第二届未来大气科学论坛

Climate change is reshaping the natural foundation of global food production. Yet, farmers' cropping decisions often lag behind these environmental shifts, leading to a misalignment between actual planting patterns and natural potential. Does this misalignment compensate for natural constraints, or does it represent an overreach that pushes beyond ecological thresholds? A systematic, global-scale quantification of this question remains absent. Here, we address this gap for the world’s four major staple crops—maize, rice, soybean, and wheat. Firstly, we simulate natural cropping potential using an optimized Maxent model based on climatic and edaphic factors. Second, we employ a Generalized Additive Model (GAM) to disentangle the effects of natural potential and observable socio-economic drivers (population, GDP, inputs, prices, etc.), defining the residuals as cropping inertia deviation —an indicator capturing misalignment driven by latent factors such as historical inertia and policy interventions. Finally, we apply a two-way fixed effects model to assess the causal impact of this deviation on crop yields. Our preliminary findings reveal three key insights. Firstly, the dominant drivers of cropping potential vary markedly across crops: precipitation dominates for maize and rice, while temperature and soil organic carbon play larger roles for soybean and wheat. Under global warming, suitable ranges for all crops are shifting poleward, with soil salinity emerging as a universal constraint. Second, the spatial pattern of misalignment exhibits strong country-level heterogeneity. Overplanting is evident in North America, Europe, and eastern China, while underplanting prevails in Sub-Saharan Africa and the margins of the South American rainforest. Third—and most importantly—the effect of misalignment on yield is highly nonlinear and context-dependent. Overplanting significantly reduces yields in countries such as China and France, yet shows positive or more complex nonlinear effects in the United States and Brazil. Interaction models further reveal that these contrasting outcomes stem from opposing mechanisms across potential gradients: overplanting may boost yields in low-potential areas through intensified inputs, but suppress yields in high-potential areas by overexploiting resources. Results from Brazil and Canada provide compelling evidence of this divergent mechanism. Overall, this study provides a global-scale, process-level assessment of cropping misalignment and its nonlinear yield effects, offering empirical support for optimizing agricultural resource allocation and designing climate-adaptive cropping policies.