The direct and ecological costs of an ant-plant mutualism

Background/Question/Methods

Mutualism theory has emphasized mechanisms that prevent cheating, but cheating is favored over cooperation only when cooperation is costly. Here, we ask: what are the direct and ecological costs of participating in an ant-plant protection mutualism? On the plant side, compelling evidence for costs requires showing that in the absence of herbivores, investing in ants reduces plant fitness. Similarly, on the ant side, if patrolling and defending a plant against herbivores is costly, colonies should have higher fitness in the absence than in the presence of herbivores.

We recently investigated the costs of interactions between the Amazonian ant-plant Cordia nodosa and its ant symbiont Allomerus octoarticulatus in a field experiment. We manipulated the presence and absence of ants and herbivores on C. nodosa saplings (n = 52) such that each plant grew for nearly a year in one of four experimental treatments: 1) ants added, herbivores present, 2) ants added, herbivores excluded, 3) ants excluded, herbivores present, and 4) ants excluded, herbivores excluded. At the end of the experiment, we measured several correlates of ant colony and plant fitness and also quantified plant investment in chemical and physical defenses, to investigate trade-offs among different types of anti-herbivore defenses.

Results/Conclusions

We found that in the presence of herbivores, plants with ants grew significantly more than plants without ants, as expected. In the absence of herbivores, however, the pattern was reversed: plants without ants grew significantly more than plants with ants, showing that hosting A. octoarticulatus is costly to C. nodosa. This experiment likely underestimates the costs of mutualism to C. nodosa because some costs are fixed, such as the costs of producing domatia, which develop whether or not ants are present. On the ant side, we found no evidence for costs. Ant colony growth did not differ between treatments, but the presence of herbivores did result in larger workers, suggesting that A. octoarticulatus may benefit directly from eating herbivorous insects. These results show that in this system, cooperation is costly to the plant but probably not to the ants. Therefore, cheating by not defending host plants is unlikely to evolve in A. octoarticulatus, but C. nodosa may be under selection to cheat by reducing its investment in feeding or housing A. octoarticulatus colonies.