Ziyan Zhou ^1,2 Xiaoli Ren ^1,2* Liang Shi ^1,2 Honglin He ^1,2,3 Li Zhang ^1,2 Xiaoqin Wang ⁴ Mengyu Zhang ^1,2 Yonghong Zhang ⁵ Yuchuan Fan ⁶

• ¹ Key Laboratory of Ecosystem Network Observation and Modeling, Institute of Geography Science and Natural Resources (CAS), Beijing, Beijing Municipality, China
• ² Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences (CAS), Beijing, Beijing Municipality, China
• ³ College of Resources and Environment, University of Chinese Academy of Sciences (UCAS), Beijing, China
• ⁴ Fujian Key Laboratory of Network Computing and Intelligent Information Processing, Fuzhou University, Quanzhou, Fujian Province, China
• ⁵ State Key Laboratory of Herbage Improvement and Grassland Agro-ecosystems, School of Life Sciences, Lanzhou University, Lanzhou, Gansu Province, China
• ⁶ Geosystems Research Institute, Mississippi State University, Mississippi State, United States

The stability of future carbon sinks is crucial for accurately predicting the global carbon cycle. However, the future dynamics and stability of carbon sinks remain largely unknown, especially in China, a significant global carbon sink region. Here, we examined the dynamics and stability of carbon sinks in China's terrestrial ecosystems from 2015 to 2100 under two CMIP6 scenarios (SSP245 and SSP585), using XGBoost and SHAP models to quantify the impact of climatic drivers on carbon sink stability.China's future terrestrial ecosystems will act as a "carbon sink" (0.27-0.33 PgC/yr), with an initial increase that levels off over time. Although the carbon sink capacity increases, its stability does not consistently improve. Specifically, the stability of carbon sinks in future China's terrestrial ecosystems transitions from strengthening to weakening, primarily occurring in areas with higher carbon sink capacity. Further analysis revealed that atmospheric vapor pressure deficit (VPD) and temperature (TAS) are the two primary factors influencing carbon sink stability, with significant differences in their impacts across different scenarios. Under the SSP245 scenario, variations in VPD (VPD.CV) regulate water availability through stomatal conductance, making it the key driver of changes in carbon sink stability. In contrast, under the SSP585 scenario, although VPD.CV still plays an important role, temperature variability (Tas.CV) becomes the dominant factor, with more frequent extreme climate events exacerbating carbon cycle instability. The study highlights the differences in driving factors of carbon sink stability under different scenarios and stresses the importance of considering these differences, along with the scale and stability of carbon sinks, when developing longterm carbon management policies to effectively support carbon neutrality goals.