Spatially-Explicitly Predicting Suitability of Three Apple Diseases in China: A Comparative Analysis of Five Species Distribution Models

Abstract

Apple Valsa Canker (AVC), Apple Ring Rot (ARR), and Alternaria Blotch on Apple (ABA) represent major threats to China's apple industry. Understanding the environmental suitability of these diseases is essential for effective orchard management and disease prevention. However, their large-scale spatial distribution and environmental interactions remain insufficiently studied. In this research, we analysed data from 1392 locations using five species distribution models—Generalised Linear Model (GLM), Generalised Additive Model (GAM), Support Vector Machines (SVM), Maximum Entropy (MaxEnt) and Random Forest (RF)—to predict the environmental suitability of these diseases across apple-growing regions in China. Model performance was evaluated using the True Skill Statistic (TSS) and the Area Under the Receiver Operating Characteristic Curve (AUC). MaxEnt and RF consistently outperformed the other models, achieving AUC values above 0.95 and TSS scores exceeding 0.78 for all three diseases. Areas with the highest environmental suitability were primarily located in the Bohai Bay, Loess Plateau and Old Course of the Yellow River regions. Among the environmental variables analysed, the mean temperature of the driest quarter and the annual maximum temperature emerged as the most influential, consistent with the physiological conditions favourable for pathogen development. The key climatic variables identified and their associated disease response curves align with established epidemiological patterns for the three diseases. By integrating ecological insights with predictive modelling, this study provides a robust foundation for targeted disease management and the development of early warning systems under changing climate conditions.