Figure 3.

The recovery of previous social associations upon the reintroduction of removed birds. (a) Removed birds' previous dyadic social associations with other birds occurring at the same feeders as them (x-axis) predicted whether they were flockmates following reintroduction (y-axis) and (b) removed birds' previous dyadic social associations with their previous flockmates (x-axis) predicted their dyadic association strength with them following reintroduction (y-axis). In both panels, solid lines and surrounding hashed area show fit and standard error, respectively, of GLMMs including individuals' identities as a random effect (see electronic supplementary material, table S2 for full model results). Boxes in (a) show the raw data interquartile range (IQR), with mid-lines denoting median and whiskers indicating the range (excluding values 1.5 times outside of IQR). Points in (b) show raw dyadic social association strengths between removed birds and their flockmates. (Online version in colour.)