A 21st Century Agricultural Apocalypse

Abstract

Industrial agriculture is among the greatest material achievements of the twentieth century, and roughly half of the world's population is fed on nitrogen fixed synthetically from the air (Erisman et al., 2008). This review argues that the achievement was won by a particular means, and that the means carries a structural cost now visible across many independent measures at once. Drawing the diagnosis from a companion framework, which holds that agriculture is the managed extraction of harvest from a living resource web whose sustainable ceiling is set by the soil's own capacity to acquire and recycle nutrients, the paper assembles the evidence that the dominant model exceeds that ceiling by substituting purchased inputs for the soil's living economy. It surveys, in turn, soil erosion and the topsoil deficit; the decline of soil organic carbon, structure, and biology; salinization and the depletion of agricultural water; the nitrogen and phosphorus cascade and its endpoint in coastal dead zones; the effects of agrochemicals on the living environment and on human health; the documented influence of commercial interests on the relevant science; persistent legacy residues that do not leave; the herbicide-tolerant cropping model and the resistance treadmill it has produced; and the systemic pressures of greenhouse-gas emissions and genetic uniformity. Throughout, the paper distinguishes what is robustly established from what is genuinely contested, declines to rest any conclusion on inflated or withdrawn figures, and grants both the achievement of the model and the real mitigations now in use. The conclusion is that the apocalypse of the title is not a dated catastrophe but the ordinary, compounding consequence of treating a self-renewing system as a dead medium, and that the diagnosis is precise in order that the response, set out in the companion solution paper, can be.

---

1. Introduction: the revolution that worked, and the bill coming due

Any honest account of modern agriculture must begin by granting what it achieved, because the achievement was real and the argument that follows does not depend on denying it. Over the twentieth century the combination of synthetic nitrogen fixed from the air, selectively bred high-yielding crops, mechanization, irrigation, and chemical control of pests and weeds raised food production faster than population grew. The synthesis of ammonia in particular changed the world, and by widely cited estimates the nitrogen it supplies now underwrites approximately half of all the food produced on Earth (Erisman et al., 2008). Famines that had recurred for millennia became, across much of the world, the exception rather than the rhythm of life. Billions of people are alive because of it.

The achievement, however, was won by a particular means, and the means is the subject of this paper. The high-yielding model did not raise the soil's own capacity to feed a crop; it bypassed that capacity, supplying the crop directly with soluble nutrients from outside the system and protecting it with chemistry, while the living machinery that a healthy soil uses to acquire, hold, and recycle its own fertility was, at best, left idle and, more often, actively dismantled. The companion framework develops this as the central claim: a soil draws its fertility through an internal recycling loop, which returns the nutrients in residues and wastes, and a biological acquisition network of roots, fungi, and microbes that releases nutrients from minerals and delivers them to plants, and the dominant model severs both. For as long as external inputs keep flowing and the soil's residual capital lasts, the bypass works and yields hold. The difficulty is that the bypass has been mistaken for the foundation, and the sections that follow are an itemization of the bill now coming due. A word on causal scope is owed at the outset, because those sections are not all consequences of the same mechanism. The severance directly produces one cluster of them, the loss of soil carbon and biology, the manufactured dependence on external inputs, and the escape of nutrients into water and air; others belong to the same chemical-intensive model without being products of the severance as such, among them the contamination and the resistance generated by the chemical regime, the finitude of mineral inputs, and the fragility of genetic uniformity. Several of these, climate foremost, then feed back to deepen the very degradation the severance begins. What follows is therefore an indictment of a model rather than of a single mechanism, and it marks the difference between what the severance causes and what merely accompanies it wherever that difference bears on the argument.

Two commitments govern what follows. The first is calibration. Where this paper makes empirical claims it keeps to what is robustly established, presents genuinely contested questions as contested rather than settled in either direction, and avoids the inflated and frequently repeated figures that have circulated on this subject, because the real account is serious enough and because an argument built on exaggeration invites the dismissal of the whole. The second is evenhandedness. The degradation processes described here vary by region, by crop, and by practice; several are improving in places; and meaningful mitigations exist and are noted where they apply. The "apocalypse" of the title is therefore not a prophecy of a catastrophe on a fixed date. It is a set of measurable, ongoing, and well-documented degradation processes, and the structural trajectory they jointly describe.

2. Soil erosion and the topsoil deficit

Soil forms slowly, over centuries to millennia, and on land cleared for cultivation it is lost far faster than it forms. The most comprehensive global assessments place erosion rates on arable and intensively grazed land at one hundred to one thousand times the natural geological background, with hilly croplands in tropical and subtropical regions reaching fifty to one hundred tonnes per hectare per year against soil formation that is one to two orders of magnitude slower (FAO and ITPS, 2015). Montgomery (2007) set out the long historical pattern, in which societies that exhausted their soils declined with them, and showed that conventional tillage erodes soil at rates that cannot be sustained against its formation.

