Prediction of Protein and Amino Acid Contents in Canola Seeds and Canola Meal with Near-infrared Spectroscopy

Abstract

Aim: The current study focused on developing and validating the near-infrared reflectance spectroscopy (NIRS) calibration models for predicting protein and amino acid contents in whole canola seeds and canola meal. This research investigated the effects of sample pre-treatments involving particle size reduction and lipid extraction and different types of spectrometers, including the diode array analyzer PerkinElmer DA7250 and the Fourier transform NIRS analyzer PerkinElmer FT9700 on the predictive performance of the NIRS calibration models.Methods: In total, 480 canola whole seed samples were selected from the 2015 and 2020 cropping year populations to produce canola meal samples and then analyze crude protein and amino acids concentrations with reference chemical methods; among those, 420 samples were assigned for constructing calibration models, while 60 samples were used for the validation study. The partial least square regression technique was used for model development and verification, performed on the Unscrambler X10.3 software with the spectra obtained from PerkinElmer DA7250 for both whole seed and meal samples and PerkinElmer 9700 for meal samples only.Results and Conclusion: The calibration models of crude protein and most amino acids except for Tryptophan, Histidine and Sulphur amino acids showed an acceptable coefficient of determinations (R2C= 0.677-0.885), while the NIR models for Tryptophan, Histidine, and Sulphur amino acids were less accurate which might require more work in the future study. Sample pre-treatments like particle size reduction and lipid extraction were found that have the potential to improve prediction ability. PerkinElmer DA 7250 was discovered with a similar performance to PerkinElmer FT9700 with no significant differences. This study indicates that the results are acceptable for screening the protein and amino acid contents in canola whole seeds and meals, this could be helpful for future quality control and the implementation of breeding strategies to enhance canola protein quality.