A critical note on meal criteria in pigs: Which behaviours do they perform during feeder visit intervals?

Abstract

The feeding behaviour of pigs can be continuously recorded using sensors, providing promising avenues for automatic monitoring of pig performance and welfare. To utilise this potential, however, the data must be cleaned and aggregated meaningfully. A common aggregation is from visits into meals, in which visits separated by intervals shorter than a meal criterion (s or min) are merged. Methods to determine the criterion and the criteria themselves vary widely between studies, and have been applied indiscriminately or only when no pen mates visited the feeder during the interval. Aggregation choices should be biologically relevant, but there is no empirical knowledge on how pigs behave during these intervals or how the intervals are influenced by feeder competition. This study had three aims: 1) test the method that classifies intervals using a three-part probability density function (short, intermediate and long intervals); 2) determine whether feeder competition differed between interval types and application methods; and 3) describe and compare the behaviours of pigs between intervals. Visit intervals were obtained from 110 barrows in ten pens with one IVOGⓇ electronic feeding station each. A three-part probability density function was fitted to the log-transformed intervals, and its fit was assessed visually. For each pig, a short and an intermediate interval were selected for behavioural observations from camera. We found that pigs had relatively more intermediate intervals (1–28 min) than cows and that the fit of the three-part function was suboptimal. Nevertheless, identified meal criteria were in similar ranges as for other species. Intermediate intervals were more often initiated by displacements than short intervals (≤1 min), and there was more aggression and less pen exploration if pen mates visited the feeder during these intervals. Short intervals reflected interruptions in feeding behaviour, shown by standing (e.g. vigilance/chewing outside the feeder) and pen exploration (e.g. rooting, searching for pellets), while intermediate intervals contained non-feeding behaviours, such as social nosing, drinking and, predominantly, lying inactive. We conclude that intermediate intervals indicated completed feeding bouts, while short intervals reflected continued feeding-focused behaviours. Therefore, only visits separated by short intervals should be merged into meals. The exact criterion depends on the dataset but may, considering the suboptimal fit of the function, be more precisely determined using other methods. Whether visits should be merged indiscriminately or only when no pen mates entered the EFS during the interval depends on whether competition effects are of interest in the study.