A civilization that loses the top of its soil faster than the bottom can build is, by definition, drawing down a stock. The most rigorous recent quantification, a high-resolution global model by Borrelli et al. (2017) and pointedly titled for the twenty-first century, estimated approximately 35.9 petagrams of soil eroded in 2012 and projected an overall increase driven by the expansion of cropland, with the largest increases expected in Sub-Saharan Africa, South America, and Southeast Asia. That same study illustrates the calibration this paper insists on: the figure most often quoted on this subject, of 75 billion tonnes eroded each year at a cost of some 400 billion United States dollars, traces to an estimate first published in 1993, and Borrelli's modern modelling is at least two times lower. This paper therefore leads with the contemporary peer-reviewed estimate and treats the older figure as historical. The Food and Agriculture Organization has separately warned that erosion could reduce crop production by approximately ten percent by 2050 (FAO, 2022). It should be noted in fairness that the trajectory is not uniform: conservation tillage, cover cropping, and residue retention measurably reduce erosion where they are adopted, and the problem is one of dominant practice rather than of agriculture as such.

3. Soil organic carbon, structure, and biology

Alongside the physical loss of soil is the loss of what makes the remaining soil fertile. The organic carbon and humified organic matter that give a soil its structure, its water-holding capacity, and much of its biological habitat are oxidized away under continuous tillage and bare fallow faster than a cropping system returns them. Tillage in particular breaks open the soil aggregates that physically protect organic matter, exposing it to microbial oxidation and releasing it to the atmosphere as carbon dioxide (FAO and ITPS, 2015). The scale of the historical loss is large: Sanderman, Hengl and Fiske (2017) estimated that agricultural land use has caused the loss of approximately 133 petagrams of carbon from the world's soils, concentrated in major cropping regions and degraded grazing lands.

The contemporary picture is consistent with that history. The Food and Agriculture Organization reports that about one third of the world's soils are degraded, that this degradation has already released up to 78 gigatonnes of carbon to the atmosphere, and that organic-carbon losses across affected croplands and grazing lands have ranged between 25 and 75 percent of the original pool (FAO Global Soil Partnership, 2017). The degradation is ongoing rather than historical: by the Food and Agriculture Organization's own land-degradation accounting, the world lost at least 100 million hectares of healthy, productive land each year between 2015 and 2019, with approximately 1.3 billion people living directly on degrading land (FAO, SDG Indicator 15.3.1). Soils hold the largest terrestrial carbon stock, on the order of 1,500 gigatonnes of carbon in the first metre, and under prevailing management they have been a net source of greenhouse gases rather than a sink. Soil biology declines with the organic matter that houses and feeds it, in both abundance and diversity, which is the acquisition network of the framework being dismantled in measurable terms. Here too the account must be balanced: soil organic carbon is partly recoverable, with credible estimates of a recoverable reserve of 21 to 51 gigatonnes of carbon under sustained restorative management, which is precisely the opening that the companion solution paper develops.

4. Salinization and the depletion of agricultural water

The model's demands on water compound the demands on soil. On a large fraction of the world's irrigated land, the salts dissolved in irrigation water accumulate where the water evaporates, until the ground turns saline and yields fall, a slow drawdown but an old and well-documented one; by conservative estimate, salinization alone has degraded approximately 82 million hectares of rainfed and 24 million hectares of irrigated cropland (FAO, 2024), and the Food and Agriculture Organization has judged the interlocking systems of soil, land, and water to be at a breaking point (FAO, 2021). The depletion of the water itself is the second face. The High Plains, or Ogallala, Aquifer, which underlies eight states of the central United States, supplies water beneath approximately 27 percent of all irrigated land in the country and yields roughly 30 percent of the groundwater used for irrigation nationally (USGS, via published syntheses). Since large-scale irrigation began in the 1940s, water levels have fallen by more than 30 metres in parts of Kansas, New Mexico, Oklahoma, and Texas; irrigation has reduced the saturated volume of the aquifer by an estimated nine percent since 1950; and natural replenishment, once the aquifer is drawn down, would take on the order of six thousand years. Modelling of the same aquifer projects that roughly 24 percent of currently irrigated land may be unable to support irrigated agriculture as depletion proceeds. Prolonged pumping has, in places, drawn mineralized water from adjacent formations and raised salinity in the aquifer itself, joining the two threads of this section. In fairness, gains in irrigation efficiency have slowed the rate of decline in several districts, but management has largely aimed at orderly depletion rather than at a sustainable yield, and a stock drawn down faster than it recharges is, again, a stock being spent.

5. The nitrogen and phosphorus cascade

Because the severed model works by flooding the ground with soluble nutrients that the living system is no longer present to capture and hold, much of what is applied is never taken up by the crop. It leaves, and the damage it does on the way out is its own chapter of the reckoning. Nitrogen is the clearest case. A large share of applied soluble nitrogen leaches into groundwater and runs off into rivers, and from there to the sea, where it drives explosive algal growth that, on dying and decomposing, strips the water of oxygen. The resulting coastal dead zones now recur at the mouths of the world's great agricultural river systems. The hypoxic zone in the Gulf of Mexico, among the largest in the world, forms every summer from nitrogen and phosphorus carried down a Mississippi River basin that drains more than 40 percent of the contiguous United States across 22 states; the National Oceanic and Atmospheric Administration estimated it at approximately 6,705 square miles, about the size of New Jersey, in 2023, against a management target of a five-year average of 1,900 square miles, with hypoxia defined as dissolved oxygen at or below 2 milligrams per litre (NOAA, 2023; Diaz and Rosenberg, 2008).

