Distinguishing plant biophysical stress responses from temporal variability using vis–NIR spectroscopy

Abstract

Visible–near infrared (Vis—NIR) spectroscopy is widely used for non-destructive plant monitoring; however, in time-series experiments, signals of interest are frequently masked by diurnal rhythms and systematic measurement drift. This limits the ability to clearly detect plant responses to stress in spectral datasets. This study presents a multi-sensor approach to monitor the onset and recovery dynamics of polyethylene glycol (PEG)-induced osmotic stress in sunflower ( Helianthus annuus L.). Initial multivariate exploration using principal component analysis (PCA) indicated that temporal and environmental variability dominated the dataset, more than treatment-related effects. The application of REP-ASCA (Reduction of Repeatability Error for ANOVA-Simultaneous Component Analysis), a design-driven multivariate approach, allowed the separation of stress-related spectral variation from temporal and measurement-related effects. This revealed a rapid divergence between control and PEG-treated plants following PEG application. The REP-ASCA scores associated with the PEG factor showed a reversible trajectory, returning towards control levels after PEG removal, consistent with the transient nature of the applied stress. This behaviour contrasted with the progressive changes observed in pigment-related indices, suggesting that the spectral components extracted by REP-ASCA primarily capture fast and reversible biophysical responses. These results highlight the potential of combining high-resolution Vis–NIR spectroscopy with design-driven multivariate analysis to disentangle treatment effects from strong temporal variability in plant time-series data.