistsproducesthesame per capita benefitto recipientsat all recipientdensities (fig. la). For simplicity considerthecase in whichthebirthrateis increasedas a resultofmutualismwhile the death rate remainsunaltered:thus equations (1) become 4 = bo- bNj + mN2 a = do + dN1. m is theper capita effectofN2 on thebirthrateof therecipient,N1. Alternatively, mutualismcould decrease the death rate. This produces an analogous mathematical derivationand will not be formallyanalyzed. The per capita change in populationdensity,birthsminusdeaths, becomes Nl11N = bo - bNj + mN2 - (do + dNj). Substituting r = (bo - do) gives Nl11N = r + mNN - (b + d)N1. (2) r and K retaintheirmeaningfromthe logisticequation in thisand the following derivations,therebyfacilitating comparisonsof theseresultsto thelogisticmodel, i.e., the situationwhenthemutualisticpartneris absent. Substituting (b + d) = ri K in (2) gives Nl11N = r + mN2 - rNj1K. Rearrangingyields Nl11N = (r + mN2)(K + mN2KIr - N&)/(K ?+ mN2KIr). (3) THE AMERICAN NATURALIST 846 a b d~~~~ b d c Cd .. // br b N).. V ../dN....N d0NN f e bo b~N N/ N~~~~N birthsand deaths with no FIG. 1.-Per capita effectsof mutualism.Dashed lines mutualistpresent;solid lines = the same whenalteredby mutualism,and dottedlines are the limits to possible increase in birthsor decrease in deaths as a result of mutualism.a, effectof mutualismon the recipient;b, c, and d, highdensity Illustratesa density-independent effects,and e and f, low densityeffects. Fromthisequationas wellas fromthegraphical analysisitcan be seenthatboth themaximum rateofincreaseandtheequilibrium densityincreasesas a resultof mutualism.This model is mathematicallyequivalent to that of Gause and Witt (1935),Vandermeer and Boucher(1978),andto Addicott's(1981)modelinwhich r-maxand K are increased by mutualism. Density-DependentModels: High DensityEffects the extentof per capita mutualisticbenefitmay varywithrecipiAlternatively, ent density. If mutualismacts to reduce density-dependent regulationat high densityor simplyhas the mostimpactwhen thepopulationunderconsiderationis at highdensitythenmodels such as the threethatfolloware applicable. MODELS OF FACULTATIVE MUTUALISM 847 In thefirstof these (fig.lb) mutualismacts symmetrically to increaseper capita birthsand to decrease per capita deaths. Mutualisticeffectsincreaselinearlywith recipientdensity.As can be seen graphicallythisresultsin an increase in equilibriumdensityabove thecarryingcapacity,butno changein the maximumintrinsic rate of increase. A saturationin mutualisticeffectsis assumed: maximumper capita birthrate cannot exceed bo and minimumper capita death rate cannot go below do. However, as birthrate approaches the line 4 = bo and death rate approaches the line a = do, equilibriumdensityincreases withoutbound. The per capita birthand death rate in this case become 4 = - bN1/(l + MbN2) do + dN1/(1 + mdN2). i= For simplicitywe let mb = md = m. Per capita change in populationdensityis given by N1/N1= bo- do - (b + d)NI/(1 + mN2) or equivalently,by Nl11N = r - (b + d)NI/(I + mN2). (4) SubstitutingrIK = (b + d) into equation (4) and rearranging yields Nl11N = r(K + mKN2 - N&)/(K + mKN2). (5) This model impliesthatmutualismincreases onlytheequilibriumdensityand that birthsand deaths are affectedsymmetrically in a mannerthatincreases linearly withrecepientdensity.Benefitssaturateat 4 = boand d = do. Equatinga withmK in equation (5) resultsin the equation commonlyseen in the literatureformutualism (Addicott 1981; May 1974, 1981; Whittaker1975). In the followingmodels, forthe sake of simplicity,only mutualisticeffectson birthswill be formallyanalyzed. The model analogous to the one above in which onlybirthsare alteredas a resultof mutualismis illustratedin figureIc. Again the birthrate cannot exceed bo. Birthand death rate are given by 4 = bo - bN1/(l + mN2) d= do + dNj, respectively.Population change withrespect to timeis given by Nl11N = bo - bN1/(l + mNN9)- (do + dNj). (6) Substitutionfor(b + d) are rearrangement yields N1/N, = r[1 - (rN1 + dKmN1N2)/rK(1 + mN2)]. (7) Again,theequilibriumdensity,butnottheintrinsicrateof increaseis increasedas a resultof mutualism. Figure Id presentsanothermode of action of mutualismwithprimaryeffectsat 848 THE AMERICAN NATURALIST highrecipientdensity.Per capita increase in birthrate as a resultof mutualismis again a linearfunctionof density.However, in this case births,expressed by 4= bo- (b - mN2)N1, can exceed boiftheimpactof individualmutualistson therecipientor thenumber of mutualistspresentis high,i.e., if mN2 is greaterthan b. Deaths are again i= do + dN1. Althoughin this model the intrinsicrate of increase does not exceed thatin the logisticmodel, it is possible for r-actualto exceed ro. Feasible