Estimating social network metrics from single-file movements in Barbary macaques, Macaca sylvanus

Abstract

Traditional methods for quantifying animal social network structure, and especially global-structural measures such as community structure, require large amounts of high-resolution data, which can be time-consuming and labour-intensive to collect. In this study, we investigated the efficacy of using a recently proposed, less effort-intensive method for collecting social association data based on the observed order of individuals in single-file movements. We used this method to estimate the social network of a group of semi-free-ranging Barbary macaques, Macaca sylvanus and then applied the Louvain community detection algorithm to estimate the community structure within the group. We validated the results by comparing them to networks informed by data from more traditional sampling methods for social network analysis, namely scan sampling and focal observations. Using Mantel tests with Spearman correlations, we found statistically significant but weak positive associations between the community assignments and dyadic association indices derived from the single-file movement data and those from the scan and focal data. Our findings do not provide convincing evidence that association data obtained from only 20 observations of single-file movements can reliably be used to estimate the strength of dyadic relationships or the composition of discrete communities within the larger group. However, the results from the community detection algorithm converged to similar estimates for the modularity value and the number of communities present in the group, irrespective of the data collection method used. We suggest that data from observations of single-file movements may not be useful for estimating fine-grained social network structure but may offer researchers a ‘quick and dirty’ method for determining whether meaningful community structure exists within their study groups as part of a pilot study and an efficient method for wildlife managers and conservationists to monitor population-level disturbances to social structure.