The Crop Deaths Argument, Counted Honestly: Which Plate Kills More Animals?

Scienceanimal agriculture crop deaths field animals least harm

Say the word “vegan” anywhere on the internet and someone will arrive with a combine harvester. Crops kill animals too, the charge goes: mice shredded in the header, rabbits under the tires, voles poisoned in the granary. So your hands aren’t clean either. In its upgraded form, the argument goes further. One large grass-fed steer feeds a person for months and dies once, while a field of wheat kills small animals by the dozen; therefore the beef eater, not the vegan, holds the true least-harm diet.

Unlike most gotchas, this one has an academic pedigree. It was published in a peer-reviewed journal by a serious scientist, and it deserves what every claim in this series gets: a fair hearing, followed by arithmetic. What makes this one unusual is that the fair hearing was already conducted, in the same journal and the same year, and the argument lost. Twenty-odd years later it still circulates as if it had won.

Where the crop deaths argument comes from

Who started the claim that vegans kill more animals?

In 2003, Steven Davis, an animal scientist at Oregon State University, published “The Least Harm Principle May Require That Humans Consume a Diet Containing Large Herbivores, Not a Vegan Diet” in the Journal of Agricultural and Environmental Ethics [1]. His model was straightforward. Drawing on the sparse literature available, he estimated that crop production kills about 15 wild animals per hectare per year, through tillage, harvest, and the predation that follows once cover is stripped away, while ruminant pasture kills about half that, 7.5 per hectare. Run all 120 million hectares of US cropland as crops for a vegan food supply and you get roughly 1.8 billion wild animal deaths; run half of it as pasture for grazing ruminants and the total drops to about 1.35 billion. His conclusion: a diet built around large grass-fed herbivores kills fewer animals than a vegan one.

Credit where it is due. Davis engaged the least-harm principle seriously instead of dismissing it, published where animal ethicists could answer him, and put his numbers on the table. The numbers are the interesting part, because they are the part that flipped.

The correction that flipped it, published the same year

Do vegans really kill more animals than meat eaters?

Gaverick Matheny’s rebuttal, “Least Harm: A Defense of Vegetarianism from Steven Davis’s Omnivorous Proposal,” ran in the same journal that same year [2]. It identified the load-bearing error in a single sentence. Davis compared deaths per hectare when the question is deaths per consumer, and a hectare of crops feeds far more people than a hectare of pasture.

The gap is not subtle. A hectare of cropland can produce on the order of 1,000 kg of plant protein; producing the same protein from grass-fed beef takes roughly ten hectares. So even granting Davis his own field-death rates, 15 per crop hectare and 7.5 per pasture hectare, the per-person accounting inverts his conclusion. Measured against the 20 kg of protein an adult needs in a year, Matheny found a vegan diet kills about 0.3 wild animals, a lacto-vegetarian diet about 0.39, and Davis’s proposed omnivorous diet about 1.5: a vegan diet works out to roughly 0.3 wild animal deaths per person per year, against 1.5 for Davis’s proposed diet. Five times fewer [2]. Matheny noted two further problems, namely that Davis counted deaths but ignored the quality of the lives lost, and ignored the animals prevented from existing; but the denominator error alone settles the arithmetic. Andy Lamey’s follow-up analysis added that Davis’s field-death rates were themselves likely overestimates, in part because they counted animals taken by predators after harvest as though the harvester had killed them [3].

The argument’s best case, the single large pasture-raised ruminant that its defenders always reach for, loses five to one using its own numbers. That correction has now been in print for more than two decades. The Australian remake of the argument, published by Mike Archer in 2011, in which grain kills more animals than rangeland beef, built partly on poisoning campaigns during mouse-plague years, recycles the same denominator problem with shakier data, and met the same fate in the literature [4].

How many animals actually die in crop harvest?

Nobody has a reliable count, and that includes both sides

Here is the part both sides should sit with, because intellectual honesty is the price of admission to this series. In 2018, Bob Fischer and Andy Lamey published “Field Deaths in Plant Agriculture” in the same journal [4], and their central finding is uncomfortable for everyone’s infographics: nobody has a defensible count.

