How Does Vegetarianism Impact Wild-Animal Suffering?

Summary

I think it's pretty unclear whether promoting vegetarianism reduces or increases total animal suffering, both when considering short-run effects on wild animals on Earth and when considering long-run effects on society's values. Compared with veg outreach, other approaches to reducing animal suffering on factory farms, such as humane slaughter, seem more clearly net positive.

Especially for smaller animals like chickens, it's plausible that the direct suffering caused to the farmed animals themselves is of a comparable order of magnitude as the size of the "error bars" of uncertainty I have about how farming chickens affects wild-invertebrate suffering, which tentatively suggests continuing to oppose chicken consumption, pending further research.

If someone insists on eating meat, I would recommend eating rainforest-raised or grass-fed beef. Rainforest-grown beef plausibly reduces net animal populations because rainforests have such high productivity. Grass-fed beef plausibly also reduces net animal populations, because cattle can eat lots of grass that would otherwise feed smaller animals, and given that less of the feed for these cattle is farmed grain than in the case of non-grass-fed cattle, the uncertain net impacts of crop cultivation loom relatively less large over the calculation.

Following are my current best guesses about the net impacts of different animal products from a negative-utilitarian standpoint. The products are ranked from best to worst.

Introduction

Utilitarian vegetarians avoid meat because they realize that animals on factory farms experience more suffering than happiness. Utilitarian vegetarians want to reduce the number of animals who have to endure horrific experiences living in cramped conditions and being slaughtered at a young age.

Many vegetarians also assume that their diet choices benefit wild animals.

Some believe wild animals have net positive lives, but this ignores the fact that r-selective reproduction dominates in most species. Some insect species have life expectancies at birth measured in days or weeks, and most deaths are quite painful. It seems likely that wild animals as a whole experience more suffering than happiness, in which case we would hope to prevent their births, just like vegetarians want to prevent the births of factory-farmed animals.

So what is the net impact of vegetarianism on wild animals? This is a very complicated question, but some considerations are discussed below.

Crop cultivation

Eating meat requires moving plant food energy up an extra trophic level in which lots of energy is lost. For this reason, vegetable protein often requires several times less crop cultivation per gram than does animal protein. Even cows that are grass-fed require lots of grazing, which has some similar effects on wild animals as crop cultivation. Actually, grazing by cows is more clearly positive than is crop cultivation, whose sign I'm still rather uncertain about.

In "Crop Cultivation and Wild Animals," I survey the pros and cons of crop cultivation for wild-animal suffering. I find it somewhat likely that growing crops like legumes/nuts, and maybe vegetables/fruits, is net beneficial to wild animals in the short run. However, growing grains/cereals might actually increase insect suffering because these crops are very productive per hectare. Farm animals are mostly fed grains/cereals (and 91.8% of grain fed to US farm animals is corn), so even though it seems likely that some crop cultivation reduces wild-animal suffering in the short run on balance, this is less clear for the crops that farm animals eat.

Climate change

Animal farming is one of the world's biggest contributors to climate change—both because it requires more energy than equivalent plant farming and because livestock emit greenhouse gases as part of their metabolism. Eating a vegan diet instead of a conventional omnivorous American diet results in 1.485 fewer metric tons of CO₂-equivalent gases per year, bigger CO₂ savings than from switching to a green car.

Once again, the question of how climate change affects wild animals on balance is sufficiently complicated that I have a separate essay for the topic: "Climate Change and Wild Animals." The conclusion is that I'm extremely uncertain of the net sign of climate change for wild-animal suffering.

In my opinion, one of the strongest reasons to expect that CO₂ emissions may increase rather than decrease total wild-animal populations (and therefore wild-animal suffering) is the so-called CO₂ fertilization effect, in which higher atmospheric CO₂ concentrations can increase plant productivity. However, a lot of the greenhouse gases from animal agriculture are in the form of methane (and, to a lesser degree, nitrous oxide). For example, Cederberg, Meyer, and Flysjö (2009) found (Fig. 9.1, p. 51) that, ignoring emissions from land-use changes, producing a kilogram of bone-free beef, exported from Brazil to Europe in 2005, released 30.8 kg CO₂-equivalent of methane, 9 kg CO₂-equivalent of nitrous oxide, and only 1 kg CO₂-equivalent of CO₂ from fossil fuels. As far as I know, methane and nitrous oxide don't contribute to CO₂ fertilization. Therefore, even if CO₂ emissions slightly increase total wild-animal populations (and hence wild-animal suffering), this doesn't necessarily mean that, in aggregate, the greenhouse gases emitted by livestock production also do so, because I assume the CO₂ fertilization effect doesn't apply for a lot of the greenhouse gases emitted in livestock production.

