View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by RERO DOC Digital Library

Ecology, 81(12), 2000, pp 3522±3529 ᭧ 2000 by the Ecological Society of America

THE EFFECTS OF MUTUALISTIC ANTS ON APHID LIFE HISTORY TRAITS

THOMAS FLATT AND WOLFGANG W. W EISSER1 Zoology Institute, University of Basel, Rheinsprung 9, 4051 Basel, Switzerland

Abstract. The relationship between homopterans and ants is generally thought to be mutualistic, as both partners seem to bene®t from an association. In aphids, previous studies have shown that ant tending improves the survival and reproduction of aphid colonies, mainly by protection of aphids from enemy attack. However, the effects of ant tending on the ®tness of individual aphids have rarely been addressed. We investigated the effects of ant tending on life history traits of aphids feeding singly on a host plant, in the absence of natural enemies. A factorial design allowed us to control for variation in the level of tending effort among individual ant colonies. The presence of workers of the ant Lasius niger had a strong positive effect on the ®tness of individuals of the aphid Metopeurum fuscoviride. Ant-tended individuals lived longer, matured earlier, had a higher rate of re- production, and a higher expected number of offspring than aphids not tended by ants. An aphid's longevity was signi®cantly correlated with the daily mean number of workers tending it. The strong dependence of aphid ®tness on the level of ant tending shows that ants can in¯uence aphid life history traits even when aphids occur singly on plants. Key words ant; ant tending; aphid; Homoptera; honeydew; Lasius niger; Metopeurum fusco- viride; mutualistic relationship; myrmecophily.

INTRODUCTION evagoras, caterpillars are protected against parasitoid Many species of phytophagous insects such as ho- wasps and arthropod predators by Iridomyrmex ants mopterans (i.e., aphids, coccids, and membracids) and (Pierce et al. 1987). Similarly, protection from preda- lepidopteran larvae (lycaenids and riodinids) are mu- tors and parasitoids is likely to be the most important tualistically associated with ant species (e.g., Herzig bene®t for the survival of tended aphid colonies since 1937, Nixon 1951, Way 1963, Carrol and Janzen 1973, predators are capable of rapidly decreasing or even Pierce and Easteal 1986, Buckley 1987a, b,HoÈlldobler eliminating whole colonies when not checked by ants and Wilson 1990). Tended herbivores are generally as- (Banks 1962, Way 1963, Buckley 1987a, b, Dixon sumed to be important food resources for many ant 1998). species, because their secretions contain energy-rich In experiments in which aphid colonies were reared nutrients (Carrol and Janzen 1973, HoÈlldobler and Wil- in the presence and absence of ants, the ant mutualists son 1990). For instance, tending ants regularly feed bene®cially affected life history characteristics of their upon the carbohydrate-rich ``honeydew'' of homopter- aphid hosts. Tending promotes reproductive perfor- ans or upon sugars and amino acids produced in the mance (e.g., El-Ziady and Kennedy 1956, Banks 1958, exocrine dorsal nectary glands of some myrmecophi- El-Ziady 1960; Buckley 1987a, b), faster developmen- lous butter¯y larvae (e.g., Buckley, 1987a, b, Fiedler tal rates (e.g., El-Ziady 1960), or colony growth (e.g. and Maschwitz 1988). Several studies have provided El-Ziady and Kennedy 1956, Buckley 1987a, b). Mech- evidence that ants bene®t from associations with her- anistically, ant-tended aphids may bene®t from in- bivores in terms of energy gain, which is thought to creased feeding and excretion rates (e.g., Herzig 1937, result in higher colony growth rates (Degen et al. 1986, Banks and Nixon 1958, Mittler 1958, Takeda et al. Pierce et al. 1987, Fiedler and Maschwitz 1988, Cush- 1982) and from the removal of their honeydew, which man and Beattie 1991). In turn, ants often act as guards reduces the mortality risk caused by fungal attack (Nix- and decrease the impact of predators and parasitoids on 1951, Buckley 1987a, b). on the ®tness of their hosts (El-Ziady and Kennedy For ant tending to be favored by selection, net costs 1956, Banks 1962, Way 1963, Banks and Macaulay to tended aphids need to be smaller than bene®ts. Al- 1967, Addicott 1979, Pierce and Mead 1981, Buckley though most studies have found an overall positive 1987a, b,VoÈlkl 1992). Ant tending may therefore con- effect of tending ants on aphids at the level of the siderably reduce mortality risks and thus provides a population, a recent study suggests that ant tending may selective advantage. In the lycaenid butter¯y Jalmenus also be costly for aphids (Stadler and Dixon 1998). In the absence of natural enemies, tended Aphis fabae Manuscript received 1 March 1999; revised 22 November feeding on plants in small groups of 10±15 individuals 1999; accepted 2 December 1999. suffered from longer developmental times and smaller 1 Corresponding author. Present address: Institute of Ecol- ogy, Friedrich-Schiller-University, Dornburger Strasse 159, growth rates, delayed offspring production, smaller go- 07743 Jena, Germany. nads, and fewer embryos. Recent studies also suggest 3522 December 2000 APHID±ANT MUTUALISM 3523 that bene®ts of mutualistic associations are density- (Drosophila melanogaster), and arti®cial ant diet dependent (Bronstein 1994). For example, the positive (``food cubes''; Keller et al. 