Elevated CO₂ Does Not Uniformly Benefit Crop Yields, and the Fertilization Narrative That Claims Otherwise Has a Funded Lineage and a Measurable Policy Cost

In 2006, Stephen Long and colleagues reported a result in Science that should have reshaped the public conversation about climate change and agriculture. After compiling data from Free-Air CO₂ Enrichment (FACE) trials across four continents, the authors concluded that yield responses to elevated CO₂ in open-field experiments were approximately half those predicted by the chamber studies that had underwritten a generation of optimistic food-security modeling. Wheat, rice, and soybean gained, on average, only around thirteen percent yield at 550 ppm atmospheric CO₂ in FACE, against figures of roughly thirty percent from Open-Top Chamber and greenhouse studies of the previous decade. For maize, a C4 crop whose photosynthesis is already CO₂-saturated at present atmospheric concentrations, the measured yield response under well-watered FACE conditions was statistically indistinguishable from zero. The paper was direct about its implication: estimates of future global food production under climate change that relied on the older chamber numbers were, in the authors' phrase, likely to be "strongly overestimated."

Twenty years and several hundred FACE-years of data later, that correction has propagated unevenly. It has reached the specialist literature. It has reached the integrated assessment models that inform climate policy, partially. It has reached the public-facing discourse, in which "CO₂ is plant food" remains a serviceable sound bite, almost not at all. This article argues four things. First, that the empirical case for a uniform, large, and durable CO₂ fertilization benefit across major food crops does not survive engagement with FACE-era evidence once C3/C4 photosynthetic differences, the full stress stack (heat, water, tropospheric ozone, nitrogen, phosphorus, pests and pathogens), and nutrient dilution effects are incorporated. Second, that the FACE numbers themselves may be ceiling estimates: recent Earth-system and multi-generational crop evidence suggests the observed fertilization response has already been declining and is not sustained across generations in the way single-season FACE trials measure. Third, that the "CO₂ is plant food" advocacy narrative is not an organic product of the scientific literature but a funded communications campaign with documented industry origins tracing to a 1991 coal-industry video, whose argumentative structure still shapes 2025 policy documents. Fourth, that the policy cost of that narrative is quantifiable: when realistic agricultural impacts are modeled, the Social Cost of Carbon more than doubles. The point is not that CO₂ does nothing for crops. The point is that what it does is far smaller, far more uneven, far more easily erased by co-occurring stressors, and far more effectively weaponized in policy advocacy than the popular framing admits.

The Chamber-to-Field Gap Was Not a Rounding Error

The early enthusiasm for CO₂ fertilization rested on an experimental architecture that systematically overestimated the effect. Greenhouses, growth chambers, and open-top chambers (OTCs) impose boundary conditions that diverge from field reality in directions that all flatter the CO₂ signal. Chamber walls reduce wind, raise humidity, and elevate air temperature above ambient. Root volumes are often restricted. Canopy light penetration is altered. Soil microbial communities and mycorrhizal associations are simplified or absent. Pest and pathogen pressure is excluded. Each of these deviations shifts the plant's physiological baseline in ways that tend to amplify the apparent benefit of extra CO₂.

FACE, developed in the 1990s and extended across wheat, rice, soybean, maize, sorghum, and tree systems over the following three decades, removed these confounds. The 2005 Ainsworth and Long meta-analysis in New Phytologist, synthesizing 120 FACE observations across twelve sites, documented a C3 crop yield response of approximately seventeen percent at target concentrations of 475–600 ppm. This was meaningfully positive but substantially below the thirty percent figure widely cited from chamber work. For C4 crops, the FACE-measured yield response under adequate water was statistically zero. A subsequent Annual Review of Plant Biology synthesis integrated FACE evidence with physiological theory and concluded that photosynthetic acclimation (the downregulation of Rubisco content in long-term-elevated-CO₂ plants) routinely erodes the instantaneous gas-exchange benefit measured in short experiments. More recent extended re-analyses of the pooled FACE dataset, including a 2024 University of Guelph working paper, estimate that the sustained yield benefit of CO₂ doubling sits in the 8–12 percent range once the full crop-meta-analysis dataset is analyzed with consistent methodology, further below the upper bounds commonly invoked in advocacy writing.

