Two ways to be a myrmecophilous butterfly: natural history and comparative immature‐stage morphology of two species of Theope (Lepidoptera: Riodinidae)

Symbiotic interactions between butterfly larvae and ants, termed myrmecophily, require a range of behavioural and morphological adaptations (ant‐organs). Here, using light and scanning electron microscopy, we describe the complete life cycle of two species of Theope (Lepidoptera: Riodinidae) that have contrasting ways of life. Theope thestias larvae are facultatively tended by several ant species, whereas Theope pieridoides have obligate symbiotic interactions with Azteca ants that inhabit a myrmecophytic tree. Morphological differences associated with their different degrees of intimacy with tending ants are visible from hatching. In T. thestias, the untended first‐instar larva has elongated bifurcated setae and a few tiny perforated cupola organs (PCOs), whereas in T. pieridoides, the ant‐tended first instar has short dendritic setae, larger and more numerous PCOs, and functional tentacle nectary organs (TNOs). Throughout ontogeny, T. pieridoides always shows more conspicuous ant‐organs than T. thestias, with the exception of balloon setae, which are larger and more numerous in T. thestias. In addition, mature T. pieridoides larvae have an anterior set of ant‐organs, including a new type, here described and termed anterior glandular openings (AGOs). Based on the behavioural responses of ants in contact with these structures, a new interpretation for the mechanism whereby Theope larvae can manipulate the behaviour of their tending ants is proposed. Until now, three ecological syndromes can be defined among Theope species: (1) oligophagous larvae with facultative myrmecophily; (2) monophagous larvae with obligate myrmecophily; and (3) polyphagous larvae with obligate myrmecophily. These results suggest that differences in the degree of specificity in the ant–plant interactions may have an important role in the evolution of host‐plant use in Theope. © 2013 The Linnean Society of London     

Behavioral strategy of a lycaenid (Lepidoptera) caterpillar against aggressive ants in a Brazilian savanna

Myrmecophily is widespread in lycaenid butterflies, in which ants receive food resources and, in turn, protect caterpillars against natural enemies. This interaction ranges from obligate myrmecophily, in which immatures are invariably associated with ants and are dependent on ants for survival, and facultative myrmecophily, in which larvae are not dependent on ants for survival, but the presence of the latter may increase larvae survival. Lycaenids also include non-myrmecophilous butterflies, which do not have positive associations with ants and have developed strategies to avoid being attacked or preyed upon by them. In this study, we examined the relationship between the lycaenid Michaelus ira and two ant species associated with Distictella elongata (Bignoniaceae). This plant has extrafloral nectaries and is patrolled by Camponotus crassus and Ectatomma tuberculatum. Morphological analyses revealed that M. ira larvae have ant organs, such as dorsal nectary organs and perforated cupolas, structures associated with myrmecophily. We performed larval exposure experiments in the field, predicting that, in the absence of myrmecophily, the butterfly larva would present strategies to avoid ant attack. Results showed that larvae were attacked by both ant species. To escape ant molestation, larvae lived and fed inside silk-sealed D. elongata flower buds. We concluded that the M. ira bud-sheltering behavior was a defensive strategy against these ant species, while the dorsal nectary organs were apparently nonfunctional. Nonetheless, myrmecophily, in general, cannot be excluded in M. ira since relationships with other ant species may exist.     

Myrmecophily in Hesperiidae. The case of Vettius tertianus in ant gardens

The larvae of the hesperiid butterfly Vettius tertianus develop by eating the leaves of Aechmea mertensii, a bromeliad epiphyte restricted to ant gardens. The relationships between ants and V. tertianus larvae highlight the preferential association of the caterpillars with Pachycondyla goeldii (Ponerinae), an ant-garden initiator. The oviposition strategy of V. tertianus may thus imply the identification of the inhabiting ant species and not only the identification of the host plant. The caterpillars neither provide secretions to the ants, nor possess defensive devices (i.e. hairs or appendices) against ants. Their activity rhythm does not isolate them from foraging workers of P. goeldii and their shelters are also attainable by the ants. Moreover, as the cuticular lipid profiles of V. tertianus larvae are clearly different from those of the ants and also from the leaf-surface of A. mertensii, acceptance is not due to mimicry between larvae and plants or ants. However, the caterpillars deposit, on the leaf they eat, silk containing a mixture of substances very similar to those found on their own cuticle. No interaction with ants was recorded during observations, even though the ant gardens were patrolled by numerous P. goeldii individuals during their activity period. But when confronted with the caterpillar, none of the tested ant species reacted aggressively. These results suggest the existence of compounds, other than cuticular lipids, responsible for the absence of aggressiveness in the ants. The case of V. tertianus is relatively new as myrmecophily within Hesperiidae has been described only once. Moreover, it preferentially involves a member of the Ponerinae, a subfamily in which interactions with other arthropods are exceptional.     

