Space use and social networks: correlated but not congruent in California ground squirrels

Space use is widely assumed to be an important predictor of social behaviour given that animals are most likely to interact with nearby conspecifics. In the age of remote-sensing technology, spatial proximity is often used as a proxy for social interaction, although this simplifying assumption has rarely been tested under field conditions. Using 5 years of spatial and social data from direct observations of nearly 200 individually marked free-living California ground squirrels, Otospermophilus beecheyi, we employed social network statistics to evaluate the role of spatial relationships in generating individual differences in sociality. As predicted, range size was positively associated with each of the four social network metrics examined, confirming that more social squirrels generally had larger ranges. The proportion of range overlap between individuals was positively correlated with the strength of their social interactions, with the strongest ties occurring within age and sex classes and with juveniles generally being more social than adults. To determine whether these outcomes varied with interaction type, we examined the effects of spatial relationships on affiliative interactions, agonistic interactions and all interactions combined. We found that spatial relationships better predicted affiliative encounters compared to agonistic encounters. Despite revealing significant links between spatial and social relationships, our models explained only a small proportion of the variation in each social network metric examined. Thus, factors other than space use must contribute to social interactions, suggesting that general assumptions regarding the effects of space use on social interactions need to be evaluated on a species-by-species basis. More generally, these findings highlight the need for explicit consideration of the spatiosocial interface and its implications for multiple aspects of animal behaviour.