Do Pigs Form Social Structures: An Application of Social Network Analysis?

As social animals, pigs in a group may form certain social structures which could be important to their welfare. The objective of this study was to use social network analysis to assess social structure and social position of pigs when grouped with littermates or non-littermates. Pigs (n = 96, initial wt = 6.93 ± 1.56 kg) weaned at 4 weeks of age were housed in 12 pens (8 pigs/pen) for 5 weeks which consisted of 6 pens of littermates and 6 pens of non-littermates. Pigs were weighed at birth, 4 weeks, and 9 weeks of age. Behavior of pigs was video-recorded for 6 h between 0900 and 1500 at 7 weeks and 8 weeks of age. Videos were scanned at 20-min intervals to register individual pigs that were lying together (1) or not (0) in binary matrices. Pigs that were lying together were considered to be socially connected. Social network analysis was performed using the UNICET software. Degree centrality (DC) and closeness centrality (CC) measure direct and indirect connections, respectively, that a pig has with its pen-mates. Network density describes how frequently pigs in a pen were lying together. Larger values for DC, CC, and network density indicate a greater degree of social connection. No differences in average DC (0.79 ± 0.23 for littermates vs. 0.83 ± 0.25 for non-littermates), CC (0.63 ± 0.34 vs. 0.73 ± 0.30), or network density (4.12 ± 0.96 vs. 4.34 ± 0.54) were found between litter origin treatments, indicating that pigs in pens of littermates connected with each other in a similar way to pigs in pens of non-littermates. Standard deviation of CC tended to be higher (P = 0.06) for littermates than non-littermates. This suggests that some littermates may form more indirect connections than other littermates, while non-littermates form similar indirect connections among themselves. Furthermore, CC was correlated with bodyweight at 4 weeks (r = 0.242, P < 0.05) and 9 weeks (r = 0.233, P < 0.05) of age. Network density was correlated with bodyweight at 9 weeks (r = 0.629, P < 0.05), 4 weeks (r = 0.559, P < 0.10) of age, and ADG (r = 0.568, P < 0.10). These data suggest that heavy pigs had more indirect connections and connected with their pen-mates more frequently than lightweight pigs. These preliminary results suggest that social network analysis may be a useful tool to measure social structure among pigs.