The larval ant-organs of Thisbe irenea (Lepidoptera: Riodinidae) and their effects upon attending ants

Observations and experiments are presented on the use oflarval ant-organs by the riodinid butterfly Thisbe irenea with emphasis on their function in myrmecophily. The results indicate that each ant organ plays a distinct role in larval-ant association and that all influence the behaviour of attending ants. Interpretations of the roles that lycaenid larval organs play in ant associations are evaluated and discussed in light of myrmecophilous riodinids. Finally, an 'enticement and binding' process, involving the concerted use of the larval organs, is proposed as the behavioural mechanism that T. irenea and other riodinid larvae use to secure the attentions of individual ants for extended periods of time.     

Ant diversity as a direct and indirect driver of pselaphine rove beetle (Coleoptera: Staphylinidae) functional diversity in tropical rainforests,Sabah, Malaysian Borneo

Pselaphinae is a species‐rich beetle subfamily found globally, with many exhibiting myrmecophily—a symbiotic association with ants. Pselaphine–ant associations vary from facultative to obligate, but direct behavioral observations still remain scarce. Pselaphines are speciose and ecologically abundant within tropical leaf litter invertebrate communities where ants dominate, implying a potentially important ecological role that may be affected by habitat disturbances that impact ants. In this study, we measured and analyzed putative functional traits of leaf litter pselaphines associated with myrmecophily through morphometric analysis. We calculated “myrmecophile functional diversity” of pselaphines at different sites and examined this measure's relationship with ant abundance, in both old growth and logged rainforest sites in Sabah, Borneo. We show that myrmecophile functional diversity of pselaphine beetles increases as ant abundance increases. Old growth rainforest sites support a high abundance of ants, which is associated with a high abundance of probable myrmecophilous pselaphines. These results suggest a potential link between adult morphological characters and the functional role these beetles play in rainforest litter as ecological interaction partners with ants.     

Mutualistic ants and parasitoid communities associated with a facultative myrmecophilous lycaenid,Arhopala japonica,and the effects of ant attendance on the avoidance of parasitism

Herbivorous insects have evolved various defensive strategies to avoid their primary enemies, parasitoids. Many species of Lycaenidae (Lepidoptera) have food‐for‐protection mutualism with ants in their larval stages, where larvae produce nectar for ants and in return ants exclude parasitoids as well as predators. Myrmecophilous relationships are divided into two categories, obligate and facultative, by degrees of myrmecophily. Although parasitoids attacking obligate lycaenids always encounter lycaenid‐specific ant species, parasitoids that use facultative lycaenids are likely to encounter diverse ant species showing various defense systems. However, we know little about the parasitoid community of facultative lycaenid larvae. In this study, we investigated the mutualistic ant and parasitoid communities of a facultative myrmecophilous species, Arhopala japonica, in seven localities in Japan. The present field observation newly recorded four ant species attending A. japonica larvae, and combined with the previous data, the number of attending ant species reached 16, which is nearly the maximum number of reported attending ant species among myrmecophilous lycaenids. However, the present study revealed that almost all parasitized A. japonica larvae were attacked by a single braconid species, Cotesia sp. near inducta. We also assessed the efficiency of facultative ant defense against the parasitoid in the laboratory and revealed that oviposition by Cotesia sp. near inducta females was almost completely hindered when A. japonica larvae were attended by ants. This suggests that the dominant parasitoid does not have effective traits to overcome defensive behavior of ants and that the female wasps oviposit mainly in A. japonica larvae without intensive attendance.     

