Chlorophyll fluorescence response mechanism to pine wilt disease infection and the potential of combining green light and <scp>SWIR</scp> bands for early diagnosis
Wenyuan Huang, Lingting Lei, Xinyi Han, Geng Wang, Zhihe Qian, Xuanhao Yan, Xiaoli Zhang
Abstract
BACKGROUND: Pine Wilt Disease (PWD) is one of the most destructive forest infectious diseases affecting pine trees. Although infected pine trees exhibit subtle physiological changes in the early stages, it is difficult to detect these changes in a timely manner using spectral reflectance alone. Consequently, accurately identifying early-infected pine forests remains a major challenge for disease monitoring. This study explored the physiological and biochemical response mechanisms of PWD at different infection stages and confirmed the application potential of chlorophyll fluorescence combined with sensitive spectral bands for early diagnosis. RESULTS: The results indicate that in the early stage of PWD, minimal fluorescence (Fo) significantly decreases, while non-photochemical quenching (NPQ) increases. These parameters can serve as sensitive indicators for early disease diagnosis. Compared with using only sensitive bands, combining chlorophyll fluorescence parameters with the least absolute shrinkage and selection operator (LASSO) selected bands significantly improved the K-nearest neighbors (KNN) model's classification performance, resulting in a 10% increase in overall accuracy and a 29% increase in early-stage precision. The LASSO-selected bands combined with chlorophyll fluorescence parameters demonstrated optimal performance in the support vector machine (SVM) model, achieved an overall accuracy of up to 96%, with an early-stage precision at 91%. CONCLUSION: This study determined key chlorophyll fluorescence parameters and diagnostic spectral bands for early detection, and proposed a novel strategy of combining sensitive bands with chlorophyll fluorescence for early identification, effectively overcoming the technical bottleneck in early-stage detection of traditional single-modality data. © 2025 Society of Chemical Industry.