青海省海晏县冷季草场枯草营养成分近红外漫反射光谱定量分析

Quantitative Analysis on Nutritional Content of Hay by Near Infrared Reflectance Spectroscopy in the Cool-season Pasture in Haiyan County Qinghai Province

该研究旨在探讨近红外光谱技术(NIRS)定量分析高原地区冷季草场枯草营养成分的可行性,实现快速、准确评估冷季枯草各营养成分含量,补充枯草期营养价值数据的空白。采集青海省海晏县冷季草场枯草样品276份,选择修正偏最小二乘法(MPLS)的回归方法,初步建立了枯草中常见营养物质含量的NIRS定量分析模型。结果表明:干物质(DM)、粗灰分(Ash)、粗蛋白(CP)、粗脂肪(EE)、酸性洗涤纤维(ADF)、中性洗涤纤维(NDF)、木质素(ADL)、钙(Ca)和磷(P)NIRS模型的交叉验证决定系数(1-VR)分别为0.9014、0.9346、0.9847、0.6165、0.9426、0.9873、0.6142、0.8931和0.8622,交叉验证标准误差(SECV)分别为0.2735、0.9156、0.5231、0.9459、1.6154、1.0945、2.3858、1.0316和0.0204,交叉验证相对分析误差(RPDCV)分别为6.8513、0.8031、5.1232、1.3381、2.2368、4.3673、3.9804、2.1660和0.8319,外部验证相关系数(RSQ)分别为0.9512、0.5183、0.9362、0.6255、0.5937、0.9044、0.9121、0.4599和0.3146。其中,DM、CP、NDF和ADL的NIRS模型预测能力较好,Ash、EE、ADF、Ca和P的NIRS模型能否进行实际预测有待研究。

To explore the feasibility of quantitative analysis on nutrient composition of cold season grassland in plateau region by near infrared reflectance spectroscopy(NIRS),276 samples of hay from cold-season grass field in Haiyan County of Qinghai Province were collected,and the regression method of modified partial least squares(MPLS)was selected.A near-infrared quantitative analysis model of common nutrient content in hay grass was established.The results showed the cross-validation determination coefficients(1-VR)of NIRS model for dry matter(DM),crude ash(Ash),crude protein(CP),ether extract(EE),acid detergent fiber(ADF),neutral detergent fiber(NDF),acid detergent lignin(ADL),calcium(Ca)and phosphorus(P)were 0.9014,0.9346,0.9847,0.6165,0.9426,0.9873,0.6142,0.8931,and 0.8622,respectively.The standard error of cross validation(SECV)were 0.2735,0.9156,0.5231,0.9459,1.6154,1.0945,2.3858,1.0316,and 0.0204,respectively.The residual predictive deviation cross validation(RPDCV)were 6.8513,0.8031,5.1232,1.3381,2.2368,4.3673,3.9804,2.1660,and 0.8319,respectively.The residual squared(RSQ)were 0.9512,0.5183,0.9362,0.6255,0.5937,0.9044,0.9121,0.4599 and 0.3146,respectively.Among them,the NIRS model of DM,CP,NDF and ADL had better prediction ability,and whether the NIRS model of Ash,EE,ADF,Ca and P could be actually predicted remains to be studied.