The implication is not that chamber evidence is worthless. It remains useful for mechanistic physiology, for screening genotypes, and for studying processes that FACE cannot resolve. The implication is that policy-relevant projections of crop response to future atmospheric composition must be anchored in field measurements, and that the early decades of climate-agriculture modeling, the period in which the CO₂-is-plant-food narrative crystallized, were calibrated to numbers that field evidence subsequently halved. Continuing to cite that older, higher figure in 2026 is not a disagreement over interpretation. It is a failure to update on twenty years of better data.

Even FACE May Overstate: The Declining Effect and the Multi-Generational Problem

The chamber-to-FACE correction is the first epistemic layer. There is a second, less discussed layer that is beginning to reshape the specialist literature: even FACE-era numbers may themselves be ceiling estimates. Two recent strands of evidence point in the same direction, and both have implications the current policy discourse has not absorbed.

The first strand comes from Earth-system modeling constrained by nitrogen-cycle observations. A 2025 PNAS analysis reports that widely used Earth-system models systematically overestimate biological nitrogen fixation (BNF), and that this misrepresentation inflates their projected CO₂ fertilization effect by roughly eleven percent. The same analysis infers, from observational carbon-flux and nitrogen-deposition records, that the global CO₂ fertilization effect has been declining over recent decades as nutrient limitations have tightened. This is not a projection about the future. It is a diagnosis of the recent past: the biosphere's capacity to convert extra atmospheric CO₂ into additional biomass appears already to be weaker than the models have assumed, and the direction of change is downward.

The second strand concerns the temporal design of FACE trials themselves. Almost all FACE experiments run for one to a few growing seasons, short enough that they capture the initial photosynthetic stimulation but too short to resolve multi-generational acclimation in the crop itself. A 2025 One Earth study addressed this gap directly for rice, the world's largest caloric staple, and reported that the CO₂ fertilization benefit on aboveground biomass, nitrogen uptake, and grain yield diminishes as the number of generations of exposure increases. The implication is structural. The canonical 10–15 percent FACE yield gain for rice is, in effect, an early-exposure measurement. What a future in which crops are grown under sustained elevated CO₂ for decades will deliver is probably lower, and the published FACE number is therefore biased upward as a long-run projection.

Taken together, these two findings point to a structural problem in the evidentiary baseline that policy documents lean on. The chamber-era thirty-percent figure was roughly twice the FACE figure. The FACE figure in turn appears to be roughly ten to fifteen percent above the sustained, multi-generational, nutrient-limited response the real biosphere will deliver. This is not a rhetorical flourish. It is the direction every systematic design refinement (from chamber to FACE, from single-season to multi-season, from nitrogen-replete to nitrogen-limited, from idealized models to observational constraints) has pushed the number. The honest summary is that the credible upper bound on a sustained, global CO₂ fertilization yield benefit for major C3 cereals under realistic conditions is meaningfully smaller than the FACE literature alone suggests, and essentially zero for C4 cereals.

C3 Versus C4: The Asymmetry the Narrative Ignores

The single most important physiological fact in this debate is one that rarely surfaces in popular writing: not all crops use the same photosynthetic pathway, and they do not respond equivalently to elevated CO₂.

This physiology has blunt consequences for a global food system. Of the four crops supplying roughly two-thirds of human dietary calories (rice, wheat, maize, soybean), three are C3 and one is C4. But the C4 crop is maize, which alone accounts for close to forty percent of global cereal production by tonnage and whose yield fuels livestock, ethanol, and processed-food industries across North and South America, China, and increasingly sub-Saharan Africa. The 2009 Leakey review in Journal of Experimental Botany, synthesizing FACE evidence specifically on the C4 question, concluded that the direct photosynthetic stimulation of C4 grain yield under elevated CO₂ is essentially nil, and that observable yield benefits in C4 systems arise only when water limitation is present, via the indirect channel of reduced stomatal conductance and improved water-use efficiency (WUE). Under well-watered conditions, maize gains nothing.

The C3/C4 asymmetry alone is sufficient to invalidate sweeping statements such as "rising CO₂ helps the world's food supply." Whether it helps depends, first, on which crop is being discussed, and second, on which co-occurring conditions prevail. A correct summary of the FACE record is not that CO₂ is plant food; it is that CO₂ is one input among many, that its effect is largest for C3 legumes and smallest for C4 cereals, and that the ranking of crops by CO₂ responsiveness does not match the ranking of crops by dietary importance.

