Global warming increases the risk of crop yield failures driven by climate oscillations

Enhancing food-system resilience is critical in the face of increasing climate variability that threatens food security. Large-scale climate oscillations are key drivers of climate conditions that disrupt agricultural productivity. However, how such effects are shifting under greenhouse warming remains unclear. Here, we integrate machine learning with process-based crop models to quantify changes in climate-oscillation-driven yield variability under warming scenarios. We find that climate change increases the dominance of the North Atlantic Oscillation (NAO) in the Northern Hemisphere and the El Niño-Southern Oscillation (ENSO) in the Southern Hemisphere, exposing an additional 5.1%–12% of global croplands to climate oscillation shocks. Negative NAO and El Niño events are projected to cause simultaneous yield losses of 2.0%–8.4% across multiple breadbaskets, while opposite phases provide weaker benefits, indicating asymmetric impacts and greater food security risks. We highlight the importance of incorporating shifting teleconnections into early-warning systems and targeted adaptation strategies to enhance global food-system resilience.