The canonical figures trace back to a handful of small studies, extrapolated far beyond what they measured. The most famous of them, a study of wood mice in English grain fields, actually found that the harvest itself killed very few mice; the population crash afterward came mostly from predation once cover was stripped away, and from mice simply dispersing to the field margins [5]. Animals flee combines more often than they die under them. Mortality varies enormously by crop, region, season, and method, and deaths from predation-after-harvest raise a genuine question of attribution, since the owl was going to eat some mouse somewhere. Fischer and Lamey’s conclusion was blunt: field-death totals could be substantially lower than the numbers circulating in these debates, and confident per-hectare figures, whether deployed by carnivore memes or by vegan counter-memes, outrun the evidence [4].

So let us not pretend to decimal places. The honest position is that the field-death numbers are soft. The conclusion, however, is not, because it never depended on the counts. It depends on the multiplier.

The multiplier: whoever eats animals orders the field twice

Why eating meat means more crops are harvested, not fewer

Livestock do not photosynthesize. Whatever dies in the production of crops, a food system that routes crops through animals multiplies it, because trophic conversion is a losing game at every step. The FAO’s own feed analysis found livestock consume one third of global cereal production and require about 3 kg of human-edible feed per kg of meat [6]. Feed crops absorb 36% of the world’s crop calories and 53% of its plant protein, and only about 12% of those feed calories ever return to a human plate [7]. US beef converts 3% of its feed calories into meat [8]. By some estimates, roughly two-thirds of the calories grown on US cropland are fed to animals rather than people [7]. However many field animals die per harvested acre, the animal-product plate commissions more harvested acres per person than the plant-based one, and then adds the slaughterhouse on top.

Which foods cause the most animal deaths per calorie?

Put the two categories together, field deaths plus slaughter deaths per calorie delivered, and the ledger orders itself. The most-cited tabulation, built on USDA production data within the Davis and Matheny framework [9], puts chicken around 251 deaths per million calories, eggs around 92, beef around 29, pork around 18, and milk around 5, against roughly 2 for plant foods. Treat the decimals as estimates; the ordering is robust, because it follows from body size and feed conversion, not from any particular mouse study.

Notice the ranking, because it embarrasses everyone’s intuitions. Chicken is the deadliest food on the table, over a hundred times the plant baseline, precisely because the bodies are small and each meal costs a life-fraction of many animals plus the feed acres that raised them. Beef, the crop-deaths argument’s champion, performs best among the meats for the same reason in reverse, one large body feeding many meals, and it still runs roughly fifteen times the plant baseline per calorie, and five times per capita in Matheny’s correction. The argument’s best case is the plant eater’s worst case, and it isn’t close.

How many animals does one person’s diet kill in a year?

Per-calorie tables are useful for comparing foods, but most people want the personal number. Here it is, and it is the most concrete way to state the whole argument.

Counting only land animals, government slaughter data divided by population puts the average American’s diet at roughly 25 to 30 animals a year, and the composition surprises people: nearly all of them are chickens, because a chicken yields so few meals. Pigs, cattle, and turkeys together account for barely one animal a year. A vegan’s figure for this column is zero. Confidence: solid, since it comes straight from USDA and FAO slaughter counts.

Now add the animals almost every version of this debate forgets. Once fish and shellfish are converted from tonnage into individual animals, the most detailed US per-capita compilation puts the average American’s annual total at roughly 400 to 580 animals, the large majority of them fish and shrimp, with wild fish caught for fishmeal and killed as bycatch making up much of the remainder. A separate global analysis reaches the same conclusion by a different route. A vegan’s slaughter figure is still zero. Confidence: medium. These are estimates built on government tonnage data using published mean-weight methods, not headcounts, and the range is wide for good reason.

So the honest headline depends entirely on whether fish are counted, and the difference is more than tenfold. State it both ways or not at all: about 25 to 30 land animals a year, or roughly 400 to 580 animals counting aquatic life, against essentially zero for someone eating plants. Note what that does to the arithmetic of the gotcha. The crop-deaths argument is a dispute about a handful of field mice. It is being deployed against a difference of several hundred animals a year.