equilibria,when theyexist, are at densitiesabove the carryingcapacity. The equation forthe populationgrowthrate is N1/N1= r - (b - mN2 + d)N1. (8) Substitutioninto equation (8) for(b + d) yields N1/N, = r(K - N1 + mN1N2)/K. (9) The termhere for the effectof mutualism,mN1N2,as expected in this case, is dependenton recipientdensityand on the densityof the mutualistpartner. Density-DependentModels: Low DensityEffects the effectof mutualismmaybe mostpronouncedat low recipient Alternatively, density.This situationmay arise when (1) the numberof mutualistsis limiting;(2) mutualismcannot compensate for limitingfactors or other density-dependent regulatoryphenomenawhichact at highdensitiesoftherecipient;or (3) thenature of the interspecificinteractionbetween two species is densitydependent,benewhenrecipientdensity ficialin effectwhenrecipientdensityis low butdetrimental is high. Two models in which maximumbenefitsfrommutualismoccur at low recipientdensitiesare presentedbelow. In thefirstof these models theper capita benefitderivedfroma givennumberof mutualistsdecreases exponentiallywith the recipient's density. This may be formulatedin termsof effectson eitherbirths(fig.1E) or deaths. For mutualism affectingonly birthrate, the birthand death rates are 4 = bo - bNj a= + mN2e-N do + dN1. Populationgrowthexpressed on a per capita basis is Nl11N = r + mN2e-'XN - (b + d)NI. (10) Substitutingfor(b + d) in equation (10) and rearranging gives N11N, = r(1 - N1/K) + mN2e-xN. (11) It is apparentboth fromfigure1E and fromthe populationgrowthequation that therate of increase exceeds r-maxat low densities.When atis largethe densityat equilibriumapproaches K; when small it exceeds K. MODELS OF FACULTATIVE MUTUALISM 849 The case in which the mutualisticbenefitdecreases linearlywith recipient density, rather than exponentially(eq. [11]), is depicted in figureIF. While enhancingthe recipient'sbirthrate at low densities,thismodel, unlikeall of the other models considered, has the propertythat at densities above the carrying capacity the birthrate is actually lower than it would be in the absence of the "mutualist." In otherwords,mutualismitself,in thiscase, is a density-dependent phenomena. The same species which is a mutualistat low recipientdensities behaves as a competitoror parasite at densities above the recipient'scarrying capacity! The birthrate in figureIF is 4 = (bo + mNN) - (b + uN,)Nj and the death rate = do + dN1. Per capita change in populationdensityis then Nl/NN = r + inN2 - (b + uN2 + d)NI. (12) When m approaches zero, this equation describes a high density mutualistic system (as in eq. [8]), while when u approaches zero it describes a densityindependentmutualisticsystem(as in eq. [2]). In orderto insurethatmutualism acts primarilyat low densities,we will considerthe special case where - mr/K and hence N1 = K. Substitutingforu and for(b + d) into equation (12) gives Nl11N = (r + mN2)(I - N1/K). (13) As in the previous model (eq. [11]), in thismodel the intrinsicrate of increase is increased by mutualism. Equating mK/rin equation (13) to (x yields one of Addicott's (1981) models. Dynamical Behavior of Two-Species Systems To examine the behavior of mutualisticsystems we must consider, in the simplestcase, pairs of interactingspecies. Paired species need not have similar responsesto mutualismwithrespectto recipientdensity.However, forsimplicity onlythose mutualisticpairs whichdo willbe examinedhere. Implicationsforpairs withqualitativelydifferent responseswithrespectto densityare consideredin the discussion. We characterizethe dynamicbehaviorof the two-speciessystemswithrespect to threecriteria:Lyapunov stability,feasibilityof equilibria,and species persistence. Local Lyapunov stability,the rate of returnto equilibriumfollowingan but smallperturbation, is characterizedin termsoftheeigenvaluesof the arbitrary linearizedsystemaround the equilibriumpoint. Eigenvalues foreach of the twospecies systems are summarized in table 1. Local stabilityanalysis alone is inadequate to characterizethebehaviorof these systemssince itdoes notindicate feasibilityof equilibria, system behavior away fromthe equilibriumpoint, or Z k = w: F v: ~~~~~~~~~~A A A VV S 0-~~~~~~~~~~~~~~~~~1 E 0 ~~ A A o._u m ~~~~~~~~~~tl tl) tl l V V MODELS OF FACULTATIVE MUTUALISM 851 species persistence.An equilibriumis mutuallyfeasibleonlyifbothspecies are at positive densities at equilibrium. Feasibility of equilibria is determinedalgebraicallyand throughstate space diagrams(figs.2-6). A systemis persistentif species do not become extinct,that is, species at positive densities remain at positive densities and any species perturbedto zero density will increase in numbersupon reinvasion. Systems in which the side solutionsare