Host-Plant Mediation of Mutualisms: Variable Outcomes in Herbivore-Ant Interactions

J. Hall Cushman

Oikos, Vol. 61, No. 1. (May, 1991), pp. 138-144.

Stable URL: http://links.jstor.org/sici?sici=0030-1299%28199105%2961%3A1%3C138%3AHMOIMV%3E2.0.CO%3B2-N

Oikos is currently published by Nordic Society Oikos.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/oikos.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org Fri Apr 20 14:13:38 2007 FORUM is intended for new ideas or new ways of interpreting existing information. It FORUM provides a chance for suggesting hypotheses and for challenging current thinking on ecological issues. A lighter prose, designed to attract readers, will be permitted. Formal FORUM research reports, albeit short, will not be accepted, and all contributions should be concise FORUM with a relatively short list of references. A summary is not required.

Host-plant mediaabn of insect mutualisms: variable outcomes in herbivore-ant interactions

J . Hall Cushman, School of Biol. Sci., Macquarie Univ., Sydney, NSW 2109, Australia

Summary. I present and evaluate the hypothesis that host plants Ants are one of the most ubiquitous mutualists found indirectly affect the fitness of many homopterans and lepidop- in nature, forming close and often mutualistic associa- terans by mediating the outcome of their interactions with ants. W o conditions must be met to support this contention. First, tions with a variety of plants, fungi, and , partic- the attractiveness of herbivores to ants must vary predictably ularly herbivorous in the Homoptera and Lep- with intra- andlor interspecific changes in host-plant quality. idoptera (Wilson 1971, Carroll and Janzen 1973. Bent- Such variation will occur if a) the chemical composition andlor quantity of ant attractants (excretions and secretions) produced ley 1977, Thompson 1982, Beattie 1985, Sudd and by herbivores varies with changes in host quality and b) ants Franks 1987, Holldobler and Wilson 1990). While herb- preferentially tend those herbivores that produce the most nu- ivore-ant associations have received considerable atten- tritionally rewarding attractants. Second, the number of tend- tion during the past century (see reviews by Hinton ing ants must have a significant effect on the fitness of herb- 1951, Nixon 1951, Way 1963, Cottrell 1984, Buckley ivores. While there is abundant evidence to support these indi- vidual conditions, few studies have examined both conditions 1987, Pierce 1987, Sudd 1987), few studies address the for a single herbivore-ant association. influence of host plants on the outcome of herbivore-ant I outline a series of testable predictions based on the host- interactions and no study outlines the mechanisms by mediation hypothesis. 1) The effects of host plants will vary with which such effects can occur. the degree to which herbivores depend on ants. 2) Hosts will affect the ability of herbivores to compete for ant mutualists. 3) In this paper, I develop the hypothesis that host Tended herbivores will exhibit host-selection behaviors that in- plants -g enerate si-g nificant variabilitv in the outcome of crease their ability to attract ants. And most importantly, 4) interactions between herbivores and their tending ants. variation in host quality will predict fluctuations in the strength and sign of herbivore-ant interactions. I present a graphical I argue that such host-plant mediation can influence model to illustrate 1) the ways in which plant-induced variation these interactions, and the selection pressures which in the value of attractants to ant colonies can be used to predict accompany them, in two fundamentally different ways. switchpoints in the behavior of ants and 2) the effects of distance First, the strength of mutualistic interactions can fluctu- to servicing ant nests and nutritional status of these colonies on +). the predictions. ate from strong to weak ( + + + t o In these situa- tions. selection pressures will be weak but consistent (i.e., in the same direction). By contrast, the sign of interactions can fluctuate from mutualism to antag- onism (+ to -). When such reversals occur, the actual An increasing number of studies demonstrate the vari- direction of selection changes and selection becomes a able and context-dependent nature of mutualistic inter- disruptive force, favoring traits (behaviors and mor- actions (see Thompson 1982, 