The disturbance has been judged to exceed the limits the biosphere can absorb. The planetary-boundaries framework, in its most recent assessment, finds that six of nine boundaries have now been transgressed, that the boundary for biogeochemical flows of nitrogen and phosphorus is among them, and that agriculture is its main driver, with annual nitrogen application running at close to 190 teragrams, more than three times the estimated safe level (Richardson et al., 2023; Steffen et al., 2015). A share of the applied nitrogen also escapes to the air as nitrous oxide, addressed in Section 11. Phosphorus adds a second and different vulnerability. It has no substitute as a plant nutrient; it is supplied from finite rock deposits geographically concentrated to a degree rare among commodities, with United States Geological Survey data placing close to 80 percent of known reserves in Morocco and Western Sahara, rising to nearly 90 percent once China is included; and the human mobilization of phosphorus has itself been estimated to exceed the planetary boundary for freshwater eutrophication by a factor of three. Specific predictions of a near-term production peak for phosphorus have been made but remain contested, and this paper rests only on the finitude and the concentration, which are not in dispute. The honest tension should also be stated: by one assessment roughly half of current global food production depends on nutrient inputs that breach these boundaries (Gerten et al., 2020), which is exactly why the problem is structural rather than a matter of simple excess, and why precision nutrient management, buffer strips, and improved nitrogen-use efficiency, real and worth pursuing, mitigate the leak without resolving the underlying dependence.

6. Agrochemicals and the living environment

The chemistry that protects the high-yielding crop acts on more than its target. The clearest documented case is the effect of systemic insecticides on insects beyond the pest. Neonicotinoids, which by value make up roughly 30 percent of the global insecticide market, are commonly applied as seed treatments and taken up systemically by the growing plant, so that the compound is present throughout its tissues, including the nectar and pollen on which pollinators feed (Godfray et al., 2014). Their use has shifted toward near-universal prophylactic seed coating: in United States maize, the share of acreage receiving an insecticide rose from approximately 35 percent before neonicotinoid seed treatments to close to 100 percent today, even as clear and consistent yield benefits have proven elusive for most crops (review in Environmental Science and Technology, 2018). A substantial body of evidence now associates neonicotinoids with harm to honeybees, wild bees, and butterflies, and with effects on aquatic invertebrate communities; a controlled outdoor experiment demonstrated a causal relationship, with field-realistic concentrations of one neonicotinoid sharply reducing the abundance and biomass of major aquatic insect orders and halving the diversity of the most species-rich freshwater family at one microgram per litre (Barmentlo et al., 2021). Against this backdrop sit the broader reports of insect decline, of which the best known found a roughly 75 percent fall in flying-insect biomass over 27 years in German protected areas (Hallmann et al., 2017); calibration requires noting that such landscape-scale declines are multi-causal, driven by habitat loss, climate, and disease as well as by pesticides, and that the attribution of any single driver from correlational data is uncertain. It is also fair to record that the European Union restricted outdoor use of three major neonicotinoids in 2018, a mitigation of exactly the kind this paper credits.

7. Agrochemicals and human health

The human-health terrain is where calibration matters most, because it is where the evidence is most uneven and most contested. The clearly established harm is acute occupational poisoning. The World Health Organization has long recognized acute pesticide poisoning as a serious occupational and public-health problem, concentrated among agricultural workers in lower-income regions where protective equipment and regulation are limited. Estimates of its global magnitude vary widely, and some recent and widely circulated figures have been formally withdrawn, so this paper relies on the recognized existence and seriousness of the problem rather than on any single contested count.

The contested terrain is chronic, low-dose exposure, and its central case is glyphosate, the most widely used herbicide in the world. In 2015 the International Agency for Research on Cancer classified glyphosate as probably carcinogenic to humans, Group 2A, on the basis of limited evidence in humans centred on non-Hodgkin lymphoma, sufficient evidence in experimental animals, and strong evidence of genotoxicity, in an assessment by seventeen scientists from eleven countries (IARC, 2015). A 2019 meta-analysis reported a 41 percent increase in the risk of non-Hodgkin lymphoma among the most heavily exposed (Zhang et al., 2019), and an earlier pooled analysis had found roughly a doubling of risk (De Roos et al., 2003). The countervailing evidence is substantial and must be stated with equal weight. The positive epidemiological signal derives in large part from case-control studies, which are vulnerable to recall bias, whereas the largest prospective study, the Agricultural Health Study of more than fifty thousand licensed applicators followed for two decades, found no statistically significant association between glyphosate and non-Hodgkin lymphoma or its subtypes (Andreotti et al., 2018). On that larger dataset, which also included unpublished studies submitted by manufacturers, the United States Environmental Protection Agency, together with the European Food Safety Authority, the European Chemicals Agency, and the Joint Food and Agriculture Organization and World Health Organization Meeting on Pesticide Residues, concluded that glyphosate is unlikely to pose a carcinogenic risk at realistic exposures. Part of the disagreement is structural: the cancer agency assesses hazard, whether a substance can cause cancer under any circumstances, while the regulators assess risk at expected exposure, and the two questions have different answers. The same Group 2A category contains red meat and very hot beverages, which illustrates that such a listing identifies a possible hazard rather than a demonstrated danger at the doses people encounter. The honest summary is that the matter is genuinely unresolved, that the cancer most consistently at issue is the one named in the litigation, and that this paper does not assert a verdict the science has not reached. The same care applies to other compounds: the organophosphate insecticide chlorpyrifos, for instance, has been the subject of regulatory restriction over concerns about neurodevelopmental effects in children, an area that is itself partly contested. The governing principle is that associations are not proof of causation, and that the well-documented occupational hazard and the contested chronic-exposure question must not be conflated.

