After empirically estimating the relation between warming temperatures and crop yield changes on the basis of observations, we apply the estimated function to output from NCAR-DOE CESM1 run under several scenarios of future emissions, each scenario explored with a medium–to–large set of initial condition ensemble members.

The presence of these initial condition ensembles allows us to robustly estimate differential impacts even in the presence of internal variability, and even for very small differences in global warming levels between the alternative futures. Nevertheless, we show that assumptions about the effects of CO₂ on crop productivity drive the significance (or lack thereof) of differential results when comparing Paris-style low warming scenarios (1.5C not exceed; 1.5C with an overshoot; 2.0C) for maize and wheat future global yields. For higher, and more separated scenarios the benefits of mitigation appear clearly instead. We also explore changes in the exposure of crops to heat extremes, characterizing future changes, and the corresponding avoided impacts, for days above critical thresholds of temperature (35C for Maize; 34C for Wheat) known to have potentially strong negative effects on plant development when occurring at a critical time in the growing season. Heat extremes have a large signal-to-noise behavior and therefore the significant benefit from remaining under a lower scenario emerges clearly and unequivocally.