Export-attributed carbon footprint of cotton production in arid China: A life cycle and driver analysis

Context Cotton is a globally traded commodity with disproportionately large environmental costs. Although its land share is modest, input use and energy demand are high in arid irrigated systems. The Tarim River Basin (TRB) in Xinjiang, China—one of the country's principal cotton-producing regions—remains underrepresented in long-term, spatially explicit carbon accounting, particularly with respect to responsibilities embedded in international trade. An assessment that connects farm-level processes, export attribution, and driver diagnosis is needed to guide equitable and efficient decarbonization. Objective Quantify the carbon footprint of cotton production in the TRB on 1 km grids for 2000–2020; develop and apply a transparent export-attribution framework that assigns to the region an equitable share of emissions embedded in cotton trade from China; and isolate principal biophysical and socioeconomic drivers using ridge regression and structural equation modeling (SEM) to identify policy-sensitive levers for mitigation. Methods We conducted a cradle-to-farm-gate life-cycle assessment (LCA) at a 1 km spatial resolution from 2000 to 2020. We compiled activity data and emission factors to derive product-, area-, and value-based indicators, including CF _ Y (kg CO₂-eq·kg −1 ), CF _ A (kg CO₂-eq·hm −2 ), and carbon economic efficiency ( CEE ; kg CO₂-eq·CNY −1 ; CNY denotes the Chinese yuan). A national-proportional export-attribution scheme was applied to allocate export-embedded carbon emissions to the TRB. Based on this allocation, trade-intensity metrics were calculated, including CF _ Y export and CEE export . Drivers were quantified through ridge regression and a confirmatory SEM spanning land use, vegetation condition, topography, climate, and socioeconomic context; direct and indirect effects were decomposed. Uncertainty was examined via sensitivity tests on key activity data and emission factors. Results and Conclusions Total emissions more than doubled from 2000 to 2020. In contrast, the intensity indicators changed differently: CF _ Y remained near 2.83 kg CO₂-eq·kg −1 , CEE decreased from 0.25 to 0.14 kg CO₂-eq·CNY −1 , and CF _ A increased from about 3900 to 5200 kg CO₂-eq·hm −2 , indicating increasing land-based emission intensity. Dominant sources were labor (34.4 %), electricity (23.8 %), and diesel (14.8 %), highlighting priorities to modernize labor structure and decarbonize irrigation energy. Export-attributed emissions directed to Asian destinations cumulatively reached 2.70 × 10 9 kg CO₂-eq, with strong heterogeneity in unit export intensity across markets. Driver analysis suggests that optimized land use and socioeconomic upgrading are associated with lower footprints, whereas increased vegetation activity and climatic variability may raise emissions in the absence of advanced management. Significance By integrating gridded LCA, a reproducible export-attribution protocol, and driver modeling, the framework pinpoints actionable hotspots, assigns responsibility along supply chains, and identifies levers—clean irrigation energy, targeted mechanization, and optimized land allocation—for rapid and equitable emission reduction in arid, export-oriented agriculture.