Spatiotemporal Dynamics and Multi-Path Driving Mechanisms of Vegetation Net Primary Productivity in Tibetan Plateau (2001–2022): Coupling Analysis of Geodetector and PLS-SEM

The Tibetan Plateau is highly sensitive to global climate change and characterized by pronounced ecological fragility, making vegetation net primary productivity (NPP) a key indicator for assessing ecosystem functioning and regional ecological security. This study aims to characterize the spatiotemporal dynamics of NPP and to disentangle the multiple natural and land-use drivers shaping its variability across the Tibetan Plateau. MODIS-derived NPP data for the period 2001–2022 are integrated with multi-source datasets on climate, topography, normalized difference vegetation index (NDVI), and land use (CLCD), and analyzed using trend and correlation analyses, land-use transfer matrices, an optimal-parameter geographical detector, and partial least squares structural equation modeling (PLS-SEM). The results indicate that NPP exhibits a significant but fluctuating upward trend (0.52 gC·m −2 ·a −1 , p < 0.01), with higher values in the southeast and lower values in the northwest, the Yunnan Plateau evergreen broadleaf and pine forest region (VA5) and the southern Himalayan montane forest region (VA6) function as high-value centers, and regions such as the Kunlun high-cold desert region (HID1) represent low-value centers. The mutual conversion between forestland and grassland and bare land constitutes a key process driving regional NPP changes, with the net expansion of forestland making a substantial contribution to NPP increases (net gain of 2606.88 TgC). Geographic detector analysis indicates that NDVI ( q = 0.741) and land use type ( q = 0.741) are the primary factors governing the spatial differentiation of NPP, followed by precipitation, slope, and temperature. Moreover, interactions between any two factors enhance their explanatory power, with the interaction between aspect and land use type exhibiting the strongest effect ( q approaching 1). PLS-SEM path analysis further quantifies the driving pathways, revealing that mean annual precipitation and land use type are the most direct drivers of NPP, while climatic and topographic factors influence NPP indirectly through their effects on vegetation cover and land use type. This study advances the understanding of the multifactorial driving mechanisms of ecosystem productivity on the Tibetan Plateau and provides a scientific basis for zoned and differentiated ecological restoration and climate adaptation strategies.