Dry matter intake prediction of Holstein heifers

TR Number



Journal Title

Journal ISSN

Volume Title


Virginia Tech


Data sets from six locations containing 631 dairy heifers and 5409 observations of individual dry matter intake (DMI) were used to develop an equation to predict DMI. Data were contributed by Pennsylvania State University, University of New Hampshire, Purina Mills, Kansas State University, University of Minnesota, and Virginia Tech. Objectives varied by location and trials within locations but all contained DMI (kg/d), BW (kg), age (wk), DM, and CP, and ADF percentages of ration DM. Daily gains (g) were calculated from consecutive BW. Season was expressed as sine wave of Julian day plus 60 d times Ï divided into 180. Average daily temperature, humidity, and wind data were collected from Minnesota and Virginia and analyzed with BW to determine their effect on DMI. Sire PTA milk, protein, and fat of heifers were recorded from all sources to determine the relationship of genetics with DMI. Diet measurements (DM, CP, and ADF percent of ration DM) were analyzed with BW to determine their impact on DMI. Two equations were developed using backwards elimination techniques. The first equation was: DMI = -12.63 + .0587(BW) - .0000264(BW2) - .000 12(BWxDM) - .000477(BWxADF) + .292(DM) - .00103(DM2) - .413(CP) + .01349(CP2) + .181(ADF) - .0025(DMxCP) - .00269(DMxADF) + .00509(CPxADF) with an R2 of .90. Because body weight and diet variables were highly correlated, a smaller model could be created with one diet variable. A smaller model would also be more practical to use if accuracy was not lost. Body weight and ADF were used for second model: DMI = -1.71 + .0429(BW) - .0000246(BW2) - .00023(BWxADF) + .032(ADF) - .00068(ADF2) with an R 2 of .87. Previous DMI prediction equations from Virginia Tech were validated using all data from other locations and had R2 of .90 and .84 with this data set. Body weight raised to the .53 power most accurately described the relationship of DMI and BW. Temperature and DMI had a quadratic relationship. Higher DMI were observed at extreme temperatures between -10 to 27° C. Humidity and temperature x humidity accounted for more variation of OMI than season and other environmental measures, but were not included in the small model due to availability of these measurements and they did not change OMI by .1 kg/SO. Sire PTA milk by groups showed differences among heifer groups although no trends were found. Dry matter percent and CP percent had linear and quadratic relationships with OMI. Maximum OMI plateaued between 75 and 950/0 OM and occurred at extreme percentages for CP between 9 and 27%. Acid detergent fiber percent had a negative linear relationship with OMI between 7 and 45%.