Understanding how cooperative interactions evolve and persist remains a central challenge in biology. Many mutualisms are thought to be maintained by ‘‘partner fidelity feedback,’’ in which each partner bases their investment on the benefits they receive.
Yet, we know little about how benefits change as mutualists vary their investment, which is critical to understanding the balance between mutualism and antagonism in any given partnership. Using an obligate ant–plant mutualism, we manipulated the density of symbiotic acacia ants (Crematogaster mimosae) and examined how the costs and benefits to Acacia drepanolobium trees scaled with ant abundance. Benefits of ants to plants saturated with increasing ant abundance for protection from branch browsing by elephants and attack by branch galling midges, while varying linearly for protection from cerambycid beetles. In addition, the risk of catastrophic whole-tree herbivory by elephants was highest for trees with very low ant abundance. However, there was no relationship between ant abundance and herbivory by leaf-feeding invertebrates, nor by vertebrate browsers such as giraffe, steinbuck, and Grant’s gazelle. Ant abundance did not significantly influence rates of branch growth on acacias, but there was a significant negative relationship between ant abundance and the number of fruits produced by host plants, suggesting that maintaining high-density ant colonies is costly. Because benefits to plants largely saturated with increasing colony size, while costs to plant reproduction increased, we suggest that ant colonies may achieve abundances that are higher than optimal for host plants. Our results highlight the conflicts of interest inherent in many mutualisms, and demonstrate the value of examining the shape of curves relating costs and benefits within these globally important interactions.