1989). Ant colonies were effect of ant tending on the growth rate of aphid pop- regularly sprayed with tap water to avoid dessiccation. ulations has been shown to decline at higher aphid Ramets of the host plant (T. vulgare) were individually densities (e.g., Breton and Addicott 1992a). These re- planted in a commercial growing medium (TKS 2, Flor- sults have two implications. First, the results suggest agard VertriebsGmbH, D-26129 Oldenburg, Germany) that bene®ts at the population level may not re¯ect in 360-mL pots several weeks prior to the start of the bene®ts at the individual level. On the one hand, ben- experiment. e®ts measured in terms of population growth rate are a secondary result of the underlying variation in in- General experimental design dividual-level ®tness, and it is generally dif®cult to Thirty-two experimental plants (height 20±40 cm) recover this underlying variation from population level were placed on four tables (ϭ blocks; eight replicates measures. On the other hand, even if life history traits per block, table size ϭ 150 ϫ 100 cm). The plants were are analyzed at the population level, density-depen- each placed on a piece of foam rubber (30 ϫ 30 cm) dence may confound measurements of individual-level and enclosed in a transparent plastic tube (25 cm in ®tness. Second, the results suggest that ant tending can diameter, 63 cm in height). The tops of these tubes be costly at the individual level in the absence of nat- were tightly covered with ®ne gauze mesh to prevent ural enemies. the escape of animals. On each table, we randomly In this paper, we investigate the effect of the presence assigned four plants to the ant-tending treatment and of honeydew-collecting workers of the ant Lasius niger four to the control (ϭant-exclusionϭ untended). The on life history traits and ®tness of individuals of the position of the plants on the tables was chosen prior aphid Metopeurum fuscoviride in the absence of natural to allocating the plants to treatment or control. Each enemies. Because of density-dependence in the inter- of the four ant colonies was assigned to one of the four action between aphid and ants, it is important to ex- tables. Workers of L. niger had access to the plants of amine the effects of ant tending on aphid ®tness not the ant-tended treatment via transparent plastic tubes only for aphid populations of various sizes, but also (1 cm in diameter,2moftube per table) that connected for individuals feeding singly on a host plant. We ad- the container housing the ant colony to the four cages. dress the question of whether ant tending is bene®cial The four cages containing the replicates from the con- at the individual level for aphids feeding singly on a trol were not connected to the ant colony. host plant, in an aphid species that is usually found The experiment was started on 24 May 1998 by as- tended by ants in the ®eld. signing one winged adult female aphid collected from plants growing outside in the garden of the Zoology METHODS Institute, Basel, to each of the 32 experimental plants. After they were allowed to reproduce for 24 h, the Study system mothers and all but one sibling were removed, leaving For our experiments we chose Metopeurum fusco- only a single offspring on each of the 32 plants. Thus, viride Stroyan, a monophagous, holocyclical aphid spe- we investigated the effects of ant tending on life history cies of the common tansy, Tanacetum vulgare L. (As- traits of single unwinged individuals that were born on teraceae), a perennial herbaceous composite from Eu- a plant, by a winged mother. The experiment was car- rope and Asia (BoÈrner 1952, Klausnitzer 1968, Mitich ried out in a climate chamber with a light:dark cycle 1992). M. fuscoviride is regularly tended by ants, of of 18:6 h, a temperature of 21ЊϮ2ЊC (mean Ϯ 1 SD) which Lasius niger (L.) is the most common (Mackauer and an approximate relative air humidity of 65%. Light and VoÈlkl 1993, Fischer 1997; W. W. Weisser, personal intensity at plant height was ϳ5000 lux. observation). Untended colonies of M. fuscoviride can regularly be found in the ®eld, particularly in places Removal of aphids by ants with moist soils that do not support colonies of the For our analysis of aphid life-span we assumed that xerothermic L. niger or other ant species (e.g., Mack- aphids not found in the enclosure had died. No escape auer and VoÈlkl 1993). We used four ant colonies col- of aphids was possible from enclosures not connected lected at a ®eld site near Basel, Switzerland, in early to the ant colonies. In the ant-tended replicates it was May 1998. Each colony had several hundred workers. possible for foraging ants to remove dead aphids. To All colonies were kept in plastic nest boxes (45 ϫ 34 test whether ants would remove dead aphids and carry ϫ 10.5 cm) or buckets (10 L in volume) coated with them to their nests we performed the following exper- ¯uon (Northern Products, Incorporated, Woonsocket, iment. We offered each ant colony ®ve bodies of M. Rhode Island, USA) and talcum powder, and ®lled with fuscoviride and, as a control, ®ve bodies of the pea natural substrate. All colonies had access to plants host- aphid Acyrtosiphon pisum (Harris) in a petri dish. Over ing M. fuscoviride aphids to provide ants with suf®cient a period of two hours we counted all remaining aphids, carbohydrate. Ants were regularly fed with mealworms in 30-min intervals. In