Vegetation restoration regulates soil carbon sequestration and erosion control in the Mu Us Sandy Land: Non-linear tipping points and driving mechanism shift (2000-2020).

Clarifying the cascading responses of soil carbon sinks and surface protection functions driven by vegetation restoration during the dryland desertification reversal process is crucial for precise regional ecological management. Focusing on the Mu Us Sandy Land from 2000 to 2020, this study integrated multi-source ecological indicators-including enhanced vegetation index (EVI), gross primary productivity (GPP), soil organic carbon density (SOCD), and soil erosion rate (ERO)-and combined spatial autocorrelation, machine learning, generalized additive models, and structural equation modeling to systematically elucidate spatiotemporal evolution patterns, non-linear threshold characteristics, and driving mechanism transitions of vegetation restoration-carbon sink-erosion cascading effects. The results showed that: (1) Above- and below-ground ecological processes exhibited significant spatiotemporal asynchrony: EVI and GPP achieved robust region-wide recovery, while SOCD responses demonstrated pronounced time lag, spatial heterogeneity, and a fragmented pattern characterized by "weak global autocorrelation and strong local mismatch". (2) Vegetation-driven effects on deep ecological functions exhibited strict non-linear threshold characteristics: an EVI range of 0.142-0.160 constituted the critical "effective window" for ecological intervention, beyond which marginal gains diminished rapidly toward saturation. (3) The regional ecosystem driving paradigm achieved a fundamental transition, as statistically inferred from SEM and RDA, from a passive degradation mode "dominated by climate-physical stress" in 2000 to an active synergistic mode of "deep regulation by the vegetation hub" in 2020, with precipitation's ecological role completely reversing from an erosive force to a system resource. This study emphasizes that dryland ecological restoration urgently requires abandoning singular scale-expansion greening paradigms in favor of precise threshold-based regulation, adhering to the principle of "determining vegetation by water availability" to guarantee long-term multi-service synergy.