unstable and thus mutualinvasibilityis possible are persistent.Side solutionsare equilibriain which only one species is at positive density,while the otheris at zero density. Unstable as well as stable systemscan be persistent.Mutualisticsystemsin which species densitiesincrease withoutbound are unstablebut persistent;presumably in these cases limitationto populationsize is extrinsicto themodel. Persistenceis analyzed fromstate space diagrams. For our model ofdensity-independent mutualism(eq. [3])thepopulationgrowth equations fortwo such mutualisticpartnersare rlNl(Kl - N, + mlK1N2/rl)/KK N1 No = For simplicity assume r1 = r^N2(K2 - No + m2K2Nl/r,)/K,. r, = r, K1 = K, = K, and ml = tni = in. The same assumptions are also used in the analyses of the other models. The Jacobian matrixthen is: [ fi/8NlIeq 6fi/8N,1eq1 8f2/8Nlleq 6f2/8N2jeq [ r/(I LmK/(I - mK/r) mK/(1 - mKir) mK/r) -r/(1 - mK/r)j The eigenvalues (A) for the matrix are found to be - r/(1 - mKir) + mK/(1 - mKI r). Simplifying, XI = -r and Xi = (-r - mK)/(1 - mK/r). Two cases arise, dependingupon the parametervalues. The firstcase is characterizedby a stable,mutuallyfeasibleequilibrium.Whenr > mK, XI is the dominant(most positive or least negative) eigenvalue. Both eigenvalues are negative,thus the systemis stable around the equilibriumpoint. The logistic model provides a frame of referencefor determiningdegree of stability.The eigenvalues for a system of noninteractingspecies governed by logisticequations both equal - r. The measure of degree of stability,the timefor returnto equilibriumfollowingperturbation or T, is givenby - l/X,whereX is the dominanteigenvalue. Returntimeforthe logisticmodel is lr. Returntimefora pair of density-independent mutualistsis likewise l/r,thusmutualismof thistype does not affectthe degreeof stabilityof the system.Furthermore, theequilibrium is a mutuallyfeasible one. N, the equilibriumdensity,equals K(1 + mK/r)/[1(mK/r)2].N is greaterthan zero when r > mK, i.e., the conditionforstabilityis identicalto the conditionformutualfeasibility.This situationcorrespondsto the state space diagram in figure2a of a persistentsystemwith a stable, mutually feasible equilibrium. The otheroutcome is an unstable,unfeasibleequilibrium.Whenr < inK thenX2 becomes thedominanteigenvalue. Since underthese conditionsXi is positive,the systemis unstable. However, thisequilibriumis also unfeasible.This situationis depicted in figure2b. The conditionforthis outcome, r < mK, is equivalent to 852 THE AMERICAN NATURALIST a. N2 b. N FIG. 2.-Isoclines for mutualisticsystems with 2 density-independent mutualistseach describedby eq. (2) or with2 highdensitymutualistseach describedby eq. (4). In a thereis a stable equilibriumwhile in b species are persistentalthoughthe system is unstable. For density-independent mutualists,a occurs when r > mK and b when r < inK. The N1 isocline has intercepts(K, 0) and (0, - rim)and slope (b + d)/mn.Switchingvalues forN1 and N, gives analogous parametersforthe N, isocline. For highdensitymutualistseach describedby eq. (4), a occurs when I > mK and b when I < mK. For the N1 isoclinesthe interceptsare (K, 0) and (0, - 1/m)and the slope is 1/mK. (b + d)<m, i.e., mutualisticbenefitis large relativeto self-damping.Although unstable, this systemis persistent.Side solutions[(O,K) and (K,O)] are unstable and as can be seen in figure2b, both species have unboundedpopulationgrowth unless limitedby some factorextrinsicto this model. We turn next to the firstmodel for high densityeffects,the case in which mutualismaffectsbirthsand deaths simultaneously,withgreaterimpactat high recipientdensities(eq. [5]). The followingJacobianmatrixis obtainedfora twospecies systemof this sort: [rtk rmK LrmK - r- The eigenvaluesgeneratedby thismatrixare -r(l ?i mK). Two cases, analogous to those of the density-independent model arise. For thelargerof the eigenvalues to be negativemK mustbe less than 1. This correspondsto a stable and feasible equilibrium.The systemis less stable, though,thanone withtwo noninteracting species or two density-independent mutualistssince returntimeto equilibrium,1/ (r - rmK), is greaterthan lhr.The linearisoclines intersectat the pointwhereN = rl(b + d - rm),thus equilibriumdensitiesare positiveas long as the stability conditions,mK < 1, hold. This situationcorrespondsto thatin figure2a. When mK is greaterthan 1, the equilibriumis unstable (saddle point) and unfeasible albeit persistent.This corresponds to figure2b. In this case populations grow withoutbound. If mK = 1, the isoclines forN1 and N2 are parallel, thereis no equilibriumpoint,and again populationsgrow withoutbound. MODELS OF FACULTATIVE MUTUALISM 853 In the second model in which mutualisticeffectsare most pronouncedat