our experiment, we could not (larvae of Tenebrio mollitor), freshly killed fruit ¯ies exclude the possibility that ants would incidentally kill 3524 THOMAS FLATT AND WOLFGANG W. WEISSER Ecology, Vol. 81, No. 12 aphids and carry them to their nest. However, this was (e.g., Fischer 1997). We calculated the rate of reaction never observed and we consider this possibility highly to disturbances of a replicate as the total number of unlikely, because ant colonies were regularly provided such reactions observed divided by aphid life span. If with sources of protein in our experiment (e.g., Way an aphid dropped off the plant it was replaced to its 1963). previous location. Effect of origin on experimental animals Statistical analysis We assumed that there were no signi®cant a priori Statistical analyses and signi®cance tests (P Ͻ 0.05) differences among the winged mothers we used to ob- were performed by following procedures described in tain experimental animals. We tested this assumption Sokal and Rohlf (1995) and using SAS v.6.12 (SAS by measuring the fresh body mass and hind tibia length Institute 1989). Prior to analysis, all data were tested of all mothers. for homogeneity of variances using the Fmax test. We checked for normality of residuals by using the pro- Aphid life history traits cedure UNIVAR implemented in SAS. If the assump- Each individual aphid was checked daily at the same tions were violated, we transformed the data as nec- time, over its whole life-span. For each individual we essary. We analyzed the effect of ant tending on aphid recorded survival and the number of offspring pro- life history traits and ®tness by performing two-way duced during the last 24 h. Any offspring produced was ANOVAs using Type III sums of squares for data from carefully removed from the plant using a ®ne brush. our randomized-block design. Ant tending (presence/ From the data we calculated age at maturity (ϭage at absence) entered the analysis as a ®xed effect, whereas ®rst reproduction, Stearns 1992), life-span (ϭage at blocks (ϭtables) were treated as a random effect. In death), lifetime fecundity and rate of reproduction our experimental design we could distinguish between (ϭlifetime fecundity/(age at death Ϫ age at maturity)). ant colony effects and other blocking (table) effects Data were collected over 48 consecutive days, that is only by comparing the performance of ant-tended and until the last aphid in the experiment had died. untended aphids separately. In the two-way ANOVAs we therefore computed main effects only. We used one- Level of ant tending way ANOVAs when separately analyzing the effects To quantify the level of ant tending we counted the of ant colony or table identity. number of ant workers tending the experimental aphid RESULTS daily when we controlled aphid survival and fecundity. For each replicate we calculated the ant tending rate Removal of aphids by ants as the number of ant workers tending an aphid averaged Workers in each of the four ant colonies removed all over the lifetime of the aphid. offered aphid bodies (5 M. fuscoviride, 5 A. pisum) within two hours of observation. Thus, foraging ants Honeydew removal are likely to remove dead aphids from an enclosure, To quantify the effect of ants on the removal of hon- suggesting that ant-tended aphids not found in the en- eydew we daily checked whether a honeydew droplet closure had died and were carried to the ant colony. was visible at the cauda of an aphid or somewhere on the plant close to the aphid. We calculated the rate at Effect of origin of experimental animals which honeydew droplets were found by dividing the There were no signi®cant differences among the total number of honeydew droplets observed during the winged mothers of experimental aphids with respect to lifetime of an individual by aphid life span. their body mass (ants: F1,22 ϭ 0.18, P ϭ 0.68; blocks: F ϭ 0.76, P ϭ 0.53) or hind tibia length (ants: F Movements and disturbances of aphids 3,22 1,21 ϭ 0.32, P ϭ 0.58; blocks: F3,21 ϭ 0.19, P ϭ 0.90). To investigate whether ant tending would affect the Thus, there were no systematic differences among the movements of the aphids on the plants we made daily animals used in the experiment that were due to dif- records of the exact position of an aphid on a sketch ferences among mothers. of the host plant. For each replicate we calculated the rate of movement as the total number of movements Effects of ant tending on aphid life history traits observed divided by life span. We predicted that un- Life-span.ÐAnt-tended aphids lived on average nine tended aphids would change their feeding position on days longer than untended aphids (Fig. 1A, ants: F1,27 the plant more often than tended ones because untended ϭ 6.45, P ϭ 0.017; blocks: F3,27 ϭ 1.64, P ϭ 0.20). M. fuscoviride cannot dispose of honeydew by kicking There were no signi®cant differences among ant col- the droplets away using a hind leg. Additionally, we onies in the tending rates of aphids (F3,12 ϭ 1.68, P ϭ recorded if an aphid moved to a different feeding site 0.23, ant-tending rate log-transformed). Aphid life span or dropped off the host plant as a reaction to the dis- was signi®cantly correlated with ant-tending rate in turbances caused by the daily checks. M. fuscoviride tended aphids (Fig. 2, r ϭ 0.80, n ϭ 16, P Ͻ 0.001, is known to have a well-developed dropping re¯ex rate log-transformed).

