Manafiazar G, McFadden T, Goonewardene L, Okine E, Basarab J, Li P, Wang Z
Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, T6G2P5, Canada.
J Dairy Sci. 2013;96(12):7991-8001. doi: 10.3168/jds.2013-6560. Epub 2013 Oct 12.
Residual Feed Intake (RFI) is a measure of energy efficiency. Developing an appropriate model to predict expected energy intake while accounting for multifunctional energy requirements of metabolic body weight (MBW), empty body weight (EBW), milk production energy requirements (MPER), and their nonlinear lactation profiles, is the key to successful prediction of RFI in dairy cattle. Individual daily actual energy intake and monthly body weight of 281 first-lactation dairy cows from 1 to 305 d in milk were recorded at the Dairy Research and Technology Centre of the University of Alberta (Edmonton, AB, Canada); individual monthly milk yield and compositions were obtained from the Dairy Herd Improvement Program. Combinations of different orders (1-5) of fixed (F) and random (R) factors were fitted using Legendre polynomial regression to model the nonlinear lactation profiles of MBW, EBW, and MPER over 301 d. The F5R3, F5R3, and F5R2 (subscripts indicate the order fitted) models were selected, based on the combination of the log-likelihood ratio test and the Bayesian information criterion, as the best prediction equations for MBW, EBW, and MPER, respectively. The selected models were used to predict daily individual values for these traits. To consider the body reserve changes, the differences of predicted EBW between 2 consecutive days were considered as the EBW change between these days. The smoothed total 301-d actual energy intake was then linearly regressed on the total 301-d predicted traits of MBW, EBW change, and MPER to obtain the first-lactation RFI (coefficient of determination=0.68). The mean of predicted daily average lactation RFI was 0 and ranged from -6.58 to 8.64 Mcal of NE(L)/d. Fifty-one percent of the animals had an RFI value below the mean (efficient) and 49% of them had an RFI value above the mean (inefficient). These results indicate that the first-lactation RFI can be predicted from its component traits with a reasonable coefficient of determination. The predicted RFI could be used in the dairy breeding program to increase profitability by selecting animals that are genetically superior in energy efficiency based on RFI, or through routinely measured traits, which are genetically correlated with RFI.
剩余采食量（RFI）是能量效率的一种衡量指标。建立一个合适的模型来预测预期能量摄入量，同时考虑代谢体重（MBW）、空腹体重（EBW）、产奶能量需求（MPER）的多功能能量需求及其非线性泌乳曲线，是成功预测奶牛RFI的关键。在加拿大艾伯塔大学（埃德蒙顿，AB）的奶牛研究与技术中心记录了281头头胎奶牛从产奶第1天到第305天的个体每日实际能量摄入量和每月体重；个体每月产奶量和成分数据来自奶牛群改良计划。使用勒让德多项式回归拟合不同阶数（1 - 5）的固定（F）和随机（R）因素组合，以模拟301天内MBW、EBW和MPER的非线性泌乳曲线。基于对数似然比检验和贝叶斯信息准则的组合，分别选择F5R3、F5R3和F5R2（下标表示拟合阶数）模型作为MBW、EBW和MPER的最佳预测方程。所选模型用于预测这些性状的个体每日值。为了考虑体储备变化，将连续两天预测的EBW差异视为这两天之间的EBW变化。然后将平滑后的301天实际总能量摄入量与MBW、EBW变化和MPER的301天预测总性状进行线性回归，以获得头胎RFI（决定系数=0.68）。预测的每日平均泌乳期RFI均值为0，范围为-6.58至8.64兆卡净能（L）/天。51%的动物RFI值低于均值（高效），49%的动物RFI值高于均值（低效）。这些结果表明，可以根据其组成性状以合理的决定系数预测头胎RFI。预测的RFI可用于奶牛育种计划，通过选择基于RFI在能量效率方面具有遗传优势的动物，或通过与RFI具有遗传相关性的常规测量性状来提高盈利能力。
