Results
All 29 individuals interacted with at least one other vulture in all three social situations: co-flight, diurnal, and nocturnal ground interactions (Fig. 3, Table S1). The centrality of individuals differed across social situations. Individuals differed in their importance across social situations in terms of their degree, strength, and PageRank. For example, a vulture that interacted with multiple vultures while co-flying did not necessarily interact with as many individuals while on the ground (Fig. 4).
The centrality of individuals in the aggregate network did not necessarily reflect their centrality in each social situation. The degree of individuals in co-flight and diurnal ground interactions were positively correlated with the degree in the aggregate network (Spearman’s correlation: co-flight: ρ = 0.78, p-value < 0.0001 and diurnal ground interactions: ρ = 0.68, p-value < 0.0001; Fig. 5A, 5G). However, degree in nocturnal ground interactions was not significantly correlated with degree in the aggregate network (Spearman’s correlation: ρ = -0.17, p-value = 0.38; Fig. 5D). Both strength and PageRank in each of the three social situations were positively correlated with the strength and PageRank in the aggregate network (Spearman correlations for strength of aggregate with: co-flight ρ = 0.44, p-value = 0.02; nocturnal ground interactions ρ = 0.8, p-value < 0.0001; and diurnal ground interactions ρ = 0.89, p-value < 0.0001; Figs. 5B, E, H and Spearman correlations for PageRank of aggregate with: co-flight ρ = 0.39, p-value = 0.04 ; nocturnal ground interactions ρ = 0.8, p-value < 0.0001; and diurnal ground interactions ρ = 0.89, p-value < 0.0001, Figs. 5C, F, I).
The correlation between centrality in the aggregate network and each social situation was often different than expected by chance when compared with the reference models. Degree both in the co-flight and diurnal ground interactions was positively correlated with the degree in the aggregate more than expected by chance (Permutation test: co-flight p-value < 0.0001 and diurnal ground interactions p-value =0.002; Fig. 6A). However, the correlation between degree in nocturnal interactions and degree in the aggregate network did not differ from what was expected by chance (Permutation test: p-value = 0.791; Fig. 6A). Strength in the nocturnal and diurnal ground interactions was positively correlated with strength in the aggregate network more than expected by chance (nocturnal ground interactions: p-value = 0.006 and diurnal ground interactions: p-value = 0.009; Fig. 6B). However, although the relationship between strength in co-flight and strength in the aggregate network were correlated (Fig. 5B) this correlation did not differ significantly from chance expectation (co-flight: p-value = 0.07; Fig. 6B). PageRank in the nocturnal and diurnal ground interactions was positively correlated with PageRank in the aggregate network, more than expected by chance (nocturnal ground interactions: p-value = 0.006 and diurnal ground interactions: p-value = 0.004, Fig. 6C). However, the positive relationship between PageRank in co-flight and PageRank in the aggregate network (Fig. 5C) was not significantly different than expected by chance (co-flight: p-value = 0.115; Fig. 6C).