Ants andPolyommatus icarus immatures (Lycaenidae) —sex-related developmental benefits and costs of ant attendance

Summary Third and fourth instar larvae and pupae of the facultatively myrmecophilous Palaearctic blue butterflyPolyommatus icarus showed no alteration in developmental time when reared in the presence of two species ofLasius ants. Sex differences were observed in larval growth and pupal weight, with males growing larger and faster. Sex-related differences also occurred in the costs and benefits of ant-attendance. Male pupal masses tended to be larger in individuals associated with ants, and their pupal weight loss was not enhanced by ant attendance. This positive developmental effect of myrmecophily is tentatively attributed to a stimulating influence of ants on caterpillar feeding behavior. In contrast, females associated with ants tended to lose more weight during the pupal stage. Hence there is evidence for developmental benefits, rather than costs, of myrmecophily in maleP. icarus immatures, whereas ant attendance appears to be more costly for females during the pupal stage. These findings are discussed in relation to data on other myrmecophilous lycaenid species. It is suggested that maintaining low-level myrmecophily and its related organs is a comparatively inexpensive evolutionary stable strategy among this butterfly group.     

The network structure of myrmecophilic interactions

1. Ants establish mutualistic interactions involving a wide range of protective relationships (myrmecophily), in which they provide defence against enemies and partners provide food rewards and/or refuge. Although similar in the general outcome, myrmecophilic interactions differ in some characteristics such as quantity and quality of rewards offered by partners which may lead to different specialisation levels and, consequently, to different network properties. 2. The aim of this study was to identify structural patterns in myrmecophilic interaction networks, focusing on aspects related to specialisation: network modularity, nestedness and taxonomic relatedness of interaction ranges. To achieve this, a database of networks was compiled, including the following interactions: ants and domatia‐bearing plants (myrmecophytes); ants and extrafloral nectary‐bearing plants (EFNs); ants and floral nectary‐bearing plants (FNs); ants and Lepidoptera caterpillars; and ants and Hemiptera. 3. Myrmecophilic networks differed in their topology, with ant–myrmecophyte and ant–Lepidoptera networks being similar in their structural properties. A continuum was found, ranging from highly modular networks and phylogenetically structured interaction ranges in ant–myrmecophyte followed by ant–Lepidoptera networks to low modularity and taxonomically unrelated interaction ranges in ant–Hemiptera, EFN and FN networks. 4. These results suggest that different network topologies may be found across communities of species with similar interaction types, but also, that similar network topologies can be achieved through different mechanisms such as those between ants and myrmecophytes or Lepidoptera larvae. This study contributes to a generalisation of myrmecophilic network patterns and a better understanding of the relationship between specialisation and network topology.     

Host-Specific Myrmecophily and Myrmecophagy in the Tropical Coccinellid Diomus thoracicus in French Guiana

A variety of arthropods, particularly insects, have developed myrmecophilous interactions with ants to gain access to resources and/or for protection. Among these myrmecophiles, only a few examples have been documented in the Coccinellidae, most of them involving species able to feed on ant-tended Hemiptera. We report here a new case of obligate myrmecophily in the coccinellid Diomus thoracicus. Larvae are invariably and exclusively found in the nests of the ant Wasmannia auropunctata and seem to rely on ant brood as their only food source. Not only do ant workers show no aggressiveness toward the D. thoracicus larvae in their behavioral interactions at the colonial level, but also at the species level; while coccinellid adults are always attacked. The integration of the larvae inside of the ant nests is based on their chemical mimicry of the host's cuticular cues. Therefore, given the presence of the D. thoracicus larvae inside of the ant's nest, their predation on Wasmannia brood and their chemical mimicry, this species can be considered a specific parasite of W. auropunctata. Overall, this new case of myrmecophily not only specifically involves a highly invasive ant species, but also provides insights into the evolution of myrmecophily and myrmecophagy in coccinellids.     

Mutualism between Thisbe irenea butterflies and ants, and the role of ant ecology in the evolution of larval-ant associations

A facultative mutualism between the riodinid butterfly Thisbe irenea and the ponerine ant Ectatomma ruidum is described from Panama. Ants protect larvae against attacks of predatory wasps, but not against tachinid parasitoids. Several potential sources of ecological variation affecting the larval survival of Thisbe irenea are noted. A preliminary means of testing the ability of larvae to appease ants is described that may be applied to all butterfly-ant systems. Observations and literature records indicate that ant taxa which tend butterfly larvae are the same taxa that tend extrafloral nectaries and Homoptera. A general hypothesis for the evolution of myrmecophily among butterflies suggests that ant taxa which utilize secretions in their diet are major selective agents for the evolution of the larval ant-organs, and hence, ant-larval mutualisms. This idea is extended to suggest how appeasement of predaceous ant taxa through the use of larval ant-organs can influence an ant-larval relationship, eventually leading to mutualism.     

