Vegetation greening and soil drying in Central Asia: Evidence for bidirectional vegetation-soil moisture feedback

In arid Central Asia (CA), variations in soil moisture (SM) strongly influence vegetation dynamics, while vegetation can affect SM through evapotranspiration. Although SM variation and its influence on vegetation dynamics have been widely documented, the bidirectional feedback between vegetation and SM remains inadequately explored. This study examined trends in Normalized Difference Vegetation Index (NDVI) and four SM layers (SM1~SM4, from shallow to deep) across CA for 1982–2022, using correlation analysis and Granger causality test to characterize their associations and bidirectional Granger-predictive feedbacks. Results indicated that: (1) significant NDVI greening occurred in eastern CA and browning in the west, while SM generally declined across much of the region; (2) NDVI was predominantly positively correlated with shallow-to-intermediate SM, but the proportion of negatively correlated areas increased with soil depth; (3) Bidirectional Granger-predictive feedback occurred across 88.4%, 79.8% and 72.2% of pixels for SM1, SM2, and SM3, respectively, the influence of SM4→NDVI dominated the deep-layer significant causality relationship; (4) coupling-trend included concurrent NDVI and SM decreases in western CA and increasing NDVI alongside declining SM in eastern. These findings demonstrated that both vegetation–soil moisture trends and relationships vary with influence of human activities and climatic conditions in such water-limited region, providing a scientific basis for climate adaptation, ecological restoration and water-source management.