And that table is the conservative comparison, because it counts only two of the nine columns below: field deaths and direct slaughter. It leaves out predator control, disease culls, on-farm mortality, pesticide wildlife kills, bycatch, dead zones, and habitat loss, every one of which falls on the animal side. The honest per-calorie picture is worse for meat than even these numbers show; the table simply happens to be the one slice that can be drawn apples-to-apples.

The full ledger the gotcha never adds up

Field deaths and slaughter are only two columns. The real account has at least nine, and the crop-deaths argument goes silent on most of them. What follows is that ledger, built wherever possible on government data and peer-reviewed journals rather than advocacy tallies. Where a number is soft, it says so, because a ledger that flags its own weak entries is the one that survives a hostile reader.

A word on what this can and cannot be. Nobody has published a single integrated body count across all of these columns, and the units do not cleanly add; a slaughtered chicken, a bycaught dolphin, and a dead zone measured in square miles are not the same kind of thing. So this is not one headline number. It is a survey of every column, each with its best available figure, showing that on all of them the animal-farming plate carries more.

1. Direct slaughter (solid)

More than 80 billion land animals are killed for food every year, about ten for every human alive [19]. To that, add farmed fish: an estimated 78 to 171 billion finfish killed annually, plus farmed shrimp and prawns in the hundreds of billions [13]. The layer industry culls 6 to 7 billion male chicks on their first day of life as standard practice, a soft but widely accepted industry-derived figure.

2. Wild capture (solid on the totals, softer per species)

Between 1.1 and 2.2 trillion wild fish are pulled from the ocean in an average year [13], roughly half of them ground into fishmeal and oil to feed other farmed animals, most of it to farmed fish but a substantial share to weaning pigs and day-old chicks [12]; so a chicken sandwich and a salmon fillet both carry ocean deaths upstream. Sharks are killed at a rate that rose from about 76 to 80 million a year between 2012 and 2019, roughly a quarter of them threatened species [20]. And a landmark analysis put global marine bycatch, the dolphins, turtles, seabirds, and unwanted fish discarded dead, at about 40% of the reported catch under its definition of unused or unmanaged catch [11]; the figure is definition-dependent and higher than the FAO’s narrower discard estimate, so treat it as an upper-range marker rather than a precise share.

3. Predator and “pest” control (solid, government data)

The USDA’s Wildlife Services program kills wild animals largely on behalf of livestock producers. In 2023 it killed 375,045 native animals, including 68,562 coyotes, 24,603 beavers, 430 black bears, 469 bobcats, 305 gray wolves, and 235 mountain lions [10]; counting invasive species, the program’s total runs to roughly 1.4 million animals a year. An earlier accounting that folds in native birds killed under blanket depredation orders reaches about 2.6 million total and 1.5 million native for 2018. It is the same program cast in a wider net, and the scope difference is why the two years look so far apart. A 2024 investigation found roughly 11,000 of these animals were killed at locations with no recorded livestock damage at all [21]. This entire column is charged to the animal-farming plate and counted by almost no one. The beaver number stings most: those are the animals that rebuild the wetlands and water tables plant agriculture would leave alone.

4. Disease-control depopulation (solid, government data)

When avian flu or swine fever reaches a facility, the response is mass killing. The current US H5N1 outbreak has killed over 185 million birds since February 2022, across all fifty states [22]; the 2014 to 2015 outbreak took another 50.5 million. In China, African swine fever killed or culled an estimated 143 million pigs in 2018 and 2019, cutting the national herd by roughly 40% [23]. That total blends disease deaths, government culling, and panic liquidation, so it is not all regulators killing healthy animals; but every death was a product of raising pigs by the hundreds of millions in the first place. These culls exist only because animals are farmed; the column does not exist on the plant side.

