In June 2025, a landmark study in Nature delivered the most comprehensive assessment of climate impact on global food production ever published. Analysing data from 12,658 regions across 55 countries, covering six staple crops that provide two-thirds of humanity’s calories, the researchers found that each additional degree Celsius of global warming reduces food production by approximately 120 calories per person per day — 4.4% of recommended consumption. At three degrees of warming, that is the caloric equivalent of every person on the planet skipping breakfast. By 2050, climate change will drag global crop yields down 8% regardless of how much emissions rise or fall. By 2100, the range extends from 11% under net-zero to 24% if emissions continue unchecked. Farmer adaptation — switching crop varieties, shifting planting dates, altering fertiliser use — offsets approximately one-third of these losses. The remaining two-thirds are structural, not fixable by individual farm decisions. A separate Stanford study published the same year found that warming and atmospheric drying have already reduced global yields of barley by 13%, wheat by 10%, and maize by 4% below what they would have been without climate trends. These are not projections. They are measurements of damage that has already occurred. Meanwhile, 800 million people are already food-insecure. The world’s population adds 80 million mouths per year. The Green Revolution’s productivity gains — the seed varieties, fertiliser techniques, and irrigation systems that prevented the famines Paul Ehrlich predicted in the 1960s — are plateauing. The gap between supply trajectory and demand trajectory is widening. This is the yield curve that bends the wrong way.
Of recommended caloric consumption per 1°C warming. 120 calories per person per day. Nature 2025 study of 12,658 regions, 55 countries, six staple crops.[1]
Current yields below what they would have been without 50 years of climate trends. Largest decline among the five major crops studied.[2]
Global yield reduction from warming and atmospheric drying. Range expansion into northern regions partially offsets losses but levels off mid-century.[2]
Current reduction. Projected to worsen to −24% by late century. Grown near equator where warming is most severe. 75% of global corn is rainfed.[3]
Of climate losses are structural and cannot be offset by farmer adaptation. Only 1/3 mitigated by switching varieties, shifting dates, altering inputs.[1]
Probability of yield declines by 2100 for each major crop except rice. Sub-Saharan Africa and South Asia face 25–30% median national yield cuts.[4]
The Nature study is the first to systematically measure how extensively real-world producers actually adapt at the global scale. Previous analyses assumed either perfect adaptation or none at all, producing wildly divergent projections. The new study found the reality in between: farmers do adapt, and their adaptations are meaningful, but they are insufficient. Switching to heat-tolerant varieties helps. Shifting planting dates helps. Adjusting fertiliser application helps. Together, these adaptations offset about one-third of climate-driven losses by end of century under high emissions. The remaining two-thirds represent a structural ceiling — losses that individual farm-level decisions cannot address because they are driven by temperature and atmospheric drying that no seed variety or planting schedule can overcome.[1]
The atmospheric drying signal is the finding that models missed. Stanford researchers found that vapour pressure deficit — a measure of atmospheric aridity and a key driver of plant water stress — increased far more in Europe and China than climate models predicted, while the US Midwest was anomalously protected. When crops experience high VPD, they close their stomata to conserve water, which reduces photosynthesis and growth. The practical consequence: even when soil moisture is adequate, dry air can reduce yields. This mechanism explains why some regions are seeing productivity losses that irrigation alone cannot fix.[2]
Between 1960 and 2000, the Green Revolution roughly doubled cereal yields worldwide through high-yield seed varieties, synthetic fertiliser, and expanded irrigation. It prevented the mass famines that demographers had predicted and enabled the world to feed a population that grew from 3 billion to 6 billion. But the productivity growth rate has been decelerating since the 1990s. The easy gains — first application of fertiliser, first irrigation of rainfed land, first introduction of improved seeds — have been captured. Further gains require precision application, genetic modification, and infrastructure investment that scale unevenly across the world’s 600 million farms.[5]
