Using phylogenetic comparative methods to gain insight into the evolution of social complexity

Abstract: Social complexity can have important evolutionary implications, and phylogenetic comparative methods have long been used as a way to assess interspecific support for hypotheses on the subject. Statistical approaches for conducting such analyses have come a long way in the last few decades, and it is now possible to use comparative data to ask sophisticated questions about the evolution of complex phenotypes while taking also uncertainty in the phylogeny into account. Here, we illustrate these advances by applying a phylogenetic analysis of covariance to two data sets addressing the evolution of social complexity in primates. First, we find that social complexity (as inferred by group size) is predicted by whether a species is diurnal and terrestrial. Second, we find that social complexity (as inferred by social organization) has increased over evolutionary time and can impact trait evolution. Particularly, we find that baculum length has been growing with body size in dispersed, pair-living and, more weakly, multi-male species, becoming relatively smaller and more variable in aggregations that are more socially stable. Significance statement: Sociality poses a particular challenge for interspecific analyses because of the many ways that its intrinsic complexity can be coded into statistical variables and models. However, with methods that incorporate phylogenies and nuanced inferences about underlying evolutionary processes, even coarse measures of social complexity can offer valuable insights in the study of behavioral trends through time. Exploring two interspecific data sets on primates, we show how simple measures of sociality, either coded as continuous or categorical variables, can be used to revisit latent evolutionary hypotheses while revealing new pieces of information that become available thanks to the added complexity of novel phylogenetic comparative methods. One example reveals that larger and more variable groups are associated with shifts in habitat and activity patterns, while the other reveals that baculum evolution might be influenced by the increasing social stability of primate aggregations.