Myrmecophily: Ants and Butterflies

The Evolution, Effects, and Maintenance of their Relationships

Jay Mann

the_jay_mann@hotmail.com

April 19, 1999

Abstract

Lycaenid butterflies compromise upwards of 30% of all butterfly species and are often found- greater than 50% of the time- involved with ants during larval development. This relationship has been around for at least fifteen-million years and in some cases the interaction has become species-specific. The basic understanding is that the lycaenid larvae provide secretions high in amino acids and carbohydrates and obtain protection from parasitism and predation from ants in return. Larvae possess three organs thought to be important in the relationship- pore cupola organs, glandular dorsal nectar organs, and eversible tentacle organs. Communication- vibration through the substrate- is also thought to play a role, though whether it is a call or more of a soothing song has not yet been determined. Studies on caterpillars in relationships with ants have had variable results. Some experiments found that ant-tended larvae experienced a reduction in growth and an increased development time, while other studies had the opposite results. A possible explanation is that some larvae compensate in their food consumption for their secretions while others do not. The relationship likely evolved first as an appeasement behavior on the part of the lycaenids who possibly later took advantage of existing behaviors in ants. Lycaenids' relationships with ants could have mediated a radiation into previously inaccessible habitats. Further research needs to address whether the relationships benefit entire ant colonies, further explore the chemistry of the caterpillars secretions, and examine the effects the relationships have on third-parties- particularly plants.

Introduction

Ants play an enormous role in terrestrial ecosystems. They are considered to be the “leading invertebrate predators” (Fielder et al. 1996). Thus one would expect to find that many of their potential prey species have adapted to them in some way. These adaptations can include behavioral modifications and camouflage- chemical and physical. In some striking cases what was likely originally simply an adaptation to pacify the ants have evolved into a mutualistic relationship. For close to two-hundred years scientists have been aware of relationships between butterflies and ants. But it was not until relatively recently that observations and speculation were tested more rigorously.

Butterfly-ant relationships have been found in the Lycaenidae and Riodinidae butterfly families but it is most striking in the Lycaenidae. The Lycaenidae make up approximately 30% of known butterflies and within the family more than half of the species have a relationship with ants during their larval development (Seufert and Fiedler 1996). But how did these relationships- which today can involve complex behaviors and are sometimes species specific- evolve? After discussing some background on interspecific relationships and describing the types of relationships that are found between ants and butterflies, the evolutionary questions will be addressed. Two ways of trying to determine the evolutionary steps in a relationship are looking at the costs/benefits of the current relationship while postulating previous functions for the traits that are involved or comparing among current day species which exhibit a range of relationships.

Interspecific relationships (background)

One major problem in the idea of mutualistic relationships is the question of cheaters. Why does not one species simply receive the benefits of the relationship without giving anything in return? What keeps the mutualism as an evolutionary stable strategy? The Prisoner’s Dilemma has traditionally been used to study species interactions. This theoretical problem considers two individuals and simplifies their choices for a given interaction. Each individual can either help the other (cooperate) or not help (defect). A payoff matrix can be constructed- the lowest “payoff” is when both defect, the highest when both cooperate, an intermediate (and asymmetrical) payoff results when one cooperates and the other defects. While one run- or iteration- of the simulation promotes the strategy of cheating, strategies that promote mutualism can be found in the Iterated Prisoner’s Dilemma (IPD). The iterated version assumes that individuals will encounter each other many times over their lifetimes and will recall recent encounters and their outcomes. But some assumptions are difficult to prove in the real world, especially so in interspecific mutualisms. A more realistic model would be one which incorporated both ecological differences between the partners and allows variable values in the payoff matrix (Doebeli and Knowlton 1998). Using this model, Doebeli and Knowlton suggested that it may be the initial step of a relationship (i.e. one individual cooperating) that hinders mutualism more than the occurrence and spread of cheaters (un-cooperators). There is more than one type of cheating, however.

The traditional type of cheating has been thought to be accepting some benefit but not reciprocating. But in some relationships it has been found that one partner can lower their reciprocation quantitatively. While working with models Sherratt and Roberts (1998) found that at a benefits to cost ratio of greater than one the individuals maintained the relationship but reduced their generosity. In another model they found that as the number of interactions between individuals increased both cooperation was more common and generosity increased. This brings up the question of individual recognition. While it is plausible to think of this occurring within a species, it is much more difficult to consider it occurring between species. But in some ant-butterfly relationships a slight modification of it has been found- individual species recognition.