The Stress Confound: CO₂ Benefit Erodes Under Realistic Conditions

FACE experiments conducted under otherwise favorable conditions (adequate water, nitrogen, phosphorus, benign temperature, low ozone) represent a physiological upper bound. They do not represent the conditions under which most of the world's food is actually grown, nor the conditions that climate change is producing. The relevant question for food security is not what CO₂ can do for an unstressed crop, but how the CO₂ response interacts with the stress stack that climate change delivers simultaneously.

Take temperature. CO₂ and heat do not combine additively in the crop physiology literature. They interact through at least three mechanisms: accelerated phenology that shortens the grain-fill window, direct heat damage to reproductive tissues during anthesis, and temperature-dependent rates of photorespiration that narrow the Rubisco selectivity margin on which the C3 CO₂ response depends. The Lobell, Schlenker, and Costa-Roberts analysis of 1980–2008 yield trends documented negative temperature effects on wheat and maize that are not offset by concurrent atmospheric CO₂ increases at observed rates. A 2025 Nature study extended this analysis forward, estimating that warming alone will likely reduce global yields by 2050 for most major crops, with probability of yield loss ranging from approximately 0.70 for sorghum to 0.95 for wheat. A parallel Stanford analysis found that rising global temperatures will dampen global food production even when farmers adapt. In each assessment, CO₂ fertilization at field-validated magnitudes partially offsets but does not reverse temperature-driven yield declines.

Water presents a subtler picture, and one the CO₂-optimist narrative exploits. Elevated CO₂ reliably reduces stomatal conductance and raises plant-level WUE, in some FACE studies by as much as fifty percent at the gas-exchange scale. The numerically important caveat is that the yield-level WUE gain is far smaller (on the order of thirteen percent in synthesis estimates) because reduced stomatal conductance also constrains CO₂ uptake and nutrient transport. Quoting the fifty-percent leaf-level figure without the thirteen-percent yield-level figure is a textbook case of selective citation. More fundamentally, in water-limited environments the stomatal water-saving can deliver a real yield benefit, but "water-limited" and "the benefit is net positive" are not the same claim. Gray and colleagues' 2016 Nature Plants study demonstrated that in drought years at the SoyFACE site in Illinois, the soybean CO₂ yield stimulation disappeared entirely, even though leaf-level stomatal conductance responded as predicted. The stomatal water-saving is real. Whether it is large enough to preserve yield under concurrent heat and drought is a quantitative question that has to be answered crop by crop, not waved away with physiology.

Tropospheric ozone (O₃) deserves its own treatment because it is both directly damaging to crops and rising in parallel with CO₂. Ozone enters leaves through the same stomata that admit CO₂, generates reactive oxygen species, damages Rubisco and photosystem II, and reduces canopy duration. The Ainsworth and Rogers 2007 synthesis in Plant, Cell & Environment estimated current global ozone-driven yield losses on the order of 6–16 percent for soybean, wheat, and maize, depending on region. More recent multi-pollutant assessments have shown that under high-ozone scenarios characteristic of parts of South Asia and China, ozone damage can negate CO₂ fertilization benefits entirely, with the net effect approaching zero or negative even for otherwise CO₂-responsive C3 cereals. Climate-agriculture assessments that report the CO₂ benefit without simultaneously accounting for the ozone penalty are presenting one side of a ledger whose other side is of the same magnitude.

Nitrogen is the constraint that most reliably converts a CO₂ stimulation into a statistical null over time. Progressive nitrogen limitation (PNL) is the empirical pattern, documented across woody and herbaceous FACE sites, by which the carbon-for-nitrogen ratio of new biomass under elevated CO₂ gradually outruns the soil's capacity to supply nitrogen. Rubisco content declines. Photosynthetic acclimation sets in. The CO₂ response compresses. This same mechanism underpins the PNAS result on declining global fertilization already discussed: nitrogen limitation is not a hypothetical future constraint, it is already operating. The implication for smallholder agriculture in sub-Saharan Africa and South Asia, where nitrogen fertilizer application rates are a small fraction of those in industrial systems, is that the CO₂ benefit they can expect is a fraction of the benefit measured at well-fertilized FACE sites in the US Midwest or European lowlands. Phosphorus limitation, widespread on weathered tropical soils, operates through analogous co-limitation to further compress the response.

