Climate change facilitates fungal pathogen expansion while driving endemic host range contractions in a tropical biodiversity hotspot

Abstract Context. Non-native plant pathogens are reshaping ecosystems globally, yet their spread and potential spatial distributions under future climate and land-use change remains underexplored, particularly in biodiversity hotspots with vulnerable hosts. In Madagascar, a vascular wilt pathogen ( Leptographium calophylli ) has been increasingly observed infecting native forest trees raising conservation concerns. Objectives. We aimed to model future distributions of both a vulnerable host tree Calophyllum paniculatum and incipient wilt pathogen Leptographium calophylli under climate and land cover change scenarios over the next 80 years to assess overlap, divergence, and implications for extinction risk. Methods. Using ensemble SDMs across Madagascar, we forecasted potential distributional ranges for both pathogen and host using the Global Climate Model (GFDL-ESM4) under three scenarios (SSP1–2.6, SSP3–7.0, SSP5–8.5) and time periods (2011–2040, 2041–2070, 2071–2100). We measured future range shifts using a habitat exposure index and potential area of occupancy. Results. The pathogen is predicted to retain 68.5% of the current projected distribution by 2100, with expansion into previously uninhabited regions. C. paniculatum is forecast to experience range contraction (65.9% loss by 2100) with persistent distributional overlap predicted across all scenarios. Conclusion. We demonstrate that future climatic conditions may facilitate fungal pathogen expansion, while simultaneously exposing vulnerable hosts within their native, endemic range to infection. The asymmetric dynamic of host range losses could intensify biodiversity loss in insular ecosystems, particularly when compounded by deforestation. This study highlights the importance of considering disease threats in biodiversity forecasts and conservation strategies, particularly in tropical systems facing rapid environmental change.