Impacts of snow cover on net primary productivity across Forest–grassland ecological zones: Hydrothermal and Phenological pathways in the greater Khingan range

Against the backdrop of climate change, variations in snow processes in cold temperate ecotones and the mechanisms by which they affect vegetation productivity remain insufficiently understood. Taking the western slope of the Greater Khingan Range in Inner Mongolia as the study area, this study integrates multi-source remote sensing and reanalysis data for the 2000–2024 hydrological years. Using trend analysis, correlation and partial correlation analyses, lagged correlation analysis, and partial least squares structural equation modeling (PLS-SEM), we systematically investigate the response of vegetation net primary productivity (NPP) to snow cover changes across forest, forest–grassland ecotone, and grassland zones, as well as the underlying spring hydrothermal and phenological mechanisms. The results showed that: (1) during the study period, the cumulative number of effective snow-covered days in the study area generally decreased, with snow cover days (SCD) declining at a rate of −0.40 d yr −1 . The snow onset date advanced overall, whereas changes in the snow end date were relatively weak. Meanwhile, NPP exhibited pronounced seasonal differentiation: in spring, areas with increasing NPP dominated, accounting for 95.5% of the total study area; in summer, areas with decreasing NPP predominated, accounting for 78.8%. These findings indicate a clear stage-specific pattern in vegetation productivity dynamics across the study area. (2) The snow–NPP relationships exhibited ecological-zone-dependent and cross-seasonal patterns. In the forest zone, a later snow cover melt date and a longer snow cover duration were closely associated with reduced spring NPP, with correlation coefficients of −0.59 between SCM and spring NPP and − 0.50 between SCD and spring NPP. In the grassland zone, antecedent snow accumulation was positively associated with subsequent productivity, with spring SWE showing correlation coefficients of 0.24 and 0.19 with summer NPP and growing-season NPP, respectively. (3) The PLS-SEM results indicated that the direct association between snow cover and NPP was relatively weak, and that snow-related effects were mainly reflected through potential pathways involving spring soil hydrothermal conditions and NDVI-derived monthly-scale phenological indicators. Across the three ecological zones, the path coefficients from snow cover to spring soil moisture ranged from 0.15 to 0.57, while those to soil temperature ranged from −0.31 to −0.50, indicating that snow cover was associated with both water-retention and cooling effects. Specifically, in the grassland zone, the dominant pathway was the positive snow cover–soil moisture–NPP linkage, whereas in the forest zone and the forest–grassland ecotone, snow-related cooling was more closely associated with monthly-scale phenological indicators and reduced early-season productivity. Overall, the snow–NPP relationships in the ecotonal region on the western slope of the Greater Khingan Range exhibited ecological-zone-dependent patterns, reflecting differences among ecosystems in the relative importance of thermal limitation and snowmelt water supply.