A novel feeding behavior index integrating several components of the feeding behavior of finishing pigs

Describing the feeding behavior of pigs is difficult given the large day-to-day variations observed for a given animal. The objectives of this study were to create an index that integrates the information from several components of feeding behavior in order to account for intra-animal variation within a day and between days, and to evaluate the capability of this index to study the impact of relevant nutritional factors affecting feeding behavior. Feed intake information from 160 pigs during the last 28 d of the growing phase from three studies was used. For each pig, the sum of the absolute values of the deviation areas between the regression line of the relative cumulative feed intake and the observed cumulative feed intake was used to calculate the weekly index measuring the irregularity of feed intake (IIFI). Spearman's correlations of IIFI with the number of daily meals (r = -0.42; P < 0.001), meal duration (r = 0.38; P < 0.01), and feed intake per meal (r = 0.41; P < 0.01) indicate that pigs with high IIFI have fewer meals of longer duration and higher feed intake compared to pigs with low IIFI. This shows that IIFI captures information from several components of feeding behavior. Analysis of variance showed no effect of treatment on feeding behavior for datasets 1 and 2. However, the correlation between IIFI and dietary levels of CP (r = 0.34) indicates that diets high in CP were associated with pigs having more irregular meals. In dataset 3, pig feeding behavior was more regular in control diets than in pigs fed fibrous diets (IIFI; 164 vs. 197, respectively; P < 0.05). Additionally, IIFI was smaller in pigs fed canola by-product diets than in pigs fed wheat by-product diets, indicating that the source of fiber may also influence the feeding behavior of pigs. In most cases, IIFI was more effective at identifying differences in feeding behavior between dietary treatments than conventional feeding behavior variables. These results show the ability of IIFI to integrate information from several conventional components of the feeding behavior of pigs and its potential to successfully evaluate the effect of nutritional factors on feeding behavior.