Novel measures to investigate the social structure and reproductive performance in a dynamic sow herd using social network analysis.

The transition from stalls to group housing has increased agonistic behaviours, with a detrimental impact on welfare and productivity. Understanding social behaviours is pivotal for improving welfare and extensive research has aimed at reducing aggression. Recently, there has been a shift in perspective to consider the impact of socio-positive interactions on the welfare of captive species. The complexities of how individuals impart effects require a technique that transitions to a group-level. Social network analysis (SNA) has developed as a valid technique in animal behaviour, although comparatively few studies have focussed on commercial animals. Understanding the stability of behaviour enables predictions of future behaviour to be made, which could inform management strategies. This thesis investigated the social structure of a dynamic breeding sow herd as a model species in preferential association and agonistic networks over three production cycles. The study examined the impact of individual behaviour on network composition, evaluating the stability of behaviour and the effect of social position on reproductive performance. SNA identified socially prominent and influential sows in every preferential and aggression network. In the preferential networks, there was an indication of discriminatory selective behaviour towards sows that were more highly connected. Although a lack of reciprocation showed no formation of sustained friendships, indicating alternative motivations for selective interactions. Prominence was not revealed as stable in the preferential association or agonistic networks. By comparison, influence was more stable in the aggression networks. An association was demonstrated between prominence, stillbirths and crushing. Results indicated a relationship between long-term stability of social position and reproductive performance. Finally, the novel approach of implementing brokerage typologies provided a unique perspective on the composition of the networks, demonstrating how sows are transferring aggression within and between identified subgroups.