3526 THOMAS FLATT AND WOLFGANG W. WEISSER Ecology, Vol. 81, No. 12

Life-span was signi®cantly correlated with lifetime (rate of reaction to disturbance in ant-tended treatment: fecundity (rS ϭ 0.86, n ϭ 18, P Ͻ 0.001; reproducing 0.14 Ϯ 0.12, untended treatment: 0.086 Ϯ 0.10 move- aphids only). The correlation remained signi®cant ments/d, ants: F1,27 ϭ 1.97, P ϭ 0.17; blocks: F3,27 ϭ when tended and untended aphids were analyzed sep- 3.97, P ϭ 0.38). arately (ant-tended replicates: rS ϭ 0.88, n ϭ 10, P ϭ DISCUSSION 0.005; untended replicates: rS ϭ 0.79, n ϭ 8, P ϭ 0.02; reproducing individuals only). Life-span was not sig- Bene®ts of mutualism should ultimately be measured ni®cantly correlated with rate of reproduction when in terms of a direct ®tness measure such as the number both tended and untended aphids were considered (rS of offspring produced (e.g, Cushman and Beattie 1991). ϭ 0.16, n ϭ 18, P ϭ 0.52) and when untended aphids For aphids, the present study is the ®rst to follow single were analyzed separately (rS ϭ 0.07, n ϭ 8, P ϭ 0.87). individuals throughout their life, both in the presence However, the correlation was signi®cant and negative and the absence of tending ants. In the ®eld, recently for ant-tended replicates, suggesting that reproductive founded aphid colonies often consist of one or a few output decreased over the life span in longer-lived individuals, in particular when the winged foundress aphids (rS ϭϪ0.8, n ϭ 10, P ϭ 0.005). Thus, long- has died (W. W. Weisser, personal observation). The lived aphids had a higher fecundity than short-lived main result of our experiment is that ant tending pos- aphids, because they lived longer. itively affected all measures of aphid ®tness. Ant-tend- There was a signi®cant positive correlation between ed individuals of M. fuscoviride lived on average 78% lifetime fecundity and ant-tending rate in tended aphids longer, needed ϳ10% less time to mature, gave birth (r ϭ 0.86, n ϭ 16, P Ͻ 0.001). This result remained to offspring at a rate that was 88% higher, and had an signi®cant when only reproducing individuals were in- expected number of offspring that was more than ®ve cluded in the analysis (r ϭ 0.80, n ϭ 10, P ϭ 0.006; times higher than that of individuals not tended by ants. rates log-transformed in both analyses). The rate of These results are in accordance with previous studies reproduction was not signi®cantly correlated with ant- showing that associations with ants can confer ®tness tending rate (rS ϭϪ0.42, n ϭ 10, P ϭ 0.22; only bene®ts to tended aphid colonies (e.g., Herzig 1937, reproducing individuals). Thus, in ant-tended aphids El-Ziady and Kennedy 1956, Banks 1958, El-Ziady the higher lifetime fecundity of individuals that were 1960, Way 1963, Buckley 1987a, b, Bristow 1991, Dix- tended more intensively was a consequence of those on 1998, Stadler and Dixon 1999). aphids living longer, i.e., reproducing for a longer pe- Although the bene®ts of ant tending to aphid colo- riod of time. nies are well documented, the selection pressures act- Honeydew removal.ÐOn average 5.7 Ϯ 5.4 honey- ing on aphid individuals are poorly understood. Pre- dew droplets (mean Ϯ 1 SD) were observed during the vious studies have found that the main bene®t for lifetime of an untended aphid, whereas only 0.75 Ϯ 1.2 aphids was protection of aphid colonies from natural droplets were counted for ant-tended individuals. This enemies (e.g., Banks 1962, Way 1963, Banks and Ma- corresponded to mean rates of 0.36 Ϯ 0.26 and 0.05 Ϯ caulay 1967, Tilles and Wood 1982, Buckley 1987a, 0.10 droplets/day for untended and ant-tended individ- b, Dixon 1998). Experiments conducted in the absence uals, respectively (ants: F1,27 ϭ 17.74, P Ͻ 0.001; blocks: of natural enemies often failed to ®nd a positive effect