5. On-farm mortality before slaughter (soft, and invisible in every official count)

A large share of farmed animals die before they ever reach the kill floor, and those deaths appear in no slaughter statistic. In the US alone, over 500 million broiler chickens die on the farm each year, from disease, from heart failure driven by the breed’s engineered growth rate, and from culling, before the survivors are trucked to slaughter [24]. That total comes from an advocacy compilation of USDA production data rather than a peer-reviewed source, so treat it as indicative; peer-reviewed cohort work puts first-week mortality near 1.4% and later mortality around 3%. Piglet pre-weaning death, transport losses, and dead-on-arrival birds add still more. The number is soft because the industry does not publish a clean total; but it is real, it is enormous, and it is counted nowhere.

6. Pesticide wildlife kills (soft, and mostly charged to feed crops)

Agricultural pesticides in the US were estimated decades ago to kill on the order of 67 million birds a year [25], a figure later authors consider conservative, and one that predates the neonicotinoid seed treatments now standard on corn and soy. Because roughly two-thirds of US crop calories are grown to feed livestock, most of that acreage, and most of the chemical load on it, belongs to animal agriculture. This is the connection the companion pesticide post documents in federal law: the legal glyphosate ceiling on animal feed runs 1,500 to 4,000 times the ceiling on the produce sold to people [26]. The bird-kill figure is old and imprecise, so flag it as such; the direction is not in doubt.

7. Aquatic dead zones and manure spills (habitat destruction, not a clean body count)

Every summer, fertilizer and manure washing out of the Midwest feed-crop belt feed an oxygen-starved dead zone in the Gulf of Mexico, 4,402 square miles in 2025, and 6,705 in 2024 [27]. NOAA measures the area of habitat rendered uninhabitable, not a body count, so this belongs in the ledger as documented ecosystem destruction rather than a number of corpses. The acute kills are more countable: in 1995 a single ruptured hog lagoon sent some 25 million gallons of waste into North Carolina’s New River, and hurricanes Floyd and Florence later flooded lagoons across the state’s east into its rivers [14]. Present the spills as documented events; treat any specific fish-kill tally as a reported estimate unless a state agency published it.

8. Habitat and biodiversity loss (solid on drivers, directional on the toll)

This is the largest column and the hardest to reduce to a number. Cattle pasture drives about 41% of tropical deforestation, and agriculture as a whole drives 90 to 99% of it [15]. Farming is the single biggest documented threat to species on the IUCN Red List. And the cumulative result is a biosphere in which farmed animals make up roughly 60% of all mammal biomass, while every wild mammal on Earth combined, every elephant, whale, deer, and mouse, is down to about 4% [16]. You cannot put a clean death count on habitat conversion; but every displaced, starved, and never-born wild animal behind those figures sits on the animal-farming side of the ledger.

9. Human deaths (solid, and rarely counted as part of the toll at all)

Food-related air pollution, much of it ammonia off manure and feed fertilizer, is estimated to cause about 12,700 US deaths a year attributable to animal-based foods, against 3,200 for plant-based [28]; by that accounting a gram of beef protein carries roughly the air-pollution mortality of 200 grams of protein from beans or grains. Antibiotic resistance was linked to 1.27 million deaths directly and 4.95 million associated worldwide in 2019 [29], and livestock consume about 73% of the world’s antibiotics [30]; though use-share is not death-share, and the post should not multiply the two. And in the pandemic’s first year, at least 269 workers died at the five largest US meatpackers [31]. Diet-attributable cancer and heart-disease deaths belong to the health post, and are contested enough to keep separate.

Nine columns. Plant agriculture registers meaningfully in exactly one of them, field deaths, and posts the lowest number there too. Every other column is animal agriculture’s alone.

Intention matters, just not the way they think

Is an accidental death the same as a deliberate one?

Suppose the numbers were even. They are not, but suppose. There would still be a difference between deaths that occur as a mitigable side effect of harvesting food and deaths that are the product. Field mortality is what Fischer and Lamey call a historically contingent problem: it falls with better harvest technology, timing, and practice, and above all it falls with acreage, which is exactly what a plant-based food system delivers: roughly 75% less agricultural land to feed everyone [17]. The kill floor cannot be mitigated out of animal farming; it is the point of animal farming. Veganism never claimed to be a zero-harm diet. The standard has always been avoiding exploitation and killing as far as possible and practicable [18], and every practicable improvement points the same direction: fewer acres, fewer blades, no slaughter line.