1988, Addicott 1984, Jor- phologies) that facilitate the association in certain sit- d a n ~19 87, Herrera 1988, Cushman and Addicott in uations and favoring very different and perhaps oppos- press). The outcome of these interactions varies in ing traits in others. space and time in response to a range of biotic and Here, I present and evaluate the mechanisms by abiotic factors. Unfortunately, very little is known which host plants can indirectly mediate herbivore-ant about the extent, mechanisms, and implications of this interactions. I then use this framework to address vari- variation, and such information is essential for devel- ous implications of host-plant mediation, with particular oping a realistic understanding of the ecology and evo- emphasis on how plants alter the strength and sign of lution of mutualism. these insect interactions. ants independently. Herbivore aggregations that form Herbivore-ant associations on plants with extrafloral nectaries will be more likely to Both homopteran-ant and lepidopteran-ant associations be located and subsequently tended by ants than aggre- are common mutualisms found in temperate and trop- gations forming on hosts without such structures. This is ical regions. In homopteran-ant associations, ants tend supported by Buckley (1983), Maschwitz et al. (1984). phloem-feeding homopterans (primarily those in the and Sudd and Sudd (1985) who show or suggest that Aphididae. Membracidae, Coccidae, and Pseudococci- ants commonly switch from visiting extrafloral nectaries dae) and harvest their excretions ("honeydew"). In lep- to visiting tended herbivores (although see Becarra and idopteran-ant associations, ants tend leaf-chewing but- Venable (1989) who claim that the reverse can also terfly larvae (those in the and Riodinidae), occur). Given that oviposition by many lycaenid species and harvest their secretions (see also DeVries and Baker is dependent upon the presence of ants (Atsatt 1981b, (1989) who document that larvae of a riodinid species Pierce and Elgar 1985, Smiley et al. 1988. Baylis 1989, consume their host's extrafloral nectar). Through their Baylis and Pierce in press), host plants with extrafloral tending, ants can provide a range of beneficial services nectaries may influence tended lepidopterans in a simi- to both homopterans and lepidopterans, particularly lar fashion. protection from natural enemies, but also increased In this paper, I examine a second and potentially feeding rates. decreased emigration, reduced fungal in- more common mechanism for indirect host mediation festations, construction of protective shelters. and pro- that involves the response of ants to herbivore excre- vision of parental care (see reviews by Hinton 1951, tions and secretions of variable composition and quan- Nixon 1951, Way 1963, Cottrell 1984, Beattie 1985, tity. Clearly, other indirect effects can also operate in Buckley 1987, Pierce 1987). such associations, but I will not discuss them because The excretions and secretions of tended herbivores they concern the ways in which host plants can in- provide an important food source for associating ant directly benefit from being infested by ant-tended ho- colonies (Way 1963, Carroll and Janzen 1973, Pierce et mopterans (Messina 1981, Buckley 1983, 1987, Fritz al. 1987, Fiedler and Maschwitz 1988. Nash 1989, Pierce 1983, Beattie 1985, Compton and Robertson 1988). 1989). The chemical composition of homopteran honey- dew closely parallels that of host-plant phloem fluids and contains a variety of sugars, organic acids, alcohols, plant hormones, salts, vitamins. amino acids, and Mechanisms and evidence amides (Auclair 1963, Dixon 1985, Klingauf 1987). Al- though not as well documented, lepidopteran secretions Two conditions must be met to document that host have also been shown to consist of numerous sugars and plants indirectly mediate the fitness of tended herb- amino acids (Maschwitz et al. 1975, Pierce 1984, 1987, ivores via ants. First, the attractiveness of herbivores to 1989. Fiedler and Maschwitz 1988). ants must vary predictably with intra- andlor interspec- ific changes in host-plant quality. Such variation will occur if a) the chemical composition andlor quantity of ant attractants produced by herbivores varies with Direct and indirect effects of plants on changes in host-plant quality and b) ants preferentially tend those herbivores that produce the most