High Yield Regions Move Poleward Less Than Expected Under Climate Change in Globally-Gridded Crop Simulations

Crop yields on currently cultivated land are projected to decline under climate change as conditions shift away from optimal. It has been assumed that some of these losses can be recovered by shifting areas of cultivation, i.e. that "breadbasket" zones may move polewards. While changes in temperature and precipitation distributions will reduce agricultural favorability in some areas, they may increase it in others, leading to shifting spatial patterns of optimum yield. The recent Global Gridded Crop Model Intercomparison (GGCMI) Phase II experiment under AgMIP allows testing for geographic shifts in maximum yield regions under possible future climate states, for all major grains and soybeans. We emulate yields under an RCP8.5 climate projection at 0.5 degree spatial resolution globally for 9 GGCMs, and calculate centroids of potential production for major sub-continental regions. While most GGCMI Phase II models show some previously uncultivatable land at high latitudes becoming viable for agriculture, in all models the peak yield centers barely shift. For example, by 2100 the center of the North American maize belt moves only 80 km poleward in the mutli-model mean, less than 15% of the distance required to maintain constant mean growing-season temperatures. As a result, mean growing-season temperatures increase by 5 K, and the zone of highest potential yield shrinks while remaining relatively fixed in location. This stasis may be the result of other latitudinally-varying factors that affect crop yields, such as seasonality or solar insolation. Although yield losses can be mitigated somewhat by adaptations that allow crops to maintain growing season length under warming, cultivar adaptations also do not significantly shift optimum yield zones. Agricultural models therefore robustly concur that climate-driven production losses cannot be offset via a shift in cultivation areas.