8. The contested science and the influence upon it

If the chronic-health question is unresolved, part of the reason is that the evidentiary record on which regulators relied has been shown, in court, to have been shaped by the manufacturer of the product under review. During the litigation over glyphosate, discovery compelled the disclosure of millions of internal company documents, the so-called Monsanto Papers. These records include instances of ghostwriting, in which company employees drafted scientific articles that were then signed by ostensibly independent researchers, and internal correspondence indicating coordination with regulatory staff over the conduct of safety reviews (court records, summarized in contemporaneous reporting). A study published in 2000 and long relied upon to support glyphosate's safety was retracted by its journal some twenty-five years later, the journal citing undisclosed involvement by the company. The litigation itself has been consequential: the first case to reach trial ended in 2018 with a 289-million-dollar award to a California groundskeeper with non-Hodgkin lymphoma; the manufacturer lost its first three trials; its parent company set aside more than 16 billion dollars against liability and agreed in 2020 to pay more than 10 billion dollars to settle approximately 100,000 claims; and the matter remains live, with a preemption question before the United States Supreme Court and a further multibillion-dollar class settlement proposed.

The proper inference from this record requires care, and it cuts in a specific direction. It does not, by itself, resolve the underlying toxicology, and it would be a mistake to treat litigation outcomes as scientific findings. Nor does the criticism run in only one direction: an external adviser to the cancer agency's working group was subsequently retained as a paid consultant by plaintiffs' attorneys in this litigation, and the agency's reliance on hazard rather than real-world exposure has drawn substantive criticism of its own. What the documented influence does establish is narrower and sufficient: that confident claims of settled safety, resting on a literature and a regulatory record demonstrably shaped in part by the interested party, deserve the scrutiny they have not always received. The contestation, in other words, is not a weakness in the case against the dominant chemical model. It is part of the case, because a body of evidence that has been manipulated cannot bear the weight of reassurance that has been placed upon it. It is worth recording that glyphosate remains the most common agricultural herbicide in the United States and is applied to most of its corn, cotton, soybean, and sugar-beet acreage, and that the manufacturer has removed glyphosate from consumer formulations while retaining it in the agricultural ones.

9. Persistent legacy residues

A further cost of the chemical model is that some of what it applies does not leave. The organochlorine insecticides, of which DDT is the archetype, were used heavily from the late 1940s through the 1970s and are characterized by low volatility, long environmental persistence, and the tendency to bioaccumulate in tissue and biomagnify up food chains; they are governed today as persistent organic pollutants under the Stockholm Convention. The persistence is not historical only: a peer-reviewed study of former orchard land found DDT metabolites still present at elevated levels in soil, earthworms, and the eggs of birds feeding on them some 26 years after an earlier survey of the same sites, the contamination still moving up the food chain decades after use ceased (Kesic et al., 2021). The principle the example establishes is that an agrochemical can outlast by generations the harvest it was applied to protect.

The contemporary instance of the same principle is the entry of per- and polyfluoroalkyl substances, the so-called forever chemicals, into farmland through the use of treated sewage sludge, or biosolids, as fertilizer. These compounds are extremely resistant to degradation, persist indefinitely, and bioaccumulate. The state of Maine found in 2019 that 95 percent of the sewage sludge produced in the state contained unsafe levels of these substances, and that farms which had spread biosolids had contaminated soil, water, crops, and animals, prompting state restrictions; industry-group estimates cited by independent analysts suggest that as many as 70 million acres of United States farmland could be affected, in a practice that federal sludge rules do not currently regulate for these compounds (Maine DEP, 2019; EWG, 2025). The acreage figures are estimates and should be cited as such, but the persistence of the compounds and their documented entry into agricultural soils are not in question.

10. The herbicide-tolerant model and the resistance treadmill

The genetic engineering of crops belongs in this account not for the reason most often supposed, but for a specific and well-documented agronomic and ecological one, and the distinction is essential. The separate question of whether approved genetically engineered foods are safe to eat has a clear scientific consensus, reached by the National Academies of Sciences and concurred in by major scientific bodies, that they are as safe as their conventional counterparts (National Academies of Sciences, Engineering, and Medicine, 2016). This paper does not dispute that consensus and makes no claim about dietary safety. The critique here is of the cropping model that herbicide tolerance entrenched.

