Development of equations to predict dry matter intake of dairy heifers

Equations to predict daily dry matter intake (DMI) of individual animals were developed with 118 Holstein heifers fed totally mixed rations (TMR). Animals grouped according to body weight (136, 227, 317 kg) were fed TMR (corn silage, ground hay, high moisture corn, soybean meal) once daily for 28 days. Rations were balanced to mean body weight according to National Research Council (NRC) recommendations for protein, vitamins and minerals, and to 85, 95, 105, and 115% of recommendation for energy. Independent variables were body weight (BWT, kg/day) daily body weight gain (GAIN, kg), environmental temperature (AMBT, °C), ration acid detergent fiber (ADF, percent of dry matter), neutral detergent fiber (NDF, percent of dry matter), net energy for maintenance and gain (NEM, NEG, megacalories), total digestible nutrients (TDN, percent of dry matter), and bulk density (BULK, g/ml as fed). Squared terms, appropriate interaction terms, and metabolic body weight (MBWT, kg^.75) were included in the model. Stepwise regression was utilized to generate two models, simplified and expanded. Simplified model was: DMI (kg/day) = -29.86 + (-.54E-05 * BWT²) + (.157 * MBWT) + (2.090 * GAIN) + (-.118 * GAIN²) + (.730 * TDN) + (-.005 * TDN²) + (-.001 * BWT * GAIN) + (-.019 * TDN * GAIN); n = 4797, r² = .59, s_y.x = 1.18. Expanded model was: DMI (kg/ day) = -1906.91 + (-0.04 * BWT) + (0.37 * MBWT) + (32.36 * ADF) + (2305.51 * NEM) + (-664.06 * NEG) + (-0.08 * AMBT) + (-0.13 * ADFSQ) + (-637.68 * NEMSQ) + (42.31 * NEGSQ) + (-5.35 * BULKSQ) + (0.001 * AMBTSQ) + (-1.56E-04 * BWT * ADF) + (8.873E-05 * BWT * AMBT) + (246.30 * NEM * NEG) + (-21.30 * NEM * ADF) + (7.83 * NEG * ADF) + (0.04 * NEG * AMBT) + (0.01 * GAIN * ADF) + (-0.01 * GAIN * AMBT); n = 4429, r² = .65, s_y.x = 1.09. Predicted intakes by simplified equation agree with those of NRC at TDN recommended by NRC and .7 kg/day. Computerized ration formulation systems were developed using simplified model. Expanded model requires further refinement prior to incorporation into a ration formulation system.