Little poleward shift of peak agricultural regions under warming

Abstract:

While agriculture is widespread on the Earth’s surface, modern agricultural production is spatially concentrated: ~90% of the main global cereals and nearly half of total global calories are produced from just ~3% of ice-free land surface area. These ‘breadbaskets’ are largely found in temperate zones where projected growing-season temperatures warm up to 8 K by the end of the century (CMIP6, SSP585). A common expectation is that high yield regions will shift poleward, following optimum temperatures, so that an appropriate adaptation strategy involves moving cultivation to new areas. The Global Gridded Crop Model Intercomparison Project (GGCMI) Phase 2 simulations under the Agricultural Model Intercomparison and Improvement Project (AgMIP) allows testing this assumption. GGCMI consists of harmonized simulations from 9 models for 5 crops (maize, rice, soybeans, and spring and winter wheat) over a wide range of climate and management inputs. Simulation results show that the peak yield regions barely shift under warming and lag behind the climate velocity in all crop models. Results are robust across regions and models: shifting the cultivation area offsets only 5-20% of temperature-driven production losses for North American maize (for example) in the SSP585 warming scenario, substantially less than other potential adaptation measures. We explore the factors underlying this stasis and separately evaluate the effects of soils, photoperiod, precipitation, seasonality, and cropping calendar. Results suggest caution in assuming agricultural responses are straightforward, and highlights the importance of global modeling for climate change impacts assessments.