ABSTRACT

It is useful to predict metabolisable energy (ME) concentration based on crude nutrients which can be determined on a laboratory scale to formulate compound feeds for pigs based on ME concentration and to control the declared concentration. In 2008 such an equation was derived premised on 290 balance experiments showing strong associations between ME predicted by digestible crude nutrients and by crude nutrients themselves. Since the suitability of a regression-based prediction equation might be strongly influenced by the number of observations, the current study aimed at 1) checking the suitability of the existing prediction equation by including more datasets and 2) deriving a revised prediction equation.

The equations were evaluated by correlation and regression analyses using the energy content calculated on the basis of crude nutrients according to the previously used (ME_S) and the newly derived (ME_Snew) equations as well as the energy content calculated on the basis of digestible nutrients (ME_D). ME_D was correlated with ME_S (r_s = 0.784; p < 0.001) and ME_Snew (r_s = 0.802; p < 0.001). The root mean square error or the adjusted r²was 0.332 MJ/kg DM or 0.830 for the regression of ME_S on ME_D, and 0.323 MJ/kg DM or 0.839 for the regression of ME_Snew on ME_D. Although the regressive evaluation for the prediction of ME revealed satisfying results, the remaining residual variation not explainable by the regression model should be considered. The minimum span of the prediction interval of the regression of ME_S or ME_Snew on ME_D covered a range of 0.65 and 0.64 MJ/kg DM, suggesting the variability of ME estimations to be expected when based on crude nutrients. The quality parameters for the newly derived equation were minimally better and the correlation coefficient between ME_D and both, ME_Snew and ME_S, was strong. Since there is also a non-negligible inaccuracy in the estimation of ME content using the newly derived equation and as the quality parameters were only slightly better, there is at this point no need to introduce the new equation. In future studies, alternative analytical methods for determining the concentration of ME in compound feeds should be considered to improve the accuracy of estimation equations.

摘要

根据粗养分预测代谢能 (ME) 浓度是有用的，粗养分可以在实验室规模上确定，以根据 ME 浓度为猪配制配合饲料并控制声明的浓度。2008 年，基于 290 次平衡实验得出了这样一个方程，该实验表明可消化粗养分和粗养分本身预测的 ME 之间存在很强的关联。由于基于回归的预测方程的适用性可能受到观测次数的强烈影响，因此当前的研究旨在 1) 通过包含更多数据集来检查现有预测方程的适用性，以及 2) 推导修订后的预测方程。

使用根据先前使用的 (ME _S ) 和新推导出的 (ME _Snew ) 方程基于粗养分计算的能量含量以及基于可消化的能量含量计算的能量含量，通过相关性和回归分析对方程进行评估。营养素（ME _D）。ME _D与 ME _S ( r _s  = 0.784; p < 0.001) 和 ME _Snew ( r _s  = 0.802; p < 0.001) 相关。对于ME _S对 ME的回归，均方根误差或调整后的r ²为 0.332 MJ/kg DM 或 0.830_D和 0.323 MJ/kg DM 或 0.839 用于 ME _Snew对 ME _D的回归。尽管对 ME 预测的回归评估显示出令人满意的结果，但仍应考虑回归模型无法解释的剩余残差变化。ME _S或 ME _Snew对 ME _D回归的预测区间的最小跨度涵盖 0.65 和 0.64 MJ/kg DM 的范围，表明基于粗养分时 ME 估计的可变性是可预期的。新推导出的方程的质量参数_稍微好一点，ME _D与两者、ME _Snew和 ME _S之间的相关系数，很强。由于在使用新导出的方程估计 ME 含量时也存在不可忽略的不准确性，并且由于质量参数只是稍微好一点，因此此时无需引入新方程。在未来的研究中，应考虑确定配合饲料中 ME 浓度的替代分析方法，以提高估计方程的准确性。