Declining Crop Yield Sensitivity to Drought and Its Environmental Drivers in the North China Plain

Drought poses a severe threat to global food security and agricultural sustainability. Despite substantial efforts to enhance crop yield tolerance to drought, the effectiveness varies spatiotemporally across different environments and management practices. In this study, we compiled long-term grain yield data alongside multiple environmental indicators, including the multiscalar Standardized Precipitation Evapotranspiration Index (SPEI), climate, soil moisture (SWC), groundwater storage (GWS), nitrogen fertilizer (Nfer), and atmospheric CO2 records. We aim to assess the variability and drivers of grain yield sensitivity to drought across the North China Plain. We found a significantly positive correlation between the interannual variability of wheat yield and SPEI over the 9-month scale, suggesting that wheat yield variations were sensitive to medium-term (>9 month) and long-term (>22 month) drought. Surprisingly, the sensitivity (SSPEI: correlation coefficient between wheat yield variations and SPEI) of wheat yield to medium-term and long-term drought has declined substantially in the past three decades. The effects of SWC, GWS, Nfer, and CO2 on SSPEI varied situationally as the duration of the drought extended. Typically, SWC primarily governed short-term (<10 month) SSPEI, with a relative weight of 38.9 ± 3.2% in explaining SSPEI variability. The decrease in medium-term SSPEI was at the expense of GWS, which contributed a relative weight of 33.7 ± 12.3% in explaining the variations. SWC, CO2, and Nfer jointly dominated long-term SSPEI variations, and the cumulative relative weight as high as 84.0 ± 6.2%. Specifically, Nfer notably enhanced the SSPEI during prolonged drought, and the anticipated enriched CO2-induced “fertilizer effect” and “water-saving effect” in decreasing SSPEI were evident during long-term drought, contrasting with CO2 enrichment-enhanced yield reductions observed in short-term drought. Our findings highlight that prediction-based practices to mitigate drought-induced yield loss and enhance agricultural sustainability, including water conservation and fertilizer addition, may differ radically depending on drought episodes.