Polyphagy and Obligate Myrmecophily in the Butterfly Hallonympha paucipuncta (Lepidoptera: Riodinidae) in the Neotropical Cerrado Savanna

This study analyzes host plant use and obligate myrmecophily in Hallonympha paucipuncta, an endemic butterfly of the Cerrado. Larvae of H. paucipuncta are polyphagous, using at least 19 species of plants in 10 families. All larvae found were being tended by Crematogaster ants. Spatial distribution of larvae and tending ants were strongly aggregated, suggesting an influence of ants on oviposition and/or larval survival. Implications of obligate myrmecophily in the evolution of polyphagy in Riodinidae are discussed.     

Cheating on a mutualism: indirect benefits of ant attendance to a coccidophagous coccinellid

Coccinellids (Coleoptera: Coccinellidae) are generally unable to prey on ant-tended prey. However, particular coccinellid species have morphological, behavioral, or chemical characteristics that render them immune to ant attacks, and some species are even restricted to ant-tending areas. The benefit gained from living in close association with ants can be twofold: (1) gaining access to high-density prey areas and (2) gaining enemy-free space. Here, the myrmecophily of Azya orbigera Mulsant (Coleoptera: Coccinellidae), an important predator of the green coffee scale, Coccus viridis (Green) (Hemiptera: Coccidae), is reported. In this paper, three main questions were studied. (1) Are the waxy filaments of A. orbigera larvae effective as defense against attacks of the mutualistic ant partner of C. viridis, Azteca instabilis F. Smith (Hymenoptera: Formicidae)? (2) Does A. instabilis reduce the rate at which A. orbigera larvae prey on scales? (3) Do A. orbigera larvae gain enemy-free space by living in close association with A. instabilis? Laboratory and field experiments were conducted to answer these questions. We found that, because of the sticky waxy filaments of A. orbigera larvae, A. instabilis is incapable of effectively attacking them and, therefore, the predation rate of A. orbigera on C. viridis does not decrease in the presence of ants. Furthermore, A. instabilis showed aggressive behavior toward A. orbigera's parasitoids, and the presence of ants reduced the parasitism suffered by A. orbigera. This is the first time that this kind of indirect positive effect is reported for an ant and a coccidophagous coccinellid. Furthermore, this indirect positive effect may be key to the persistence of A. orbigera's populations.     

A comparative analysis of morphological and ecological characters of European aphids and lycaenids in relation to ant attendance

Ants are a major environmental factor for many insect species. For example, aphids and lycaenids have evolved an array of associations with ants ranging from obligate myrmecophily to the avoidance of contact. Here we (1) analyze the predictive power of different ecological and morphological traits for explaining the strength of the association between ants and aphids/lycaenids and (2) contrast different taxonomic levels with respect to the variance explained by ant attendance. Data come from a literature survey including 112 species of aphids and 103 species of lycaenids from Europe. For aphids, feeding on woody plant parts is positively associated with ant attendance, while a high degree of mobility, feeding in isolation, and the possession of wings in the adult stage are negatively associated with ant attendance. In lycaenids, feeding on inflorescences and feeding on Fabaceae host plants is closely associated with ant attendance, while living in forests bears a smaller likelihood to establish mutualistic relationships. Body size always appeared to be a poor predictor for the degree of ant attendance. Overall, in both insect groups less than 10% of the variation in the ecological traits recorded is explained by the different modes of ant association. When decomposing the variance in traits explained by ant attendance at different taxonomic levels, aphids and lycaenids show contrasting results. In aphids, most variance in the degree of ant attendance is explained at the subfamily level and least at the species level. The opposite is true for lycaenids, where most variance is explained at the lowest taxonomic level. Possible mechanisms explaining these different patterns of associations with ants are suggested.     

