Evaluation of pasture herbage intake equations based on dairy cow behavior recorded with the RumiWatch system

Abstract

Accurate estimation of individual feed intake is essential for calculating feed efficiency, planning diets, monitoring cow herds, and managing grazing cows. This study aimed to evaluate the performance and applicability of estimation equations developed to predict pasture herbage DMI (PHDMI) in dairy cows using behavioral traits recorded and scored by the RumiWatch system. The study had 4 primary objectives: (1) to compare the behavioral characteristic outputs of 2 versions of the RumiWatch converter (0.7.3.31 and 0.7.3.36); (2) to validate existing PHDMI estimation equations using an independent dataset of Swiss Fleckvieh and Holstein cows; (3) to assess the suitability of the equations for estimating the intake of fresh herbage and hay fed indoors; and (4) to determine whether the RumiWatch halter impedes feed intake. Nine behavioral characteristics relevant to the PHDMI estimation equations were evaluated with both converters and found to be consistent, with negligible differences in the results. The validation of 6 of the PHDMI estimation equations appeared to provide individual intake predictions for grazing cows with low mean bias and may be suitable for herd-level assessments. For individual cow PHDMI estimation, the utility appeared limited, as only a very small proportion of PHDMI variability was explained by the equations. However, the calculated root mean square error of ~15% is in the acceptable range, according to the literature. As expected, the equations largely underestimated the intake of fresh forage and hay fed indoors, negating their suitability for estimating the intake of combinations of grazed herbage and barn-fed forages. A comparison of hay intake in cows with or without the halter showed no evidence that wearing a correctly fitted RumiWatch halter reduced hay intake. These results highlight the potential of the RumiWatch system and the associated estimation equations to support herd-level PHDMI and underscore the need for further testing to improve its performance in diverse feeding environments.