Comparison of equations for predicting voluntary intake by dairy cows

TLDR: Rations for dairy cows are often formulated using predictions of voluntary feed intake calculated from regression equations, but the more complex equations of Vadiveloo & Holmes (1979) and of Lewis (1981) give consistently the best predictions.

Abstract: Summary Rations for dairy cows are often formulated using predictions of voluntary feed intake calculated from regression equations. The accuracy of the predictions of dry-matter intake by seven equations is investigated. Comparisons are made when live weight is taken to be the observed weekly mean (MLW), the observed live weight after calving (CLW) or an estimated breed weight accompanied by a notional pattern of live-weight change (BLW). Data recorded on a British Friesian dairy herd at the Grassland Research Institute fed mostly silage ad libitum and various supplements are used. The mean square prediction error (MSPE) is calculated for each week and summarized over the whole experimental period. The least MSPE's are 2·1, 2·8 and 2·4 (kg D.M.)2 for comparisons using MLW, CLW and BLW respectively. The Ministry of Agriculture, Fisheries and Food (1975) equation involving only live weight and milk yield performs well and would be useful when only the limited measurements of live weight and milk yield are available, but the more complex equations of Vadiveloo & Holmes (1979) and of Lewis (1981) give consistently the best predictions. The importance of using the original definitions of the variables when applying a predictive equation is illustrated. Even so, the large errors found in the predictions of intake points to the need for further research.