Can Crops Keep Up With a Hotter Planet?

Two mechanisms dominate. First, heat speeds up a plant’s development, shortening the weeks it spends filling grain, so even moderate warming means smaller harvests. Researchers at the International Rice Research Institute found that rice yields fell by about 10% for each 1°C rise in night-time minimum temperature during the dry season. Second, extreme heat at flowering damages pollen. A few days above roughly 35°C during that window can leave maize cobs and rice panicles partly empty, no matter how good the rest of the season was. That is why averages understate the risk: much of the damage comes from short, extreme events.

With warming of about a quarter of a degree per decade, a 6% loss per degree and a third of losses avoided by adaptation, warming takes about 0.1 percentage points of wheat yield a year (Derived). Ray et al. estimated historical wheat gains of about 0.9% of base-year yield per year; this is not a measurement of current breeding alone. The average race is therefore being won, but by less than it looks, because yield growth was already too slow: Ray et al.’s historical projection produced gains of roughly 38–67% from 2008 to 2050, below its doubling benchmark. That benchmark is dated and is not a forecast of present food demand. Warming makes a slow race slower. Climate exposure, continued agronomic gains and adoption of suitable varieties all decide the outcome.

Harvests can rise while climate change still makes food scarcer

Two comparisons can both be true. Yields in 2050 may be higher than today because varieties, irrigation and management improve. They may also be lower than they would have been in a world without additional warming. The second gap is the climate loss. Positive net growth therefore does not mean climate damage has been avoided. It means other improvements have outweighed it on the chosen average. Access to affordable food then depends on population, diets, land use, trade and household incomes as well as tonnes harvested per hectare. Climate Impact Lab researchers’ interpretation

The 4.4% headline needs its denominator. The 2025 analysis translates modeled losses across six staple crops into about 120 kilocalories per person per day for each additional degree of warming, equivalent to 4.4% of present recommended daily consumption. It is not a universal 4.4% yield penalty for every crop or every farm. The study incorporates observed adaptation and projects how it changes with climate and income. Applying another adaptation discount to this already-adapted estimate would count that benefit twice. Hultgren et al., Nature, 2025

The global calorie estimate and individual crop-yield sensitivities have different units, baselines and adaptation assumptions. They illustrate scale but cannot be substituted directly in one equation.

The last mile is the seed shop

The fastest yield gain available in much of Africa may not need any new science at all. Maize varieties grown in sub-Saharan Africa averaged about 15 years old in one study, against 3–5 years in the US, Latin America and Asia, so most farmers plant seed bred for an earlier, cooler climate. Where public breeders and seed companies have pushed replacement, average variety age has fallen quickly. Every year of variety age shaved off is yield gained, and heat and drought tolerance arrive with it.

The farmer needs a reliable variety, not just a higher mean yield

Ray and colleagues’ 2013 growth rates came from historical, linear yield trends, measured relative to a base-year yield. They include management improvements as well as breeding and are not measurements of current genetic gain. Their 2.4% annual benchmark was a non-compounding rate for doubling production from 2008 to 2050 without adding land. From a 2026 baseline, doubling by 2050 would require 4.17% of starting yield added each year, or 2.93% compounded annually. Neither is a universal food-demand target. The calculator now uses a linear projection to match the historical convention. Ray et al., PLOS ONE, 2013

A global average also conceals bad years and unequal access to adaptation. A variety that raises normal-year yield can still fail under heat at flowering or severe drought. Farmers need performance across sites and seasons, affordable seed and traits that fit their growing window and market. Faster laboratory breeding does not remove field testing or seed multiplication. CIMMYT’s maize-variety selection work illustrates the information problem: vague altitude and maturity labels can leave farmers unsure which seed fits local conditions. Weather services and seed systems therefore connect directly; a crop improvement becomes useful only when farmers can choose and grow it successfully. CIMMYT’s maize variety selector; Zhao et al., PNAS, 2017

Yield projections are scenarios, not forecasts. Regional extremes, irrigation limits and prices can worsen food security even when the global yield index rises.

Who is building what

Public breeding centres, trait developers and the rules and tools that set how fast new varieties reach farmers. Search the record, or filter by role.

Yield claims from small plots or single seasons often shrink in farmers’ fields. The test of a trait is consistent gains across many locations, years and stress conditions.

Four numbers to watch

First, annual yield growth for the main staple crops, especially in the tropics. Second, the average age of varieties in farmers’ fields, a direct measure of how fast improvements arrive. Third, yield losses in extreme heat years compared with earlier heatwaves of similar intensity. Fourth, the number of heat- and drought-tolerant varieties released and adopted, not just developed.