The use of a multivariate analysis to study between-goat variability in feeding behavior and associated rumen pH patterns

Abstract

This study examined feeding patterns and their relation to rumen pH. The measured variables were the cumulative time spent chewing (eating or ruminating) and the mean rumen pH, per 20-min intervals (46 intervals altogether). A total of 333 feeding patterns, defined as time-series of measurements for 1 animal and 1 given day during 15h and 20min following the afternoon feeding, were collected from 12 stall-housed dairy goats fed a total mixed ration twice daily. The first objective of this study was to analyze the within-day variability of feeding behavior to characterize different feeding patterns. The second objective was to analyze, for each goat, the between-day variability of its feeding behavior. The third objective was to study the influence of feeding behavior on rumen pH. A multivariate method (principal component analysis) was performed on the time spent eating or ruminating during each of the 46 intervals. The factor score plots generated by the principal component analysis highlighted the evolution in time of chewing activities. They also were used to determine 4 different types of feeding behavior in the extreme patterns of the first 2 axes. These feeding patterns were associated with different rumen pH patterns in goats with the same total dry matter intake and mean rumen pH. Some goats had a major eating period immediately after feeding, followed by a period of rumination; this pattern was associated with a marked decrease in rumen pH and a pH nadir below 6.0. Other goats performed alternate periods of eating and ruminating throughout the studied period. The latter feeding pattern was associated with a shorter total duration of rumination, but a more stable rumen pH than the feeding pattern described previously. Feeding pattern was quite constant between days for some goats, but highly variable for others. However, the continuum observed between these 4 groups shows that many intermediate types of behavior can be observed. In conclusion, characterizing animals according to their feeding behavior by using a principal component analysis may provide information on the animal's predisposition to rumen acidosis.