nutrition- tended herbivores ally rewarding attractants. Second, the number andlor Host plants can directly and indirectly affect the fitness attentiveness of ants must have a significant effect on the of their ant-tended herbivores. Phloem fluids directly fitness of herbivores. affect the survival, growth, and reproduction of homop- terans (Auclair 1963, Dixon 1985) and the chemical composition of leaf tissue directly influences the fitness Plant-mediated variation in herbivore of lepidopterans (Mattson 1980, Crawley 1983, Hill and Pierce 1989). Other plant characteristics, such as sur- attractiveness to ants face type and architecture, can directly affect the ability Influence of host plants on herbivore attractants of herbivores to acquire food resources (Juniper and The influence of plants on homopteran honeydew is Southwood 1986). well established. Many studies show that genotypic and While a number of studies evaluate the direct effects phenotypic variation in host plants (within individual of host plants on herbivores (Price et al. 1980, Crawley plants, within species, and among species) can affect the 1983, Strong et al. 1984), few have considered the in- amount of honeydew produced as well as the identity direct effects of host plants on herbivore fitness via ant and concentration of sugars, amino acids, and other associates. Thcre are at least two mechanisms by which compounds in these excretions (Mittler 1958a.b, host plants can have such effects. First, as noted by Auclair 1963, Dixon 1985, Holt and Wratten 1986, Buckley (1987), host plants bearing extrafloral nectaries Klingauf 1987). can mediate herbivore-ant interactions by attracting By comparison, the relationship between host-plant quality and the composition and quantity of lepidop- tending ants and herbivore fitness. While many species teran secretions is poorly understood. To my knowl- of homopterans and lepidopterans benefit from ant edge, only one study h a ~ ~ ~ r o v i dieredc t evidence that tending and have evolved a variety of morphological host plants influence secretion characteristics. Fiedler and behavioral traits which encourage this association (1990)showed that the ability of a lycaenid (Polyomma- (reviews by Way 1963, Blackman 1974, Heie 1980, At- rus icurus) to secrete ant attractants was significantly satt 1981a. Cottrell 1984, Pierce 1984,1985,1987,Buck- reduced when their larvae fed on a woody host plant ley 1987, Sudd 1987). the precise relationship between (Robinia pseudacacia). Baylis (1989) and Baylis and ant tending level and herbivore fitness is less clear (e.g.. Pierce (in press) have shown that larvae of the lycaenid are 10 ants per aggregation more beneficial to the herb- Jalmenus evagorus are more attractive to ants when ivores than 6?). This situation results from a strong bias feeding on fertilized host plants. Presumably. the mech- towards performing ant-exclusion rather than ant-re- anism for this result is that larvae are able to produce duction and ant-enhancement experiments. This prac- more nutritionally rewarding secretions when feeding tice seems particularly surprising given that the com- on high-quality host plants. Pierce (1984, 1985) also plete exclusion of ants rarely occurs in nature. More hypothesized such effects. realistically, the number of ants tending herbivores var- ies in space and time in response to numerous factors, Variation in urrractants and ant recruitment including plant-induced variation in the excretions and Numerous studies support the contention that ant secretions of herbivores. recruitment varies predictably with differences in the Two studies have shown that ant tending levels signif- composition and quantity of attractants produced by icantly affect homopteran fitness. First. Cushman and tended herbivores. Many of these studies demonstrate Addicott (1989) presented correlative data suggesting that ants can detect variation in the identitv and concen- that colonies of the aphid Aphis varians acquired great- tration of sugars and amino acids in artificial baits (Hangartner 1969. Taylor 1977, Kiss 1981, Crawford and Rissing 1983. Lanza and Krauss 1984, Sudd and Sudd 1985, Pierce 1989, J . H. Cushman, unpubl. data). Four studies provide more direct evidence that the recruitment response of ants is strongly influenced by the nutritional content and quantity of herbivore attractants. Working with the lycaenid Jalmenus evago- rus. Pierce (1984. 