Calibration of NIRS for the Estimation of Fatty Acids in Brassica Juncea

This study was done to test whether near infrared reflectance spectroscopy (NIRS) could be used as a quick substitute for measuring quality characteristics, i.e., a fatty acid profile over gas liquid chromatography. Near infrared reflectance spectroscopy is a multi‐trait technique. In the present study, the fatty acid profiles of seeds of 200 genotypes of Brassica juncea were analysed by gas liquid chromatography. Near infrared reflectance spectra of intact seeds of the same samples were collected (400–2500 nm) on a NIR systems Model 6500 spectrophotometer. The spectra were subjected to scanning, mathematical processing, and statistical analysis using Win ISI software. Data were scored to remove redundancy. Spectra were treated as outliers with H > 3.0 (global H) and similar samples with H < 0.6 (neighbourhood H). Cross validation of the spectra was done using a modified partial least‐square method to develop the calibration equation. The calibration equation had good R² values for oleic acid (R² = 0.91), linoleic acid (R² = 0.83), and erucic acid (R² = 0.88). The internal validation was done to test the goodness of fit of the developed equation. The equation provided reliable estimations of these traits in internal validation with R² values of 0.77, 0.68, 0.81 for the above quoted fatty acids, respectively. The external validation results also showed higher R² values for oleic acid (0.89), linoleic acid (0.69), and erucic acid (0.90). The equation was less reliable for linolenic acid, which had R² values of 0.53 in cross validation, 0.25 for internal validation, and 0.20 for external validation. The results indicated that NIRS could be used to rapidly determine oleic acid, linoleic acid, and erucic acid in intact B. juncea seeds.