David Logue

Colorado State University, Fort Collins Colorado

Trees of the genus Cecropia (Moraceae) frequently house Azteca (Hymenoptera) colonies in their hollow stems and trunk segments. The ants feed their larvae proteinaceous Müllerian bodies produced by the plant, and tend mealybugs (Coccidae), which feed on the tree’s phloem. Though Cecropia exhibits numerous apparent adaptations that encourage this symbiosis, there has been long-standing debate over whether or not the ants exploit their host tree, thereby rendering the relationship between ants and plant a parasitism. I herein review the literature on this system in an attempt to relate Doebeli and Knowlton’s model of biological mutualism to the Cecropia-Azteca-Coccidae system. Costs to the trees include the production of Müllerian bodies, phloem loss to mealybugs, and increased woodpecker (Picidae) attack. Trees benefit from reduced shading, epiphyte load and herbivory and the associated freedom from production of allelopathic chemicals and alkaloids. They also derive nitrogen from frass and ant carcasses. The ants gain few to no direct benefits from defending their host tree, since they do not consume encroaching plants, are easy targets for woodpeckers, and suffer high rates of young nest failure due in part to their aggressive lifestyle. They do, however, benefit indirectly from a healthy host tree, which provides nest sites (though some Azteca are capable of building nests outside of trees), and nutrition for their larvae and the mealybugs. The mealybugs, which lose a substantial portion of their caloric intake to the ant colony, receive protection, reduced parasite load, a food source free of harmful secondary compounds, and a low-cost means of dispersal. I conclude that all parties receive net benefits from this association, and that exploitation would be maladaptive for each of the parties involved.

Distributed throughout the Neotropics, trees of the genus Cecropia are symbiotic with Azteca ants (Fig.1). Most species of Cecropia associate with these ants (Vasconcelos and Casimiro, 1997), however

not all individuals contain ant colonies. Conversely, certain species of Azteca are thought to be obliged to live within Cecropia trunk segments (Janzen, 1968) and may gather most of their nutrients from their host (Vasconcelos and Casimiro, 1997; but see Sagers et al., 2000). Azteca larvae derive a major component of their energetic needs in the form of the protein-rich Müllerian produced by special organs at the petioles of each leaf. Additionally, the trees provide shelter for the ants, which dwell in hollow chambers in the trunk of the tree.

There is a third member of this symbiosis, a mealybug (Coccidae) that the ants tend in their nests inside the tree. The mealybugs feed exclusively on the phloem of the tree. When palpitated by their host ants, mealybugs excrete carbohydrate-rich honeydew, which the ants ingest (Beattie 1985).

In order to explore the fitness dynamics of this complex relationship, I will present a theoretical framework for modeling the evolution of biological mutualism, discuss the parameters of the model for Azteca and Cecropia, and make conclusions about the cost-benefit dynamics of the system.

Biological mutualisms pervade terrestrial tropical ecosystems, and are present to a lesser degree throughout the world (Boucher 1985). Evolutionary theorists, however, have struggled to develop models that predict the evolution and stability of mutualism. Historically, an oversimplification of natural selection theory led to the hypothesis that those who "cheat," or exploit the resources of their symbiotic partner, thereby reducing the benefit to the partner and increasing their own immediate benefit, will increase their survival relative to those who "play by the rules." Following this argument to its logical extreme, one must conclude that mutualisms are highly unstable and any cheating mutants would drive the system away from mutualistic symbiosis. Given the vast numbers of mutualistic symbioses on earth and their apparent longevity (as evidenced by morphological specializations), the "advantaged cheater" hypothesis is not a satisfactory model for the evolution of mutualism.

Trivers (1971) attempted to explain reciprocal altruism (the mechanism that maintains mutualism) in terms of the Prisoners’ Dilemma. His theory predicts that altruism, the voluntary payment of one’s own fitness costs, will be selected for if the following criteria are met: (1) both organisms derive benefits from the association, (2) associations between individuals are long term, (3) dissolution of the symbiosis would result in increased costs to re-establish a symbiosis with the same or another individual, (4) reciprocation occurs repeatedly throughout the lives of the symbionts, and (5) both symbionts pay roughly equivalent costs and derive roughly equivalent benefits. These constraints, particularly number five, are not consistently present in biological mutualisms. As an example, Cecropia is capable of surviving and reproducing without Azteca, yet the ants are obligate mutualists to the tree. Clearly, the fitness benefits for the two parties are not equal.

Axelrod and Hamilton (1981) improved upon Trivers’ model by iterating the prisoners’ dilemma and allowing the players to switch strategies at each iteration. Though this model is robust in it’s ability to describe human sociology, it is only marginally effective in describing biological mutualisms. As Doebeli and Knowlton (1998) point out, this model makes two assumptions that are frequently violated in biological mutualisms: (1) the players are in direct competition for resources with their partners, and (2) the payoffs associated with each strategy (cheat, cooperate, etc.) are constant.