The demand curve has not decelerated. The world adds approximately 80 million people per year. Rising incomes in Asia and Africa are shifting diets toward more resource-intensive foods: meat, dairy, and processed products that require more grain per calorie consumed. The FAO estimates the world needs roughly 50% more food by 2050 to feed a projected 9.7 billion people. The supply curve is bending downward while the demand curve bends upward. The intersection of these curves is the structural food security risk that this case documents.[5]
The regional asymmetry compounds the problem. The Nature study found that the greatest losses will occur in modern-day breadbaskets with currently favourable climates — places like the US Great Plains, the European wheat belt, and southern Brazil — that have limited present adaptation because they have not yet needed it. Meanwhile, losses in low-income regions, while proportionally smaller as a share of global production, are devastating for food security because these populations have no buffer. Sub-Saharan Africa and parts of South Asia face 25–30% median national yield cuts by late century. Cassava — a staple for hundreds of millions in the tropics — is a standout vulnerability.[1][4]
The invisible casualties are already visible in commodity markets. Coffee, cocoa, oranges, and olives have all experienced supply challenges and significant price increases driven by climate variability. The 2025 cocoa price spike — futures exceeding $12,000 per tonne, triple the historical average — was driven by drought and disease in West Africa. Orange juice prices hit records after citrus greening disease, exacerbated by warming, devastated Florida’s crop. These are the leading indicators: speciality crops with concentrated growing regions that signal what will eventually happen to staples grown on wider footprints.[2]
| Dimension | Evidence |
|---|---|
| Quality / Yield (D5)Origin · 75 At Risk | Yield degradation is the origin because the cascade flows from the physical reality of declining crop productivity. Nature 2025: −120 cal/person/day per °C (−4.4% of recommended consumption). By 2050: −8% global yields. By 2100: −11% (net-zero) to −24% (high emissions). Stanford/PNAS: barley −13%, wheat −10%, maize −4% already. Atmospheric drying (VPD) increasing faster than models predicted. 70–90% probability of decline by 2100 for all staples except rice. Maize −24% by late century. Cassava highly vulnerable. Coffee, cocoa, oranges, olives already showing supply challenges. Farmer adaptation offsets only 1/3 of losses. The remaining 2/3 are structural.[1][2][3] |
| Operational (D6)Origin · 70 At Risk | UN declared “global water bankruptcy” in January 2026. 70% of major aquifers in long-term decline. 40% of irrigation water from aquifers being drained. Agriculture accounts for 72% of global freshwater withdrawals. Over 80% of global croplands could face water scarcity by mid-century. India loses up to a foot of groundwater per year from irrigation pumping. Halting groundwater depletion would add 26M people at risk of hunger by 2050. Infrastructure — irrigation, storage, cold chain, transport — is degrading in the regions that need it most. The operational dimension is co-origin because water is the non-substitutable input: without it, no seed variety, fertiliser, or technology can produce food.[6][7][8] |
| Customer / Population (D1)L1 · 78 | 800 million people currently food-insecure. 2 billion face severe water scarcity at least one month per year. Population adds 80M/year, reaching 9.7B by 2050. Rising incomes shifting diets toward more resource-intensive foods. Sub-Saharan Africa and South Asia face 25–30% median national yield cuts. Cassava — staple for hundreds of millions — is highly vulnerable. One-quarter of global crops grown in areas where water is highly stressed or unreliable. The customer dimension scores highest at L1 because the downstream impact is measured in human lives, not market share. When the yield curve bends, the first casualties are the 800 million who are already hungry.[1][4][8] |
| Revenue / Markets (D3)L1 · 72 | Cocoa futures exceeded $12,000/tonne in 2025 (3× historical average). Orange juice at record prices. Coffee supply constrained. Fertiliser prices rose 15–20% in 2025. India banned wheat exports for 4 years (2022–2026). Russia restricts fertiliser exports. China cut urea exports 90% in 2024. Food price volatility is the financial transmission channel: when yields decline, prices spike, and the spike cascades through commodity markets, consumer budgets, sovereign debt, and political stability. The 2010 Russian wheat crisis contributed to the Arab Spring.[9][10] |