1989) detected a positive correlation between the number of ants tending pupae and the amino acid concentrations of their secretions (espe- cially serine). Edingcr (1985)found that the ant Campo- notus noveboracensis was more aggressive when tending large colonies of the aphid Chaitophorus populicola, presumably because these aggregations produced larger quantities of honeydew. In addition. C. noveboracensis more aggressively tended colonies of C. populicola than other aphid species producing less rewarding honey- dew. Third. Fiedler (1990) showed that larvae of the lycaenid Polyommarus icarlts were more attractive to m" LOW sPI sp2 HIGH ants when feeding on herbaceous legumes than a woody Nutritional content andlor quantity species. Fourth, Bristow (in press) has shown that feed- of herbivore attractants ing sites used by aphid colonies strongly influence their ability to attract ants. Colonies of the aphid Aphis nerii Fig. 1. Proposed relationships for how plant-induced variation on oleander (:Veriuin oleander) were more frequently in the nutritional content and quantity of rewards provided to ants. in association with other factors, predicts the behavior of tended by ants when feeding on flower buds than on leaf ants towards tended herbivores. When herbivores feed on low- tips. I'his pattern occurred within individual host plants, quality host plants and produce rewards of low nutritional independent of aphid colony size, and throughout the content andlor quantity, ant colonies are predicted to acquire season. greater fitness by acting as predators than as mutualists. Al- ternatively, when herbivores feed on high-quality hosts and produce rewards of high nutritional content and/or quantity, ant colonies are predicted to gain more by acting as mutualists. The first switchpoint (SP,) represents a hypothetical level of Ant tending levels and herbivore fitness reward at which ant colonies act as predators of herbivores feeding on host plants far away from their nests and mutualists Variation in herbivore attractants and recruitment be- of herbivores on hosts close to their nests. The second switch- havior of ants is important only if there is a positive point (SP?) represents the level of reward after which ant correlation between the number or attentiveness of colonies with high protein demand act as mutualists. er benefit (presumably protection from natural ene- Implications of host-plant mediation mies) when the ants Formica cinerea and E fusca were in greater abundance. Second, by performing ant-re- If host plants are significant mediators of insect interac- duction experiments, Cushman and Whitham (in press) tions, then there are a number of testable predictions demonstrated that greater numbers of the membracid that can be made which have important implications for Publiliu rnodesta survived and matured into adults when the ecology and evolution of herbivore-ant associations. tended by increased numbers of the ant F: altipetens. The first prediction based on the host-mediation hy- Greater membracid fitness was associated with reduced pothesis is that the indirect influence of host plants will abundance of the predatory salticid spider Pellenes sp. vary with the degree to which herbivores depend on Information on the relationship between lepidopte- ants. Effects should be minimal for herbivores that form ran fitness and the density of tending ants is less clear. facultative associations with ants and influential for The only direct evidence comes from Baylis (1989) and more obligate species. Second, as emphasized by Cush- Baylis and Pierce (in press) who demonstrated that the man and Addicott (1989, in press). host plants should lycaenid Jalmenus evagorus acquires greater benefit influence the ability of tended herbivores to compete from an increased abundance of tending ants (Iridomyr- with neighboring plants and herbivores for ant mu- nlex sp.). In addition, Pierce (1984. 