A four‐tier warning framework for pine wilt disease: integrating species distribution models and smart sensing approaches

Abstract BACKGROUND Pine wilt disease (PWD), caused by the pine wood nematode (PWN, Bursaphelenchus xylophilus ), is one of the most destructive diseases of coniferous forests. Early warning of PWD outbreaks is critical for timely and effective control strategies to suppress disease spread and potentially achieve eradication. However, interference from multiple stressors has hindered the development of effective stand‐level early warning systems. To address this, we developed a four‐tier operational framework that integrates species distribution modeling and smart sensing approaches (unmanned aerial vehicles, Internet of Things‐based proximal sensing, and artificial intelligence technologies) for stand‐level PWD risk alerts. The framework can provide sequential alerts (blue, yellow, orange, and red), each corresponding to increasing evidence of infestation. RESULTS The framework incorporates three core models: a MaxEnt‐based species distribution model, a host tree species and health status detection model, and an insect vector detection model. The MaxEnt models for PWN and its insect vector achieved an AUC (area under the receiver operating characteristic curve) exceeding 0.92, demonstrating reliable performance in predicting suitable climatic conditions for PWD. The host tree species and health status detection model achieved an mAP0.5 (mean average precision at an Intersection over Union threshold of 0.5) of 97.5%, and the insect vector detection model reached an mAP0.5 of 99.8%. Applied to 81 forest stands, the framework successfully distinguished risk levels. Among 34 stands with discolored host trees, 16 were confirmed as PWD‐infested, all originating from red alert stands, validating the accuracy of risk identification. CONCLUSION This four‐tier framework provides a feasible solution for stand‐level early warning of PWD. By minimizing interference from non‐PWD stressors, it improves detection accuracy while lowering labor and financial costs. Its integration of multiple frontier technologies offers an innovative, scalable tool for forest health monitoring in high‐risk areas. © 2025 Society of Chemical Industry.