Oumeng Qiao, Honglei Che, Buchun Liu, Enke Liu, Xurong Mei, Meng Zhang
Large-scale climate modes such as ENSO influence crop production, yet the pathways linking remote variability to local losses remain unclear. Here, we link the El Niño-Southern Oscillation to winter wheat risk across China using heading-harvest hot-dry-windy hours from hourly ERA5-Land and 1981-2020 county-level yields. ENSO does not shift exposure uniformly; it redistributes probability mass along the HDW-hour distribution in a region-dependent way, changing the likelihood of high-exposure seasons. This imprint widens and left-shifts the yield-anomaly distribution and thickens the negative tail, consistent with thresholded, convex damage under late-season compound stress. Mediation decomposition indicates spatial asymmetry: HDW-mediated effects are clearest in southern latitude bands, whereas northern ENSO-yield variation is more strongly associated with competing non-HDW pathways. A targeted northern diagnosis identifies late frost as the strongest adverse non-HDW pathway, counterbalanced by a spring-precipitation pathway of comparable magnitude but opposite sign. A structural equation model supports a layered cascade in which ENSO intensifies extreme heat, the dominant proximal driver of HDW exposure, while irrigation intensity and soil organic carbon primarily damp exposure-to-loss translation.