Cattle grazing

When cattle eat vegetation on pasture fields, they presumably leave less vegetation left over to be eaten by bugs (and other organisms). Hence, I think it's at least, say, ~60% likely that cattle grazing on the whole reduces bug populations (and hence, bug suffering) on pasture fields. These articles explore this topic further.

Rainforest loss

One article claims that "80% of Amazon Deforestation Stems from Cattle Ranching". Another piece puts the figure at 70%. Given that rainforests are home to some of the highest densities of wild-animal populations (and hence wild-animal suffering) on Earth, this appears to be a consideration in favor of beef consumption.

Eutrophication

Manure from factory farms can leach into rivers and contribute to eutrophication in water bodies. The sign of eutrophication on wild-animal suffering is unclear.

Wild-caught fish

Promoting vegetarianism typically involves promoting avoidance of seafood (although in some cases, people move toward eating fish upon learning about factory farming of land animals). The impacts of wild-caught fishing on wild-animal suffering are complex, and I don't have a strong opinion on whether they're good or bad on balance.

Fishmeal

Poultry and pigs are sometimes fed wild-caught forage fish. In 2002, 22% of fishmeal and fish oil were fed to poultry, 24% to pigs, and 46% to aquaculture. That leaves only 8% for other uses (some of which include pet food), which suggests that beef is probably not a major consumer of fishmeal? How we evaluate fishmeal depends on the net impact of capturing small wild-caught fish in general.

Despite the above figures, humans still eat most fish of the fish harvest: "Humans ate 130.8 million tons of fish in 2011. The remaining 23.2 million tons of fish went to non-food uses such as fishmeal, fish oil, culture, bait, and pharmaceuticals." Another source claims a slightly higher figure: that 36% of wild-caught fish become fishmeal. The worldwide amount of forage fish fed to pigs and chickens is 6 times the amount of fish eaten by Americans.

How much fish is fed to a single chicken? I assume feeding fishmeal to chickens is mainly something that happens in developing nations. About 10 billion chickens are killed in the US per year. Suppose the total number of chickens fed fishmeal is in the same ballpark: ~10¹⁰. The amount of fishmeal fed to chickens is, by the above numbers, roughly: (23.2 million tons) * 0.22 = 5 million tons = 10¹⁰ pounds. So it looks like a single chicken is fed on the order of ~1 pound of fish.

There's also a risk that in the future, an increasing number of farms will begin using ground-up insects as feed. Farming those insects would significantly increase total suffering.

Insects as feed

Insect farming has been proposed not just to feed fish but also to feed land farm animals. This article says "If we want to continue our customs of farming and eating animal products on our limited resources, we may have to look to novel alternatives like black soldier fly farming." van Huis et al. (2013) (p. xiv): "Recent high demand and consequent high prices for fishmeal/soy, together with increasing aquacultural production, is pushing new research into the development of insect protein for aquaculture and poultry."

Byrne (2018): "Global fast food giant, McDonald’s, is exploring the potential of protein derived from both insects and seaweed as alternatives to soy in chicken feed. [...] Insect meal for use in chicken feed would appear to have more potential than seaweed meal in the short-term, said Robinson."

Human lifespans

Veg*ans often claim that a veg*an diet will extend your lifespan. I'm skeptical of nutrition studies in general, so I don't know if this is true, but if veg*ism does add, say, 1-2 years to a person's life, that slightly increases that person's environmental impact, which probably reduces wild-animal suffering on balance. I would guess the effect is pretty small, though. For example, say lifelong veg*ism adds 1 year to a lifespan of 75 years. Then it increases a person's environmental impact by ~1/75 = 1.3%. In contrast, choosing veg*an food probably changes a person's total environmental impact by at least like 10-30% or something (I'm just making those numbers up).