F3,27 ϭ 0.16, P ϭ 0.92). Honeydew droplets were ob- of ant tending on aphid reproduction or fecundity served only in six tended individuals and the correlation (Banks 1958, Takeda et al. 1982, Bristow 1984, Breton between aphid life span and the rate of honeydew ap- and Addicott 1992b). For the facultatively tended aphid pearance was not signi®cant in tended aphids (rS ϭ Aphis fabae cirsiiacanthoides, Stadler and Dixon Ϫ0.12, n ϭ 16, P ϭ 0.65). However, there was a sig- (1998) even reported a detrimental effect of ant tending ni®cant positive relationship between aphid life span and on ®tness-related traits such as developmental time or the rate at which honeydew droplets were found in un- mean relative growth rate of individuals. tended aphids (rS ϭ 0.64, n ϭ 16, P ϭ 0.007). There are several differences between previous stud- ies and our study. First, all previous experiments reared Movements and disturbances of aphids aphids in smaller or larger groups. Because the bene- There were no signi®cant differences in the rates of ®cial effects of ant tending on aphid ®tness may depend movement between ant-tended (0.33 Ϯ 0.16) and un- on the size of the aphid population (e.g., Breton and tended aphids (0.29 Ϯ 0.20 movements/d) (ants: F1,27 Addicott 1992a), effects of group size could not be

ϭ 0.57, P ϭ 0.46; blocks: F3,27 ϭ 1.0, P ϭ 0.41). This excluded in previous studies. The results of Breton and is in contrast to our expectation that ant tending reduces Addicott (1992a) suggest that ant tending may bene®t the number of movements of aphids on the plant. small aphid populations or individual aphids directly, The daily check of the replicates rarely caused a possibly through a stimulation of their feeding rate, measurable disturbance to the aphids. Aphids in the and that the decline of direct effects of ants on aphids ant-tended treatment moved or dropped off the host is probably due to a decline in the relative number of plant during checks on average only every seventh day, ants tending aphids at higher aphid densities. Second, and in untended replicates on average every 11th day different aphid species may respond differently to the December 2000 APHID±ANT MUTUALISM 3527 presence of ants. For instance, successful colonies of improving their feeding rate (e.g., Herzig 1937, Banks A. f. cirsiiacanthoides are often found in the absence and Nixon 1958, Mittler 1958, Takeda et al. 1982). of ants (Stadler and Dixon 1998), whereas ®eld ob- Correlational evidence for this hypothesis could be ob- servations suggest that M. fuscoviride depends more tained by measuring both feeding rates and life span on the presence of ants (W. W. Weisser, personal ob- in an experiment. Aphids may also feed more contin- servation). This might be a consequence of species- uously when tended by ants; the presence of ants may speci®c differences in investment into securing tending represent a ``tranquilizing'' cue for aphids (e.g., Way by ants (see discussion below). Third, selection pres- 1963), indicating, for example, that the risk of preda- sures exerted by natural enemies may differ for dif- tion is likely to be low. Finally, ant tending may reduce ferent aphid species. We can expect selection to favor the risk of fungal attack (Nixon 1951, Buckley 1987a, ant tending only if the costs to tended individuals are b). Although three untended aphids were visibly at- smaller than the bene®ts. If predation pressure is high, tacked by fungus originating from honeydew droplets the advantages of being protected by ant workers may prior to their death in the experiment, we were unable outweigh any negative effects ants might have on aphid to con®rm the hypothesis that untended aphids change life history in the absence of natural enemies (cf. Stad- their feeding sites more often than tended aphids to ler and Dixon 1999). In the case of M. fuscoviride, avoid close contact with excreted honeydew. Thus, it aphids bene®t from ant tending even in the absence of remains unclear whether protection from fungal attack natural enemies. Thus, to understand aphid±ant mu- is an important cause for improved survival of aphids tualism it is necessary to study the effect of ants on in the presence of ants. aphid ®tness both in the presence and absence of nat- The strong effect of L. niger on the life history traits ural enemies. Testable predictions concerning the effect of M. fuscoviride suggests that the aphids suffer from of ants on the life history of a particular aphid species strong ®tness losses if they are not tended by ants. will therefore depend on the importance of natural en- Because homopterans often have to compete for the emies in the life cycle of this species. services of ants (Addicott 1978, 1985, Cushman and Aphid individuals were regularly tended by up to