Pests and pathogens add a further asymmetry. Plants grown under elevated CO₂ often have altered tissue chemistry, including lower protein content and shifted secondary-metabolite profiles, which changes their palatability and nutritional value for herbivorous insects. Chamber and FACE studies have documented cases in which chewing insects compensate for lower leaf nitrogen by consuming larger quantities of leaf area, raising effective herbivory. The literature is heterogeneous across crop-pest pairs, but the central point stands: CO₂ does not act on crops in a biotic vacuum, and several of the most important interactions are not beneficial for yield.

The Nutrient Dilution Tax: A Benefit That Comes Out of Nutritional Content

Even where elevated CO₂ delivers a durable biomass or grain-yield gain, that gain is rarely nutritionally neutral. A recurring and well-replicated finding from both chamber and FACE studies is that the protein, zinc, iron, and B-vitamin content of grain declines under elevated CO₂. The decline is crop- and cultivar-dependent, but its direction is consistent enough that meta-analyses recover it across species and systems.

The 2014 Myers and colleagues analysis in Nature pooled data from 41 cultivars of wheat, rice, maize, soybean, field pea, and sorghum grown at seven FACE sites. At CO₂ concentrations projected for mid-century, zinc concentrations in wheat and rice grain fell by roughly 9 and 3 percent respectively, iron by 5 and 5 percent, and protein by 6 and 8 percent. Legume and C4 responses were attenuated or absent, consistent with their underlying physiology. Subsequent work has extended the finding to B vitamins in rice: the 2018 Zhu and colleagues study in Science Advances documented significant declines in vitamins B₁, B₂, B₅, and B₉ across 18 rice cultivars grown under FACE in China and Japan. The largest synthesis to date, published in 2025 in Global Change Biology, pooled approximately 59,000 samples across 43 crop species and confirmed the pattern at a scale previous analyses had not reached: elevated CO₂ reduces concentrations of protein, iron, zinc, and multiple micronutrients across the majority of food crops examined, with the direction of effect consistent across cultivars and regions.

The Smith and Myers 2018 Nature Climate Change analysis translated these grain-composition shifts into population-scale nutritional exposures. Under a business-as-usual CO₂ trajectory, the authors estimated that by 2050 an additional 175 million people would be at risk of zinc deficiency, 122 million at risk of protein deficiency, and 1.4 billion women of reproductive age and children under five at increased risk of iron deficiency. These are numbers of the same order of magnitude as the yield benefit claimed for CO₂, operating in the opposite direction on the outcome that actually matters for human health, which is not tonnes of grain but micronutrients delivered to populations whose diets are already marginal.

A correct policy summary is therefore not "CO₂ boosts grain yield and we should count that against the harms of climate change." It is "CO₂ delivers a small yield gain for some C3 crops under some conditions, accompanied by a consistent decline in grain nutritional quality whose health consequences fall on populations already bearing the largest micronutrient-deficiency burden." These two effects do not cancel arithmetically. They operate on different outcomes. Treating them as a single net number, which a surprising amount of climate-optimist writing does, is a category error.

Where the Fertilization Narrative Has Scientific Grounding

An evidence-based critique requires stating clearly what the CO₂ fertilization effect actually is, not only what it is not. The strongest version of the case for taking CO₂ benefit seriously rests on several findings that the FACE record does support, along with one observational study that policy advocates lean on heavily and that an honest accounting must engage with directly rather than dismiss.

Legumes respond well. Soybean, pea, lentil, chickpea, and cowpea consistently show 10–20 percent biomass or seed-yield gains under elevated CO₂ in chamber and FACE work, and the effect appears more durable over seasons than the equivalent response in cereals. The physiological mechanism is well understood: grain legumes obtain nitrogen through symbiotic fixation by rhizobial bacteria, which partially decouples them from the soil nitrogen limitation that imposes progressive acclimation in cereals. The additional carbon from elevated CO₂ can flow into supporting enhanced nodulation and fixation, converting photosynthetic gain into biomass more efficiently than in nitrogen-demanding cereals. This is a genuine, repeatable finding. In a world where pulses are an underappreciated component of sustainable, protein-dense, low-input agriculture, it matters.

