近红外漫反射光谱法测定青贮玉米品质性状的研究

Prediction of Ensiled Maize Quality Traits with Near Infrared Diffuse Reflectance Spectroscopy (NIDRS)

【目的】研究利用近红外漫反射光谱法(NIDRS)测定青贮玉米的体外干物质消化率(IVDMD)、中性洗涤纤维(NDF)、酸性洗涤纤维(ADF)、粗蛋白(CP)和粗脂肪(EE)含量的可行性。【方法】以普通、高油和超高油玉米全株和秸秆的青贮样为材料,采用光谱的主成分空间技术和偏最小二乘回归法(PLS)。【结果】所建立的IVDMD、NDF、ADF、CP和EE的校正模型的交叉验证决定系数(R2cv)分别为0.9133、0.9764、0.9789、0.9254和0.7294,外部验证决定系数(R2val)分别为0.8879、0.9455、0.9635、0.9387和0.7333,各项误差(RMSEE、RMSECV和RMSEP)为0.24(CP)～2.23(NDF)。【结论】利用近红外漫反射光谱法测定青贮玉米品质性状是完全可行的,该结果可满足畜牧业对青贮饲料品质快速分析的需要,对青贮玉米育种材料的快速鉴定筛选具有重要的意义。

[Objective] This study was conducted to investigate the feasibility of measuring in vitro dry matter digestion （IVDMD）, neutral detergent fiber （NDF）, acid detergent fiber （ADF）, crude protein （CP） and ether extract （EE） concentrations in ensiled maize by near infrared diffuse reflectance spectroscopy （NIDRS）. [Method] The ensiled stover and whole plant samples from normal oil maize, high oil maize and super oil maize were used to establish NIDRS calibration models of ensiled maize with the techniques based on the principal component space of spectra and partial least square regression （PLS）. [Result] The results showed that the determination coefficients of cross validation （R^2cv） and validation （R^2val, in parentheses） of these yielded models were 0.9133 （0.8879）, 0.9764 （0.9455）, 0.9789 （09635）, 0.7294 （0.7333） and 0.9254 （0.9387） for IVDMD, NDF, ADF, EE and CP respectively. The root mean square error of estimation, cross validation and prediction （RMSEE, RMSECV and RMSEP） ranged from 0.26 for CP to 2.23 for NDF. [ Conclusion ] It has been demonstrated that it is feasible to use NIDRS as a rapid and an accurate technique to predict ensiled maize quality traits. These calibration models could meet the needs of silage quality evaluation in livestock industry and screening of various samples in silage maize breeding.