Despite continued net forest cover expansion in Japan, forests are increasingly affected by multiple and interacting disturbance processes, including climate extremes, topographic instability, and human pressures. Understanding where forests are vulnerable and which factors drive these patterns is essential for climate-resilient forest management and effective biodiversity conservation. In this presentation, we introduce a national-scale assessment of tree loss disturbance and its possible disturbance drivers in Japan. We further evaluate vulnerable protected areas that are exposed to tree loss disturbance in the current environmental conditions.

In this study, we developed a multi-factor Random Forest model to map tree loss susceptibility across Japan and to identify the key environmental and anthropogenic drivers of forest disturbance. Tree loss between 2001 and 2021, derived from Landsat-based satellite observations, was used as an indicator of forest disturbance resulting from both natural and human-induced processes. To explain these patterns, we compiled 25 predictor variables representing physical and climate-related conditions (e.g., topography, soil properties, precipitation, evapotranspiration), biological factors (e.g., deer presence and forest cover type), human footprint factors (e.g., road networks, agricultural land, built-up areas, solar panel density, population density), and hazard-related factors (e.g., typhoon wind exposure, landslide hazard). All datasets were harmonized at a 250-m spatial resolution to enable national-scale analysis.

The Random Forest model showed strong predictive performance (AUC = 0.862) and revealed that tree loss in Japan is primarily driven by physical and climate-related factors, particularly slope angle, elevation, typhoon wind speed, evapotranspiration, precipitation, solar radiation, slope aspect, population density, and landslide hazard. These results highlight the importance of geomorphic instability and compound climate hazards in shaping forest disturbance regimes.

The vulnerability of protected areas was higher in smaller sites, including Quasi-National Parks (IUCN Categories II/V) and Prefectural Wildlife Protection Areas (IUCN Category IV), due to their lower protection status, which reflects differences in protection and management intensity under Japan’s national designations.

Our national-scale, integrated tree loss assessment provides a valuable tool for assessing forest disturbance and vulnerability of protected areas. These findings are directly relevant to Japan’s 30×30 biodiversity strategy, climate change adaptation planning, and long-term forest resilience and carbon-neutrality goals.