Addicott 1989, 1991), M. fuscoviride is expected to four or ®ve ants at a time in our experiment. There invest in securing attendance by ants. A recent study were no signi®cant differences in the mean number of by VoÈlkl et al. (1999) shows that the honeydew of M. workers tending the aphids at the level of the ant col- fuscoviride differs signi®cantly both from the phloem ony. However, one interesting result is the strong pos- content of T. vulgare and from the honeydew of three itive correlation between the rate of ant tending and other aphid species that also live on tansy. M. fusco- aphid survival (Fig. 2). There was also a positive cor- viride produces more honeydew per unit of time, and relation between honeydew production and survival, its honeydew contains a high proportion of trisaccha- and between life span and lifetime fecundity for un- rides that are not found in the phloem sap but are strong tended aphids. These results are consistent with the attractants for workers of L. niger (Kiss 1981). In hypothesis that individual aphids vary in ``quality'' in choice tests with colonies of several aphid species pre- some way such that ``high-quality'' aphids feed more, sent, ant workers did preferentially visit plants with produce more honeydew, live longer and have a higher colonies of M. fuscoviride (VoÈlkl et al. 1999). Although lifetime fecundity than ``low-quality'' aphids. In the at present it cannot be excluded that the high rate of presence of ants, high-quality aphids are tended more honeydew production and the provision of trisaccha- intensively by ants, which may further improve their rides in the honeydew are by-products of physiological survival for some unknown reason. In our experiment, processes unrelated to the maintenance of mutualistic variation in aphid quality may have been caused by relationships, it is possible that they at least partly rep- variation in feeding rates, feeding sites on plants, plant resent an investment of M. fuscoviride into the rela- quality, or possibly some intrinsic differences among tionship with attending ants. individuals. To identify the underlying causes of the Current theory suggests that mutualistic interactions variation in aphid quality, experiments could be per- are best viewed as reciprocal exploitations that nev- formed that assess the quality of the phloem sap and ertheless provide net bene®ts to each partner (Bronstein other correlates of host plant quality. Aphid feeding 1994, Herre et al. 1999). To unravel the con¯ict of rate and the production, composition, and concentra- interests in mutualistic systems requires an identi®- tion of the excreted honeydew of individuals could also cation of the costs and bene®ts at a biologically relevant be measured (e.g., VoÈlkl et al. 1999). However, the scale of organisation, which, in most cases, is the level proximate causes of improved survival due to the level of the individual (Boucher et al. 1982, Boucher 1985, of ant tending may be more dif®cult to identify, because Templeton and Gilbert 1985, Herre et al. 1999). The quantitative measures of internal physiological param- dependence of aphid ®tness on the level of ant tending eters may have to be estimated. in addition to the possibility that aphids have to com- Tending may improve aphid survival for several rea- pete for the services of ants by de novo synthesis of sons. Removal of honeydew by ants possibly allows ant-attracting substances (cf. VoÈlkl et al. 1999), does, individuals of M. fuscoviride to increase life span by however, also point to a potential con¯ict of interests 3528 THOMAS FLATT AND WOLFGANG W. WEISSER Ecology, Vol. 81, No. 12 between the mutualistic partners. Whenever ants stop Pages 104±119 in C. R. Huxley and D. F. Cutler, editors. tending a colony because more pro®table sources of Ant±plant interactions. Oxford University Press, Oxford, UK. carbohydrates are available or the needs of the colony Bronstein, J. L. 1994. Conditional outcomes in mutualistic change, the abandoned aphids will suffer high ®tness interactions. Trends in Ecology and Evolution 9:214±217. costs. The costs associated with attracting ant workers Buckley, R. C. 1987a. Interactions involving plants, Ho- might be another reason why so few aphids are ant- moptera, and ants. Annual Review of Ecology and System- tended (Bristow 1991). This hypothesis would be sup- atics 18:111±135. Buckley, R. C. 1987b. Ant±plant±homopteran interactions. ported if future studies show that both the level of ant Advances in Ecological Research 16:53±85. tending and the ®tness gain to the tended individual is Carrol, C. R., and D. H. Janzen. 