Social interaction, and not group size, predicts parasite burden in mammals.

Although parasitism is often considered a cost of sociality, the evidence is mixed, possibly because sociality is multivariate. Here we contrast the dependence of parasitism costs on major social variables such as group size and social structure, as measured by network metrics. We conduct two robust phylogenetic meta-analyses, comprising 43 published results for studies with group size and 32 results with social structure metrics. This is the first meta-analytical test of this hypothesis for mammals as a whole. Contrarily to theoretical expectations and previous meta-analyses, there is no relationship between group size and parasitism, but we find conflicting results when analysing different aspects of sociality. Our analysis reveals that social structure is connected to parasite load, possibly because contact between group members, and not group size, is linked to parasite transmission. While more intensely interconnected groups facilitate parasite transmission, large groups are frequently fragmented into smaller, weakly connected subgroups. Strong social modularisation should thus be favoured by natural selection to hamper parasite overload. Future empirical studies should focus on specific parameters of social network structure and on parasite transmissibility. If social structure can evolve fast, even culturally, then host/parasites evolutionary games enter into a whole new fast dynamics, and animal conservation studies should take advantage of this possibility. [ABSTRACT FROM AUTHOR]