Spatiotemporal variations and driving mechanisms of carbon storage in Central Asia: Insights from the PLUS-InVEST models and machine learning.

Against the backdrop of global climate change and rapid socioeconomic advancement, significant land use/cover changes(LUCC) in Central Asia have profoundly impacted terrestrial ecosystem carbon storage(CS). However, the assessment and spatiotemporal dynamics of CS in Central Asia remain inadequately understood. This study systematically examined the spatiotemporal dynamics of LUCC and CS in Central Asia from 1990 to 2020, and anticipated CS in 2030 under 3 SSP-RCP scenarios using an combined structure consisting of the land use harmonization 2(LUH2) dataset, the patch-generating land use simulation(PLUS) model, and the integrated valuation of ecosystem services and tradeoffs(InVEST) model. Additionally, the extreme gradient boosting(XGBoost) model-Shapley(SHAP) values was employed to identify the elements impacting geographical distinction of CS. The findings show the following: (1)there was a net rise of 0.02 Pg in total CS in Central Asia between 1990 and 2020. From 1990 to 2010, extensive deforestation and urban sprawl led to a 0.1 Pg reduction in CS. However, post-2010, forest regeneration and large-scale conversion of unused land to grassland contributed to a 0.13 Pg increase in CS. (2)Between 2020 and 2030, forest expansion under the SSP126 and SSP245 scenarios is projected to enhance total CS by 0.03 %(0.01 Pg) and 0.17 %(0.08 Pg), respectively. Conversely, under the SSP585 scenario, substantial declines in both forestland and grassland are expected to result in a pronounced 1.67 % loss in CS. Moreover, while grassland undergoes a notable reduction under SSP126(-1.82 %), it experiences a expansion under SSP245(0.06 %). Consequently, the total CS exhibits a more substantial increase under SSP245 than under SSP126, SSP245 scenario is more favorable for enhancing CS in Central Asia. (3)Soil temperature(ST) is the most critical factor impacting the spatial heterogeneity of CS in Central Asia, followed by the normalized difference vegetation index(NDVI). This study explores a suitable path for Central Asian countries to optimize land use planning, increase ecosystem CS and achieve sustainable development, and also provides a reference for arid and semi-arid regions to enhance their carbon sequestration capacity.