Types of ant-butterfly relationships

When a caterpillar is found to be attended by ants- usually during its larval stage- the number of ants can vary from one to fifteen. The chief protection the ants provide is against parasitism, while the caterpillar usually (but not always) provides secretions- often of amino acids and carbohydrates- to the ants. Ant-butterfly relationships can be classified as three main types: facultative- or unspecific relationships where a caterpillar can interact with a number of different ant species (but does not necessarily need to in order survive), obligates- where the caterpillar depends on a specific ant-partner for survival, and those where no relationships are present (Seufert and Fiedler 1996). The most common relationship found is the facultative one in which the caterpillars are tended by ants after their third of forth instar. It is interesting to speculate why it is not until a later instar that the relationship starts, because often larvae are vulnerable to parasitism during all their instars; some possible reasons include reduced predation at an early age, an inability of the caterpillar to energetically afford to provide secretions for the ants, and/or the organs used in the relationship being undeveloped in the caterpillar. In some cases, the larvae pupate inside an ant nest, or rather a nest outpost. Pupae have sometimes been found to continue the relationship they had as larvae with the ants- i.e. provide secretions, but this is not common.

Some butterflies have been observed to lay eggs only on those host plants that also have a particular ant species nearby. This implies that perhaps the use of the particular plant has been facilitated by the ant relationship- i.e. ants allowed the butterfly to use a new host plant. But how are the ants and butterflies recognizing, maybe even communicating with each other?

Mechanisms of the relationships and communication

Chemicals and smell play a large role among insects, thus it is not surprising to find that lycaenid caterpillars possess myrmecophilous organs- highly specialized epidermal glands. These glands are surrounded in most lycaenid larvae by pore cupola organs which are antennated by ants “intensively” upon first encounters; the glands are thought by some to release ant brood signals in some species, amino acids in other species (Fiedler et al. 1996). But likely the most important organ in the relationship is the caterpillar’s glandular dorsal nectar organ. It has been observed that after the ants drum around the gland with their antennae the gland delivers droplets; these droplets have been found to contain amino acids and carbohydrates (Fiedler et al. 1996). It is thought that some caterpillars are able to distinguish differences in antennal drumming- i.e. will only provide food to the “right type of ants.” It is these droplets which are thought to be the “reward” the ants receive for their guarding/ protection- for the ants will attack predators and parasitic wasps that approach the caterpillar. Another organ many lycaenid caterpillars possess that is thought to play a role in their relationships with ants are eversible tentacle organs. The waving of these tentacles causes the attendant ants to exhibit alarm behavior, thus it is thought that they may emit ant alarm pheromones (Fiedler at al 1996). Perhaps even more striking was the observation that caterpillars may call ants via substrate-vibrations. But some alternative thoughts are that the vibrations reinforce rather than initiate the relationships (Fiedler et al. 1996). In addition to looking at the mechanisms of the relationships, it is important to look at the costs and benefits.

Cost/benefit analyses of lycaenid-ant relationship

Animals’ behavioral decisions are usually thought of as responding to a number of tradeoffs. One of the largest tradeoffs is between time spent foraging and predator/parasite avoidance. Ants may help caterpillars by reducing both parasitism and predation thus allowing them to expand their foraging- both spatially and temporally. An early study on the subject of reduced parasitism by ants was by Pierce and Mead (1981) who studied G. lygdamus. They found that those larvae from which ants were experimentally kept away from were both parasitized more and had a lesser survival rate (implied from a reduction of their numbers in censusing). One interesting observation they had was that “individually marked ants in the field were observed to be remarkably constant in their attention to a given larva.” This individual recognition satisfies one of the major assumptions of the IPD. Further studies examined not just survival of larvae but their fitness as well- as measured by developmental time and weight.

Cushman et al. (1994) examined the hypothesis that both species benefit from the relationship (i.e. it is a true mutualism). They studied the rather advanced relationship between Iridomyrmex nitice and Paralucia aurifera- the larvae spend the day in the chambers of the ants (located at the base of the larvae’s host plants) and come out to feed at twilight. The chambers they examined contained up to 20 larvae, 10 pupae and are thought to be ant nest outposts. To assess the benefit to the ants the secretions were analyzed and found to contain 13-15 amino acids and glucose. In ants the worker-caste in particular is thought to require sugar for foraging and Cushman et al. stressed the need to examine whether increased sugar results in increased foraging and thus protein for the entire ant colony. They measured the ant mass and survival rates both before and after the experiment in nests treated without anything, with an artificial diet, and with lycaenid larvae; they found that ant survival, but not mass was increased by the presence of lycaenids. They found that the lycaenid larvae “were 31-76% heavier, developed 37% faster, and commonly completed one or two few instars than larvae reared with ants” (Cushman et al. 1994). These lead to a shorter generation time which can be a very important competitive advantage. As far as the reasons for the faster growth rate Cushman et al. postulated two possible mechanisms- the improved conditions a shelter provides (mainly increased humidity) or abnormal behavior in untended larvae- i.e. less time spent feeding; they feel the second mechanism is the more important one. Some alternative hypothesis were discussed by Wagner and del Rio (1997).

