219 Social network analysis of tail-biting behavior in growing-finishing pigs with intact tails.

Abstract Understanding the dynamics of tail biting in pigs, such as which pigs are involved and what roles these pigs play in this damaging behavior, can help us develop management strategies to minimize tail biting outbreaks. We evaluated social structures and positions of pigs involved in tail biting outbreaks (TBO) using social network analysis. Pigs (n=252, initial weight=28.9±3.4 kg) with intact tails were grouped (7 pigs/pen, 36 pens) based on litter origins (littermates, half-group littermates, non-littermates). Pigs were housed in barns with slatted floors (0.98m2/pig) for 12 weeks until market weight (123.1±11.9 kg). Behavior of pigs was recorded continuously using the NUtrack Livestock Monitoring System. Tail injury was evaluated weekly to monitor TBO. Thirteen pens (littermates=3 pens, half-littermates=5 pens, and non-littermates=5 pens) experienced TBO in which at least one pig in a pen had blood on the tail caused by tail biting. Videos for these pens, recorded the day before and during the first TBO, were manually reviewed to document tail biting events and identify the roles of the pigs involved. The top 25% of pigs responsible for the most tail-biting incidents were classified as tail-biters and others as non-biters. Network metrics were calculated using RStudio. Data were analyzed using Glimmix and NPAR1WAY procedures. No difference in network metrics at the pen-level (all P ≥ 0.128; Table 1) was detected among litter origins. Average density ranged from 0.27 to 0.33, indicating that 27 to 33% of possible pairs (dyads) of pigs engaged in tail biting. Among these pairs, 17 to 37% (reciprocity) were biting each other. Out-, in- and all-degree centralizations were closer to 0 than 1, indicating that tail biting was not performed mainly or received by few pigs. At the pig-level, no difference was detected in unweighted or weighted centralities among litter origins (all P ≥ 0.32). Compared to non-biters, tail-biters had greater unweighted out-degree, closeness, and betweenness, and greater weighted out-edge strength and betweenness centralities (all P ≤ 0.04, Table 2), indicating that tail-biters bit more pigs, were more connected to other tail-biters, and committed more tail biting events. No difference was detected in unweighted in-degree, weighted in-edge strength, out- and in-closeness centralities between tail-biters and non-biters. Both tail-biters and non-biters were bitten about 3 times (in-edge strength) by 1.76 pigs (unweighted in-degree centrality) on average, suggesting that tail-biters were bitten as often as non-biters. Gilts tended to have greater (all P ≤ 0.08) unweighted closeness and betweenness, weighted out-edge strength and betweenness centralities than barrows, suggesting that gilts were more closely connected with other pigs that were involved in tail biting and performed more tail biting than barrows. These results demonstrate that social network analysis can provide insight into the dynamics of tail biting among pigs.