| Regulatory / Policy (D4)L1 · 60 | No global coordination on food security equivalent to the CHIPS Act for semiconductors. Agricultural subsidies in developed nations distort markets and discourage efficiency. WTO unable to enforce trade rules during crises (India’s wheat ban violated commitments without consequence). Climate policy (Paris Agreement) does not directly address agricultural productivity. Export bans are the default crisis response — India, Russia, China all restrict food exports when domestic supply is threatened — exacerbating global shortages. The regulatory dimension scores lower than other L1s because the methodology is genuinely weak: the policy architecture for global food security does not exist at the scale required.[10] |
| Employee / Workforce (D2)L2 · 55 | Average US farmer age: 58. US farms declined 7% from 2017 to 2023 (1.89M remaining). 50% of California’s tractor jobs unfilled. Rural-urban migration reducing agricultural workforce globally. 84% of the world’s 600M+ farms are smallholder (<2 hectares), operated by aging populations with limited access to technology or capital. The employee dimension captures the human capital crisis: the people who grow the world’s food are aging out and not being replaced, while the technology that could compensate is inaccessible to the majority.[5] |
-- The Yield Curve: 6D At-Risk Cascade
-- Agriculture Cluster Case 1 of 4 (UC-107, UC-108, UC-109, UC-110)
FORAGE yield_curve_degradation
WHERE caloric_loss_per_degree_c > 100
AND yield_decline_by_2050 > 0.05
AND adaptation_offset_ratio < 0.40
AND food_insecure_population > 700_000_000
AND aquifer_decline_pct > 0.60
AND population_growth_annual > 70_000_000
AND green_revolution_gains_plateauing = true
AND vpd_increase_exceeds_models = true
ACROSS D5, D6, D1, D3, D4, D2
DEPTH 3
SURFACE yield_curve
DIVE INTO structural_degradation
WHEN climate_warming_continues AND water_bankruptcy_declared AND productivity_plateau AND demand_growing
TRACE at_risk_cascade
EMIT at_risk_signal
DRIFT yield_curve
METHODOLOGY 70 -- Green Revolution legacy, precision agriculture (UC-108), AI weather forecasting (UC-086-091), seed R&D, irrigation technology, climate science
PERFORMANCE 20 -- yields declining despite adaptation, water bankruptcy, 800M food-insecure, no global coordination, export bans, subsidy distortions, smallholder exclusion
FETCH yield_curve
THRESHOLD 1000
ON EXECUTE CHIRP at_risk "Nature 2025: -120 cal/person/day per degree C. -8% yields by 2050. Barley -13%, wheat -10%, maize -4% already. Adaptation offsets 1/3 only. UN declared global water bankruptcy January 2026. 70% of aquifers declining. 800M food-insecure. Population +80M/year. Green Revolution plateauing. VPD increasing faster than models in Europe and China. 70-90% probability of yield decline by 2100 for all staples except rice. The supply curve is bending down while the demand curve bends up. The intersection is structural food insecurity for billions."
SURFACE analysis AS jsonRuntime: @stratiqx/cal-runtime · Spec: cal.cormorantforaging.dev · DOI: 10.5281/zenodo.18905193
Farmer adaptation is real and meaningful. Switching varieties, shifting planting dates, adjusting fertiliser — these offset approximately one-third of climate losses. But two-thirds remain structural. No individual farm decision can lower the atmospheric temperature or restore depleted aquifers. The two-thirds wall is the boundary between what markets can solve and what requires systemic intervention. That intervention does not exist at the required scale.
The greatest absolute losses will occur not in the tropics but in today’s breadbaskets — the US Great Plains, European wheat belt, southern Brazil. These regions have favourable current climates and limited present adaptation precisely because they have not yet needed it. When the climate shifts, they will face the largest yield shocks because they start from the highest baseline. The developing world suffers more in relative terms, but the global food supply depends on the breadbaskets. Both are at risk.
Atmospheric drying — measured by vapour pressure deficit — is increasing faster than climate models predicted in Europe and China. Even when soil moisture is adequate, dry air causes plants to close their stomata, reducing photosynthesis and growth. This means irrigation alone is not a solution: you can water the roots but the atmosphere is too dry for the leaves. This mechanism explains yield losses that previous analyses attributed solely to heat, and it changes the adaptation calculus significantly.
The weather pentalogy (UC-086–091) documented the AI revolution in forecasting. Those cases are upstream of this one: better forecasts help farmers make better decisions about planting, irrigation, and harvest timing. But forecasts cannot change the climate — they can only improve the response to it. The yield curve documented here is what the forecasts are measuring. AI weather will help at the margin (UC-088 showed 15–20% yield gains from precision timing). It cannot solve the two-thirds wall.
One conversation. We’ll tell you if the six-dimensional view adds something new — or confirm your current tools have it covered.