Spatiotemporal heterogeneity and multi-scenario simulation of vegetation carbon sinks under the influence of ecological engineering: a case study in the Three-River Headwaters Region, China

Clarifying the impact of ecological engineering on vegetation carbon sink (VCS) and predicting future VCS under ecological protection constraints is of significant importance for formulating ecological protection policies and addressing climate change. This study uses Net Ecosystem Productivity (NEP) as the core indicator and establishes a multi-scenario coupled simulation framework of “ecological engineering effects—SSP-RCP scenarios—PLE-SEM driver decomposition” by dynamically linking the effects of ecological engineering implementation at different stages. Taking the Three-River Headwaters Region (TRHR) in China as an example, this study reveals the phased differences in VCS during different ecological engineering periods (2000−2022) and the evolution patterns of their source-sink distribution, and also predicts the VCS under various scenarios for the carbon peak year (2030) and the carbon neutrality year (2060). The results show that from 2000 to 2022, vegetation NEP in the TRHR exhibited a fluctuating upward trend with a multi-year average of 32.3 gC·m −2 ·a −1 . The boundary between carbon sources and sinks was distinct, with the southeast functioning as carbon sinks and the northwest as carbon sources; the carbon sink area was 1.38 times that of carbon source area. During ecological engineering implementation, the increasing trend of vegetation NEP was significantly higher than before and after implementation, with the second phase showing a more pronounced increase than the first phase. However, ecological engineering had limited influence on the source-sink pattern, which remained dominated by stable carbon sources and sinks. For 2030 and 2060, the spatial distribution characteristics of vegetation NEP under the three SSP-RCP scenarios showed obvious similarity to the historical period. Notably, mean NEP differed significantly across scenarios: SSP5-8.5 exhibited the highest NEP, while SSP1-2.6 showed the lowest. From 2030 to 2060, NEP increased across all three scenarios, with SSP1-2.6 demonstrating the largest increment (∼7.83 gC·m −2 ) and strongest carbon sink potential, whereas SSP5-8.5 showed the smallest increment (∼0.95 gC·m −2 ) with limited carbon sink potential. Ecological constraints generated positive carbon sink benefits across all scenarios, enhancing carbon sequestration by 8.65 gC·m −2 ·a −1 (SSP1-2.6), 4.65 gC·m −2 ·a −1 (SSP2-4.5), and 3.72 gC·m −2 ·a −1 (SSP5-8.5) in 2060, with the constraint effect exhibiting an inverse relationship with baseline NEP levels. PLE-SEM analysis indicated that vegetation factors had the strongest direct effect on NEP, while topography dominated through indirect effects via climate and vegetation. These findings provide important references for optimizing regional ecological protection and restoration strategies and formulating carbon sequestration enhancement pathways.