That model commits the field to blanket chemical use. With the introduction of herbicide-tolerant crops in 1996, glyphosate use rose steeply; in the United States it climbed from approximately 4,000 tonnes in 1987 to more than 80,000 tonnes in 2007, and an analysis of the first sixteen years of genetically engineered crops attributed to them a net increase of some 404 million pounds of pesticide once the insecticide reductions from insect-resistant varieties are netted out (Benbrook, 2012; 2016). The predictable biological consequence followed. Repeated reliance on a single mode of action selected for weeds that survive it: glyphosate-resistant Palmer amaranth emerged in the southeastern United States in 2006 and spread across the country, and more than forty weed species worldwide have now evolved resistance to glyphosate, against a longer list resistant to one or more herbicides. The industry response has been to engineer crops tolerant of older herbicides, including 2,4-D and dicamba, that are considered more toxic to human health and the environment than glyphosate; these compounds are volatile, and in 2017 complaints of drift damage to neighbouring fields were numerous enough that Arkansas and Missouri banned dicamba spraying for part of the season. Weeds are now evolving resistance to these herbicides in turn, with one population of Palmer amaranth confirmed surviving 2,4-D at eighteen times the labelled rate, and the next commercial varieties stack tolerance to as many as five herbicide groups. The treadmill loops back on the earlier sections: resistant weeds drive a return to tillage, which accelerates the erosion and organic-matter loss of Sections 2 and 3.

The gene-flow concern is also real and documented. The escape of an engineered trait into wild or feral plant populations has been recorded in the peer-reviewed literature: glyphosate-resistant creeping bentgrass expressing an introduced gene was found established outside its cultivation area in central Oregon, the first such documented escape into wild populations in the United States (Reichman et al., 2006), and feral populations of genetically engineered canola have been found to be large, widespread, and persistent across the northern United States, some carrying combinations of traits never planted together commercially, which implies recombination in the wild (Schafer et al., 2011). A related and frequently misstated grievance concerns organic and non-engineered growers. The claim that such growers have been sued merely for inadvertent cross-pollination does not survive examination, and this paper does not repeat it; the documented and legitimate grievance is that contamination of a crop by a neighbour's engineered pollen can threaten its organic certification and its access to markets that require freedom from engineered material, placing the burden of a problem on the party that did not create it.

Gene flow of this kind has a feature that distinguishes it from the other harms in this account and warrants a different standard of judgment. A released, self-propagating genetic trait cannot be recalled: unlike a chemical that degrades or a practice that can be discontinued, an escaped transgene that establishes in wild or weedy populations is for practical purposes irreversible, and its long-term ecological consequences are difficult to bound in advance and may not become apparent until the trait is widely established. The absence of demonstrated harm to date is therefore not, by itself, evidence of safety; it may equally reflect that the relevant effects are slow, diffuse, or not yet investigated, since absence of evidence is not evidence of absence. This is precisely the circumstance for which the precautionary principle was articulated: where an action carries a risk of serious or irreversible harm and the science remains incomplete, the burden of demonstrating safety rests most defensibly on the party introducing the novelty rather than on the public exposed to it. That principle is separate from, and does not contradict, the dietary-safety consensus stated at the opening of this section; it concerns the ecological release of a self-replicating trait into shared environments, not the eating of the food, and on the former question the conservative course places the onus of proof where the irreversibility lies.

11. The systemic picture: climate and genetic uniformity

Two further pressures operate at the level of the whole system. The first is climate, and it stands in a reinforcing relationship to the rest of this account rather than beside it. Agrifood systems account for about one third of total anthropogenic greenhouse-gas emissions, and within the farm gate crop and livestock production contribute more than half of human methane emissions and about three quarters of human nitrous-oxide emissions (FAO, via Tubiello et al.; Crippa et al., 2021). Nitrous oxide is roughly 270 times as potent as carbon dioxide per unit mass and is driven principally by nitrogen fertilizer and manure, which ties the nutrient overuse of Section 5 directly to warming, while the soil-carbon losses of Section 3 make the land a net source rather than a sink under prevailing management. The relationship runs in both directions, which is what makes it systemic rather than parallel: the warming and the altered rainfall that follow stress the soil biology and the cropping systems described here, and the share of emitted carbon dioxide that the oceans absorb acidifies them, degrading the marine ecosystems at the far end of the nutrient runoff of Section 5. Climate is in that sense not a separate cost but a limb of the same degrading loop, an output of the model that returns as a pressure upon it.

The second pressure is genetic uniformity. The high-yielding model rests on a narrow set of elite cultivars grown over vast areas, and the erosion of crop genetic diversity that accompanies it leaves the food supply more exposed to a pest or pathogen able to exploit the common genotype, the historical precedents for which, from the Irish potato failure to the 1970 southern corn leaf blight in the United States, are well known. Neither pressure is independent of the soil story; they are the same model's costs accounted in other currencies.

12. The trajectory and the structural limit

None of the trends above sets a date, and this paper does not pretend to. The claim is structural rather than calendrical, and it is the stronger for being so. A system that runs by mining capital can continue for as long as the capital lasts and the inputs keep coming, and the year of failure is neither predictable nor necessary to the argument. What can be stated with confidence is the shape of the trajectory and the limit toward which it runs, and that limit has several faces.