high recipientdensity(eq. [7]) the Jacobianis L br~m/(b + d + d~nA br ml(b + d + dinN + d + dmnA)21 Ijj1;b2mnl(b - .' The eigenvalues forthis matrixare - r ? brl2n/(b+ d + dmN)2. The dominant + d + dmnN)2.The eigenvalueis negativeand the systemis stable if - r > brl-2m/(b eigenvaluesthusdepend on N. There are two equilibria:N = (rm - b - d)/21nd + L(b + d)2 + 2rin(d - b) + 'ml]"12/nd. These correspond to the two points of intersectionof the curvilinearisoclines seen in figure3, one a stable, feasible equilibriumand the other an unstable, unfeasiblesaddle point. In this system, then,thereis always a stable feasibleequilibrium.The systemis less stable than one with a pair of noninteractingspecies since Tr.> Ilr. Since the feasible equilibriumis stable and thereare unstable side solutions,this systemis persistent. The thirdmodel withmutualisticeffectsoccurringprimarily at highself-density (eq. [9]) has the possibilityof unlimitedbirthrate. Here the Jacobianis [7r ]2 min m&2] - I' and the eigenvalues are -r ? inN2, where N = (b + d)/2mn? [(b + d)2 4rin]12/2rn. This systemwould tend toward instabilitysince wheneverInN2 exceeds the intrinsicrateof increase thedominanteigenvalueis positive.The return timeunderconditionswhere stabilityholds, 1/(r- miN2), also exceeds thatfora noninteracting species pair. The parametervalues determinewhetherthereare two real equilibria,one real equilibrium,or no real equilibrium.When (b + d)lm > 4K, thereare two real solutions:theone at lowerdensities{N = (b + d)12mn [(b + d - 4rm]l1/2m}is stable whilethe one at higherdensities{N = (b + d)l 2m + [(b + d)2 - 4rm]112/2m} is unstable. In effect,as seen in figure4a, thisis a persistentsystemin which, above a criticaldensityforboth species, population numbersincrease withoutbound unless limitedby some factorextrinsicto the model. The second case arises in theinfinitely unlikelysituationthat(b + d)lm = 4K. The result is a single unstable, feasible equilibrium(fig. 4b) in which the ifN1 and N. are above N. In thethirdcase, (b + populationsincrease indefinitely d)/m<4K. Here the species are also persistentand populationdensitiesincrease withoutbound, but thereare no real equilibria(fig.4c). Thus for all three models for which per capita mutualisticbenefitsto the recipientspecies increase withincreased density,species are always persistent, but stabilityis less than thatfornoninteracting species. In contrast,the models in which mutualismacts primarilyat low recipient densityexhibitgreaterstability.The Jacobianformutualisticspecies pairs which THE AMERICAN NATURALIST 854 N2 each describedby eq. (6). Thereis FIG.3.-The isoclinesfor2 highdensitymutualists is also unfeasible. The N1 The unstableequilibrium alwaysa stable,feasibleequilibrium. N1 = rld andN, = -(b + d)l isoclinehas intercepts (K, 0) and(0, - I/m)andasymptotes md. as a function ofincreasdecreasing benefits ofmutualism eachhaveexponentially density(eq. [11])is ingrecipient L -xmN2e N- (b - d) 1 Nme-N LNme-N OwmNIe- N- (b + d)] + rN/K) ? Nie-'xN, The eigenvalues for this matrix equal -(mN-cxeor equivalently,- rN/K - (No- + 1)Nme-IV. Since thisis a mutualisticsystem actingprimarily we assumeN - K. In addition,to assurethatthisis a mutualism that ourselvesto valuesof(x> 1/K,insuring at low recipient density,we restrict effect as K is approached.Undertheseconditions themutualistic dropsoffrapidly N ( > 1 and X < - rN/K< - r. This means thatX < - r and returntimeto the is inall cases morerapidthanthatfornoninteracting species. feasibleequiibrium 5 inwhichitcan be seenthatthereis alwaysa The isoclinesare depictedinfigure ofthissortis notonlypersistent, but point.Mutualism stable,feasibleequilibrium enhancessystemstability. atlowrecipient In theothermodelinwhichmutualistic benefits occurprimarily effectsat densities interaction producesdetrimental densities,the interspecific exhibitstability above thecarrying capacity(eq. [13]). Pairsof suchmutualists thesameas thoseforthemodeljust described. whichare qualitatively properties The Jacobianmatrixforthissystemis -rI mK - 0 r - mKJ and the eigenvalues both equal (-r - mK). This systemis always stable and returntimeto equilibrium, 1/(r+ mK), is less than1hr.In otherwords,this like and the one discussed, alwayshas a stable,feasibleequilibrium just system, a. N2 0I1 S N 111 b~ ~~. ~ C. I N2 N2 0 v ~ ~ = 10 C4 C. ~N2 =0 N N1 FIG. 4.-The isoclines for2 highdensitymutualistseach describedby eq. (8). There are 2 equilibria when (b + d)lnm> 4K, the one at lower densities stable and the one at higher densitiesunstable,as depictedin a. When(b + d)/n = 4K thereis an unstableequilibriumas in b and when (b + d)/rn< 4K thereis no equilibriumand the systemis unstable.as in c. In all cases species are persistent.The interceptforthe N, isocline is (K, 0) and the asymptotes are N, = 0 and N, = (b + d)lnm. 856 THE AMERICAN NATURALIST N2 0 low densitymutualists, byeq. (10). Thereis FIG.5.