Comparison of rumination activity measured using rumination collars against direct visual observations and analysis of video recordings of dairy cows in commercial farm environments

Automated systems for monitoring the behavior of cows have become increasingly important for management routines and for monitoring health and welfare. In the past few decades, various devices that record rumination have been developed. The aim of the present study was to compare rumination activity measured with a commercially available rumination collar (RC) against that obtained by direct visual observations and analysis of video recordings in commercial dairy cows. Rumination time from video recordings was recorded by a trained observer. To assess observer reliability, data were recorded twice, and the duration of recorded behaviors was very similar and highly correlated between these 2 measurements (mean = 39 ± 4 and 38 ± 4 min/2 h). Measurements of rumination time obtained with RC when compared with analysis of video recordings and direct observations were variable: RC output was significantly positively related to observed rumination activity when dealing with animals housed indoors (trial 1 video recordings: slope = 1.02, 95% CI = 0.92-1.12), and the limits of agreement method (LoA) showed differences (in min per 2-h block) to be within −26.92 lower and 24.27 upper limits. Trial 1 direct observations: slope = 1.08, 95% CI = 0.62-1.55, and the LoA showed differences to be within −28.54 lower and 21.98 upper limits. Trial 2: slope = 0.93, 95% CI = 0.64-1.23, and the LoA showed differences to be within −32.56 lower and 19.84 upper limits. However, the results were poor when cows were outside grazing grass (trial 3: slope = 0.57, 95% CI = 0.13-1.02, and the LoA showed differences to be within wider limits −51.16 lower and 53.02 upper). Our results suggest that RC can determine rumination activity and are an alternative to visual observations when animals are housed indoors. However, they are not an alternative to direct observations with grazing animals on pasture and its use is not advisable until further research and validation are carried out.