1985) hypothesized tualist~. that the fitness of lycaenid larvae in general is increased The predictions that I discuss in this section suggest by attracting more tending ants. that host-plant selection by ant-tended herbivores will be an important and complex process. Herbivores asso- ciating with ants must select hosts that simultaneously maximize their nutritional intake and their ability to Overall evidence of host-plant mediation attract ants. Indirect effects of host plants on habitat While many studies have addressed various aspects of selection should be especially important for herbivores the hypothesis that host plants mediate herbivore-ant with extreme dependencies on ants, such as the scale interactions (see conditions above), only one study pro- Saissetia zanzibarensis (Way 1954) and the lycaenid Jal- vides clear evidence for this possibility in a single herb- menus evagorus (Pierce 1987). For example, as men- ivore-ant association. With field experiments, Baylis tioned earlier. female J. evagorus prefer to oviposit on (1989) and Baylis and Pierce (in press) documented that fertilized hosts. as such hosts make their larvae more 1) fifth-instar larvae of the lycaenid Jalmenus evagorus attractive to ants (Baylis 1989, Baylis and Pierce in attract a greater number of ants when feeding on fertil- press), and tended lycaenids in general exhibit strong ized host plants ( decurrens) than on unfertilized preferences for nitrogen-rich host plants (Pierce 1984. controls and 2) larvae attracting more ants survive bet- 1985). ter than larvae receiving less attention. Baylis and The fourth prediction of the plant-mediation hypo- Pierce also demonstrated that J. evagorus more com- thesis concerns variability in the strength and sign of monly oviposited on fertilized host plants (see discus- herbivore-ant interactions. The behavior of ants to- sion in Pierce 1989, Pierce et al. in press). Although wards herbivores will be determined in part by the these results were obtained by manipulating host-plant nutritional content and quantity of rewards that they quality experimentally, Auclair's (1963) review of the receive from herbivores. I predict that differences in the literature suggests that similar variation also occurs nat- rewards will cause significant variation in the strength urally. and sign of herbivore-ant interactions (i.e., whether An additional study provides less direct but more ants act as mutualists or predators). In Fig. 1. I present general support for the host-mediation hypothesis. a simple graphical model that illustrates these predicted Pierce (1984. 1985) presented data suggesting that host- relationships. When herbivores feed on low-quality host plant quality has strongly influenced the evolution of plants and offer attractants of low nutritional content ant-tended lycaenids. She documented that tended ly- andlor quantity, ant colonies will acquire greater fitness caenid larvae usually feed on nitrogen-fixing and/or pro- by acting as predators rather than mutualists. Conver- tein-rich host plants whereas non-tended species do not sely, when herbivores feed on high-quality host plants exhibit such patterns. Pierce goes on to hypothesize that and the nutritional content and/or quantity of their re- selection has favored the use of protein-rich hosts by wards is high, ant colonies will gain more by acting as tended lycaenids because their larvae (and sometimes mutualists (i.e., tending herbivores and harvesting their pupae) produce secretions that are more attractive to excretions/secretions). ants and that the increased attention results in greater The value of herbivore attractants to ant colonies will fitness. be a complex function of various factors, in addition to host-mediated variation in the nutritional content and quantity of attractants. These include the distance of tended herbivores from servicing ant colonies (McEvoy 1979, Sudd 1983), nutritional requirements of ant colo- nies, age and size of ant colonies, and the availability of alternative food sources, such as insect prey, extrafloral pendent upon the physical and biological setting in nectar (Buckley 1983), and the attractants produced by which they occur (Thompson 1988, Cushman and neighboring herbivores (Cushman and Addicott 1989, Whitham 1989, Cushman and Addicott in press). Here, Cushman and Whitham in press). These factors should I have argued that host plants represent an important interact with the content and quantity of rewards to and largely overlooked form of biological setting that change the position of curves predicted in Fig. 1, varies sufficiently to alter the outcome of herbivore-ant thereby producing a more complex range of ant behav- interactions, potentially explaining significant variation iors. For example, I predict that the curve will shift in the occurrimce and strength of mutualism in these