Mehtiö T, Rinne M, Nyholm L, Mäntysaari P, Sairanen A, Mäntysaari E A, Pitkänen T, Lidauer M H
Natural Resources Institute Finland (Luke), Green Technology, Jokioinen, Finland.
Valio Ltd., Farm Services, Valio, Finland.
J Anim Breed Genet. 2016 Apr;133(2):115-25. doi: 10.1111/jbg.12183. Epub 2015 Sep 27.
This study was designed to obtain information on prediction of diet digestibility from near-infrared reflectance spectroscopy (NIRS) scans of faecal spot samples from dairy cows at different stages of lactation and to develop a faecal sampling protocol. NIRS was used to predict diet organic matter digestibility (OMD) and indigestible neutral detergent fibre content (iNDF) from faecal samples, and dry matter digestibility (DMD) using iNDF in feed and faecal samples as an internal marker. Acid-insoluble ash (AIA) as an internal digestibility marker was used as a reference method to evaluate the reliability of NIRS predictions. Feed and composite faecal samples were collected from 44 cows at approximately 50, 150 and 250 days in milk (DIM). The estimated standard deviation for cow-specific organic matter digestibility analysed by AIA was 12.3 g/kg, which is small considering that the average was 724 g/kg. The phenotypic correlation between direct faecal OMD prediction by NIRS and OMD by AIA over the lactation was 0.51. The low repeatability and small variability estimates for direct OMD predictions by NIRS were not accurate enough to quantify small differences in OMD between cows. In contrast to OMD, the repeatability estimates for DMD by iNDF and especially for direct faecal iNDF predictions were 0.32 and 0.46, respectively, indicating that developing of NIRS predictions for cow-specific digestibility is possible. A data subset of 20 cows with daily individual faecal samples was used to develop an on-farm sampling protocol. Based on the assessment of correlations between individual sample combinations and composite samples as well as repeatability estimates for individual sample combinations, we found that collecting up to three individual samples yields a representative composite sample. Collection of samples from all the cows of a herd every third month might be a good choice, because it would yield a better accuracy.
本研究旨在获取关于通过近红外反射光谱（NIRS）扫描不同泌乳阶段奶牛粪便斑点样本预测日粮消化率的信息，并制定粪便采样方案。利用NIRS从粪便样本中预测日粮有机物质消化率（OMD）和不可消化中性洗涤纤维含量（iNDF），并以饲料和粪便样本中的iNDF作为内部标记物来预测干物质消化率（DMD）。使用酸不溶性灰分（AIA）作为内部消化率标记物作为参考方法来评估NIRS预测的可靠性。在产奶量约50、150和250天（DIM）时，从44头奶牛采集饲料和复合粪便样本。通过AIA分析的奶牛特异性有机物质消化率的估计标准差为12.3 g/kg，考虑到平均值为724 g/kg，这一数值较小。在整个泌乳期，NIRS直接预测粪便OMD与AIA测定的OMD之间的表型相关性为0.51。NIRS直接预测OMD的重复性低且变异性估计值不够准确，无法量化奶牛之间OMD的微小差异。与OMD相反，通过iNDF预测DMD的重复性估计值，特别是直接预测粪便iNDF的重复性估计值分别为0.32和0.46，这表明开发针对奶牛特异性消化率的NIRS预测方法是可行的。使用20头奶牛的每日个体粪便样本组成的数据子集来制定农场采样方案。基于对个体样本组合与复合样本之间相关性的评估以及个体样本组合的重复性估计，我们发现采集多达三个个体样本可得到具有代表性的复合样本。每三个月从一群奶牛中所有奶牛采集样本可能是个不错的选择，因为这样能获得更高的准确性。