-The isoclinesfor2_______\Z%/r;2=o 11 ~each described N always a stable, feasible equilibriumand an unstableequilibrium.The systemis thus stable are(K,0) is (K FortheN1 isoclinetheintercepts andpersistent. (.. and - rlm),theminimum - tua, - rKange[I-K]) andthereis an asymptote at N =g e b N2 m ehnistictiey mutualis o isc thsriedb m geneato of3)ah syset The. ouTcoe ofa e fore2lo muuais on [10] [13])ep inoprtn fetof h of models(eqql.[2,h4] [6] [8],in oforecipaien densinty.Acnalsso species. pairstoug stbiriths andordathsaseate fntion ) deninty repnsseucdts thrbenchneea geeal pasutten of mutualism:(i. wthesimilabriu a densitybreipintmndo ofhlo densmityrcipientsrfchingh mtho lse characeriti or e o behaviion vimprtn of density efecssthe dhensity-independentreipechnts.the oftwo-speciesmutualistic systemscan be seenfromtable1. as follows.(1) In thosesystemswheremaxThese resultscan be summarized MODELS OF FACULTATIVE MUTUALISM 857 density, mutualism exertsa stabilizing imalbenefits are incurred at low recipient beingequal,return timetoequilibrium effect. In otherwords,all otherparameters is morerapidforsuch two-speciessystemsthanforsystemsof noninteracting Thisresultis species.Furthermore, thereis alwaysa stable,feasibleequilibrium. of a generalnature,applyingbothto cases in whichthe impactof mutualism decreasesexponentially withrecipientdensity(eq. [11]) and to thosein which effects density,detrimental mutualistic benefits decreaselinearlywithrecipient whenrecipient densityexceedsK (eq. resulting fromtheinterspecific interaction at highselfprimarily benefit [13]). (2) Mutualisticsystemsin whichrecipients densityexhibitverydifferent behavior.Two speciessystemsbasedon equations and unstable, on theparameter values,eitherpersistent (4) or (8) are,depending or stablebutless so thanthecomparablesystemwitha pairof noninteracting species.For a pairof specieseach describedby equation(6), thereis alwaysa whichlikewise,is less stablethanthatfornoninstable,feasibleequilibrium, mutualism (eq. [2]) eitherdoes not teracting species. (3) Density-independent affecttherelativestability of thesystemas comparedto thebehaviorof noninon the system, depending teracting species,orresultsinan unstablebutpersistent parameter values. inall ofthesemodelsresultsin speciespersistence. In otherwords, Mutualism systemsarealwaysmutually invasibleand do two-species facultative mutualistic occursforpairsof notlead to speciesextinctions. In thosecases whereinstability unbounded recipients, it reflects highdensityrecipients or density-independent The persisratherthanpopulations decreasing to zerodensity. population growth and has also beennotedby Vandermeer tenceofunstablefacultative mutualists topopulation sizewouldarisefromfactors Boucher(1978).Presumably limitation incompetitive instabilextrinsic tothemodelsinsuchcases. In contrast, systems, on andmutualinvasibility is conditional ityis associatedwithspeciesextinction, values(MacArthur 1972). parameter that systems mutualistic The assumption is madeintheanalysesoftwo-species arecontinupair.Becauseeigenvalues r,K, andmareidentical forthemutualistic values ous functions oftheparameters ofa model,forsmallchangesinparameter resultshold theeigenvalueswillchangea limitedamount.Thus,ourqualitative overa rangeofparameter valuesin whichmutualistic speciesare notidentical. withmixed we didnotanalytically determine thebehaviorofsystems Although different responsesto models(i.e., wheremutualistic partners havequalitatively mutualism withrespectto self-density), we can makesomegeneralpredictions simulations aboutthebehaviorofsuchsystems.The modelsusedincomputer by Addicott(1981) are mathematically analogousto threeof the modelsderived one fora highdensityrecipient, recipient, herein,one fora density-independent and one fora low densityrecipient.In our workwe showthatthedynamical behaviorof the systemdiffersbetweenthese classes of modelsand exhibits commonattributes withineach class ofmodels.It is thuspossibleto extrapolate of fromAddicott'ssimulations patterns withmixedmodelstogenerate ofsystems witha thebehaviorof mixedmodels.A combination of a low densityrecipient resultsin systems or witha highdensityrecipient density-independent recipient of whichare morestablethanthatfornoninteracting species.The combination 858 THE AMERICANNATURALIST and density-independent highdensityrecipients recipients does notproducethis effect. Thus,mutualists whichaccruethegreatest percapitabenefits frommutualism at low self-density enhancesystemstability, bothwhentheyoccurwitha partnerwithqualitatively similarresponsesto