Water-use efficiency gains are real and consequential in arid and semi-arid cropping systems. The stomatal closure response to elevated CO₂ is reliable and occurs in both C3 and C4 plants. Where water is the binding constraint on yield, this can deliver benefits even for crops whose direct photosynthetic response is small. Sorghum, pearl millet, and maize grown under imposed drought conditions have shown measurable yield advantages under elevated CO₂ relative to ambient-CO₂ droughted controls. For dryland agriculture in the Sahel, the Indian semi-arid tropics, and the Australian wheat belt, this is not a trivial effect, and an honest accounting must include it, while keeping in view the leaf-to-yield gap (a fifty-percent plant-level WUE gain corresponds to roughly thirteen percent at the yield level).

Rice under well-watered, well-fertilized paddy conditions delivers one of the most consistent cereal-level CO₂ responses in the literature, with yield gains on the order of 12–15 percent at 550 ppm under favorable temperatures. The One Earth multi-generational finding complicates but does not eliminate this result: the effect is real in the early generations that dominate the published FACE record, and durable over shorter horizons. For regions where irrigated rice systems dominate and are unlikely to be heat-limited below species thresholds, the initial benefit is a real additive effect on food supply, qualified by the expectation that it attenuates as exposure accumulates.

The strongest observational study that CO₂-optimist policy documents invoke is Schlenker and Taylor's NBER working paper 29320, which uses satellite-derived productivity indices over US field crops to infer positive effects of rising CO₂ on yields. This is legitimate research using a data modality distinct from FACE, and any honest steelman must acknowledge that it exists and that it points in a supportive direction for the fertilization hypothesis. The relevant objection is not to the paper itself but to how it is cited: as a stand-alone foundation for policy conclusions that require the full ledger (nutrient dilution, temperature interaction, geographic distributional effects, multi-generational attenuation) before any net judgment can be defended. A single observational study in a temperate, well-managed agricultural region is evidence. It is not the entire evidence.

The honest summary is that none of these findings rescue the simple narrative. Each is conditional. The legume response assumes adequate phosphorus, functional rhizobia, and non-extreme temperatures. The WUE benefit is most pronounced precisely in the systems where temperature stress is most likely to be concurrent. The rice response collapses under heat stress above roughly 34°C during flowering and attenuates across generations. The Schlenker-Taylor observational signal is real but regionally circumscribed. What the steelman case establishes is that the FACE and observational record contains real positives that a comprehensive climate-agriculture assessment must include. What it does not establish is a general case for treating CO₂ as a net benefit to the world's food supply under the climate trajectory we are actually on.

The Industry Origin Story: How a Real Effect Became a Funded Narrative

Up to this point, the critique has engaged the CO₂ fertilization claim as a scientific proposition. This is necessary but incomplete, because the claim does not circulate primarily as a scientific proposition. It circulates as a policy narrative, and that narrative has a specific, traceable, funded history that is usually absent from both the popular and the specialist literature.

In 1991, the Western Fuels Association, a consortium of coal-producing utilities, spent approximately 250,000 US dollars to produce a video titled The Greening of Planet Earth, narrated by climatologist Sherwood Idso. The explicit purpose, documented in the association's internal materials, was to reframe atmospheric CO₂ from a pollutant to be regulated into an agricultural input to be welcomed. A 1992 sequel, The Greening of Planet Earth Continues, was produced and, according to later reporting, circulated favorably within the George H.W. Bush White House. The Greening Earth Society, which distributed and amplified this messaging through the 1990s and 2000s, was a nonprofit front group established by the Western Fuels Association for that purpose. This lineage is not speculation or inference. It is documented in industry archives and in subsequent investigative and academic work.

What made the campaign effective, and what distinguishes it from simple misinformation, is that it was built on a kernel of genuine science. The videos featured real scientists and cited real experiments. But the editorial choices were systematic: context about heat stress, drought interaction, weed competition, pest response, and nutrient dilution was omitted; individual findings were presented without the qualifiers the underlying literature attaches to them; and the distinction between controlled-chamber experiments and open-field behavior was never surfaced. Documentary reconstruction of the campaign has shown that these omissions were not accidents of compression but a consistent editorial pattern.