Displacing other foods

When people eat more meat, they generally eat less of certain other foods (e.g., tofu, soy milk, beans). So one more effect of meat consumption is reduced consumption of certain other plant foods. (Thanks to Vidur Kapur for this point.) The impact of this on wild-animal suffering is unclear.

Substitution among meats

kbog (2019): "If I buy 1kg of chicken, some people will buy less chicken, and some of those people will buy more of other things (e.g. turkey)." This kind of substitution effect seems most clear for very similar products, such as wild-caught vs farmed fish. But it probably applies to some extent to most meats.

This consideration tends to move estimates of the net impact of different animal products closer together. For example, suppose you thought beef causes a +5 change in total utility per kg produced, while chicken causes -5 per kg produced. If your buying 1 kg of beef causes other people to buy, say, 0.5 kg less beef, and if, say, this results in a 0.25 kg increase in chicken consumption, then the actual net impact of your buying the beef is (1 kg - 0.5 kg) * (+5 / kg) + (0.25 kg) * (-5 / kg) = +1.25, not +5. (This calculation ignores the substitution effects that the 0.25 kg increase in chicken consumption will itself have.)

In principle this substitution effect could even apply to vegetarian foods. For example, maybe if you buy a vegan burrito, it slighty increases the price, making other people more likely to buy a meat burrito instead. On the other hand, especially for fledgling industries, one can imagine the effect working in the reverse direction: maybe if you buy vegan burritos, it becomes more worthwhile for food companies to produce them and for supermarkets to stock them. As a result, more consumers will have access to them and will buy them instead of meat burritos.

Promoting general concern for animals

I don't know the net directional impact of vegetarianism on suffering in the wild. Even if vegetarian diets do increase animal suffering by creating more wild animals, it's plausible that organizations promoting vegetarianism do more good than harm. Exposure to the cruelties of factory farming is one way in which many people are first introduced to the topic of animal suffering in general, and such concern can spill over into other domains, perhaps including suffering in nature. After all, if animals on factory farms would be better off not existing, then if conditions in the wild are for some animals just as miserable, then those animals would be better off not existing as well. And even if wild-animal lives were not on the whole negative, it might still be possible—perhaps much farther off in the future—to improve their welfare as is done for farm animals, such as by shifting from ecosystems filled with small, short-lived creatures that die young to ecosystems with larger, longer-lived animals.

General concern for animal suffering is crucial if humans are to make wise choices with respect to wild animals once more advanced technologies arrive, and promoting vegetarianism based on reducing suffering seems generally likely to cultivate such sympathies.

However, advancing vegetarianism from, say, the perspective that humans have no right to interfere with animals or that meat causes destruction of intrinsically valuable nature could be counterproductive. Indeed, I find it likely that vegetarianism increases support for wilderness conservation on balance, and the negative impacts of wilderness preservation on wild-animal suffering could be substantial. This concern suggests that the indirect, memetic effects of vegetarianism could very well be net harmful.

Do wild-animal considerations dominate farm-animal suffering?

Clearly large numbers of wild animals, especially insects, are affected by animal farming. But how does the impact on wild-animal suffering that meat production entails (whether positive or negative) compare with the suffering of the farmed animals themselves?

Gaverick Matheny and Kai M. A. Chan took a stab at this question in reverse in "Human Diets and Animal Welfare: the Illogic of the Larder", where they assumed for the sake of argument that both factory-farmed and wild-animal lives were positive. Ignoring distinctions among animals regarding brain complexity, they found (Table 2, p. 586) that—when just considering wild mammals and birds—cows and pigs reduced wild-animal life-years more than the number of life-years added by farming, but for chickens, on-farm life-years dominated wild-animal ones. So if we assume that both farm animals and wild animals have net negative lives, then this calculation suggests that farming of at least broiler chickens (and egg-laying hens) increases total suffering even after the wild-animal impacts of crop cultivation are considered. That said, this calculation ignored cold-blooded vertebrates, as well as invertebrates.

How does the calculation look if we include insects in the calculation? I'll try two different estimates:

Bottom-up estimate

The National Chicken Council estimates that a single chicken, with a typical slaughter age of 48 days, weighs about 6.2 pounds while alive and is produced with a feed-to-meat weight ratio of 1.91. That implies about 12 pounds of feed. In the US, most of that feed is corn, plus some soybeans and other things, but I'll focus on corn, and other crops should be have similar numbers within at most an order of magnitude.

