Figure 5

Social networks by season. Nodes represent adult females and the thickness of edges corresponds to the SRI value. Isolates appear in the upper left corner of each network. Each row corresponds to a season and each column to a type of network. 'Ego' shows the ego-networks of selected subjects, who are indicated by black circles with colored borders. Node colors correspond to each Ego's Girvan-Newman cluster assignments in T1. These colors are maintained through all seasons to allow comparison across seasons (actual cluster designations in other seasons are not shown). In subsequent seasons, individuals colored gray are those who did not appear as Ego's companions in any preceding season covered by this study, although they may have associated prior to January 2007. Labeled nodes indicate those who associated with the subject in nearly every season. By the fifth season, networks clearly consist primarily of individuals who previously associated with the subject, even if not all were present in every season. 'Residents' shows all residents. One can track the coherence of each cluster over time. Some clusters from T1 maintain their integrity (e.g. brown) whereas others do not (e.g. light blue); associations among clusters also change over time, most notably the large pink and dark green clusters. Clusters are connected via just a few bridging individuals. 'Population' shows the full social network for each season with sample sizes reported in Table 2. Social networks constructed from real data have distinct structure, whereas those constructed from randomized data do not (see Additional file 5: Figure S5).