Rice feeds more than half the world's population, and projections of how climate change will affect it have varied widely. Part of the reason, according to work led from the College of Urban and Environmental Sciences at Peking University with international collaborators, is that earlier projections focused on rising average temperatures and handled extreme heat poorly. The team compared 214 observations from field-warming experiments worldwide against simulations from seven crop models.
Temperatures above 30 degrees Celsius emerged as a key driver of yield loss, and at more than 70 per cent of experimental sites heat exposure explained over half of the observed yield change. The vulnerability is concentrated in particular growth stages: extreme heat during flowering and grain filling reduces grain-setting rates and disrupts grain development, damage that a season-average temperature simply cannot represent.
Correcting the models against the field evidence changes the projections substantially. Under one degree of global warming the models had projected a global rice yield decline of 3.8 per cent plus or minus 1.7; after correction the estimate rose to 8.1 per cent plus or minus 2.3. The largest revisions fell in South and Southeast Asia, where projected losses for Pakistan, India and Bangladesh rose from 2.1 to 6.6 per cent. The study appeared in Science Advances.
Beyond the revised numbers the paper offers a method for improving global crop models using field-based evidence of heat vulnerability, which is the more transferable contribution. It also sharpens the case for specific adaptation measures: heat-tolerant varieties, adjusted planting schedules, improved irrigation and better soil management, each of which targets exposure during the vulnerable reproductive window rather than seasonal averages.