Farmland Transformation Reshapes Ecosystem Services Through Multi-Level Landscape Reconfiguration: Evidence from China’s Songnen Plain

Global agricultural expansion has profoundly reshaped terrestrial landscapes, yet the ecological consequences of farmland transformation across multiple spatial scales remain insufficiently understood. This study investigates how farmland dynamics influence ecosystem service (ES) provision in China’s Songnen Plain by integrating long-term land-use trajectories, remote sensing data, and landscape ecological analysis. Using annual 30 m land-use/land-cover datasets from 2000 to 2020, we identified diverse farmland transformation pathways, including stable cultivation, reclamation, abandonment, fallow, afforestation, and urban encroachment. The InVEST model was applied to quantify three key ecosystem services: food supply, soil conservation, and carbon storage, while grid-level Partial Least Squares Regression (PLSR) was employed to identify their major drivers. Results revealed substantial spatial and temporal heterogeneity in ecosystem services. Food supply and soil conservation generally increased during the study period, whereas carbon storage exhibited a declining trend. Landscape-level characteristics, particularly Shannon diversity index (SHDI), aggregation index (AI), and patch density (PD), were identified as dominant factors regulating ecosystem service variation. Class-level indicators, including farmland area (FA) and mean patch area (MPA), also showed significant but comparatively weaker effects. This study provides new insights into the ecological consequences of intensive farmland transformation and highlights the importance of optimizing multi-scale landscape configurations to balance agricultural production and ecosystem sustainability.