The distribution and density of a lycaenid butterfly in relation to Lasius ants

Summary Larvae and pupae of lycaenid butteflies are often associated with ants: this is usually a mutualism in which ants guard the lycaenids from natural enemies, and the lycaenid larvae and pupae provide sugars and amino acids for the ants. A possible consequence of the interaction is spatially correlated ant and lycaenid distributions, but the phenomenon is poorly documented. We examined the lycaenid Plebejus argus, which is tended by Lasius ants. Within habitat patches, P. argus eggs, larvae and pupae were all spatially associated with Lasius. On a larger scale, the densities of butterflies in different habitat patches and populations, and whether the butterfly was present or not, were correlated with Lasius ant densities. The association of P. argus with Lasius ants is consistent among populations, and occurs at several spatial scales. Other aspects of the ecology of P. argus are more variable.     

Species‐specific levels of ant attendance mediate performance costs in a facultative myrmecophilous butterfly

Trophobiont butterfly larvae offer caloric rewards to ants through specialised glands and, in return, gain ant‐derived protection from natural enemies. Thus, from the larva's perspective, the major cost of myrmecophily comprises the reward production. Larvae of the butterfly Parrhasius polibetes (Stoll) (Lycaenidae) are facultatively tended by several ant species, which might differ in the intensity of tending behaviour. The performance costs (development time, survival, pupal mass and adult dry mass) of P. polibetes are examined when tended by two ant species differing in size and foraging strategies (Camponotus melanoticus Emery and Camponotus crassus Mayr), along with the corresponding intensity of tending behaviour towards late instars. Untended larvae serve as controls. Larvae tended by C. melanoticus take longer to pupate compared with both C. crassus and control larvae. By contrast, pupae whose larvae are tended by C. crassus are lighter than control larvae but do not differ from those tended by C. melanoticus. No differences are found in the adult stage, indicating compensation in all cases. Both at short‐ and long‐term scales, C. melanoticus tends larvae of P. polibetes more intensely than C. crassus. The increase in tending activity of C. melanoticus presumably delays the development time of larvae tended by this ant species. The results of the present study show that tending intensity varies depending on the ant species, and that P. polibetes has compensatory mechanisms to minimise myrmecophily costs, regardless of tending intensity. To the authors' knowledge, this is the first experimental evidence that intensity of ant‐tending behaviour is species‐specific and affects performance in a trophobiont insect.     

Juxtaposition of the generative cell and vegetative nucleus in the mature pollen grain of amaryllis (Hippeastrum vitatum)

Summary Computer-generated, three-dimensional reconstructions from serial ultrathin sections were used to investigate the spatial organization and extent of association between the generative cell and vegetative nucleus within the mature pollen grain of amaryllis. In all cases examined, the highly lobed vegetative nucleus was found in close proximity and positioned laterally to the elongated, oval shaped generative cell. Numerous projections of the vegetative nucleus come to within 53 nm of the inner vegetative cell plasma membrane which surrounds the generative cell. These areas of close association may continue transversely around the generative cell for a distance of up to 4 m. Although an association exists between the generative cell and vegetative nucleus of the mature pollen grain, it is apparent that several changes must take place after pollination in order to achieve the high amount of close contact that occurs between the vegetative nucleus and the numerous terminal cell extensions of the leading sperm in the pollen tube of amaryllis (Mogensen 1986). Thus, this study demonstrates that the spatial organization among components of the male germ unit in the mature pollen grain does not necessarily reflect relationships that ultimately exist among these components within the pollen tube.     

Differences in worker caste behaviour of Oecophylla smaragdina (Hymenoptera: Formicidae) in response to larvae of Anthene emolus (Lepidoptera: Lycaenidae)

The larvae of Anthene emolus (Lycaenidae) cannot survive in the wild without their associated ant Oecophylla smaragdina (Formicidae) . The ants groom the lycaenid larvae for secretions and, in turn, protect them from parasitoids and predators. Both major and minor worker ants tend larvae, although, in the present study, the frequencies of tending by different castes proved to be significantly different. Ants also tended late instars significantly more than young larvae. O. smaragdina ants of the same colony were observed fighting when A. emolus larvae were present. Aggression was exhibited by the major workers against the minor workers that tried to tend secreting larvae. Major ants precluded the minors from the most nutritious secretions, leaving minors to tend the less productive, early instars. On a few occasions, aggression was unchecked and major workers killed the larvae they were tending (presumably accidentally). When larvae were experimentally reared with only one caste of ant, the resulting butterflies showed no differences in dry weight. Mortality was high when larvae were reared with only minor ants or in the absence of ants.  © 2006 The Linnean Society of London, Biological Journal of the Linnean Society , 2006, 88 , 391–395.     