mutualism withrespectto selfdensityand whentheyoccur witha mutualistic partnerhavinga qualitatively different response. How frequently do eachoftheseclassesofmutualists occurinnature?Thedata at the existing presenttime,although somewhat limited, suggestthateachofthese classes of mutualists and that low are perhapsmore occur, densityrecipients commonthanhighdensityrecipients. This is nottoo surprising since,in many mutualistic systems,mutualism cannotoverridetheeffectof all otherdensitydependent factorslimiting facilipopulation growth. Speciesforwhichmutualism tatesnutrient procurement wouldbe a goodplace to lookfortheoccurrenceof highdensityrecipients. Predictions aboutwhichdensityresponsesshouldoccur withgreatestfrequency in mutualistic systemsare difficult to makebased solely inthispapersinceourmodelsarenotevolutionary on themodelspresented ones. The conclusionthatcertainclassesofinteractions areinfrequent sincepopulation dynamicmodelsare unstable,thoughsometimesasserted,can be misleading when modelsdo not take into accounttemporalor spatialvariability, other interspecific interactions, age structure, or evolutionary This is considerations. truein the case of facultative particularly mutualism sinceinstability does not Effectsof recipientdensityin real mutualistic implyextinction. systemshave beenat leastpartially in thefollowing characterized cases. In some systemsit appearsthatinterspecific interaction at low is beneficial densitiesand detrimental at highrecipient densities,in a mannersimilarto that modeledin equation(13). Addicotthas foundthatin ant-aphid associationson Epilobiumaugustifolium aphidsbenefit moststrongly fromthepresenceofantsat and thatantscan actuallybe detrimental low-density at highaphiddensitywhen thelatterare experiencing mimfoodshortage (Addicott1979).CertainMullerian at low ics such as Heliconiusspecies are thoughtto interactmutualistically densitiesbutnotat highdensities.Presumably a criticaldensityofbutterflies is necessaryforpredators to learnthesebutterflies aredistasteful. Athighdensities, however,theinterspecific interaction since is detrimental ratherthanmutualistic forresourcesthenoccurs(Gilbert1983). competition Othermutualistic are probablyadvantageous deninteractions at low recipient sitiesand less effective butnotharmful at higherdensities.Such a situation can occurwhendensity-dependent limitations on birthrateor survivalare not removedthroughmutualism. It mayalso arise whencompetition formutualistic partnersnormally at low recipient occurs,promoting higherper capitabenefits density.One possiblecase involvesthe extrafloral nectarybearingcomposite, Helianthellaquinquenervis, at subalpineelevations(Inouyeand Taylor1979).It has also beenpostulated thatpollinators arelimiting incertainpollination systems (Bierzychudek 1981;Heinrich1975;Levin and Anderson1970;Mosquin1971; Reader1975;Waser1978).In thecase ofpollination systems thisdoes notalways meanthatlowerplantdensitywouldincreasethe probability to of visitations MODELS OF FACULTATIVE MUTUALISM 859 individualplants since higherdensitymay be needed to attractpollinatorsto a patch or to maintainpollinatorconstancy(Heinrich 1979a; Levin 1979). Density effectsare known to be importantin pollinationsystems. Pollinator choice, constancy,and movementpatternsare dependenton plantdensity(Levin and Kerster 1969; Carpenter1976; Gill and Wolf 1977; Levin 1978; Schaal 1978; Heinrich 1979b) and these in turnaffectsuccess of gamete transferforthe plant (Levin and Kerster1969; Levin 1972; Frankieand Baker 1974;Frankieet al. 1976; Price and Waser 1979). Plant densitycan have positive(Platt et al. 1974; Beattie 1976; Silander 1978) or negative(Antonovicsand Levin 1980) effectson pollination success. Plant density also affectssuccess of seed dispersal for plants. Sanguinaria canadensis seeds, for instance, are dispersed longer distances at lower plant densities(Pudlo et al. 1980). Data on the effectof densityon seed dispersal are limited.The interactionof seed-dispersalbehavior,geneticstructureof the plant population, distributionof safe sites, and plant densitywould be expected to influencethe effectof plant densityon success of seed dispersaland subsequent germination. Furthercharacterizationof real mutualisticsystemson the basis of density responses and comparisonsbetween these systemswithrespect to stabilityand persistencepropertieswould be a richand potentiallyrewardingarea forresearch. We conclude fromthe models presentedthatfacultativemutualisticsystemsare persistentunder all conditionsconsidered. Furthermore,mutualism,even using the stringent propertiesof Lyapunov stability,need notbe destablizing,and when greatestper capita benefitsaccrue at low recipientdensities,mutualismenhances stability.The