Lying in spatial proximity and active social behaviours capture different information when analysed at group level in indoor-housed pigs

Social preferences, and long-term social associations in the form of social bonds, are commonly assessed through observation of affiliative behaviour and spatial proximity between group members. The aim of this study was to investigate the relationship between active social behaviours and lying in proximity in indoor-housed pigs in order to evaluate their suitability as indicators of social preferences. Social behaviours were recorded by scan sampling over five weeks (75 scans per pig pre-weaning and 240 scans per pig post-weaning), and spatial proximity was recorded while pigs were lying (over six weeks; & GE;10 observations per pig per week) with 23 groups pre-weaning and 12 groups post-weaning. At weaning, pigs were mixed with unfamiliar conspecifics, thus causing social instability, which was compared to the socially stable weeks pre-weaning and 3 weeks post weaning. Behaviour and spatial proximity were statistically analysed at group level using mixed models, with as predictor variables observation week (age) and group size (8-15 pigs); and ambient temperature in the model for spatial proximity. Nose-to-nose contact (1.3% of scans), allogrooming (0.5%) and agonistic behaviour (1.4%) were observed infrequently, whereas 'other non-agonistic social behaviour' was observed in 8.2% of the scan samples. Nose-to-nose contact, agonistic behaviour and other non-agonistic social behaviour peaked in the week after regrouping (all p < 0.001). Allogrooming increased from the 3rd week of life until around 1% (of scans); and occurred more frequently in larger groups (p = 0.03). For spatial proximity, pigs were observed in 48.2% of the observations lying in full body contact, in 42.8% partly in body contact, and only in 9% of cases lying alone. They were lying in 57.8% of the cases head-to-head. Pigs were less frequently observed lying in full body contact and head-to-head at higher temperatures, while lying more in part body contact (p < 0.05); and at 7 weeks of age they were lying more at distance from each other than when younger (p < 0.001). Active social behaviours all significantly correlated with each other (r = 0.67-0.84) but did not correlate with lying in proximity or lying orientation. Various methods have been proposed by others to aggregate affiliative behaviour and spatial proximity into an index, but the biological relevance of this in indoor-housed pigs is questionable as active social behaviours and lying in proximity appear to capture different and uncorrelated information when analyzed at group level.