Wagner and del Rio (1997) first discussed how some studies have found that larvae experience a reduction in size and a prolonged development time when they interact with ants while other studies have shown the opposite effects. One hypothesis for the increased growth is an overcompensation of sorts, with tended larvae adjusting their food consumption in order to energetically afford the secretions (perhaps some overcompensate, some do not); at alternative idea is increased digestive efficiency in tended larvae. Wagner and del Rio examined the increased digestive efficiency hypothesis by looking at fecal matter in larvae and examining the digestion success of pollen (which is a common food). They found no evidence for an increased efficiency but proposed another explanation- tended larvae had lower activity levels and thus spent less energy. The question can be raised of why the ants do not simply eat the entire caterpillar. A possible explanation is that an amount of energy equivalent to the caterpillar can be obtained via secretions over the life-time of the caterpillar (Fiedler et al. 1996). While current costs and benefits are certainly important, it is also important to examine how the relationships began.

Evolution of the relationships

Devries and Poinar (1997) discuss an amber fossil they found of a caterpillar that possessed the major organs that are used today in relationships with ants- tentacle nectary organs, balloon setae, and vibratory papillae. The fossil was dated at 15-20 million years old and this long relationship between butterflies and ants is illuminating in a number of ways. It can explain both the large number of species in the Lycaenidae family as well as the intricacies of their relationships today. But what caused the organs to develop in the first place?

At first caterpillars likely simply appeased ants or camouflaged themselves with chemicals. But the pressure of parasitism on caterpillars may have selected for stronger relationships. In those lycaenids who are normally associated with ants but have had the ants experimentally removed or kept away, parasitism is extremely high. Thus one can speculate whether the caterpillar had a previous adaptation against parasitism that is now gone, or simply lived in a habitat with less parasitic risk. Historically, those butterfly species that due to some (perhaps at the time) novel organ or secretion were attended by ants and were less parasitized likely left more offspring. In addition a relationship with ants could open new habitats/host plants- by a reduction of parasitism in a habitat where it was previously too high to survive. The new habitat could also contain no or few other species of caterpillars and thus less competition. This can lead to speciation (e.g. an isolated population protected by ants in a novel environment) and perhaps explains the large number of lycaenid species- a radiation may have occurred as a result of their involvement with ants.

The lycaenid caterpillars that evolved relationships with ants likely took advantage of inherent behaviors of the ants. Perhaps they evolved after ant-aphid relationships (where the ants will feed off aphid secretions in return for protection) had been around for awhile. Another exploitation could have been a structural one- some ant species have outposts, or satellite nests. Wagner found that Hemiargus isola larvae that entered Formica perpilosa nests spent less time on the ground searching for pupation sites (than other lycaenids), were tended by ants in the nest, and had a lower incidence of predation rate while pupating (Wagner 1995). Once again it is interesting to ask why the ants simply do not eat the pupae themselves, perhaps they do not to insure further generations of secretion providers. Some relatively recent speculation has involved the honesty of the caterpillars secretions.

In their relationships, caterpillars may initially have provided food secretions, but later stopped. This can be taking advantage of the observation (likely behavior) that “ants are known to guard plants secreting extra floral nectar” (Pierce and Mead 1981). The paper goes on to describe the discovery of several lycaenid species that while tended by ants, do not possess secretion glands. Perhaps conditions changed so that evolutionary pressures selected against the energetically expensive glands. Since evolution is a continuous process and ant-butterfly relationships have existed for a significant length of time, it is not surprising to find that parasitic wasps have adapted themselves to sometimes pass the ant guards- with camouflage and mimicry. Another way of examining the evolution of ant-butterfly relationships is the comparative approach.

Comparative approach: comparisons among different parasitoid and lycaenid species

Baumgarten and Fiedler (1998) looked at three different types of parasitoids of two lycaenid species. They found that the parasitoids attacked differently sized/aged caterpillars and that there could be some risk in leaving the host (i.e. if ants are still around when the parasitoids emerge). A common feature of the parasitoids was that they preferably attacked young host caterpillars which usually are unattended by ants. Interestingly, parisitized caterpillars are often still tended by ants. They postulated that hyperparasitism (where one parasite itself is parasitized) may be a factor that prevents the parasites from destroying the myrmecophilous (ant-attracting) properties of their hosts. This idea was reinforced by their observations that ants still were somewhat attracted to the caterpillars 1-2 days after the parasitoid had left. Konrad Fiedler was also involved in an evolutionary study that looked a three lycaenid butterflies (Seufert and Fiedler 1996).