response to mutualismwithrespectto recipientdensityis thusone key parameterin determiningthe dynamicalbehaviorof mutualisticsystems. SUMMARY Six populationmodels of facultativemutualismare formulatedin termsof per capita births and deaths. Each explicitlyconsiders the per capita impact of mutualismwith respect to recipientdensity.The models include one with per capita benefitsof mutualismindependentof recipientdensity,three models of density-dependentmutualismwith effectsmost pronounced at high recipient density,and two models withmutualisticbenefitsmostpronouncedat low recipient density.The dynamicbehavior of systemsof species pairs withlike density mutualism responses is analyzed. The model incorporatingdensity-independent resultsin a systemthat,dependingon parametervalues, is eitherunstableor has stabilityequal to thatof systemsof noninteracting species. The threemodels with highdensitymutualismeach resultin systemsthatare eitherunstableor thatare species. The two stable, but less so thancomparable systemswithnoninteracting models withlow densitymutualismeach resultin systemsthatare always stable, and always more stable thancomparablesystemswithnoninteracting species. All of these mutualisticsystems,whetherstable or unstable,exhibitspecies persistence. 860 THE AMERICAN NATURALIST ACKNOWLEDGMENTS We thankP. Davis, D. Barnes, C. Galen, M. Oldfield,B. Pierson,D. Marshall, J. Gray, and L. Williams for theirparticipationin stimulatingdiscussions. We would also like to thank E. Pianka, M. Uyenoyama, C. Galen, N. Waser, L. Gilbert,C. Toft,D. Boucher, and two anonymousreviewersforhelpfulcritiques of the manuscript.This articleis based on the Master's thesis of C. L. W. LITERATURE CITED Addicott,J. F. 1979. A multispeciesaphid-antassociation: densitydependence and species-specific effects.Can. J. Zool. 57:558-569. 1981. Stabilitypropertiesof two-species models of mutualism:simulationstudies. Oecologia 49:42-49. Antonovics,J.,and D. A. Levin. 1980.The ecological and geneticconsequences of density-dependent regulationin plants. Annu. Rev. Ecol. Syst. 11:411-452. mutualismand commensalism.Proc. Symp. Batra, L. R., ed. 1979. Insect-fungussymbiosis:nutrition, Second InternationalMycological Congress. Wiley, New York. Beattie, A. J. 1976. Plant dispersion,pollinationand gene flowin Viola. Oecologia 25:291-300. Bentley,B. L. 1977. Extrafloralnectariesand protectionby pugnaciousbodyguards.Annu. Rev. Ecol. Syst. 8:407-427. Bieryzchudek,P. 1981. Pollinatorlimitationof plantreproductiveeffort.Am. Nat. 117:838-840. Burns, R. C., and R. W. F. Hardy. 1975. Nitrogenfixationin bacteria and higherplants. SpringerVerlag, New York. Carpenter,F. L. 1976. Plant pollinatorinteractionsin Hawaii-pollination energeticsof Metrosideros collina (Myrtaceae). Ecology 57:1125-1144. Cooke, R. 1977. The biologyof symbioticfungi.Wiley, New York. Dean, A. M. 1983. A simple model of mutualism.Am. Nat. 121:409-417. chlorophyllpigmentsof the reefDustan, P. 1979. Distributionof Zooxanthellae and photosynthetic buildingcoral Monastrea annularis Ellis and Solander in relationto depthon a West Indian coral reef. Bull. Mar. Sci. 29:79-95. Faegri, K., and L. van der Pijl. 1979.The principlesof pollinationecology. 3d. ed. Pergamon,Oxford. Frankie,G. W., and H. G. Baker. 1974. The importanceof pollinatorbehaviorin the reproductionof tropicaltrees. An. Inst. Biol. Univ. Nac. Auton. Mex. Ser. Bot. 45:1-10. Frankie,G. W., P. A. Opler, and K. S. Bawa. 1976. Foragingbehaviourof solitarybees: implications foroutcrossingof neotropicalforesttree species. J. Ecol. 64:1049-1057. Gause, G. F., and A. A. Witt. 1935. Behavior of mixed populations and the problem of natural selection. Am. Nat. 69:596-609. and plants. Gilbert,L. E. 1975. Ecological consequences ofa coevolved mutualismbetweenbutterflies Pages 210-240 in L. E. Gilbertand P. H. Raven, eds. Coevolution of animals and plants. Universityof Texas Press, Austin. 1983. Coevolution and mimicry.Pages 263-281 in D. Futuymaand M. Slatkin,eds. Coevolution. Sinauer, Sunderland,Mass. Gill, F. B., and L. L. Wolf. 1975. Foraging strategiesand energeticsof east Africansunbirdsat mistletoeflowers.Am. Nat. 109:491-510. Goh, B. S. 1979. Stabilityin models of mutualism.Am. Nat. 113:261-275. Goreau, T. F., N. I. Goreau, and T. J. Goreau. 1979. Corals and coral reefs. Sci. Am. 241:124-136. Handel, S. N. 1978. The competitiverelationshipof threewoodland sedges and its bearingon the evolutionof ant-dispersalof Catex pedunculata. Evolution 32:151-163. Heinrich,B. 1975. Bee flowers:a hypothesison flowervarietyand bloomingtimes.Evolution29:325354. 1979a. "Majoring" and "minoring"by foragingbumblebees,Bombus i'agans: an experimental analysis. Ecology 60:245-255. MODELS OF FACULTATIVE MUTUALISM 861 and patternsof movementin foragingbumblebees. Oecologia . 