Résumé
Spectroscopy and hyperspectral data are widely recognized non-destructive tools for monitoring plant health, enabling real-time detection of both biotic and abiotic stresses [1]. From a decision-making perspective, real-time assessment represents a transformative approach that saves time and resources while promoting more sustainable and environmentally friendly agricultural practices. However, a key challenge lies in selecting and applying the right analytical method capable of effectively unraveling the complexity of such data and delivering actionable insights for rapid decision support. Hydric stress (drought stress) is one of the most important abiotic factors limiting crop growth and development worldwide. It disrupts key physiological processes such as photosynthesis, transpiration, and nutrient uptake, ultimately compromising plant performance. In this study, we implemented a controlled environment experimental setup to better understand plant responses to drought-like conditions. To this end, polyethylene glycol (PEG) was used in our experimental setup, as it is commonly employed to simulate water stress under reproducible conditions. The experiment was conducted over a five-day period to assess the effects of PEG-induced hydric stress and subsequent recovery in sunflower plants. Three out of six plants were treated with PEG, while the others served as untreated controls for the entire duration of the study, providing a baseline for comparison. The PEG treatment was applied over a period of two days, after which it was discontinued to allow for the observation of recovery dynamics. Throughout the five-day experiment, Visible-Near Infrared (VNIR) spectral data were collected from all plants at nine time points to monitor changes in their spectral signatures associated with the pre-stress, stress, and recovery phases. To analyze the resulting spectral data, we applied multivariate statistical approaches to extract patterns over time. Principal Component Analysis (PCA) was first used to explore the main sources of variance and visualize the plant responses to stress [2]. While PCA provided initial insight into the temporal dynamics of the stress response, it lacked the ability to isolate the effects of treatment and time. Therefore, we applied ANOVA–Simultaneous Component Analysis (ASCA), which is a method specifically designed to separate structured variation in experimental data [3]. By separating the contributions of PEG treatment, measurement time, and their interaction, ASCA enabled a clearer interpretation of the spectral responses associated with hydric stress progression and recovery. These results demonstrate the potential of ASCA to effectively separate and interpret treatment and temporal effects in spectral data under controlled hydric stress conditions. The clear distinction observed between stressed and control plants, along with the indication of recovery pattern following PEG removal, highlights the potential of this approach for real-time stress assessment.