Spatial non-stationarity and nonlinearity in multi-scale drivers of net ecosystem productivity on the Qinghai-Tibetan Plateau: An MGWR-OPGD-PLS-SEM integrated framework

The Qinghai–Tibet Plateau (QTP) is a critical yet vulnerable carbon sink. However, a mechanistic, multi-scale understanding of the drivers of its net ecosystem productivity (NEP)-a direct metric of carbon sink strength-remains limited, particularly regarding nonlinear effects and critical thresholds. This knowledge gap constrains adaptive ecosystem management under climate change. To address this, we integrated the Optimized Parameter-based Geographical Detector (OPGD), Multiscale Geographically Weighted Regression (MGWR), and Partial Least Squares Structural Equation Modeling (PLS-SEM) to analyze the QTP (2000−2020). Results show: (1) NEP increased significantly (0.62 gC·m −2 ·yr −1 ) with a stable southeast-high, northwest-low gradient; its centroid shifted southwestward post-2010, indicating spatial reorganization. (2) Critical thresholds were quantified for elevation (3512, 4507 m; corresponding to vegetation/permafrost transitions) and grazing intensity (∼0.3 sheep unit·ha −1 ; marking a compensatory growth threshold). (3) NDVI, vegetation coverage, and elevation were core drivers, with the strongest interaction between elevation and vegetation. The “climate → vegetation → NEP” pathway dominated regional carbon flux. (4) Approximately 73% of core carbon sink areas exhibited anti-persistent trends, signaling high degradation risk. This study clarifies the synergistic controls and thresholds governing the QTP carbon sink, providing scientific support for alpine grassland conservation and regional carbon neutrality strategies. • Integrated framework reveals spatial and nonlinear NEP drivers. • NEP increased across QTP with a clear spatial gradient. • Vegetation and elevation dominate NEP variability. • Elevation and grazing thresholds regulate carbon dynamics. • Most regions show unstable future carbon sink trends.