It would be inappropriate to apply either of these assumptions to the Cecropia-Azteca-Coccidae symbiosis. As an example, the first assumption is violated in that one of the primary benefits to Cecropia from Azteca is the nitrogen contained in the ants’ fras. Fras is waste matter for which the ants have no use, therefore it would be inappropriate to model this system as if the ants were giving up a valuable resource for the benefit of the tree. We can observe the failure of the second condition if we consider that a "cooperative" coccids’ costs may considerably exceed its benefits until the nest is raided by coccid predators, at which point the coccids experience massive benefits from their association with the defensive ants.

Doebeli and Knowlton have developed a model of the evolution of interspecific mutualisms in which neither direct competition nor constant payoffs are assumed (1998). Their model instead assumes that (1) symbionts require different but overlapping ecological conditions, (2) randomly generated mutant symbionts can play new strategies, and (3) what a symbiont invests in one iteration is determined by the amount that it received in the prior iteration. As a result, this model (but not either of the models discussed above) allows for gradually increasing investments and rewards, which was almost certainly the case in the evolution of the Cecropia-Azteca-Coccidae system. Simulation runs of the Doebeli and Knowlton model showed that large fluctuations in the costs paid and benefits received by each partner do not destabilize the mutualism. Further, their model predicts that there will be great spatial heterogeneity in regard to costs paid and benefits received among given symbionts.

Doebeli and Knowlton enumerated six factors that favored the evolution of mutualism, three of which will be discussed in this paper: (1) Large population, (2) many mutualistic interactions per generation, and (3) high maximum benefit. Their model predicted increasing instability in relation to the following factors: (1) High costs, (2) non-local dispersal, (3) asymmetry in generation time, and (4) asymmetry in mutation rate or magnitude. This paper will focus on the first factor, as the latter three are impossible to estimate given the current state of knowledge.

For clarity sake, I will analyze this three-way symbiosis as a two-way mutualism between Azteca and Cecropia. For reasons that will be made clear below, I believe that the trees derive little or no benefit from the coccids; therefore the costs incurred on the tree by the coccids may be considered costs incurred by the ants. An in depth description of the mutualism between the ants and the mealybugs is beyond the scope of this paper. Therefore, I will briefly summarize this relationship at the end of the chapter.

The Doebeli-Knowlton model predicts that large population sizes will favor the evolution of biological mutualism. Both Azteca and Cecropia are abundant in lowland neotropical light gaps (Caroll 1983; Sagers et al. 2000; pers. obs.). Studies have found that between 47% and 85% of Cecropia house Azteca colonies (Vasconcelos and Casimiro 1997, Schupp 1986). Individual species of Cecropia differ widely in their propensity to house ants, as would be expected if investment in mutualism varied among species.

Evidence abounds that mutualistic encounters between the symbionts are nearly continuous. Janzen (1968) showed that Azteca ants incurred heavy damage on vines that encroached on their host tree within five days. The Coccids suck phloem from Cecropia throughout the day and the honeydew produced by the Coccids may be an important source of carbohydrates for adult Azteca workers (Carroll and Janzen 1973). Further, the continuous deposition of fras (ant feces) inside the trunk of Cecropia, provides 93% of the host tree’s nitrogen intake (Sagers et al. 2000).

Having satisfied two of Doebeli and Knowlton’s conditions for a stable mutualism, we may explore the third condition: high maximum benefits for both Cecropia and Azteca (Table 1).

In an elegant study, Sagers et al. used isotope analysis to prove that the flow of nutrients between Cecropia and Azteca, vastly favors the tree. This cost-benefit transaction demonstrates the need to consider the separate ecological needs of the players when modeling this system. The nitrogen provided by the ants is a byproduct of their metabolism and depositing it in the tree costs the ants nothing while delivering potentially large benefits to the tree. Further, this study is the first to prove that ants derive nutrients from sources other than their host tree, which provides a rationale for including separate ecological needs for the ants and the tree in our model.

Janzen (1968) was the first to suggest that Azteca acts as an allelopathic agent of Cecropia, meaning the ants inhibit competition from other plants. He found that across several species of Azteca and Cecropia, ants chewed encroaching vines, resulting in trees with ants carrying a significantly lighter vine load than unoccupied trees. Schupp (1986) performed an ant removal experiment and found that while 25% of trees had vines clinging to them at the beginning of the study, after two months, ant-occupied trees had no vines but 60% of unoccupied trees had vines. The fitness costs of vines (damage induced by weight and reduced growth from shading) can be great, and many plants produce allelopathic chemicals as a defense against vine encroachment. As Janzen pointed out, the allelopathic activity of Azteca may be more robust and resistant to co-evolution than any that could be provided by a chemical.

