Individual personality predicts social network assemblages in a colonial bird

Abstract

Animal personalities manifest as consistent individual differences in the performance of specific behavioural expressions. Personality research has implications for zoo animal welfare, as it can further our understanding of how captive individuals may differ in their resource use and provide insight into improving individual and group social health. For group living species, personality may enable assortment based on similar behaviour and influence an individual's interactions with conspecifics (e.g. social support). This research aimed to document how personality traits (aggressive, exploratory, submissive) influenced the social network structure of highly social animals in a captive environment. Data were collected from separate flocks of captive Caribbean (Phoenicopterus ruber) and Chilean flamingos (Phoenicopterus chilensis) to identify relationships between birds and examine opportunities for social support. The flocks associated non-randomly, and in both cases, personality was a substantial predictor of network structure. Personality also predicted key elements of Caribbean flamingo social role (degree, betweenness and average association strength) conflict outcome, and propensity to provide social support, however these patterns were not replicated within the Chilean flamingo network. While both species appear to assort by personality, the broader relationship between personality and social role may vary depending on species and context.

Figure 1

Comparison of the lag (blue) associations rates (the observed probability that two individuals which are associating during a particular sample session (day) will associate again in the future) and the null (green) associate rates (the expected lag association rate if the birds were randomly associating) for the Caribbean (a) and Chilean (b) flocks. In both cases, the lag association rates were statistically higher than the null associate rates (Caribbean, p = 0.026; Chilean, p < 0.001) suggesting non-random assortment.

Figure 2

The Bray–Curtis similarity matrices which compared how similar individuals were in their behaviour (aggressive, submissive, exploratory) were visualised for both the Caribbean (a) (2D stress = 0.14) and Chilean (b) (2D stress = 0.15) using non-metric multidimensional scaling (NMDS) using PRIMER 6 . This plotted each bird’s score in a 2D space and grouped them according to similarity. Birds which expressed similar behavioural patterns are located closer to each other on the graph. Hierarchical cluster analysis was applied to generate percentage threshold of similarity. All the Caribbean birds were at least 34% similar in their behaviour, while all the Chilean birds were at least 29% similar in their behaviour (represented by the outermost grey circle in each panel respectively). The 50% (green, solid circles) and 80% (blue, dashed circles) similarity thresholds are displayed to better identify sub-groups of birds which display increasingly similar behavioural patterns. Male (purple, ▲), female (green, ▼) and birds of unknown sex (grey, ●) are distinguishable by colour and/or shape.

Figure 3

Criteria for the recording of fight outcome in the collection of social support data.