There is the face of diminishing returns, in which each additional unit of input buys less additional yield on a degrading base, so that costs rise while output stalls. There is the face of fragility, the characteristic exposure of a vast, uniform, high-input monoculture to a pest, a pathogen, a drought, a heat wave, or a disruption in the supply of the inputs it cannot do without; a system optimized for output under stable conditions is brittle when conditions move, and conditions are moving. There is the face of arithmetic, the plain impossibility of drawing down a finite stock without end: soil eroding faster than it forms, organic matter spent faster than it is returned, biological capacity dismantled rather than maintained, a finite and concentrated phosphorus supply, and an aquifer pumped beyond recharge describe a balance sheet that cannot be run indefinitely whatever the price of inputs. And there is the face the planetary-boundaries assessment makes explicit, that several of these pressures have already crossed the limits judged safe at the global scale, that agriculture is the leading driver of several of them, and that crossing multiple boundaries at once produces cascading and systemic risk rather than independent problems (Richardson et al., 2023). The apocalypse of the title is not a single catastrophe on a fixed day. It is the ordinary, compounding consequence of treating a living, self-renewing system as a dead medium to be supplied from outside, carried to its end. A trajectory of that kind does not require a prophecy. It requires only that it not be changed.

13. Not a counsel of despair: the turn to restoration

It would be a misreading of everything above to conclude that agriculture is doomed, or that the answer is less food, or a retreat to a pre-industrial past that fed far fewer people far less reliably. That is not the argument. The argument is that one particular model, the model that severs the soil's economy and substitutes purchased inputs for it, is the thing that mines capital and runs toward the limit, and that a model is a choice and not a fate. The diagnosis is precise precisely so that the response can be.

And the response exists. The companion record shows that fertile, carbon-rich, durable soil has been built by hand on the very poorest ground, that it has lasted for a thousand years and more, and that it is still being made today; the companion mechanistic case sets out how such a soil can carry itself, becoming a self-stabilizing state rather than a perishable input that must be purchased again each season. The way out, therefore, is not to dose a dead medium more cleverly but to rebuild the living economy that the model dismantled, restoring the two severed channels so that the soil resumes acquiring and recycling its own fertility and inputs become a means of restoration rather than a permanent dependency. The stakes for getting this right are real without requiring inflation: global soil degradation is well documented and consequential, much of the lost soil carbon is in principle recoverable, and a credible, mechanistically grounded route to building durable, self-stabilizing fertile soil would matter at planetary scale. Restoration is not, however, a claim to abolish the nutrient requirement itself: feeding a population of the present size demands large flows of reactive nitrogen under any model, and rebuilding the soil's living economy reduces the leakage, the losses, and the dependence rather than removing the underlying demand. The honest claim is that it closes the leaks and shrinks the bill, not that it renders the inputs unnecessary. That route is the subject of the companion solution paper, to which this diagnosis hands off.

14. Conclusion

The bill that industrial agriculture is now being handed is the cost of a method, not the cost of feeding the world. The method fed the world by bypassing the soil rather than building it, and a bypass run long enough becomes a drawdown, and a drawdown has an end. The soil is eroding faster than it forms; its organic carbon and its biology are being spent rather than replenished; its irrigation water is being pumped beyond recharge and salted in the process; the soluble nutrients that prop up its yields leak into the water and the air, feed the coastal dead zones, and exceed the limits the biosphere can absorb; the chemistry that protects its crops takes a documented toll on pollinators and a contested but unresolved one on human health, on an evidentiary record shown to have been shaped by the interested party; some of its residues do not leave; its dominant transgenic model has locked the field into escalating chemical use and bred the weeds that defeat it; and the curve of diminishing returns and rising fragility bends toward a limit that no amount of additional input can move. The apocalypse is not inevitable, because it is the endpoint of a trajectory, and trajectories can be left. Leaving this one means rebuilding what was severed, and how that is done, and how it can be done at scale, is where this account turns from the problem to the answer.

---

A note on claim calibration

This paper makes claims of differing evidential strength, and treats them differently. The table records the principal calibration decisions.