down when tended herbivores feed on host plants that associations. are far away from servicing ant nests and shift up when infested hosts are close to nests. Thus, when herbivores Acknowledgements - I thank J. Brown. A. Harvey, L. Hughes. feed on host plants that result in the reward content and J. Maschinski, N. Moran, P. Price, V. Rashbrook, and espe- cially J. Addicott, A . Beattie, and T. Whitham for valuable quantity specified by the first switchpoint in Fig. 1(SP,), discussion and comments regarding this project. My work has the model predicts that ants will act as predators of been supported by funds from the National Science Founcia- herbivores on hosts located far away from their nest and tion (BSR-86-04983 to T. G . Whitham and BSR-8815544 to mutualists of herbivores on hosts close to their nest. In J.H.C.), U.S. Department of Agriculture (84-CRCR-1-1443 to T.G.W.), NATO Postdoctoral Fellowship Program, and Mac- addition, I predict that ant colonies with a high demand quarie University. for protein will act as mutualists only at extremely high content andior quantity of rewards (after SP?). Ants can switch between acting as predators and mu- tualist~o f tended homopterans, although variation in host-plant quality has not been considered to be the References driving mechanism for these switchpoints (Way 1963, Sudd 1987). Pontin (1958, 1978) reported that colonies Addicott, J. F, 1984. Mutualistic interactions in populations and communities. - In: Price, P. W., Slobodchikoff. C. N. of the ant Lasius flavus rely on the honeydew of root- and Gaud, W . S. (eds), A new ecology: novel approaches feeding aphids during early sunimer but then prey on to interactive systems. Wiley, New York, pp. 4371155. the same aphid species latter in the growing season. Atsatt. P. R. 1981a. Lycaenid and ants: selection for Although not evaluated by Pontin, this switch in behav- enemy-free space. - Am. Nat. 118: 638454. - 1981b. Ant-de~endenfto od selection bv the mistletoe but- ior may be explained by seasonal changes in the honey- terfly Ogyris dmary/lis (Lycaenidae). :Oecologia (Berl.) dew requirement of ant colonies and/or plant-induced 48: 60-63. changes in the content or quantity of honeydew. Auclair. J. L. 1963. Aphid feeding and nutrition. - Ann. Rev. Ent. 8: 439490. Edinger (1985) provides additional evidence for be- Baylis, M. 1989. The role of nutrition in an ant-lycaenid-host havioral switchpoints in ants. He found that the ant plant interaction. - Ph.D. Thesis. Univ. of Oxford, Oxford. Lasius rzeoniger preyed on a tended aphid species pro- - and Pierce, N. E . In press. The effect of host plant quality ducing small quantities of honeydew more frequently on the survival of larvae and oviposition behaviour of an ant-tended lycaenid butterfly. - Ecol. Ent. than it preyed on the more productive aphid Chaitopho- Beattie. A . J . 1985. The evolutionary ecology of ant-plant mu- roils populicola. In addition, Edinger elegantly showed tualisms. - Cambridge Univ. Press, New York. that when C. populicola was experimentally deprived of Becarra, J. X. I. and Venable, D . L. 1989. Extrafloral nectar- phloem fluids (by girdling the vascular tissue of its ies: a defense against ant-Homoptera mutualisms? - Oikos 55: 276280. host), L. neoniger preyed on them twice as often as Bentley, B. L. 1077. Extrafloral nectaries and protection by controls. pugnacious bodyguards. - Ann. Rev. Ecol. Syst. 8: 407- Way (1954) also demonstrated that ants switch be- .a- . tween acting as homopteran mutualists and predators. Blackman. R . L. 1974. Aphids. - Ginn and Co., London. When honeydew appeared to be in excess of its needs, Bristow, C. M. In press. Why are so few aphids ant-tended? - In: Cutler, D . F. and Huxley. C. R. (eds). Ant-plant inter- the ant Oecophylla longinoda began to prey on the actions. Oxford Univ. Press. Oxford. scales (Saissetiu zanzibarensis) that it normally tends. In Buckley, R . 