1973. Ecology of foraging a function of the investment on the part of the aphid. by ants. Annual Review of Ecology and Systematics 4:231± 257. ACKNOWLEDGMENTS Cushman, J. H., and J. F. Addicott. 1989. Intra- and inter- We thank Cesare Baroni Urbani and Else-Juliette Fjer- speci®c competition for mutualists: ants as a limited and dingstad for their advice and help in rearing ants, Marcel limiting resource for aphids. Oecologia 79:315±321. Kaiser for providing fruit ¯ies, Christian Braendle for help Cushman, J. H., and J. F. Addicott. 1991. Conditional inter- with the experiment, and Wolfgang VoÈlkl, Bernhard Stadler actions in ant±plant±herbivore mutualism. Pages 92±103 and Jean-Christophe Simon for stimulating discussions. Com- in C. R. Huxley and D. F. Cutler, editors. Ant±plant inter- ments by Dieter Ebert, Melanie Fischer, Deborah Gordon, actions. Oxford University Press, Oxford, UK. Tad Kawecki, Ian Sanders, John Sloggett, Bernhard Stadler, Cushman, J. H., and A. J. Beattie. 1991. Mutualisms: as- Steve Stearns, Wolfgang VoÈlkl, and two anonymous referees sessing the bene®ts to hosts and visitors. Trends in Ecology greatly improved the presentation of our results. T. Flatt was and Evolution 6:193±195. supported by the Emilia Guggenheim-Schnurr Foundation Degen, A. A., M. Gersani, Y. Avivi, and N. Weisbrot. 1986. and the Swiss Study Foundation, and W. W. Weisser was Honeydew intake of the weaver ant Polyrhachis simplex supported by the Novartis-Stiftung, the Roche Research (Hymenoptera: Formicidae) attending the aphid Chaito- Foundation, and the Swiss Nationalfonds (grant no. 3100± phorus populialbae (Homoptera: Aphididae). Insectes So- 053852.98). ciaux 33:211±215. LITERATURE CITED Dixon, A. F. G. 1998. Aphid ecology. Chapman and Hall, London, UK. Addicott, J. F. 1978. Competition for mutualists: aphids and El-Ziady, S. 1960. Further effects of Lasius niger L. on Aphis ants. Canadian Journal of Zoology 56:2093±2096. fabae Scopoli. Proceedings of the Royal Entomological So- Addicott, J. F. 1979. A multispecies aphid±ant association: ciety London (A) 35:30±38. density dependence and species-speci®c effects. Canadian El-Ziady, S., and J. S. Kennedy. 1956. Bene®cial effects of Journal of Zoology 57:558±569. the common garden ant, Lasius niger L., on the black bean Addicott, J. F. 1985. Competition in mutualistic systems. aphid, Aphis fabae Scopoli. Proceedings of the Royal En- Pages 217±247 in D. H. Boucher, editor. The biology of tomological Society London (A) 31:61±65. mutualism: ecology and evolution. Croom Helm, London, Fiedler, K., and U. Maschwitz. 1988. Functional analysis of UK. the myrmecophilous relationships between ants (Hyme- Banks, C. J. 1958. Effects of the ant, Lasius niger (L.), on noptera: Formicidae) and lycaenids (Lepidoptera: Lycaen- the behaviour and reproduction of the black bean aphid, Aphis fabae Scop. Bulletin of Entomological Research 49: idae). II. Lycaenid larvae as trophobiotic partners of ants 701±714. Ða quantitative approach. Oecologia 75:204±206. Banks, C. J. 1962. Effects of the ant Lasius niger (L.) on Fischer, M. 1997. Hierarchien im Mutualismus zwischen Las- insects preying on small populations of Aphis fabae Scop. ius niger und verschiedenen honigtauproduzierenden Blat- on bean plants. Annals of Applied Biology 50:669±679. tlausarten. Diploma thesis. University of Bayreuth, Ger- Banks, C. J., and E. D. M. Macaulay. 1967. Effects of Aphis many. fabae Scop. and of its attendant ants and insect predators Herre, E. A., N. Knowlton, U. G. Mueller, and S. A. Rehner. on yields of ®eld beans (Vicia faba L.). Annals of Applied 1999. The evolution of mutualisms: exploring the paths Biology 60:445±453. between con¯ict and cooperation. Trends in Ecology and Banks, C. J., and H. L. Nixon. 1958. Effects of the ant, Lasius Evolution 14:49±53. niger L., on the feeding and excretion of the bean aphid, Herzig, J. 1937. Ameisen und BlattlaÈuse. Zeitschrift fuÈr an- Aphis fabae Scop. Journal of Experimental Biology 35: gewandte Entomologie 24:367±435. 