The evolution of intergroup cooperation

Abstract

Sociality is widespread among animals, and involves complex relationships within and between social groups. While intragroup interactions are often cooperative, intergroup interactions typically involve conflict, or at best tolerance. Active cooperation between members of distinct, separate groups occurs very rarely, predominantly in some primate and ant species. Here, we ask why intergroup cooperation is so rare, and what conditions favour its evolution. We present a model incorporating intra- and intergroup relationships and local and long-distance dispersal. We show that dispersal modes play a pivotal role in the evolution of intergroup interactions. Both long-distance and local dispersal processes drive population social structure, and the costs and benefits of intergroup conflict, tolerance and cooperation. Overall, the evolution of multi-group interaction patterns, including both intergroup aggression and intergroup tolerance, or even altruism, is more likely with mostly localized dispersal. However, the evolution of these intergroup relationships may have significant ecological impacts, and this feedback may alter the ecological conditions that favour its own evolution. These results show that the evolution of intergroup cooperation is favoured by a specific set of conditions, and may not be evolutionarily stable. We discuss how our results relate to empirical evidence of intergroup cooperation in ants and primates. This article is part of a discussion meeting issue 'Collective behaviour through time'.

Keywords: coalitions; dispersal; intergroup conflict; intergroup cooperation; kin selection; peace.

Figure 1.

Conceptual overview of model. (a) Life cycle of modelled organism; (b) example of a fully occupied patch. Here groups are shown with size 2, but the model allows for any size group. (c) During the establishment and maturation phase, groups of pre-reproductives interact with probability y, and if they interact, one group is eliminated with probability σ.

Figure 2.

Selection gradient, frequency of patch types and force of selection as a function of long-distance dispersal, d, and the success rate of disruption, σ, for different types of patches. In patches with three resident groups, disruption is disfavoured because all groups are related to each other. In all other patch types, disruption may be favoured when dispersal is low and success rate is high. Disruption is most favoured in patches with two resident groups and one incoming group, because the two resident groups are related to each other but unrelated to the incoming group, and because the frequency of these patches is high for the range of parameter values where disruption is favoured. Parameter values: m = 0.6, c = 0.1, n = 2.

Figure 3.

Force of selection, fitness benefit of disruption, average relatedness within groups and average relatedness between groups as a function of long-distance dispersal, d, and the rate of disruption, σ, for different patch types, assuming fertility benefits. When disruption of a neighbouring group generates fertility benefits, the fitness benefits of disruption increase with dispersal. In such cases, direct fitness benefits outweigh indirect fitness benefits, and disruption is most favoured in patches with three (unrelated) groups when dispersal is high. High dispersal allows mothers to export the additional fertility benefit, and therefore avoid competition among kin offspring. Parameter values: c = 0.1, f₀ = 1, f₁ = 2, n = 2.

Figure 4.

Force of selection, relatedness within groups, and relatedness between groups as a function of long-distance dispersal, d, and the success rate of disruption, σ, for intermediate movement between local groups (m = 0.6) and for low movement between local groups (m = 0.1) in patches with two resident groups and one incoming group. Within the region above the dashed line in (a) the behaviour is cooperative, while below the dashed line the behaviour is altruistic. Little movement between local groups increases relatedness within groups but reduces relatedness between groups. By contrast, intermediate movement between groups equalizes relatedness within and between groups, which tends to favour altruism between groups, especially when rates of disruption are relatively low. Parameter values: c = 0.1, n = 2. (Online version in colour.)