PREDICTING AD-LIBITUM DRY-MATTER INTAKE AND YIELDS OF JERSEY COWS

Two data files were used that contained weekly mean values for ad libi tum DMI of lactating Jersey cows along with appropriate cow, ration, a nd environmental traits for predicting DMI. One data file (n = 666) wa s used to develop prediction equations for because that file represent ed a number of separate experiments and contained more diversity in po tential predictors, especially those related to ration, such as forage type. The other data file (n = 1613) was used primarily to verify the se equations. Milk protein yield displaced 4% FCM output as a predicti on variable and improved the R(2) by several units but was not used in the final equations, however, for the sake of simplicity. All equatio ns contained adjustments for the effects of heat stress, parity (1 vs. >1), DIM >15, BW, use of recombinant bST, and other significant indep endent variables. Equations were developed to predict DMI of cows fed individually or in groups and to predict daily yields of 4% FCM and mi lk protein; equations accounted for 0.69, 0.74, 0.81, and 0.76 of the variation in the dependent variables with standard deviations of 1.7, 1.6, 2.7, and 0.084 kg/d, respectively. These equations should be appl ied to the development of software for computerized dairy ration balan cing.