Myrmecophilous butterflies utilise ant–treehopper associations as visual cues for oviposition

1. Selection of a safe oviposition site is important for herbivorous insects whose immature stages have limited mobility. Female herbivores rely on environmental cues for this choice, and presence of natural enemies or mutualistic partners may be important in this process. 2. Some butterflies have mutualistic interactions with ants (myrmecophily), in which caterpillars offer a nutritional liquid and gain protection against natural enemies. Participants in butterfly–ant mutualisms may utilise signals to initiate interactions, but the use of visual cues by ovipositing myrmecophilous butterflies remains uncertain. 3. Larvae of facultatively myrmecophilous Parrhasius polibetes (Lycaenidae) feed on Schefflera vinosa, and females prefer to oviposit near aggregations of the ant‐tended treehopper Guayaquila xiphias, where caterpillars survive better due to increased ant attendance. Given the conspicuousness of ant–treehopper associations, it was investigated whether butterflies use them as visual cues for oviposition and, if so, which participants of the association are used as cues: ants, treehoppers, or both. 4. Experiments using dried insects on paired branches revealed that females visually recognise ants and ant–treehopper associations, using them for egg‐laying decisions. However, presence of a treehopper aggregation alone had no effect on oviposition choices. 5. This is a first insight into the importance of visual discrimination for ovipositing myrmecophilous butterflies. The results show that facultative mutualisms can be important enough to promote a behavioural adaptation (visual detection of ants) reinforcing the interaction. Our research highlights the importance of the behavioural interface within complex multispecies systems.     

Aphid-mediated coexistence of ladybirds (Coleoptera: Coccinellidae) and the wood ant Formica rufa: seasonal effects, interspecific variability and the evolution of a coccinellid myrmecophile.

It is generally believed that most homopteran-eating insects avoid ant-tended colonies of Homoptera, due to the ant aggression they encounter there. However, because homopteran colonies which are ant-tended often persist for longer than untended colonies, some homopteran-eaters may utilise ant-tended Homoptera when untended colonies are scarce. Furthermore, a few homopteran-eaters are myrmecophilous, habitually coexisting with ants. To investigate these phenomena, a study was made of aphids and aphidophagous coccinellids (ladybirds) on Scots pine, Pinus sylvestris , growing in areas foraged and unforaged by the wood ant Formica rufa . The non-tended aphid Schizolachnus pineti exhibited a marked population decline in late summer but persisted in both areas at very low density. Facultatively tended Cinara aphids exhibited higher population densities when associated with F. rufa , and remaining colonies of these aphids were only found associated with ants in late summer. Coccinellids exhibited considerable interspecific variability in their level of association with F. rufa , and there was some evidence of an increase in certain species' frequencies of occurrence with the ant when Cinara aphids were all ant-tended, in late summer. Coexistence with ants appears to be associated with either an intolerance of low aphid densities, in Coccinella septempunctata and Harmonia quadripunctata , or with extreme dietary specialisation, in Myzia oblongoguttata . Similar factors to those which bring C. septempunctata into contact with ants were probably of importance in the initial stages of the evolution of myrmecophily of its congener, Coccinella magnifica .     

Evolution of influence: signaling in a lycaenid-ant interaction

Some phytophagous insects gain defense from natural enemies by associating with otherwise potentially harmful top predators. Many lycaenid butterfly caterpillars are involved in such interactions with ants: larvae provide carbohydrate rewards from the dorsal nectary organ (DNO) to associated ants in return for protection from natural enemies. The stability of these interactions involves signals that identify the lycaenid caterpillar as a mutualist. However, larvae of some lycaenid species, such as Lycaena xanthoides, are found in close association with ants but do not possess the reward producing DNO. Evaluating the relationship in a phylogenetic framework, we show that the association between L. xanthoides and ants likely evolved from a non-ant-associated ancestor. Behavioral trials also show that L. xanthoides larvae are capable of influencing ant behavior to increase ant tending when faced with a simulated predator attack, without providing DNO-derived rewards to ant associates. These results demonstrate that the DNO is not necessary to maintain associations between lycaenid larvae and ants. Third-party interactions may affect the evolution of mutualisms and consideration of underlying evolutionary history is necessary to understand contemporary species associations.