Developing a regression equation model to predict individual water intake in Holstein growing cattle using reticulorumen temperature

The aim of this study was to develop an equation model for predicting individual water intake (WI) in Holstein growing cattle by calculating the area under the curve (AUC) incorporating reticulorumen temperature from a wireless bolus sensor. A total of 21 female Holstein growing cattle aged 11.3 ± 2.06 mo and weighing 346 ± 5.2 kg were used in this study. Bolus sensors were fitted to all animals, and water was given at 09:00 am, 01:00 pm, and 05:00 pm at a temperature of 20 ± 1.4 °C for 1 h. The actual WI was recorded by weighing the volume of residual water in the bucket each time the animals drank water. The accuracy of the regression equation model was evaluated by mean absolute percentage error (MAPE) analysis. In study 1, prior to conducting an evaluation of the correlation between AUC and the actual WI, the determination of the end-temperature point on the 30 graphs generated by the reticulorumen temperature change in response to the WI behavior of 3 cattle was necessary. We observed that a recovery rate of 98% compared to the start-temperature point was the optimal point with a goodness-of-fit R2 of 0.8067. In study 2, considering that the reticulorumen temperature may be influenced by factors of feeding and space conditions, the coefficient determination in the regression between AUC and actual WI was performed using 12 cattle divided into 4 groups according to whether they were restricted or unrestricted in feed and space. The investigation revealed no significant difference (P > 0.05) in the change of reticulorumen temperature after WI in cattle housed in either free or confined feed and space. Based on these data, we established a regression equation model to forecast WI by combining AUC and actual WI (forecasted WI = 0.0632 × AUC + 3.8936, R2 = 0.7473, MAPE = 26.01%). In order to confirm the regression equation model for forecasting WI, the MAPE was introduced to assess the accuracy of the forecasted WI compared to the actual WI, yielding a value of 75.39% (study 3). In conclusion, the bolus sensor can be used to measure reticulorumen temperature changes in response to water consumption to predict individual WI in Holstein growing cattle and monitor ruminant health.