The narrative's persistence over three decades is not accidental either. It exploits a structural asymmetry in science communication. The claim "CO₂ helps plants grow" is simple, memorable, and partially true. The accurate rebuttal ("under controlled conditions with adequate water and nitrogen, short-term experiments show yield gains for some crops, but these benefits are offset by simultaneous temperature increases, nutrient dilution, multi-generational acclimation, and geographic inequity") is accurate and unwieldy. Industry-funded advocacy has consistently exploited this asymmetry, and it continues to appear in regulatory comments, opinion writing, and policy documents today.

The most recent and most consequential instance is the United States Department of Energy's July 2025 Critical Review of Impacts of Greenhouse Gas Emissions on the US Climate. The report asserts that CO₂ fertilization has a stronger beneficial effect on agriculture than was known when integrated assessment models such as DICE and FUND were originally parameterized, and uses this assertion to argue for softer climate policy. Carbon Brief's detailed fact-check of the report documents more than one hundred false or misleading claims, including the central claim that the fertilization benefit offsets warming damages at policy-relevant scales. The Harvard Salata Institute's public comment notes specifically that the nonlinear adverse impact of rising temperatures will be felt most acutely in regions that are already warm, and that the CO₂ fertilization benefit the report leans on is limited and conditional. Science Feedback's review characterizes the report as choosing bias over science and as excluding well-established countervailing evidence, including the nutrient-dilution and multi-stress literatures this article has summarized.

It is useful, in this connection, to note what the IPCC actually says. Working Group II of AR6 presents CO₂ fertilization as one factor among many and explicitly acknowledges that temperature increases, precipitation changes, extreme events, and nutrient-quality reductions will dominate agricultural outcomes in most scenarios. The policy advocacy uses of "the IPCC says" in the CO₂ fertilization debate frequently invert this emphasis, presenting as the IPCC consensus what is in fact a selective extraction from it. The distinction between what the IPCC literature actually supports and what advocacy documents claim it supports is one of the clearest places where scientific and rhetorical uses of the same vocabulary diverge.

Four Narratives, Four Failures

The gap between the FACE-era scientific record and the public uses of CO₂ fertilization is not random. It appears in four specific policy contexts, each of which selectively presents the evidence to support a preferred conclusion.

Climate-Optimist Advocacy

The most visible instance is the argument, advanced in popular climate-skeptic and climate-lukewarmer writing, that CO₂ fertilization is a global greening of sufficient magnitude to offset agricultural harms from warming. The strong version of this claim cites satellite-derived leaf-area-index trends attributing a global greening signal to rising CO₂. The observational greening is real. The inference that it translates into a proportional crop-yield benefit, however, is unsupported by FACE work for the reasons already canvassed: the largest responders (C3 legumes, trees in temperate forests) are not the same as the crops that feed most people, and the stress stack compresses the signal that does exist. As the preceding section documented, this line of argument is not an independent reading of the satellite and field evidence. It is a direct rhetorical descendant of the Western Fuels campaign and its successors, and the 2025 DOE report is its most recent institutional instantiation. A responsible summary of the satellite record and the FACE record reads them as complementary pieces of evidence about different outcomes. A selective summary treats the satellite greening as a reason for complacency about the agricultural consequences of continued warming, which the yield literature does not support.

Food-Security Modeling in Integrated Assessment Models

A second failure mode, less visible but more consequential, lives inside the integrated assessment models that shape national and multilateral climate policy. Many of these models carry CO₂ fertilization parameters that were calibrated to the chamber-era literature and have been updated only partially toward field-validated values. A 2020 Nature Food multi-model analysis comparing crop-model projections with FACE observations concluded that a subset of widely used models continues to over-predict CO₂ yield benefit, particularly for maize, and that this systematically biases downstream food-price and food-availability projections in an optimistic direction. The quantitative policy consequence is not marginal. A 2017 Nature Communications analysis estimated that incorporating more realistic agricultural impact modeling, accounting for the fact that CO₂ fertilization does not offset warming damages at the scale earlier IAMs assumed, shifts agricultural impacts in Social Cost of Carbon calculations from a net benefit of approximately 2.70 US dollars per tonne to a net cost of approximately 8.50 US dollars per tonne, with the total SCC more than doubling as a result.

