Decomposed efficiency and driving mechanisms of agricultural carbon emissions: Evidence from the Yangtze River Economic Belt, China

Under the context of global climate change, reconciling food security with low-carbon development poses a complex spatial data challenge for agricultural basins worldwide. Existing evaluations of Agricultural Carbon Emission Efficiency (ACEE) typically rely on aggregate indicators, treating the agricultural sector as a “black box” and lacking the computational resolution to reveal structural ecological trade-offs among internal production factors. To overcome this data limitation, this study constructs a high-resolution computational decomposition framework applied to the Yangtze River Economic Belt (YEB), China (2003−2022). By integrating the Directional Distance Function (DDF) for multi-dimensional factor deconstruction, Standard Deviational Ellipse (SDE) for trajectory tracking, and Partial Least Squares Regression (PLS-R) for complex network modeling, we systematically uncover the spatiotemporal dynamics and systemic drivers of ACEE. The results demonstrate that: Abdi (2010) (1) Total-factor efficiency evolved nonlinearly, and SDE tracking visualizes a spatial drag effect driven by the decline in resource efficiency in the mid-stream regions. Charnes et al. (1978) (2) The structural decomposition exposes a severe ecological asymmetry: while labor and machinery efficiencies improved rapidly due to technological substitution, fertilizer and irrigation efficiencies remained in a state of low-value lock-in. This reveals a critical “Masking Effect” in current agricultural transformations, where capital-driven data statistically conceals deep-seated resource intensity. Chen and Kou (2014) (3) PLS-R modeling identifies systemic spatial trade-offs: urbanization and the urban-rural income gap drive technological substitution, while spatial heterogeneity—manifested as land fragmentation in mountainous terrains—induces a “Scale Paradox” that hinders ecological economies of scale. This study provides a transferable data deconstruction and diagnostic protocol for evaluating multidimensional ecological performance in global agricultural basins. • Multi-dimensional decomposition unveils the “Masking Effect” in efficiency. • SDE trajectory tracking visualizes a “Spatial Drag” from resource inefficiency. • PLS-R modeling identifies systemic ecological trade-offs in complex networks. • Spatial heterogeneity analysis exposes a “Scale Paradox” in mountainous basins. • The integrated computational framework provides a transferable diagnostic protocol.