THE DEVELOPMENT OF FEED INTAKE PREDICTION EQUATIONS FOR LACTATING DAIRY COWS
Data collected during two long-term experiments were analyzed by stepwise multiple regression to develop feed intake prediction equations. In one experiment (Sire comparison trial (SCT)) cows were individually fed complete rations ad libitum while in the other (lifetime silage trial (LST)) roughage was fed ad libitum and grain according to production. Each experiment covered three lactations. Milk production and feed intake were recorded daily, milk composition biweekly (LST) or monthly (SCT) and weight and feed composition monthly. The data included 100 SCT and 79 LST records representing 47 and 34 different cows, respectively. Only data from lactations of at least 280 days were used. The SCT cows were older (30 vs. 24.6 mo at first calving. Dry matter (DM) intake of the SCT cows increased linearly between lactations 1 and 2 but changed very little between lactations 2 and 3. Dry matter intake by the LST cows increased linearly over the three lactations. Each data set was used to formulate two prediction equations. Regression equations using 11 independent variables gave R2 values of 0.84 for both data sets. Equations, using five selected independent variables (lactation number, milk production, milk fat percent, body weight, and acid detergent fibre (ADF) content of ration) gave R2 values of 0.78 (SCT) and 0.77 (LST). The SCT equation described a negative relationship between ADF content and intake, while the LST equation showed a positive relationship.