Integrating Remote Sensing and Machine Learning to Project Global Habitat Suitability and Productivity of Chinese Fir...

Chinese fir (Cunninghamia lanceolata) is China's most widely planted industrial plantation species, yet productivity declines have been reported in several regions. Climate change is likely to intensify these risks by simultaneously reshaping climatic suitability and limiting sustainable net primary productivity (NPP), but their combined effects have not been quantified in a global, multi-model framework. Here, we integrate ecological niche models (ENMs) with multiple machine-learning models for NPP, calibrated using 3139 occurrence records, MODIS-derived NPP, and 37 climate-soil covariates. Future projections are driven by an ensemble of 13 CMIP6 GCMs under SSP245 and SSP585. Across scenarios, suitable habitat is projected to contract in the current core region of southern China while expanding poleward, with new suitability in North China, the eastern United States, and South America. By 2081-2100, habitat losses account for 16%-18% of the current suitable area, partly offset by gains in newly suitable regions equivalent to 35%-45% of the current suitable area. Within today's planting footprint, total NPP is projected to decline by 6%-12% (≈1.3-5.6 × 10⁹ t·year^-1) relative to the current total NPP under the same footprint. In contrast, tracking future suitable zones under an idealized assisted-migration scenario could potentially increase total NPP by 15%-20% relative to the current total NPP. Warm-season precipitation and temperature-regime variability (annual range and isothermality) emerge as dominant controls, highlighting coupled hydrothermal constraints. This integrated assessment provides strategic evidence for prioritizing climate-forward plantation siting.