It has been shown that Azteca are effective at defending their host tree by reducing overall herbivory, and leaf cutting by leaf-cutting ants (Schupp 1986; Vasconcelos and Casimeros 1997). This defense provides measurable benefits to the tree, both in terms of growth and in early reproduction (Schupp 1986). It has been suggested that young Cecropia accrue greater benefits, though no one has performed a study comparing young and old plants.

Table 1: Both Azteca and Cecropia benefit greatly from their symbiotic association. The costs of this association are not of sufficient magnitude to destabilize the mutualism.

Much work has been done on the mechanism of defense, due in large part to de Andrade and Carauta’s (1982) observation of abundant herbivores on certain Azteca inhabited Cecropia. Mechanical disturbance, tree wounds, and the presence of volatiles released by damaged leaves release defensive behaviors by the ants including ejecting and eating herbivorous insects (Agrawal 1998; Agrawal and Dubin-Thaler 1999). Schupp (1986) demonstrated that Azteca are effective in mitigating nocturnal damage by coleopterans, but had no effect on Homopteran or cecidomyiid gall fly activity. The ants’ poor defense against these herbivores may have to do with the way in which Homopterans and gall flies penetrate, rather than chew, leaves. Agrawal’s (1998) work demonstrates that pinpricks applied to a Cecropia leaf induce far less defensive activity than do cuts applied to the leaf. I hypothesize that the ants do not receive the stimulation necessary to respond defensively to penetrating herbivores.

While it seems apparent that the ants benefit from the housing provided by the tree, it would be quite difficult to test this hypothesis since the species of Azteca that live inside Cecropia have never been found to nest outside of Cecropia. There are species of Azteca that do not live in Cecropia, but comparisons between these and their Cecropia-dwelling kin would be tenuous at best. At the very least, living in the hollow nodes, or "domatia," of Cecropia trees saves the ants the expense of hollowing out or building their own nest.

The ants’ most significant benefit from their mutualistic association with Cecropia comes in the form of nutrients. The tree produces Müllerian bodies that provide the ants with protein, lipids, and glycogen, a substance rarely found in plants whose sole function in Cecropia may be to nourish the ants (Beattie 1985). The tree also ‘allows’ mealybugs to suck its chemically undefended phloem, with which the bugs produce honeydew. The Azteca induce the mealybugs to excrete the honeydew and they then consume this sugar and amino acid-rich fluid. Interestingly, Cecropia are unusual among trees with advanced ant-defense mutualisms in that they do not have extrafloral nectaries (EFN’s). In summary, Azteca and Cecropia both receive significant, fitness enhancing benefits from their association, fulfilling the third of Doebeli and Knowlton’s criteria for stable biological mutualism.

Cecropia pays minor fitness costs for its mutualism with Azteca. Though it produces a large portion of the ant’s caloric intake, Cecropia’s investment in Müllerian bodies and phloem, may be relatively minor. O’Dowd (1980) estimated that Ochroma pyramidale, another neotropical ant plant, expends only 1% of the energy required to make each leaf on pearl bodies that feed defending ants. If this is the case for Cecropia, it appears that they are getting a fitness bargain. Further, ailing Cecropia continue to produce Müllerian bodies, indicating that the benefits provided by a healthy colony exceed the cost of feeding them, even under stress. The cost incurred by the mealybugs is unestimable with current data. While it may be argued that coccids act as vector for disease, one must consider that Azteca queens may bring a coccid with them when they colonize a tree (Beattie 1985). If this is the case , there is only a single opportunity for contamination in the life of the mutualism between a tree and an ant colony. Carroll (1983) describes an increased incidence woodpecker (Picidae) attack on trees containing Azteca. He does not specify the extent of this damage, but indicates that it may be limited to certain regions and certain times of year.

Though it is clear that ants pay fitness costs as a result of their mutualism with Cecropia, these costs are difficult to quantify given the current state of knowledge. Carroll’s (1983) observation of increased woodpecker attack on ant-occupied Cecropia suggests that ant colonies, as well as trees, suffer from these attacks more often than would be expected if they lived outside of the tree. Carroll, however, neglects to consider the ant-predators that are unable to attack Azteca because of its association with Cecropia. A cursory survey of Costa Rican mammals reveals five ant predators, none of which are capable of cracking the Cecropia’s trunk (life histories of mammals from Janzen 1983). Despite the lack of empirical evidence, it seems likely that the predator avoidance benefits of living in Cecropia outweigh the costs incurred by woodpecker attacks.

