Validation of behavioural sampling techniques for 20-25 kg pigs during transport

Abstract

Monitoring behaviour during transport is challenging due to the design of vehicles and often limited deck height. The aim of the present study was to validate behavioural sampling techniques during transport. This study used data from four journeys, constituting a subset of 32 experimental journeys in a commercial vehicle transporting weaners of 20–25 kg for 23 h. In each journey, behaviour was observed in two compartments holding pigs, each with a deck height of either 60 cm or 90 cm.Body posture (upright, lying, sitting) was recorded using instantaneous scan sampling at 1 min intervals (considered ‘true’ occurrence). Based on these, less frequent sampling intervals were created (5, 10, 15, 30, and 60 min) and the resulting percentages of behaviour per hour were compared. Occurrences of inter-pig aggressive events were recorded using continuous sampling. It was investigated whether the occurrence of aggression when observed during selected recording sessions per hour (20, 30 and 40 min/h) differed from observing the whole hour throughout the journey (the ‘true’ occurrence).Two methods were used to compare sampling techniques. First, estimates of body posture and aggression were pairwise compared with the ‘true’ occurrence using regression metrics (RMSE, MAE, R2). The results showed that instantaneous scan sampling of body posture at 15 min intervals or less did reflect the ‘true’ values, whereas for aggression use of shorter recording sessions of 20 min/h generally did not reflect the ‘true’ values.Secondly, similar results were achieved by using mixed effects logistic regression. These models showed that sampling every 30 min differed from the ‘true’ values, whereas scanning in 5-, 15- or 60 min intervals did not. Compared with sampling the whole hour, the total duration of aggression was significantly underestimated in all three recording sessions (20, 30 and 40 min/h), while the total frequency of aggression was only significantly underestimated in 20 min/h recording session.Based on results of this study, the use of instantaneous scan sampling intervals of 15 min or less is considered valid for analyses of body postures during long distance transport of 20–25 kg pigs. Due to the sporadic occurrence of aggressive interactions, continuous sampling using recording sessions of 20 min cannot be recommended. Using longer recording sessions of 30 or 40 min/h should be approached with caution dependent on the specific scientific context.