Because IAM output feeds into the economic and policy assumptions behind nationally determined contributions, adaptation finance, and trade negotiations, a calibration error at this point in the chain is not confined to the academic literature. It propagates into operational climate policy. The direction of the bias is consistent: optimistic. The magnitude, for the specific case of agricultural impact accounting, is more than a doubling of the estimated social cost of each tonne emitted.

Adaptation and Breeding Complacency

A third failure is what the breeding community sometimes calls the "CO₂ will save us" heuristic. Public breeding programs in low- and middle-income countries have, in some instances, deprioritized selection for heat tolerance, drought tolerance, and nutritional quality on the implicit assumption that CO₂ fertilization will lift yields sufficiently to buy time for other interventions. This reasoning is most damaging where the underlying crops are least CO₂-responsive. Maize-dominant agricultural systems in eastern and southern Africa cannot be rescued by a CO₂ effect that FACE work shows to be absent under their conditions. Cultivars selected for heat resilience, phenological plasticity, and nutritional density do not self-generate; they require sustained public investment. Every year in which that investment is reduced on the assumption of a CO₂ dividend is a year the climate-agriculture gap widens rather than closes. A 2026 Frontiers in Climate analysis makes the stronger point that yield projections hinging on CO₂ fertilization as a partial offset are structurally fragile, in the sense that the assumptions they require (adequate nitrogen, benign temperature, negligible ozone, single-generation FACE validity) are unlikely to jointly hold across the regions where adaptation planning is most urgent.

Carbon-Sink Claims for Agriculture

A fourth and increasingly common failure is the claim, made in various agricultural-offset and regenerative-agriculture marketing contexts, that elevated CO₂ plus altered management will convert agricultural land into a durable net carbon sink, partially compensating for industrial emissions. The soil-carbon literature here is nuanced, but the claim that CO₂ fertilization itself (independent of management) converts cropland into a net sink is not well supported. FACE studies that directly measured soil organic carbon changes under elevated CO₂ have reported responses ranging from small positive to net zero across a range of cropping and grassland systems, with no evidence for a durable, large sink effect proportionate to the emissions being offset. The PNAS declining-BNF result further undercuts simple sink claims: if the biosphere's sustained capacity to convert CO₂ into additional biomass is already weakening under nutrient limitation, the sink-expansion story built on that conversion is weaker than its proponents describe. Where real soil-carbon gains occur, they are largely attributable to changes in tillage, cover cropping, and residue management, not to the direct CO₂ pathway that carbon-credit markets frequently invoke.

These four failure modes share a common structure. Each takes a real but conditional finding from the CO₂ literature (a greening signal, a modest yield gain, a WUE benefit, a soil-carbon increment) and extrapolates it to a scale or context the evidence does not support. Each erases the stress stack, the C3/C4 distinction, the nutrient dilution cost, or all three. Each serves a policy preference that is harder to defend once the conditions are reinstated. Calling this pattern "scientific misinformation" is not overwrought. It is the appropriate term for systematic selective presentation of evidence in service of policy conclusions the full evidence does not support.

For Students and Public Readers: How to Read a CO₂ Fertilization Claim

Given how easily the CO₂ story lends itself to misrepresentation, a small number of diagnostic questions can separate credible claims from rhetorical ones. First: what crop is being discussed, and is it C3 or C4? A claim that cites global staple-crop yields without distinguishing wheat, rice, and soybean from maize and sorghum is elided physiology. Second: what experimental architecture generated the cited numbers? If the figure is north of twenty percent, it almost certainly originated in chamber or OTC work and has not been replicated at comparable magnitude in FACE. Third: what stressors are included? Any CO₂-benefit figure presented without specifying the concurrent temperature, ozone, water, nitrogen, and phosphorus conditions is a ceiling estimate, not a realistic projection. Fourth: is the nutritional dimension addressed? A yield-only analysis is incomplete for crops feeding populations with existing micronutrient deficiencies. Fifth: whose projection is it, and is the projection single-season or multi-generational? A single FACE result from a well-fertilized temperate site, presented as a global long-run estimate, is a category error. Sixth, and this is new in 2026: what is the funding and argumentative lineage of the claim? If a policy document cites CO₂ fertilization as a net benefit in the language of the Western Fuels campaign, that is diagnostic. Not every invocation of plant physiology is industry-funded, but the specific rhetorical pattern of "CO₂ is plant food, therefore climate policy is less urgent" has a traceable history and should be flagged as advocacy when it appears.