703±711. HoÈlldobler, B., and E. O. Wilson. 1990. The ants. Springer- BoÈrner, C. 1952. Aphidae Europae Centralis. Mitteilungen Verlag, Berlin, Germany. der ThuÈringer Botanischen Gesellschaft 4:1±184. Keller, L., D. Cherix, and P.Ulloa-Chacon. 1989. Description Boucher, D. H. 1985. The idea of mutualism, past and future. of a new arti®cial diet for rearing ant colonies as Irido- Pages 1±28 in D. H. Boucher, editor. The biology of mu- myrmex humilis, Monomorium pharaonis and Wasmannia tualism: ecology and evolution. Croom Helm, London, UK. auropunctata (Hymenoptera; Formicidae). Insectes So- Boucher, D. H., S. James, and K. H. Keeler. 1982. The ecol- ciaux 36:348±352. ogy of mutualism. Annual Review of Ecology and System- Kiss, A. 1981. Melizitose, aphids and ants. Oikos 37:382. atics 13:315±347. Klausnitzer, B. 1968. Zur Kenntnis der Entomofauna von Breton, L. M., and J. F. Addicott. 1992a. Density-dependent Tanacetum vulgare L. und Artemisia vulgaris L. Wissen- mutualism in an aphid±ant interaction. Ecology 73:2175± schaftliche Zeitschrift der Technischen UniversitaÈt Dresden 2180. 17:19±21. Breton, L. M., and J. F. Addicott. 1992b. Does host-plant Mackauer, M., and W. VoÈlkl. 1993. Regulation of aphidiid quality mediate aphid±ant mutualism? Oikos 63:253±259. wasps: Does aphidiid foraging behaviour or hyperparasit- Bristow, C. M. 1984. Differental bene®ts from ant attendance ism limit impact? Oecologia 94:339±350. to two species of homoptera on New York ironweed. Jour- Mitich, L. W. 1992. Intriguing world of weedsÐTansy. Weed nal of Animal Ecology 53:715±726. Technology 6:242±244. Bristow, C. M. 1991. Why are so few aphids ant-tended? Mittler, T. E. 1958. The excretion of honeydew by Tubero- December 2000 APHID±ANT MUTUALISM 3529

lachnus salignus (Gmelin). Proceedings of the Royal En- Stearns, S. C. 1992. The evolution of life histories. Oxford tomological Society London (A) 33:49±55. University Press, Oxford, UK. Nixon, G. E. J. 1951. The association of ants with aphids Takeda, S., K. Kinomura, and H. Sakurai. 1982. Effects of and coccids. Commonwealth Institute of Entomology, Lon- ant tending on the honeydew excretion and larviposition don, UK. of the cowpea aphid, Aphis craccivora Koch. Applied En- Pierce, N. E., and S. Easteal. 1986. The selective advantage tomology and Zoology 17:133±135. of attendant ants for the larvae of a lycaenid butter¯y, Glau- Templeton, A. R., and L. E. Gilbert. 1985. Population ge- copsyche lygdamus. Journal of Animal Ecology 55:451± netics and the coevolution of mutualism. Pages 128±144 462. in D. H. Boucher, editor. The biology of mutualism: ecology Pierce, N. E., R. L. Kitching, R. C. Buckley, M. F. J. Taylor, and evolution. Croom Helm, London, UK. and K. F. Benbow. 1987. The costs and bene®ts of coop- Tilles, D. A., and D. L. Wood. 1982. The in¯uence of car- eration between the Australian lycaenid butter¯y, Jalmenus penter ant (Camponotus modoc) (Hymenoptera: Formici- evagoras, and its attendant ants. Behavioral Ecology and dae) attendance on the development and survival of aphids Sociobiology 21:237±248. (Cinara spp.) (Homoptera: Aphididae) in a giant Sequoia Pierce, N. E., and P. S. Mead. 1981. Parasitoids as selective agents in the symbiosis between lycaenid butter¯y larvae forest. Canadian Entomologist 114:1133±1142. and ants. Science 211:1185±1187. VoÈlkl, W. 1992. Aphids or their parasitoids: Who actually SAS Institute. 1989. SAS/STAT user's guide: basics. Version bene®ts from ant tending? Journal of Animal Ecology 61: 6.04. SAS Institute, Cary, North Carolina, USA. 273±281. Sokal, R. F., and F. J. Rohlf. 1995. Biometry. W. H. Freeman, VoÈlkl, W., J. Woodring, M. Fischer, M. W. Lorenz, and K. New York, New York, USA. H. Hoffmann. 1999. Ant±aphid mutualism: the impact of Stadler, B., and A. F. G. Dixon. 1998. Costs of ant attendance honeydew production and honeydew sugar composition on for aphids. Journal of Animal Ecology 67:454±459. ant preferences. Oecologia 118:483±491. Stadler, B., and A. F. G. Dixon. 1999. Ant attendance in Way, M. J. 1963. Mutualism between ants and honeydew- aphids: why different degrees of myrmecophily? Ecological producing Homoptera. Annual Review of Entomology 8: Entomology 24:363±369. 307±344.