Play fighting social network position does not predict injuries from later aggression between pigs

Play fighting is common in young mammals. Proposed functional benefits include enhanced motor, social and cognitive development whereby, in some species, improved physical skill and opponent assessment are expected. This could reduce social conflict later in life and may contribute to improved welfare in managed animals. Using pigs as a model we examined whether variation in play fighting involvement before weaning was predictive of the number of injuries received from aggressive interactions later in life. Piglets (n=239) experienced either pre-weaning socialisation (n=135; joining two neighbouring litters at 2 weeks old; ‘socialised’) or remained in their litter group (‘control’; n=104). Sows were kept in farrowing crates. Play fighting interactions were extracted from video between the ages of 14-26d. Social network analysis quantified the centrality position of pigs in the play fighting network of their pen. Centrality (degree centrality, Eigenvector, betweenness, clustering coefficient) in play fighting networks quantified metrics such as a pig’s tendency to select play partners who themselves had experienced play with many partners. The effect of social network parameters on the number of skin lesions at 11 weeks of age was assessed 24h after pigs were mixed into new groups (n=8 socialised groups; 94 pigs; 5 control groups; 63 pigs) and at 3 weeks post-mixing (14 wk age). Prior experience of other tests which may have led to variation in fighting experience did not affect the test outcome and was omitted from subsequent analysis. Assortment analysis showed that in 2 out of 6 socialised groups, piglets positively assorted by litter (P<0.001), indicating a strong preference to play with their own littermates. At the pen level, play network structures were similar in socialised and control groups. Large variation existed in play fighting engagement and individual network position between pigs (e.g. 1-19 play partners; median clustering coefficient 0.78, range 0.19-1.00 which is expressed on a 0 to 1 scale). Mixed model analysis showed that pigs with many play partners had more lesions at 14 (F1,46=4.11, p=0.05) but not 11 wk. Centrality metrics that described the animal’s position within its play network (e.g. whether it preferentially played with experienced play partners) did not affect skin lesions (F1,88=0.06-1.76; p=0.18-0.95). These results suggest that play fighting does not reduce costs of fights in immature pigs. Future work should explore if it promotes social cohesion or reduces conflict between adults where fitness costs of aggression may be higher. The work was approved by the UK government and adhered to ISAE ethical guidelines.