1983. Interaction between ants and membracid the laboratory. Way (1954) was also able to experi- bugs decrease growth and seed set of a host plant bearing mentally induce and suppress ant predation of scales by extrafloral nectaries. - Oecologia (Berl.) 58: 132-136. - 1987. Ant-plant-homopteran interactions. - Adv. Ecol. manipulating the availability of honeydew. When Res. 16: 53-85. honeydew was in abundance, ants acted as predators Carroll, C. R . and Janzen, D. H. 1973. The ecology of foraging while they acted as mutualists when honeydew was in ants. - Ann. Rev. Ecol. Syst. 4: 231-257. short supply. This result suggests that the model in Fig. Compton. S. G . and Robertson, H . G. 1988. Complex interac- tions between mutualists: ants tending homopterans protect 1 can be amended to include a fourth curve which never fig seed set and pollinators. - Ecology 69: 1302-1305. (or rarely) rises above the switchpoint line at zero. Cottrell, C. B. 1984. Aphytophagy in butterflies; its relation- When honeydew requirements of an ant colony become ship to myrmecophily. - Zool. J. Linn. Soc. 79: 1-57. saturated, no value of reward will be sufficient to deter Crawford, D. L. and Rissing. S. W. 1983. Regulation of recruitment by individual scouts of Formica oreas Wheeler workers from acting as predators. (Hymenoptera: Formicidae). - Insectes Sociaux 30: 177- The outcome of species interactions is strongly de- 183. Crawley, M. J . 1983. Herbivory. - Univ. of California Press, McEvoy, P. B. 1979. Advantages and disadvantages to group Berkeley, CA. living in treehoppers (Homoptera: Membracidae). - Misc. Cushman. J . H. and Addicott, J. F. 1989. Intra- and interspec- Publ. Ent. Soc. Am. 11: 1-13. ific competition for mutualists: ants as a limited and limiting Messina, F. J . 1981. Plant protection as a consequence of an resource for aphids. - Oecologia (Berl.) 79: 315-321. ant-membracid mutualism: interactions on goldenrod (Sol- - and Whitham, T. G. 1989. Conditional mutualism in a idago sp.). - Ecology 62: 1433-1440. membracid-ant association: temporal, age-specific, and Mittler, T. E . 1958a. Studies on the feeding and nutrition of density-dependent effects. - Ecology 70: 104&1047. Tuberolachnus salignus (Gmelin) (Homoptera, Aphidi- - and Addicott. J. F. In press. Conditional interactions in dae). 11. The nitrogen and sugar composition of ingested ant-plant-herbivore mutualisms. - In: Cutler, D. F. and phloem sap and excreted honeydew. - J. Exp. Biol. 35: Huxley, C. R. (eds). Ant-plant interactions. Oxford Univ. 7 4 8 4 . Press, Oxford. - 1958b. Studies on the feeding and nutrition of Tuherolach- - and Whitham, T . G. In press. Competition mediating the nus salignus (Gmelin) (Homoptera, Aphididae). 111. The outcome of a mutualism: protective services of ants as a nitrogen economy. - J. Exp. Biol. 35: 62fw638. limiting resource for membracids. - Am. Nat. Nash, D . R. 1989. Cost-benefit analysis of a mutualism be- DcVries. P. J. and Baker, I. 1989. Butterfly exploitation of an tween lycaenid butterflies and ants. - Ph.D. thesis, Univ. of ant-plant mutualism: adding insult to herbivory. - J. New Oxford, Oxford. York Ent. Soc. 97: 332-340. Nixon, G. E. J. 1951. The association of ants with aphids and Dixon, A. F. G . 1985. Aphid ecology. - Blackie, London. coccids. - Commonwealth Institute of Entomology. Lon- Edinger, B. B. 1985. Conditional mutualism in three aphid- don. tending ants. - Bull. Ecol. Soc. Am. 66: 168. Pierce, N. E. 1984. Amplified species diversity: a case study of Fiedler. K. 1990. Effects of larval diet on myrmecophilous an Australian lycaenid butterfly and its attendant ants. - qualities of Polyomn~arusi carus caterpillars (: In:. Vane Wright, R. I. and Ackery, P. R. (eds), The biol- Lycaediae). - Oecologia (Berl.) 83: 284-287. ogy of butterflies. Academic Press, London. pp. 197-200. - and Maschwitz. U. 1988. Functional analysis of the myr- - 1985. Lycaenid butterflies and ants: selection for nitrogen- mecophilous relationships between ants (Hymenoptera: fixing and other protein-rich food plants. - Am. Nat. 125: Formicidae) and lycaenids (Lepidoptera: Lycaenidae). - 888-895. Oecologia (Berl.) 75: 204206. - 1987. The evolution and biogeography of associations be- Fritz. R. S. 1983. Ant protection of a host plant's defoliator: tween lycaenid butterflies and ants. - In: Harvey, P. H. and consequence of an ant-membracid mutualism. - Ecology Partridge, L. (eds), Oxford surveys in evolutionary biology 64: 789-797. (Volume 4). Oxford Univ. Press. Oxford. pp. 89-116. Hangartner. W. 1969. Structure and variability of the indi- - 1989. Butterfly-ant mutualisms. - In: Grubb, P. J. and vidual odor trail in Solenopsis geminata Fabr. (Hymenop- Whittaker, J. B. (eds), Blackwell, Oxford, pp. 299-324. tera. Formicidae). - Z. Vergl. Physiol. 62: 111-120. - and Elgar, M. A. 1985. The influence of ants on host plant Heie, 0 . E . 