Development and evaluation of equations for prediction of feed intake for lactating Holstein dairy cows

Improved prediction equations for dry matter intake (DMI) of Holstein cows that consume high energy diets were developed using regression techniques applied to a comprehensive database. The equations for predicting DMI, which were dependent on parity, accounted for the effects of milk yield, milk protein, body weight (BW), BW change, days pregnant, ambient temperature, relative humidity, and night cooling. A simplified prediction equation of DMI for farm application was developed and based on milk protein yield and BW at calving. An ambient temperature and a lag adjustment factor for early lactation were developed to improve accuracy of prediction of DMI of dairy cows in early lactation. The developed equations for DMI were evaluated against six independent data files. These equations accounted for 55 to 98% of the variation of the weekly group DMI of the independent validation data. The remainder of the variation in intake was attributed to diet, management, and undescribed animal factors. The equations developed in this study had a mean proportional bias of 5.6% and a mean square prediction error of 5.45 kg2/d. Predicted intake using the new equations was within 3 to 8% of actual intake. The new equations must be applied to situations in which Holstein dairy cows are fed highly digestible diets because dietary fill effects are not considered in these equations. The relationship of milk protein yield and DMI warrants further investigation.