Take the argument seriously and it converts you

What would you eat if you actually minimized animal deaths?

Here is the test that separates a sincere argument from a conversation-ender. Anyone who genuinely ranked their diet by animal deaths per calorie would act on the nine-column ledger above. They would quit chicken and eggs first, since those top every count; treat beef as damage control at fifteen times the plant baseline; and land, finally, on grains, fruits, vegetables, and legumes, the bottom of every column, including the one the argument is named after. Followed all the way down, the crop-deaths argument is an argument for veganism with extra steps. The people who deploy it almost never eat as though they believe it, which tells you what it is actually for. It was never a way of counting deaths. It is a way of ending conversations about them.

The bottom line

The field-death numbers are soft; the structure is not. Whatever the true count per harvested acre, the plate with animal products on it commissions more harvested acres, then adds more than 80 billion slaughtered land animals a year, over a trillion wild-caught fish, the predator-control kills, the disease culls in the hundreds of millions, the half-billion broilers that die before slaughter, the bycatch and the fishmeal, the lagoons and the dead zones, a wild world compressed to 4% of mammal biomass, and thousands of human deaths from the air it fouls and the drugs it burns through. The plant-based plate carries the smallest share of the one column the gotcha points at, and essentially none of the other eight.

For the acre-by-acre production math, see Two Acres. For the feed statistic this argument usually travels with, see The 86% Meme, Debunked. For the chemical load those extra acres carry, and the farmworkers who absorb it, see The Pesticide Argument. For what the freed land becomes, see Rewilding the Plate.