Claim	Status as treated here
Erosion exceeds soil formation on tilled cropland; global erosion is large and rising with cropland expansion	Well established (FAO and ITPS, 2015; Montgomery, 2007; Borrelli et al., 2017). The dated 75 Pg / 400 billion USD figure is deliberately set aside in favour of the modern peer-reviewed estimate.
Cultivation has caused large losses of soil organic carbon; about one third of soils are degraded	Well established (Sanderman et al., 2017; FAO Global Soil Partnership, 2017). The recoverability of much of that carbon is also stated, in fairness.
Aquifer overdraft and irrigation salinization are drawing down agricultural water	Well established (USGS syntheses; FAO, 2021), with efficiency gains noted as a partial mitigation.
Nutrient runoff drives eutrophication and coastal dead zones; N and P planetary boundaries are transgressed with agriculture the main driver	Well established (NOAA, 2023; Diaz and Rosenberg, 2008; Richardson et al., 2023; Steffen et al., 2015).
Phosphorus is finite and geographically concentrated	Well established. Specific "peak phosphorus" dates are contested and are not relied upon.
Neonicotinoids harm pollinators and aquatic insects	Strong and growing evidence, causal in controlled aquatic experiments; broad landscape insect declines are multi-causal and attribution from correlational data is treated as uncertain.
Glyphosate causes cancer	Presented as a genuine, unresolved dispute with both sides stated at equal weight: IARC's 2A hazard classification and the positive case-control meta-analyses (Zhang et al., 2019; De Roos et al., 2003) against the null prospective cohort (Andreotti et al., 2018) and the negative risk assessments of EPA, EFSA, ECHA, and JMPR, with the hazard-versus-risk distinction and conflict-of-interest criticism noted on both sides. No verdict is asserted. The documented influence of the manufacturer on the safety record is treated as the substantive, citable point.
Acute occupational pesticide poisoning is a serious global problem	Well established (WHO). A widely cited recent global figure is avoided because it has been retracted.
The herbicide-tolerant model increased herbicide use and selected resistant weeds	Well established (Benbrook, 2012, 2016; weed-science literature).
Genetically engineered foods are unsafe to eat	Not claimed. The scientific consensus that approved GE foods are as safe as conventional is stated explicitly (National Academies, 2016); the paper's critique is agronomic and ecological.
Transgenes escape into feral and wild populations	Documented (Reichman et al., 2006; Schafer et al., 2011). The "sued for cross-pollination" claim is rejected as unsupported; the certification-and-market grievance is retained. The precautionary argument is applied to this risk alone: it concerns the irreversible ecological release of a self-replicating trait, is held explicitly separate from the dietary-safety consensus, and asserts where the burden of proof lies, not demonstrated harm.
Persistent residues (DDT, PFAS) enter and remain in farmland	Well established for persistence and bioaccumulation; PFAS-affected-acreage figures are flagged as industry-derived estimates.
The model runs toward a structural limit	A structural claim about trajectory, explicitly not a dated prediction.

---

References

Citations are tiered. Entries under "Verified during preparation" were checked against current sources while this paper was written. Entries under "Established literature" are standard references whose fine-grained bibliographic details (volume, page, and in two flagged cases the lead author and year) should be confirmed against the originals before publication. The body deliberately avoids inflated and frequently repeated figures, such as fixed predictions of a small number of remaining harvests, that do not survive scrutiny, and it does not rely on the retracted global estimate of acute pesticide poisonings.

Verified during preparation

• Borrelli, P., Robinson, D. A., Fleischer, L. R., Lugato, E., Ballabio, C., Alewell, C., et al. (2017). An assessment of the global impact of 21st century land use change on soil erosion. Nature Communications, 8, 2013.
• Sanderman, J., Hengl, T., & Fiske, G. J. (2017). Soil carbon debt of 12,000 years of human land use. Proceedings of the National Academy of Sciences, 114(36), 9575–9580.
• Richardson, K., Steffen, W., Lucht, W., Bendtsen, J., Cornell, S. E., Donges, J. F., et al. (2023). Earth beyond six of nine planetary boundaries. Science Advances, 9(37), eadh2458.
• FAO & ITPS. (2015). Status of the World's Soil Resources (SWSR) — Main Report. Food and Agriculture Organization of the United Nations and Intergovernmental Technical Panel on Soils.
• FAO Global Soil Partnership. (2017). Global Symposium on Soil Organic Carbon: key messages and outcome documents. Food and Agriculture Organization of the United Nations.
• FAO. (2021). The State of the World's Land and Water Resources for Food and Agriculture: Systems at Breaking Point. Food and Agriculture Organization of the United Nations.
• FAO. (2022). Soils statement, Global Forum for Food and Agriculture. Food and Agriculture Organization of the United Nations.
• FAO. (2024). The State of the World's Land and Water Resources for Food and Agriculture (SOLAW 2025): conservative estimates of cropland degraded by salinization (82 Mha rainfed; 24 Mha irrigated). Food and Agriculture Organization of the United Nations.
• FAO. SDG Indicator 15.3.1: Proportion of land that is degraded over total land area (SDG Indicators Data Portal; loss of ≥100 Mha of productive land per year, 2015–2019; ~1.3 billion people on degrading land). Food and Agriculture Organization of the United Nations.
• International Agency for Research on Cancer. (2015). Glyphosate, in IARC Monographs Volume 112: Some Organophosphate Insecticides and Herbicides. World Health Organization.
• Zhang, L., Rana, I., Shaffer, R. M., Taioli, E., & Sheppard, L. (2019). Exposure to glyphosate-based herbicides and risk for non-Hodgkin lymphoma: a meta-analysis and supporting evidence. Mutation Research / Reviews in Mutation Research, 781, 186–206.
• Andreotti, G., Koutros, S., Hofmann, J. N., Sandler, D. P., Lubin, J. H., Lynch, C. F., Lerro, C. C., De Roos, A. J., Parks, C. G., Alavanja, M. C., Silverman, D. T., & Beane Freeman, L. E. (2018). Glyphosate use and cancer incidence in the Agricultural Health Study. Journal of the National Cancer Institute, 110(5), 509–516.
• Godfray, H. C. J., Blacquière, T., Field, L. M., Hails, R. S., Petrokofsky, G., Potts, S. G., et al. (2014). A restatement of the natural science evidence base concerning neonicotinoid insecticides and insect pollinators. Proceedings of the Royal Society B, 281(1786), 20140558.
• Reichman, J. R., Watrud, L. S., Lee, E. H., Burdick, C. A., Bollman, M. A., Storm, M. J., et al. (2006). Establishment of transgenic herbicide-resistant creeping bentgrass (Agrostis stolonifera L.) in nonagronomic habitats. Molecular Ecology, 15(13), 4243–4255.
• Schafer, M. G., Ross, A. A., Londo, J. P., Burdick, C. A., Lee, E. H., Travers, S. E., et al. (2011). The establishment of genetically engineered canola populations in the U.S. PLoS ONE, 6(10), e25736.
• Crippa, M., Solazzo, E., Guizzardi, D., Monforti-Ferrario, F., Tubiello, F. N., & Leip, A. (2021). Food systems are responsible for a third of global anthropogenic greenhouse gas emissions. Nature Food, 2, 198–209.
• National Oceanic and Atmospheric Administration. (2023). Gulf of Mexico "dead zone" measurement and forecast. NOAA / USGS Mississippi River and Gulf of Mexico Hypoxia Task Force.
• Benbrook, C. M. (2016). Trends in glyphosate herbicide use in the United States and globally. Environmental Sciences Europe, 28, 3.
• Benbrook, C. M. (2012). Impacts of genetically engineered crops on pesticide use in the U.S. — the first sixteen years. Environmental Sciences Europe, 24, 24.
• Barmentlo, S. H., Schrama, M., de Snoo, G. R., van Bodegom, P. M., van Nieuwenhuijzen, A., & Vijver, M. G. (2021). Experimental evidence for neonicotinoid driven decline in aquatic emerging insects. Proceedings of the National Academy of Sciences, 118(44), e2105692118. https://doi.org/10.1073/pnas.2105692118
• Environmental Science and Technology. (2018). Review: Environmental risks and challenges associated with neonicotinoid insecticides. Environmental Science and Technology, 52(6). [Confirm lead author and pagination.]

