Animal social networks are robust to changing association definitions.

The interconnecting links between individuals in an animal social network are often defined by discrete, directed behaviours, but where these are difficult to observe, a network link (edge) may instead be defined by individuals sharing a space at the same time, which can then be used to infer a social association. The method by which these associations are defined should be informed by the biological significance of edges, and therefore often vary between studies. Identifying an appropriate measure of association remains a challenge to behavioural ecologists. Here, we use automatically recorded feeder visit data from four bird systems to compare three methods to identify a social association: (1) strict time-window, (2) co-occurrence in a group, and (3) arrival-time. We tested the similarity of the resulting networks by comparing the repeatability and sensitivity of individuals' social traits (network degree, strength, betweenness). We found that networks constructed using different methods but applying similar, ecologically relevant definitions of associations based on individuals' spatio-temporal co-occurrence, showed similar characteristics. Our findings suggest that the different methods to construct animal social networks are comparable, but result in subtle differences driven by species biology and feeder design. We urge researchers to carefully evaluate the ecological context of their study systems when making methodological decisions. Specifically, researchers in ecology and evolution should carefully consider the biological relevance of an edge in animal social networks, and the implications of adopting different definitions. Supplementary Information: The online version contains supplementary material available at 10.1007/s00265-025-03559-7.