By Eric Ralls, Earth.com staff writer

Rice kernels swell over a few short weeks. For every day above 86°F (30°C) in that window, a rice plant loses 1.1 to 1.8 percent of its yield, and on cooler days it loses almost nothing.

But the computer models behind the world’s food forecasts simulate almost none of that damage.

Yiwei Jian and colleagues at the Helmholtz Centre for Environmental Research in Leipzig and at Peking University measured the gap.

When the researchers corrected the models, the world loses 8.1 percent of its rice to 1.8°F (1°C) of warming instead of 3.8 percent.

Rice feeds more than half the people alive. Earlier teams put the loss anywhere between 2.3 and 8.3 percent for that much warming, and the new figure sits near the worst of them.

Warm seasons and hot afternoons

A warmer season and a hot afternoon damage a rice plant differently.

Throughout a warmer growing season, the crop develops faster and has less time to build tissue. It also burns more sugar simply staying alive, leaving less available to fill the grain.

Heat during flowering is more direct. It damages enzymes and reproductive tissue outright, and the plant can never recover that loss.

Researchers building global forecasts have mostly treated the two as one effect of temperature.

Christoph Müller of the Potsdam Institute for Climate Impact Research is one of the authors. When people check a summer forecast, he said, “we’re looking for the daily high, not the average temperature.”

Heating rice in real fields

The researchers compared 214 matched pairs of plots, one heated and one left alone, both grown the same way. Infrared heaters warmed every heated plot, all of them in open, irrigated fields.

At Jian’s own 12 sites across China’s rice regions, the heaters ran all season and lifted plot temperatures by 0.9°F to 7.2°F (0.5°C to 4°C).

The team left out trials that warmed the plots only at night. Rice loses yield on warm nights too, but more slowly, mostly because the plant burns more of its own sugar after dark.

The results varied enormously from site to site. At some locations, the crop lost 31.3 percent of its yield for every 1.8°F of warming, while at others it gained 6.6 percent.

At more than 70 percent of the sites, days above 86°F did most of the damage.

Where the grain goes missing

Plants heated during flowering and while the kernels swelled grew nearly as much material as unheated ones. They just put less of it into the grain.

At the 12 Chinese sites the team tracked the plants closely enough to see the steps. Under heat while the seed heads were forming, fewer seeds set, and less nitrogen moved out of the stems and into the grain.

Heat earlier in the season cost less. Above 95°F (35°C), while the plant was still putting out new shoots, photosynthesis and leaf area both dropped.

Those plants grew 1.4 percent less material per day of exposure and lost 1.5 percent of their yield, well under what the same heat cost at flowering.

Testing models against reality

Jian’s team ran the same analysis on seven of the global crop models used for food-security forecasts. Most of them simulated almost no response to hot days.

Three did simulate a loss, but a small one. Between 86°F and 95°F, simulated rice lost 0.1 to 0.4 percent per day, against 0.9 percent in the real plots.

These models already account for temperature. They simulate smaller plants through weaker photosynthesis and a shorter growing season.

What almost none of them simulate is flowers that fail to set seed, even though that is where the crop lost most of its grain.

Ting Sun and colleagues at Nanjing Agricultural University reported the same gap in 2021, after finding that rice models underestimated the losses from short heat spells.

The team corrected the model output with the field measurements, then ran the corrected version forward.

Under 1.8°F of warming, the estimated loss for Pakistan, India and Bangladesh rose from 2.1 to 6.6 percent. For Thailand, the Philippines and Myanmar it rose from 3.7 to 7.7 percent. The two estimates for China and Japan differ by under two percentage points.

What flooded fields might change

Every one of these experiments was irrigated, and in every one the researchers changed temperature and nothing else. That is what makes the numbers clean, and it is also what makes them narrow.

They describe what heat does on its own, not what a farmer will bring in.

What matters is the temperature of a rice plant’s leaves, not the air above them. Water evaporating from a flooded paddy cools those leaves.

Jian’s team built the global estimate from air temperature, and irrigated fields, which grow 75 percent of the world’s rice, may be cooler than that.

Rainfed rice goes the other way. The team found rainfed systems more sensitive to heat in the model runs, though nobody has settled by how much.

Rice in different regions may not share the same 86°F threshold either, especially as farmers adapt.

Researchers have not yet measured how much cooling reaches a rice flower in a flooded field on a 95°F afternoon. Until someone does, Jian’s team has only air temperature to work from.

Future models will have to simulate flowers failing to set seed, rather than simply growing smaller plants. The authors say that will take time.

The full study was published in the journal Science Advances.

Published Date: August 4, 2026