Updating predictions of dry matter intake of lactating dairy cows

摘要

Our objective was to model dry matter intake (DMI)by Holstein dairy cows based on milk energy (MilkE),body weight (BW), change in BW (ΔBW), body conditionscore (BCS), height, days in milk (DIM), andparity (primiparous and multiparous). Our databaseincluded 31,631 weekly observations on 2,791 cows enrolledin 52 studies from 8 states of the United States,mostly in the Upper Midwest. The means ± standarddeviations of these variables were 24 ± 5 kg of DMI,30 ± 6 Mcal of MilkE/d, 624 ± 83 kg of BW, 0.24 ±1.50 kg of ΔBW/d, 3.0 ± 0.5 BCS, 149 ± 6 cm height,and 102 ± 45 DIM. Data analysis was performed usinga mixed-effects model containing location, studywithin location, diet within study, and location andcow within study as random effects, whereas the fixedeffects included the linear effects of the covariates describedpreviously and all possible 2-way interactionsbetween parity and the other covariates. A nonlinear(NLIN) mixed model analysis was developed using a2-step approach for computational tractability. In thefirst step, we used a linear (LIN) model componentof the NLIN model to predict DMI using only datafrom mid-lactation dairy cows (76–175 DIM) withoutincluding information on DIM. In the second step, anonlinear adjustment for DIM using all data from 0 to368 DIM was estimated. Additionally, this NLIN modelwas compared with an LIN model containing a fourthorderpolynomial for DIM using data throughout theentire lactation (0–368 DIM) to assess the utility ofan NLIN model for the prediction of DMI. In summary,a total of 8 candidate models were evaluated asfollows: 4 ways to express energy required for maintenance(BW, BW~(0.75), BW adjusted for a BCS of 3,and BW~(0.75) adjusted for a BCS of 3) × 2 modelingstrategies (LIN vs. NLIN). The candidate models werecompared using a 5-fold across-studies cross-validationapproach repeated 20 times with the best-fitting modelchosen as the proposed model. The metrics used forevaluation were the mean bias, slope bias, concordancecorrelation coefficient (CCC), and root mean squarederror of prediction (RMSEP). The proposed predictionequation was DMI (kg/d) = [(3.7 + parity × 5.7) +0.305 × MilkE (Mcal/d) + 0.022 × BW (kg) + (−0.689+ parity × −1.87) × BCS] × [1 – (0.212 + parity ×0.136) × exp~((−0.053 × DIM))] (mean bias = 0.021 kg, slopebias = 0.059, CCC = 0.72, and RMSEP = 2.89 kg),where parity is equal to 1 if the animal is multiparousand 0 otherwise. Finally, the proposed model wascompared against the Nutrient Requirements of DairyCattle (2001) prediction equation for DMI using anindependent data set of 9,050 weekly observations on1,804 Holstein cows. The proposed model had smallermean bias and RMSEP and higher CCC than the NutrientRequirements of Dairy Cattle equation to predictDMI and has potential to improve diet formulation forlactating dairy cows.