Figure legends:
Figure 1: Social networks across multiple situations. A hypothetical example of social interactions among vultures in different social situations: co-flying in blue, nocturnal ground interactions, (i.e., co-roosting) in green, and diurnal ground interactions (e.g., co-feeding) in yellow. Solid lines within each social situation indicate interactions within the social situation and black-dotted lines between social situations connect occurrences of the same individual. Dashed orange lines connect individuals to food sites to show how spatial proximity can be used to infer social interactions, for example when co-feeding. An aggregate network at the top, in gray, combines all interactions from the different social situations.
Figure 2. Inferring interactions from movements. (A) Location of study site - small red square - on a Satellite image of Earth (from GoogleEarth). (B) Satellite image (from GoogleEarth) of the study area in southern Israel (the Judean and Negev deserts) with example trajectories of vultures that interacted in two different social situations. Blue lines are movement trajectories of vultures that engaged in co-flight. The word ‘co-flight’ appears next to the two regions of their trajectories that overlapped in time and space - deeming them co-flight interactions. Green lines are movement trajectories of two vultures that roosted at the same roost overnight and therefore are considered to have a nocturnal ground interaction. The small green polygon to the bottom right of the word ‘Roost’ is the roost in which they both spent the night. Note that these two vultures did not engage in co-flight interactions. (C) Photo of a griffon vulture with a GPS tag attached to its back (circled) and wing tags (photo credit: Tovale Solomon).
Figure 3. Social networks in different social situations.Networks of 29 individually tagged vultures during the breeding season (December 2020 - June 2021). From left to right the situations include co-flight (blue), nocturnal ground interactions (green), diurnal ground interactions (yellow), and an aggregate network that combines interactions in all three situations (gray). Nodes depict individual vultures and the position of each node (individual) is maintained in all four networks. As an example, the identities of three individuals are specified in all four networks. Lines connecting nodes indicate that individuals interacted within a social situation, and line thickness corresponds to association strength.
Figure 4: Individuals differ in their social position across social situations. Annular representation of the interaction strength of 29 tagged vultures in the breeding season of 2021. Three rings represent the three social situations: co-flight (blue hue), nocturnal ground interactions (green hue), and diurnal ground interactions (yellow hue). The outer ring represents the aggregate network (gray hue). Darker shades indicate a higher rank of interaction strength. Each slice in the ring corresponds to one individual. Some individuals may be important in one social situation but not in others. For example, of the three individuals that are highlighted, J15w has a strength that is highly ranked in the diurnal ground interactions but not in the other social situations. Similarly, J36w is highly ranked in the nocturnal ground interactions but not in other social situations. Finally, individual J06w is highly ranked in co-flight interactions but not in other situations. Similar plots for degree and PageRank are provided in Fig. S1.
Figure 5. Relationship between centrality in the aggregate network and each social situation. Correlation between centrality measures (degree (A, D, G), strength (B, E, H), and PageRank (C, F, I)) in the aggregate network (x axes) and each of the three social situations (y axes): (A-C) co-flight, (D-F) nocturnal ground interactions, and (G-F) diurnal ground interactions. Each point is an individual vulture and points were jittered (0.4) along the x and y axes to improve readability in the plots of degree (A, D, G). Lines represent the correlations and the shading around the line is the 95% confidence interval computed using ggpubr() function of the ggplot R package (Wickham, 2011) with the lower confidence limit being 1±0.95/2 percentiles.
Figure 6. Comparing observed and expected correlations.Observed (diamonds) and randomized reference distribution (violins) of Spearman’s correlation coefficient (ρ ) between three centrality measures: degree (A), strength (B), and PageRank (C) in each of the three social situations: co-flight (blue), nocturnal ground interactions (green), and diurnal ground interactions (yellow). Vertical black lines on the violin plots depict the 95% quantiles of the reference distribution. Asterisks to the right of each plot denote statistically significant differences between the observed values and chance distribution using a two-sided test. Violin plots were created using ggplot2 (Wickham, 2011).