References

  1. Davis, S. L. (2003). The least harm principle may require that humans consume a diet containing large herbivores, not a vegan diet. Journal of Agricultural and Environmental Ethics, 16(4), 387–394.
  2. Matheny, G. (2003). Least harm: a defense of vegetarianism from Steven Davis’s omnivorous proposal. Journal of Agricultural and Environmental Ethics, 16(5), 505–511.
  3. Lamey, A. (2007). Food fight! Davis versus Regan on the ethics of eating beef. Journal of Social Philosophy, 38(2), 331–348.
  4. Fischer, B., & Lamey, A. (2018). Field deaths in plant agriculture. Journal of Agricultural and Environmental Ethics, 31(4), 409–428.
  5. Tew, T. E., & Macdonald, D. W. (1993). The effects of harvest on arable wood mice. Biological Conservation, 65(3), 279–283.
  6. Mottet, A., de Haan, C., Falcucci, A., Tempio, G., Opio, C., & Gerber, P. (2017). Livestock: On our plates or eating at our table? A new analysis of the feed/food debate. Global Food Security, 14, 1–8.
  7. Cassidy, E. S., West, P. C., Gerber, J. S., & Foley, J. A. (2013). Redefining agricultural yields: from tonnes to people nourished per hectare. Environmental Research Letters, 8(3), 034015.
  8. Shepon, A., Eshel, G., Noor, E., & Milo, R. (2016). Energy and protein feed-to-food conversion efficiencies in the US and potential food security gains from dietary changes. Environmental Research Letters, 11(10), 105002.
  9. Animal Visuals (2009). Number of animals killed to produce one million calories in eight food categories. An advocacy-site tabulation built on USDA data within the Davis and Matheny framework; treat point estimates as approximate.
  10. Center for Biological Diversity summary of USDA APHIS Wildlife Services data for 2023; see also the annual APHIS Program Data Report tables.
  11. Davies, R. W. D., Cripps, S. J., Nickson, A., & Porter, G. (2009). Defining and estimating global marine fisheries bycatch. Marine Policy, 33(4), 661–672.
  12. Cashion, T., Le Manach, F., Zeller, D., & Pauly, D. (2017). Most fish destined for fishmeal production are food-grade fish. Fish and Fisheries, 18(5), 837–844.
  13. Mood, A., & Brooke, P. Estimates of global numbers of fish caught annually (fishcount.org.uk); peer-reviewed version: Estimating global numbers of fishes caught from the wild annually from 2000 to 2019, Animal Welfare, 33, e6.
  14. Burkholder, J., et al. Impacts of industrial animal production on rivers and estuaries, American Scientist (documenting the 1995 New River hog lagoon failure); and NOAA Gulf of Mexico hypoxia monitoring.
  15. Pendrill, F., et al. (2019). Agricultural and forestry trade drives large share of tropical deforestation emissions. Global Environmental Change, 56, 1–10, as compiled in Our World in Data, “Drivers of Deforestation”; and Pendrill, F., et al. (2022). Disentangling the numbers behind agriculture-driven tropical deforestation. Science, 377(6611). The 40% versus 41% variation across sources is rounding within the same dataset.
  16. Bar-On, Y. M., Phillips, R., & Milo, R. (2018). The biomass distribution on Earth. PNAS, 115(25), 6506–6511.
  17. Poore, J., & Nemecek, T. (2018). Reducing food’s environmental impacts through producers and consumers. Science, 360(6392), 987–992.
  18. The Vegan Society. Definition of veganism (established 1944).
  19. FAO. FAOSTAT: livestock primary, annual global slaughter statistics.
  20. Worm, B., et al. (2024). Global shark fishing mortality still rising despite widespread regulatory change. Science, 383(6679), 225–230.
  21. NPR Investigations (2024). How a federal program kills wildlife for private interests.
  22. USDA APHIS. Highly Pathogenic Avian Influenza detections in commercial and backyard flocks, February 2022 to present.
  23. USDA Economic Research Service. How China’s African swine fever outbreaks affected global pork markets; and FAO situation updates on African swine fever in Asia and the Pacific.
  24. Animal Equality (2024). Mortality in broiler chickens, an advocacy compilation of USDA NASS production data; peer-reviewed mortality rates in The sustainability gap of US broiler chicken production, Royal Society Open Science.
  25. Pimentel, D., et al. (1992). Environmental and economic costs of pesticide use. BioScience, 42(10), 750–760.
  26. US EPA. Glyphosate tolerances, 40 CFR § 180.364; see the companion post, The Pesticide Argument the Carnivore Movement Can’t Answer.
  27. NOAA. Gulf of Mexico hypoxic zone monitoring: 4,402 square miles in 2025; 6,705 square miles in 2024.
  28. Domingo, N. G. G., et al. (2021). Air-quality-related health damages of food. PNAS, 118(20), e2013637118.
  29. Murray, C. J. L., et al. (2022). Global burden of bacterial antimicrobial resistance in 2019. The Lancet, 399(10325), 629–655.
  30. Van Boeckel, T. P., et al. (2017). Reducing antimicrobial use in food animals. Science, 357(6358), 1350–1352. Livestock account for about 73% of global antimicrobial use; use-share is not death-share.
  31. US House Select Subcommittee on the Coronavirus Crisis (2021). Data on coronavirus infections and deaths among meatpacking workers.
  32. Sethu, H. How many animals does a vegetarian save? Counting Animals. A detailed independent compilation of USDA and FAO production data into per-capita death counts, including fish and shellfish; an advocacy-adjacent analysis rather than a peer-reviewed one, so treat the totals as well-documented estimates.
  33. Animal Charity Evaluators. Effects of Diet Choices (2021). A global estimate of animals spared per plant-based year, built on FAO data across farmed land vertebrates, farmed fishes, and wild-caught fishes.
  34. Archer, M. (2011). Ordering the vegetarian meal? There’s more animal blood on your hands. The Conversation. The Australian restatement of the least-harm argument, addressed in Fischer and Lamey (2018).

Last updated: October 6, 2026 at 8:49 AM EDT

Animal Deaths Per Million Calories Chicken Tops the List Not Beef 1 768x429

Leave a Review

Footer preview