Moisture Constraints on Realized CO₂-Associated Productivity Gains in Xinjiang Drylands

Abstract. Rising atmospheric carbon dioxide (CO₂) can enhance terrestrial productivity, but how CO₂-associated productivity responses are expressed across water-limited ecosystems remains uncertain. Here, we developed a model-based dual-pathway framework comprising an explicit CO₂-associated productivity component and a water-use-efficiency (WUE)-related carbon–water response. Using paired standard CASA and CO₂-CASA simulations, seven-scenario attribution experiments, and complementary carbon–water diagnostics, we evaluated net primary productivity (NPP) changes across Xinjiang, China, during 2001–2020. Baseline CASA reproduced the major spatial and interannual variability of MOD17 NPP. Mean annual NPP in 2018–2020 exceeded that in 2001–2003 by 10.24 g C m⁻² yr⁻¹. Under the isolated-component framework, vegetation-state change contributed 9.41 g C m⁻² yr⁻¹ (91.9 % of the total modeled change), whereas the analytically isolated explicit CO₂ component contributed 1.11 g C m⁻² yr⁻¹ (10.8 %) and climate contributed −0.53 g C m⁻² yr⁻¹ (−5.2 %). Long-term CO₂–WUE coupling was strongest under intermediate moisture conditions, with class-level associations strongest within 100–400 mm and a nonlinear fitted maximum near 200–300 mm. However, first-difference analysis showed little interannual CO₂–WUE association, while April–October precipitation was more closely associated with interannual NPP variability than annual precipitation. Positive productivity trends also persisted after excluding areas potentially influenced by persistent agriculture. Together, these results support a model-based dual-pathway interpretation of CO₂-associated productivity responses while identifying moisture context and temporal scale as key constraints on their expression and interpretation.