Spatio-Temporal Dynamics and Driving Mechanism of Ecosystem Services Under Ecological Restoration in the Kubuqi Desert, Northern China

Desertification is an ever-growing global ecological and environmental problem. With the implementation of various ecological restoration initiatives, vegetation cover in many desert regions has increased substantially. Consequently, it is essential to understand the dynamics of ecosystem services (ESs) in desert ecosystems to better inform environmental management. This study integrates the InVEST model, RWEQ model, Spearman correlation analysis, trade-off and synergy coefficient method, and the Partial Least Squares Path Model (PLS-PM) to systematically assess the spatio-temporal dynamics and underlying driving mechanisms of five key ESs in the Kubuqi (KBQ) Desert, northern China. Specifically, the application of PLS-PM enables the identification of latent pathways, indirect effects, and multi-step causal relationships, which traditional correlation-based methods fail to capture. The results show that the KBQ Desert underwent substantial land use changes from 2000 to 2020: sandy land decreased by 2697.83 km2, grassland increased by 1864.15 km2, and cropland and urban land expanded by 519.15 km2 and 257.74 km2, respectively. ESs exhibited divergent trajectories. habitat quality (HQ), carbon sequestration (CS), soil conservation (SC), and water yield (WY) all showed overall increases, with WY and SC increasing particularly strongly, whereas Sand-fixation service (G) displayed a fluctuating trend. Over the past two decades, HQ–CS, HQ–G, and CS–G have shown moderately strong synergies, while CS–WY has exhibited a pronounced trade-off, and SC–G and SC–CS have displayed relatively weaker trade-offs. The spatial distribution results of trade-off and synergy relationships show that the KBQ Desert is dominated by a synergy relationship, and the main synergy relationship combinations are CS–HQ, CS–SC, and HQ–SC. The correlation coefficients between other ES pairs are generally low. Additionally, this study identifies key pathways through the PLS-PM method, such as PRE → NDVI → ES and LU → NDVI → ES, revealing the complex interactions between precipitation (PRE), land use (LU), and vegetation dynamics. The findings show that land use (LU) consistently exerts a strong negative impact on CS, while PRE and NDVI have a significant positive effect on WY. These pathways deepen our understanding of how climate and anthropogenic factors affect ESs, particularly the influence of temperature (TEMP) on evapotranspiration (ETP), which in turn affects WY. Additionally, the impact of NDVI on wind–sand fixation (G) and SC varies over time, with vegetation dynamics playing a particularly enhanced role in 2010 and 2015. These findings highlight the impact of ecological restoration and land management on regional ESs changes. A comprehensive understanding of the interactions between climate factors, LU, and vegetation dynamics will help in developing more effective intervention strategies.