1979b. Resource heterogeneity 40:235-245. Heithaus, E. R., D. C. Culver, and A. J. Beattie. 1980. Models of some ant-plantmutualisms.Am. Nat. 116:347-361. Howe, H. F. 1977. Bird activityand seed dispersalof a tropicalwet foresttree. Ecology 58:539-550. Inouye, D. W., and 0. R. Taylor, Jr. 1979. A temperateregion plant-ant-seedpredator system: consequences of extrafloralnectarsecretionby Helianthellaquinquenervis.Ecology 60:1-7. Janos, D. P. 1980. Mycorrhizaeinfluenceon tropicalsuccession. Biotropica 12 (suppl.):56-64. Janzen,D. H. 1966. Coevolution of mutualismbetween ants and acacias in CentralAmerica. Evolution 21:620-637. Karakshian, M. W. 1975. Symbiosis in Paramecium bursaria. Pages 145-173 in D. H. Jenningsand D. L. Lee, eds. Symbiosis, Symposia of the Society for ExperimentalBiology XXIX. CambridgeUniversityPress, Cambridge. Karakshian, S. J. 1970. InvertebratesymbioseswithChlorella. Pages 33-52 in K. L. Chambers,ed. Biochemical coevolution. Proc. 29th Annual Biol. Colloquium. Oregon State University Press, Corvallis, Oreg. zooxanthellae on the growthof Kinzee, R. A., III., and G. S. Chee. 1979. The effectsof different experimentallyreinfectedhosts. Biol. Bull. (Woods Hole) 156:315-327. Lange, R. T. 1967. Bacterial symbiosiswithplants.Pages 99-170 in S. M. Henry,ed. Symbiosis.Vol. I. Academic Press, New York. in naturalpopulationsofLithosperLevin, D. A. 1972. Plantdensity,cleistogamyand self-fertilization mum caroliniense. Am. J. Bot. 59:71-77. 1978. Pollinatorbehavior and the breedingstructureof plant populations.Pages 133-150 in A. J. Richards, ed. The pollinationof flowersby insects. Academic Press, New York. 1979. Pollinatorforagingbehavior:geneticimplicationsforplants.Pages 131-153 in 0. Solbrig and P. H. Raven, eds. Topics in plantpopulationbiology.Columbia UniversityPress, New York. Levin, D. A., and W. W. Anderson. 1970. Competitionfor pollinatorsbetween simultaneously floweringspecies. Am. Nat. 104:455-467. gene dispersalin Liatris. Am. Nat. 103:61Levin, D. A., and H. W. Kerster. 1969. Density-dependent 74. MacArthur,R. H. 1972. Geographicalecology. Harper & Row, New York. May, R. M. 1974. Stabilityand complexityin model ecosystems,2d ed. PrincetonUniversityPress, Princeton,N.J. 1981. Models for two interactingpopulations. Pages 78-104 in R. M. May, ed. Theoretical ecology: principlesand applications. 2d ed. Sinauer, Sunderland,Mass. Meyer, F. H. 1967. Mycorrhiza and other plant symbioses. Pages 171-255 in S. M. Henry, ed. Symbiosis. Vol. I. Academic Press, New York. time.Oikos Mosquin, T. 1971. Competitionforpollinatorsas a stimulusforthe evolutionof flowering 22:398-402. Platt,W. J., G. R. Hill, and S. Clark. 1974. Seed productionin a prairielegume. Oecologia 17:55-63. Price, M. V., and N. M. Waser. 1979. Pollen dispersaland optimaloutcrossingin Delphiniumnelsoni. Nature 277:294-296. Pudlo, R. J.,A. J. Beattie,and D. C. Culver. 1980. Populationconsequences of changesin an ant-seed mutualism.Oecologia 146:32-37. Reader, R. J. 1975. Competitiverelationshipsof some bog ericads formajorinsectpollinators.Can. J. Bot. 53:1300-1305. Richards,A. J., ed. 1978. The pollinationof flowersby insects. Academic Press, New York. Schaal, B. A. 1978. Density-dependentforagingon Liatris pycnstachya.Evolution 32:452-454. Silander,J. A. 1978. Density-dependentcontrolof reproductivesuccess in Cassia biflora.Biotropica 10:292-296. Taylor, D. L. 1975. Symbioticdinoflagellates.Pages 267-278 in D. H. Jenningsand D. L. Lee, eds. Symbiosis,Symposia of the Society forExperimentalBiologyXXIX. CambridgeUniversity Press, Cambridge. Tinker,P. B. H. 1975. Effectsof vesicular-arbuscularmycorrhizason higherplants.Pages 325-350 in 862 THE AMERICAN NATURALIST D. H. Jenningsand D. L. Lee, eds. Symbiosis, Symposia of the Society for Experimental Biology. XXIX. CambridgeUniversityPress, Cambridge. Travis, C. C., and W. M. Post, III. 1979. Dynamics and comparativestatics of mutualisticcommunities.J. Theor. Biol. 78:553-571. Vandermeer,J. H., and D. H. Boucher. 1978. Varieties of mutualisticinteractionsin population models. J. Theor. Biol. 74:549-558. van der Pijl, L. 1972. Principlesof dispersal in higherplants. 2d ed. Springer-Verlag, New York. Waser, N. M. 1978. Competitionfor hummingbird pollinationand sequential floweringin two Colorado wildflowers.Ecology 59:934-944. Wheelright,N. T., and G. H. Orians. 1982. Seed dispersalby animals: contrastswithpollen dispersal, problemsof terminology,and constraintsof evolution.Am. Nat. 119:402-413. Whittaker,R. H. 1975. Communitiesand ecosystems. Macmillan,New York. Wilson, E. O., and W. H. Bossert. 1971. A primerof populationbiology.Sinauer, Sunderland,Mass.
© Copyright 2026 Paperzz