傅里叶近红外反射光谱法快速测定大豆脂肪酸含量

目前大豆脂肪酸育种需要进行大量的气相色谱数据分析,因此建立近红外光谱(NIRS)快速测定脂肪酸组分技术具有重要意义。文章以108个中国大豆Glycine max(L.)Merr.]品种或品系为材料,以傅里叶近红外光谱(FT-NIRS,4 000~12 500 cm-1)与气相色谱(GC)技术相结合,采用偏最小二乘(PLS)回归和交叉验证法,探讨利用FT-NIRS技术预测脂肪酸含量的可行性。依据OPUS 5.0软件针对不同脂肪酸组分筛选出最佳NIRS光谱区域为6 101.9~5 446.5 cm-1。交叉验证结果显示大豆主要脂肪酸组分,如油酸(C18∶1,R2CV=0.94)、亚油酸(C18∶2,R2CV=0.87)、亚麻酸(C18∶3,R2CV=0.85)和总饱和脂肪酸(C16∶0 C18∶0,R2CV=0.88)的预测准确率较高。外部验证结果证明大豆油酸预测模型的决定系数最高(R2val=0.91),其预测均方根误差(RMSEP)为2.47 g.kg-1干重,RMSEP/SD的比值为0.29,可保证大豆油酸辅助育种的准确性;而棕榈酸、硬脂酸、亚油酸、亚麻酸和总饱和脂肪酸的预测决定系数范围为0.66~0.76,RMSEP范围为0.37~2.74 g.kg-1干重,RMSEP/SD比值范围为0.47~0.53,表明可以进行大豆脂肪酸组分含量的初步筛选。该研究进一步证明利用FT-NIRS技术预测大豆主要脂肪酸组分是稳定可行的。

Rapid Determination of Fatty Acids in Soybeans [Glycine max (L.) Merr.] by FT-Near-Infrared Reflectance Spectroscopy

Current breeding programs dealing with fatty acid (FA) concentrations in soybean Glycine max (L. ) Merr.] require large numbers for gas chromatographic analyses, thus it is important to develop a method for rapid determination of fatty acid by Near-Infrared Reflectance spectroscopy (NIRS) in soybeans. The objective of this work was to study the potential of fourier-transform near-infrared reflectance spectroscopy (FT-NIRS) to estimate the fatty acid concentrations in Chinese soybean varieties. One hundred and eight of soybean cultivars or lines (the calibration set: 64; the external validation set: 44) were scanned within 4000-12500 cm(-1) of wavenumbers using a standard sample cup by NIRS machinery, and analyzed the fatty acids by gas chromatograph (GC) methods. Equations were developed using partial least squares (PLS) regression and cross validation for multivariate calibration in this study. The optimal spectral region was selected from 6101.9 to 5446.5 cm(-1) based on the OPUS 5.0 software. Cross validation results showed that major FA components such as oleic acid (R2(CV) = 0.94), linoleic acid (R2(CV) = 0.87), linolenic acid (R2(CV) = 0.85), and total saturates (R2(CV) = 0.88) were accurately determined by the proposed equations as compared with the reference data obtained by the GC method. External validation results also demonstrated that equation for oleic acid had the highest predictive ability R(2)val = 0.91), root mean square error of predication (RMSEP) value was 2.47 g x kg(-1) dry weight, the ratios of RMSEP to the standard deviation (SD) was 0.29, which was usable for quality assurance application. Moreover, equations for palmitic acid, stearic acid, linoleic acid, linolenic acid, and total saturates were predicted with the determination coefficients ranging from 0.66 to 0.76, RMSEP values from 0.37 to 2.74 g x kg(-1) dry weight, and RMSEP/SD values from 0.47 to 0.53, which could be used for sample screening. Therefore, we confirmed that a reliable estimation of the major fatty acid components is possible by using NIRS technique in soybeans.