Seufert and Fiedler (1996) looked at the idea that mutualism could cause co-speciation. They studied three species of butterflies that sharer similar larval host plants- an obligate, facultative, and a species that was not involved with ants. The variables they measured were oviposition behavior, developmental times, adult body mass, and parasitism. An interesting result was that the unattended butterfly larvae had the least amount of parasitism, perhaps due to physical adaptations that reduce parasitism- but suggesting something lost in the other two species.

Seufert and Fiedler thought that the unattended species had a secondary reduction of their myrmecophilous organs. They postulated that his can occur under one or more conditions: as a result of a specialization on a host plant avoided by most lycaenids, in a place with a reduced presence of ants, and/or endophytic larvae feeding habits. Perhaps this particular lycaenid species is spending its resources on digestion of secondary compounds rather than secretion production.

Further Study

There are a number of areas where further research would expand the understanding of the relationships between ants and butterflies. When studying the benefits the ants receive from their relationship with butterflies most studies have concentrated on the health of those workers tending the caterpillar(s). But as Cushman et al. (1994) pointed out, the fitness of the entire ant colony is the variable that needs to be studied. Cushman et al. also stressed the need to examine the additional partners each member of a mutualistic relationship may have (e.g. ants often have many relationships ).

There needs to be more work done on the chemistry of the communication- both on what chemicals the pore copola organs release (also whether they vary among lycaenid species) and in return how sensitive the larvae are to these chemicals. In one lycaenid species it was found that the larvae could follow an ant chemical trial and it would interesting to see if this adaptation/ability exists in other lycaenid species.

It would be interesting to compare facultative versus obligate lycaenid species- to try to determine which evolved first and also compare their relative success. While the obligates often have a higher survival rate and can dominate a particular plant species it can be argued that the facultatives are able to colonize/survive in more types of habitats. Modeling with parameters obtained from species in the field could be useful in trying to deduce the evolution of the two types.

One possible complication that does not appear to have received attention is the possibility of ants choosing among caterpillars- i.e. which they attend. This could be a possible mechanism for enforcing honesty among the larvae’s secretions. Additional work on the relationships between adult butterflies and ants would be interesting since oviposition behavior has been observed to be influenced in some species by the presence of ants. It would be nice to explore what other interactions the two (adult butterflies and ants) may have.

The effects that butterfly-ant relationships have on competition between species needs to be examined. Lycaenid larvae can gain a competitive advantage both among Lepidoptera and other species. Lastly a related area is the effect of the relationship on third parties- the influence of the partnership on plants seems particularly interesting because it can result in a number of conflicting things. The presence of ants- due to the presence of an associated lycaenid larvae on a plant- can reduce herbivory by other insects, increase herbivory due to increased survival of the lycaenid larvae, yet increase pollination due to increased survival of lycaenid through the pupation stage. Seems a fruitful avenue for further research.

References

Baumgarten, H.T. & Fiedler, K. 1998. Parasitoids of lycaenid butterfly caterpillars: different patterns in resource use and their impact on the hosts’ symbiosis with ants. Zoologischer Anzeiger 236: 167-180.

Cushman, J.H., Rashbrook, V.K. and Beattie, A. J. 1994. Assessing benefits of both participants in a lycaenid-ant association. Ecology 75(4): 1031-1041.

DeVries, P.J. & Poinar, G.O. 1997. Ancient butterfly-ant symbiosis: direct evidence from Dominican amber. Proceedings of the Royal Society of London Series B- Biological Sciences 264, 1385: 1137-1140.

Doebeli, M. & Knowlton, N. 1998. The evolution of interspecific mutualisms. Proceedings of the National Academy of Sciences of the USA Vol. 95: 8676-8680.

Fiedler, K., Holldobler, B. and Seufert, P. 1996. Butterflies and ants: the communicate domain. Experientia 52: 14-24.

Pierce, N.E. & Mead, P.S. 1981. Parasitoids as selective agents in the symbiosis between lycaenid butterfly larvae and ants. Science 211: 1185-1187.

Seufert, P. & Fiedler, K. 1996. Life-history diversity and local co-existence of three closely related lycaenid butterflies (Lepidoptera: Lycaenidae) in Malaysian rainforests. Zoologischer Anzeiger 234: 229-239.

Sherratt, T.N. & Roberts, G. 1998. The evolution of generosity and choosiness in cooperative exchanges. Journal Theoretical Biology 193: 167-177.

Wagner, D. 1995. Pupation site choice of a North American lycaenid butterfly: the benefits of entering ant nests. Ecological Entomology 20: 384-392.

Wagner, D and del Rio, C.M. 1997. Experimental tests of the mechanism for ant-enhanced growth in an ant-tended lycaenid butterfly. Oecologia 112: 424-429.