Nonlinear maize yield responses to drought drive escalating regional vulnerabilities under climate change in China

Drought poses a growing threat to global maize production, with risks projected to intensify under climate change. However, comprehensive, long-term quantification of crop yield responses to drought, particularly their nonlinear across diverse environments, remains limited. Here, we combined generalized additive models and threshold regression to unveil the critical role of Vapor Pressure Deficit (VPD) in determining maize yields in China from 1981 to 2020. Our results demonstrate that yield responses to VPD are strictly nonlinear, governed by distinct regional thresholds: 0.82 kPa in the Northeast (NE) and 0.91 kPa in the North China Plain (NCP). Beyond these thresholds, a stark regional divergence emerges: in the NE, high VPD significantly amplifies water stress, leading to precipitous yield declines at a rate of 4.52 Mg ha^-1 per kPa. Conversely, the NCP exhibits greater resilience, with losses limited to 1.88 Mg ha^-1 per kPa. Structural equation modeling reveals that the heightened sensitivity in the NE is driven by tighter land-atmosphere coupling. Future projections under four Shared Socioeconomic Pathways (SSP1-2.6, 2-4.5, 3-7.0, and 5-8.5) indicate that threshold exceedance will become widespread, with probabilities rising to 54-86% in the NE and 48-85% in the NCP by the late century. Correspondingly, under the SSP5-8.5 scenario and assuming stationary climate-yield relationships, these conditional projections estimate that VPD-induced yield losses could reach up to 12.0% (95% CI: 7.12-18.71%) in the NE-nearly double the 6.18% (95% CI: 3.51-10.30%) projected for the NCP. These findings underscore the necessity of integrating nonlinear threshold-based indicators into agricultural risk management and climate adaptation strategies to safeguard national food security.