In 2006, former U.S. Vice-President Al Gore predicted that we only had 10 years to stave off our carbon dioxide doom, with plummeting yields in Africa, the Himalayas melting and other doomsday scenarios happening by 2016. 

None of that came to pass, agriculture is doing better than ever, but another round of computer estimates claims that will change eventually. The models were confirmed by successfully predicting the past - a hallmark of numerical models before they can set about predicting the future - and they believe that though the ecosystem is complex, using just a few critical factors like temperature and CO2 and irrigation and fertilization they can comfortably predict a decline in U.S. food if warming occurs.

Yield losses of 20 percent by 2100

Burning fossil fuels elevates the amount of CO2 in the air. More actually CO2 leads to better water efficiency in plants, since they lose less water for each unit of CO2 taken up from the air, but there can be too much of a good thing. The additional CO2 fertilization in the simulations does not alleviate the drop in yields associated with high temperatures above about 86°. Obviously increased natural gas and the resulting less coal have begun to eat into the runaway emissions caused by the Clinton's administration's ban on nuclear power that led to huge coal upsurges in the first place, but there is room for progress. The new estimate doesn't seem to account for progress.

They state that for every single day above 86 degrees, maize and soybean plants can lose about 5 percent of their harvest. Their simulations found that rather small heat increases beyond that threshold can result in abrupt and substantial yield losses. They believe those temperatures will be more frequent if emissions are not slashed in the U.S. and can severely harm agricultural productivity. They estimate harvest losses from elevated temperatures of 20 percent for wheat, 40 percent for soybean and almost 50 percent for maize without efficient emission reductions. 
Since the U.S. is one of the largest crop exporters, that would impact food security in poor countries.

More water would obviously fix the problem, and that is a simple matter of energy, but the researchers keep technology static, so no CRISPR plants, no cheaper energy that would make the 98 percent of water unavailable usable, and project more problems. That is in defiance of the history they did not model. American science has boosted agriculture a lot since 1980, it is impossible to even fathom what we will accomplish by 2100.