Cheating interactions favor modularity in mutualistic networks

Abstract

A fundamental fact about mutualisms is they are often explored by species that explore resources and services provided by individuals without providing any benefit. The role of these cheaters on the evolutionary dynamics of mutualisms has long been recognized, but cheaters may not only affect the species they explore. Because mutualisms form networks that often involve dozens to hundreds of species in a given site, indirect effects generated by cheaters may cascade through the network, reshaping trait evolution. Here, we study how harboring cheating interactions can influence coevolution in mutualistic networks. We combine a coevolutionary model, data on empirical networks of mutualisms, and numerical simulations to show that the higher frequency of cheating interactions can lead to the formation of groups of species phenotypically similar to each other but distinct from other groups of species, leading to higher trait disparity. The clustered trait patterns generated by cheaters, in turn, change the patterns of interaction in simulated networks, fostering the formation of modules of interacting species. Our results indicate that cheaters of mutualisms can contribute to generate phenotypic clusters in mutualisms, counteracting selection for convergence imposed by mutualistic patterns, and favoring the emergence of modules of interacting species.