Rapid detection of malondialdehyde in herbicide-stressed barley leaves using spectroscopic techniques

Malondialdehyde (MDA) is an important indicator for research of plant senescence and resistance. Traditional detection method is complex and time-consuming. In this study, near infrared spectroscopy was used to detect the malondialdehyde (MDA) in herbicide-stressed barley leaves as a convenient, non-invasive and rapid method. A total of 75 barley leave samples were collected for near infrared spectral scanning. Seven spectral preprocessing methods were compared for a better prediction performance, including Savitzky-Golay (SG) smoothing, standard normal variate (SNV), multiplicative scatter correction (MSC) and so on. The optimal partial least squares (PLS) model was obtained for the detection of MDA in barley leaves. The latent variables (LVs) extracted by PLS were also applied as input variables to develop leas squares-support vector machine (LV-LS-SVM) model. PLS, MLR and LS-SVM models were developed using EWs selected by regression coefficient. The correlation coefficient (r) and root mean square error of prediction (RMSEP) were applied as the indices of model assessment. The results indicated that LV-LS-SVM mode was better than PLS model, and the LV-LS-SVM model by SNV and MSC preprocessing methods achieved the same prediction performance with higher correlation, which r and RMSEP were 0.9383 and 10.4598. An excellent prediction precision was achieved. The results demonstrated that near infrared spectroscopy was successfully applied for the rapid and high accurate detection of MDA in herbicide-stressed barley leaves, and this supplied a new approach for on field monitoring and resistance detection of biological information of barley.