1980. The Aphidoidea (Hemiptera) of Fennoscan- selection by Jalmenus evagorus. a myrmecophilous butter- dia and Denmark. I. - Scandinavian Science Press, Klam- fly. - Behav. Ecol. Sociobiol. 16: 209-222. penborg. Denmark. - , Kitching, R. L., Buckley, R. C., Taylor, M. F. J. and Ben- Herrera, C. M. 1988. Variation in mutualisms: the spatio-tem- bow, K. F. 1987. The costs and benefits of cooperation poral mosaic of a pollinator assemblage. - Biol. J. Linn. between the Australian lycaenid butterfly. Julmenus evago- Soc. 35: 95-125. rus, and its attendant ants. - Oecologia (Berl.) 21: 237-248. Hill. C. J. and Pierce, N. E. 1989. The effect of adult diet on - ,Nash, D. R., Baylis. M. and Carper, E. R. In press. Varia- the biology of butterflies. 1. The common imperial butter- tion in the attractiveness of lycaenid butterfly larvae to fly. Jalrnen~tse vagorus. - Oecologia (Berl.) 81: 249-257. ants. In: Cutler, D. F. and Huxley, C. R. (eds). Ant-plant Hinton, H. E. 1951. Myrmecophilous Lycaenidae and other interactions. Oxford Univ. Press. Oxford. Lepidoptera- a summary. - Proc. Lond. Ent. Nat. Hist. Pontin, A. J . 1958. A preliminary note on the eating of aphids Soc. 1949-50: 11 1-175. by ants in the genus Lasius. - Ent. Monthly Mag. 95: Hi)lldobler, B. and Wilson, E. 0. 1990. The ants. - Harvard 154-155. Univ. Press, Cambridge, MA. - 1978. The numbers and distribution of subterranean aphids Holt, J. and Wratten, S. D. 1986. Components of resistence to and their exploitation by the ant Lasiusflavus. - Ecol. Ent. Aphis fahae in faba bean cultivars. - Ent. exp. appl. 40: 3: 203-207. 3 5 3 0 . Price, P. W . , Bouton, C. E., Gross. P.. McPheron. B. A . and Jordano, P. 1987. Patterns of mutualistic interaction in pollina- Thompson. J. N. 1980. Interactions among three trophic tion and seed dispersal: connectance, dependence asym- levels: influence of plants on interactions between herb- metries, and coevolution. - Am. Nat. 129: 657477. ivores and their natural enemies. - Ann. Rev. Ecol. Syst. Juniper. B. E . and Southwood, T . R . E . (eds) 1986. Insects and 11: 41-65. the plant surface. - Edward Arnold, London. Smiley. J. T.. Atsatt, P. R. and Pierce. N. E . 1988. Local distri- Kiss. A. 1981. Melezitose, aphids and ants. - Oikos 37: 382. bution of the lycaenid butterfly. Julmenus evagorus, in re- Klingauf. F. A . 1987. Feeding. adaptation and excretion. - In: sponse to host ants and plants. - Oecologia (Berl.) 76: Minks. A. K. and Harrewijn. P. (eds), Aphids: their biol- 41&422. ogy. natural enemies and control (Volume A). Elsevier, Strong. D. R.. Lawton, J. H. and Southwood, R. 1984. Insects New York. pp. 225-253. on plants: community patterns and mechanisms. - Harvard Lanza. J. and Krauss. B. R. 1984. Detection of amino acids in Univ. Press, Cambridge, MA. artificial nectars by two tropical ants, Leptothorax and Sudd. J. H. 1983. The distribution of foraging wood-ants (For- Monomorium. - Oecologia (Berl.) 63: 423-425. mica lugubris Zett.) in relation to the distribution of aphids. Maschwitz. U.. Wust, M. and Schurian, K. 1975. Blaulings- - Insectes Sociaux 30: 298-307. rupen als Zuckerlieferanten fiir Ameisen. - Oecologia - 1987. Ant aphid interaction. - In: Minks, A . K . and Harre- (Berl.) 18: 17-21. wijn, P. (eds), Aphids: their biology, natural enemies and - . Schroth, M., Hanel, H . and Pong, T. Y. 1984. Lycaenids control (Volume A). Elsevier. New York. pp. 335-365. parasitizing symbiotic plant-ant partnerships. - Oecologia - and Sudd. M. E . 1985. Seasonal changes in the response of (Berl.) 64: 78-80. wood-ants to sucrose baits. - Ecol. Ent. 10: 89-97. Mattson. W. J. 1980. Herbivory in relation to plant nitrogen - and Franks. N. R. 1987. The behavioural ecology of ants. - content. - Ann. Rev. Ecol. Syst. 11: 119-161. Blackie, London.

OIKOS 6 1 i ~IYYI) 143 Taylor. F. 1977. Foraging behavior of ants: experiments with Way. M. J. 1954. Studies of the association of the ant Oeco- two species of myrmecine ants. - Behav. Ecol. Sociobiol. 2: phylla longinoda and the scale insect Saissetia zanzibaren- 147-167. sis. - Bull. Entomol. Res. 45: 113-134. Thompson. J.N. 1982. Interaction and coevolution. - Wiley, - 1963. Mutualism between ants and honeydew-producing New York. Homoptera. - Ann. Rev. Ent. 8: 307-344. - 1988. Variation in interspecific interactions. - Ann. Rev. Wilson, E . O . 1971. The insect societies. - Harvard Univ. Ecol. Syst. 19: 65-87. Press. Cambridge, MA.