Social networks in primates: smart and tolerant species have more efficient networks

Abstract

Network optimality has been described in genes, proteins and human communicative networks. In the latter, optimality leads to the efficient transmission of information with a minimum number of connections. Whilst studies show that differences in centrality exist in animal networks with central individuals having higher fitness, network efficiency has never been studied in animal groups. Here we studied 78 groups of primates (24 species). We found that group size and neocortex ratio were correlated with network efficiency. Centralisation (whether several individuals are central in the group) and modularity (how a group is clustered) had opposing effects on network efficiency, showing that tolerant species have more efficient networks. Such network properties affecting individual fitness could be shaped by natural selection. Our results are in accordance with the social brain and cultural intelligence hypotheses, which suggest that the importance of network efficiency and information flow through social learning relates to cognitive abilities.

Figure 1. Networks of four different species depict variations between groups and network efficiencies (Global noted as E1 and Average Dyadic noted as E2).

Size and colour of nodes are linked to individual centrality. The bigger and the bluer the node, the higher the centrality. CI indicates centralisation index and Q is for modularity. We chose the four groups for the variances in networks measures. As a consequence, the group size of these four examples are not representative of the mean species group size. We acknowledged C.S. for permission to use photographs.

Figure 2. Influence of socio-species variables on network properties.

All linearized models with statistical values are detailed in the supplementary information. (a.) Neocortex ratio is positively correlated with Global Efficiency. (b.) Centralisation index is negatively correlated with Average Dyadic Efficiency. (c.) Modularity is positively correlated with centralisation index. (d.) Group size is negatively correlated with Global Efficiency.

Figure 3. Representation of the dynamic relationship between social networks, efficiency (information or disease flow) and individuals.

Individual characteristics influence social networks through their effects on social relationships, and also network efficiency through variation both in individual behaviour and individual preferences for sociality. These are emergent properties because the network is more than the sum of individual interactions; therefore its properties are not directly traceable by studying only behavioural interactions. As feedback, network efficiency could influence the behaviour of individuals to be more central in the network or favour information flow. Selective pressures (ecological or social) have direct effects on how individuals interact, associate and on the overall social network, and thus on sociality and efficiency. These three different levels have a direct effect on individual fitness, which influences individual characteristics through natural selection. This overall schema shows how natural selection at the individual level can favour upper-level structure such as social networks and their efficiency.