Established literature (confirm fine-grained details before submission)

• Erisman, J. W., Sutton, M. A., Galloway, J., Klimont, Z., & Winiwarter, W. (2008). How a century of ammonia synthesis changed the world. Nature Geoscience, 1, 636–639.
• Montgomery, D. R. (2007). Dirt: The Erosion of Civilizations. University of California Press.
• Diaz, R. J., & Rosenberg, R. (2008). Spreading dead zones and consequences for marine ecosystems. Science, 321(5891), 926–929.
• Steffen, W., Richardson, K., Rockström, J., Cornell, S. E., Fetzer, I., Bennett, E. M., et al. (2015). Planetary boundaries: guiding human development on a changing planet. Science, 347(6223), 1259855.
• Rockström, J., Steffen, W., Noone, K., Persson, Å., Chapin, F. S., Lambin, E. F., et al. (2009). A safe operating space for humanity. Nature, 461, 472–475.
• Gerten, D., Heck, V., Jägermeyr, J., Bodirsky, B. L., Fetzer, I., Jalava, M., et al. (2020). Feeding ten billion people is possible within four terrestrial planetary boundaries. Nature Sustainability, 3, 200–208.
• Hallmann, C. A., Sorg, M., Jongejans, E., Siepel, H., Hofland, N., Schwan, H., et al. (2017). More than 75 percent decline over 27 years in total flying insect biomass in protected areas. PLoS ONE, 12(10), e0185809.
• De Roos, A. J., Zahm, S. H., Cantor, K. P., Weisenburger, D. D., Holmes, F. F., Burmeister, L. F., & Blair, A. (2003). Integrative assessment of multiple pesticides as risk factors for non-Hodgkin's lymphoma among men. Occupational and Environmental Medicine, 60(9), e11.
• National Academies of Sciences, Engineering, and Medicine. (2016). Genetically Engineered Crops: Experiences and Prospects. The National Academies Press.
• Cordell, D., Drangert, J.-O., & White, S. (2009). The story of phosphorus: global food security and food for thought. Global Environmental Change, 19(2), 292–305.
• Tilman, D., Cassman, K. G., Matson, P. A., Naylor, R., & Polasky, S. (2002). Agricultural sustainability and intensive production practices. Nature, 418, 671–677.
• United States Geological Survey. Mineral Commodity Summaries: Phosphate Rock (most recent edition) and High Plains (Ogallala) Aquifer water-level and storage assessments. [Confirm editions.]
• Maine Department of Environmental Protection. (2019). Findings on PFAS in sewage sludge and biosolids; and Environmental Working Group. (2025). Sewage-sludge PFAS contamination of farmland: analysis. [Public-record and analysis sources; confirm details.]
• Court records and contemporaneous reporting on the glyphosate litigation, including the disclosed internal company documents ("Monsanto Papers"), the 2018–2020 trial outcomes and settlement, and the subsequent retraction of a 2000 safety review by Regulatory Toxicology and Pharmacology. [Matters of public record; specific citations to be added before publication.]
• Kesic, R., Elliott, J. E., Drouillard, K. G., et al. (2021). Continuing persistence and biomagnification of DDT and metabolites in northern temperate fruit orchard avian food chains. Environmental Toxicology and Chemistry, 40(12), 3379–3391. https://doi.org/10.1002/etc.5220