Résumé
Understanding the association between agroecosystem factors across multiple scales is essential for sustaining production in agroecosystems under disease pressure. Cocoa swollen shoot virus (CSSV) is a devastating and currently uncontrollable epidemic, posing the greatest threat to cocoa production in West Africa. The present study investigates the associations between CSSV incidence, meteorologic conditions, soil properties, agroforestry variables and landscape composition at different scales, using Self-Organizing Maps, an analysis method that handles nonlinearity and complex variable interactions. This study was based on data collected between 2021 and 2023 in 150 cocoa plots representing diverse cocoa-growing conditions in Côte d'Ivoire, which were clustered according to factor similarity. We found that low disease incidence was primarily associated with more frequent events of extreme precipitation and lower variation in temperature. On the contrary, a high incidence of CSSV was linked to a higher density of trees hosting the virus. We drew on existing knowledge of disease epidemiology, mealybug vector biology and cocoa tree physiology to interpret these results. Additionally, plots with low disease incidence were predominantly surrounded by non-host land, suggesting that large-scale management strategies could help mitigate CSSV incidence by promoting non-host land use at the landscape level.