Estimation of Acetolactate Synthase Activity in Brassica napus under Herbicide Stress Using Near-Infrared Spectroscopy

Abstract

The appropriateness of near-infrared spectroscopy was investigated for the fast and noninvasive estimation of acetolactate synthase (ALS) activity in oilseed rape (Brassica napus L.) leaves, which is important for studying the effects of the new herbicide propyl 4-(2-(4,6-dimethoxy-2-pyrimidinyloxy)benzylamino)benzoate (ZJ0273) on plant metabolism and physiological regulation. A total of 248 leaf samples were collected at different leaf positions and ZJ0273 concentrations. A complete comparison was performed on the preprocessing methods, which included smoothing, multiplicative scatter correction, standard normal variate, derivatives, detrending, and direct orthogonal signal correction (DOSC), as well as linear calibrations, i.e., multiple linear regression (MLR) and partial least squares (PLS), and nonlinear calibration (least squares-support vector machine, LS-SVM). In all linear (MLR and PLS) models, the optimal models were the successive projections algorithm (SPA)-MLR (DOSC) and SPA-PLS (DOSC) models, with the same prediction performance (rp = 0.938 and RMSEP = 6.587). The best prediction model was the nonlinear SPA-LS-SVM (MSC) model, with rp = 0.988 and RMSEP = 2.844. A direct linear function using raw and DOSC preprocessed spectra at 2094 nm was developed and obtained a good prediction performance. These results will be helpful for the design of ALS detection sensors and in-field monitoring systems for the growth status of oilseed rape. The results are also useful for study of the inhibitory effects, metabolic pathways, and environmental residues of the new herbicide ZJ0273 during the growth of oilseed rape.