Carroll (1983) found that Azteca react aggressive to other colonies of ants attempting to colonize the same tree. Assuming that this aggression is a result of evolutionary pressure to colonize the entire tree as soon as is possible, one would conclude that Azteca colonies pay high costs (worker loss or colony failure) during nest establishment. To some extent, these costs to young queens may be mitigated by the protection that the young queens enjoy from non-ant predators. In a pilot study, Benjamin et al. (1999) found evidence that young trees may be more likely than older trees to host several colonies, though they do not specify the number of colonies found in the young trees. If future research quantifies the number of unsuccessful colonies for each successful colony, we will be better equipped to comment on the costs of intraspecific aggression resulting from the Azteca’s association with Cecropia.

To summarize, costs of the mutualism between Cecropia and Azteca may be small, as predicted by Doebeli and Knowlton’s model. However, the uncertainty surrounding the ant-paid costs incurred by high aggression keeps us from making a firm conclusion in this matter.

The mealybugs primary benefits of mutualism with Azteca are protection from predators and abiotic factors, a durable and high quality food source, dispersal, and hygiene (Beattie 1985). In turn, the coccids lose a high proportion of their dietary intake to the ants.

Azteca’s relationship to the coccids is much like the relationship between a subsitance farmer and his cows. The ants derive a great deal of their dietary intake from the coccids, but expend some resources caring for them. The one brief and informal study directly addresses the benefits of coccid tending to Azteca (Benjamin et al. 1999). The results of this study indicate that only the most successful Azteca colonies tend mealybugs. Further work is needed to determine whether this trend holds with larger sample sizes. It would also be interesting to explore the cause of this correlation through experimental removals of coccids.

Doebeli and Knowlton’s model of the evolution of biological mutualisms correctly predicts that the mutualism between Cecropia, Azteca, and Coccidae is evolutionarily stable. Though I was unable to consider all parameters of the model, I suggest that those that were considered provide an acceptable test of the model. The model’s assumption that symbionts have different but overlapping ecological needs is an especially important and widely applicable condition missing from other models of reciprocal altruism and mutualism. Interestingly, the model’s novel prediction that costs and benefits will fluctuate widely over a spatial scale is upheld in this system (e.g. compare Janzen 1968 to Schupp 1986).

Though much work has been done on the Cecropia-Azteca-Coccidae system, there are many gaps in the scientific literature. I suggest further research on the following: (1) the early stages of tree colonization, (2) how the ants acquire mealybugs, (3) differences in costs and benefits across trees of different ages, and (4) how the ants acquire outside nutrients.

Agrawal AA (1998) Leaf damage and associated cues induced aggressive and recruitment in a Neotropical ant-plant. Ecology 79(6): 2100-2112.

Agrawal AA and BJ Dubin-Thaler (1999) Induced responses to herbivory in the Neotropical ant-plant association between Azteca ants and Cecropia trees: response of ants to potential induced cues. Behav Ecol Sociobiol 45: 47-54.

Axelrod R and WD Hamilton (1981) The evolution of cooperation. Science 211: 1390-1396.

Beattie AJ (1985) The Evolutionary Ecology of Ant-Plant Mutualisms. Cambridge press, Cambridge.

Benjamin A, Giambruno M, Lyke M, Pyle AB, and Warner R (1999) A lesson in ABC’s of La Selva (Ants, Bugs, Cecropia). www.woodrow.org/teachers/esi/1999/costarica/projects/group2/ants/

Boucher DH, ed. (1985) The Biology of Mutualism: Ecology and Evolution. Oxford University Press, New York.

Carroll CR (1983) Azteca in Costa Rican Natural History, ed. Daniel Janzen. University of Chicago Press, Chicago.

Carroll CR and DH Janzen (1973) Ecology of foraging in ants. Annual Review of Ecology and Systematics. 4: 231-57.

De Andrade JC and JPP Carauta (1982) The Cecropia-Azteca association – a case of mutualism? Biotropica 14(1): 15.

Doebeli Mand N Knowlton (1998) The evolution of interspecific mutualisms. Proceedings of the National Academy of Sciences 95: 8676-8680.

Janzen DH (1968) Allelopathy by Myrmecophytes: The ant Azteca as an allelopathic agent in Cecropia. Ecology 50(1): 147-153.

O’Dowd DJ (1980) Pearl bodies of a neotropical tree, Ochroma pyramidale: ecological implications. American Journal of Botany 67(4): 543-549.

Sagers CL, Ginger SM, Evans RD (2000) Carbon and nitrogen isotopes trace nutrient exchange in an ant-plant mutualism. Oecologia 123:582-586.

Schupp EW (1986) Azteca protection of Cecropia: ant occupation benefits juvenile trees. Oecologia 70: 379-385.

Trivers RL (1971) The evolution of reciprocal altruism. The Quarterly Review of Biology 46: 35-57.

Vasconcelos HL and AB Casimiro (1997) Influence of Azteca alfari ants on the exploitation of Cecropia trees by a leaf-cutting ant. Biotropica 29(1): 84-92.