None of these questions requires specialized physiology to ask. They require willingness to treat the CO₂ fertilization claim as a specific empirical hypothesis rather than a rhetorical flourish. Posed in this way, most popular versions of the claim do not survive contact with the published record.

Conclusion: Two Policy Futures, One Empirical Record

Twenty years of FACE data have converged on a picture that is more complicated than the popular narrative and, in important ways, more sobering. Elevated atmospheric CO₂ does, under favorable conditions, deliver a measurable yield and biomass benefit to C3 crops, particularly legumes, and modest water-savings benefits to C4 crops in water-limited settings. These effects are real, quantifiable, and worth including in any serious accounting of the climate-agriculture relationship. What the evidence does not support is the rhetorical use of these effects to suggest that rising CO₂ will substantially offset, let alone reverse, the agricultural harms of concurrent warming, drought intensification, ozone increase, nutrient dilution, and now the additional evidence that the fertilization response itself is weaker in multi-generational and nitrogen-limited conditions than single-season FACE trials captured. The FACE numbers are smaller than the chamber numbers. The sustained numbers are likely smaller than the FACE numbers. The C4 number is close to zero. The stress stack compresses the C3 number further. The nutrient dilution cost partially negates whatever biomass benefit survives. The integrated assessment models that still propagate the older, higher estimates into food-security projections carry a documented optimistic bias that, corrected, more than doubles the Social Cost of Carbon.

Two policy futures are defined by how this evidentiary picture is handled. In the first, CO₂ fertilization continues to be treated as a meaningful offset to warming damages in IAMs and regulatory documents. Agricultural adaptation investment is deprioritized because projections appear more favorable than they are. Food-security planning in the Global South is calibrated to a rosier baseline. The nutritional dimension of climate impacts remains invisible in calorie-focused modeling. The SCC remains systematically underestimated, weakening the economic case for mitigation. Planning failures compound over decades, and by the time the gap between projected and realized agricultural performance becomes undeniable, the adaptation window has narrowed substantially. In the second, integrated assessment models are re-parameterized to reflect realistic CO₂ fertilization magnitudes (accounting for nitrogen limitation, multi-stress interactions, multi-generational attenuation, and nutrient dilution), nutritional quality is incorporated alongside caloric yield, and agricultural impact in SCC calculations reflects the net-cost estimates the post-2017 literature supports. The resulting food-security projections are more alarming. They are also more accurate. They generate stronger, better-grounded arguments for both mitigation and targeted adaptation investment. The political cost of the second path is real. The long-run cost of the first is larger and, past a point, irreversible.

The honest position is that CO₂ fertilization is a real but modest and conditional benefit whose magnitude has been substantially overstated in public discourse, whose distribution across crops and regions is sharply uneven, whose multi-generational sustainability is increasingly in doubt, and whose nutritional consequences are negative for precisely the populations most at risk. None of this vindicates a blanket dismissal of CO₂ physiology. It vindicates a more precise and more conditional accounting than the current discourse usually offers, and a sharper separation between the peer-reviewed science (which is generally careful and qualified) and the advocacy narrative (which is neither, and which carries a documented funded history that every serious policy analyst should know). Climate-agriculture policy that treats CO₂ benefit as a general credit against warming harm is working from a scientific picture that has been outdated for roughly two decades and from a rhetorical script that has been recognizably the same for more than three.

The case for taking atmospheric CO₂ seriously as a driver of climate risk does not rest on denying plant physiology. It rests on describing plant physiology accurately, and describing the policy uses of plant physiology honestly. The CO₂ fertilization effect, correctly summarized, is not a counterweight to the agricultural harms of continued warming. It is, at best, a small, conditional, fragile, and diminishing partial offset to a much larger set of concurrent stressors, and the confidence with which it is often invoked in policy contexts is a function of the narrative's funding history more than of its evidentiary strength. Writing about it as anything more than that is not caution. It is confusion, and the cost of that confusion is borne by the people whose food supply depends on our getting the physiology, and the political economy around the physiology, right.