Corn yields 171 bushels per acre-year times 56 pounds per bushel = 9600 pounds per acre-year = 24,000 pounds per hectare-year. So feeding a single chicken over its life requires on the order of 12/24,000 = 0.051% of a hectare-year. However, I'm worried this number is too low, since a good portion of the weight of corn is from corn cobs, which I assume aren't used as part of chicken feed? So let's try an alternate estimate based on calories.

This article reports 1070 calories in a half chicken, which is 2140 in a whole one. Another page estimates 1490 calories in a whole chicken. These numbers don't include eviscerated organs, the head, feathers, etc., so the calories in a live chicken are presumably a bit bigger—maybe something like ~2500-3000? I'll assume 3000, and I'll approximate the calories-in-to-calories-out ratio for chickens as 2:1, giving 6000 calories of corn needed. Corn yields "roughly 15 million calories per acre" per year, which is 37 million calories per hectare-year. So one chicken requires something like 6000/(37 million) = 0.016% of a hectare-year. This is actually lower than my previous 0.051% estimate. Let's split the difference and call it 0.03% of a hectare-year.

In this discussion, I estimated there might be ~10⁷ insects per hectare on potential crop land. So on the portion of land that's used to feed a chicken, there might be (~10⁷ insects/hectare)*(0.03% hectare-years/chicken) = ~3000 insect-years/chicken. Suppose that in expectation, farming crop land either increases or decreases insect populations by 5%. (The actual difference is probably greater than this, but given uncertainty about even the sign of crop cultivation's impact on insect welfare, the expected impact after a reasonable examination will be closer to 0%.) Then crop cultivation can be expected to create or prevent something like (5%) * (3000 insect-years/chicken) = ~150 insect-years/chicken. Since there may be several insect life cycles per year depending on the species, this might be something like, say, ~300 insects (either created or prevented) per chicken.

Top-down estimate

This estimate is far more hand-wavy, but it's nice to have a second look at the numbers, since many of the figures used above could be off by a few orders of magnitude. One study reported that invertebrate populations declined on average by 45% in recent decades. That number could be too high given that more extreme estimates are more likely to get news coverage. On the other hand, humans have been around for more than a few decades, so maybe the total impact of human activity on invertebrate populations is something like a 45% drop. Given that crop cultivation occupies 11% of Earth's land area, growing crops might contribute a nontrivial portion to the drop (or might increase net insect populations—it's unclear). Let's say the "error bars" on crop cultivation's impact are 1% of the world's total insect population, i.e., 1% * 10¹⁸ = 10¹⁶ insect-years per year. Suppose chicken farming accounts for (just making up a number), 1% of that amount, giving error bars for chicken farming of ~10¹⁴ insects impacted. Worldwide, there are ~18 billion chickens at any given time. So the insect error bars per chicken-year would be something like (~10¹⁴ insect-years)/(~2 * 10¹⁰ chicken-years) = ~10⁴ insect-years (either created or prevented) per chicken-year. If the typical insect lives about as long as the typical farmed chicken, this would also be ~10⁴ insect lives created or prevented per chicken. This is about an order of magnitude higher than the bottom-up estimate, but the input parameters are just guesses, so I'm actually surprised it's so close to the bottom-up estimate.

Weighing insects vs. chickens

My intuitive feeling is that 10² to 10³ insects' worth of suffering matters about equally as the suffering of a broiler chicken over 48 days of life and a potentially agonizing slaughter. So chicken suffering and the error bars on insect suffering from crop cultivation seem to be pretty close and are probably within a few orders of magnitude.

How about comparing on a per-neuron basis? Carl Shulman estimates that chickens have about 4/1,350 times the brain weight of humans, and since "In terms of neurons per gram of brain mass chickens are very close to humans", this implies about the same fraction of neurons. Humans have 86 billion neurons, so chickens probably have about (86 billion)(4/1,350) = 250 million neurons. Fruit flies have on the order of 100,000 neurons. So the ratio of chicken to insect neurons is on the order of 10³—about the same as the ratio of (error bars on insects affected)/chicken.

If we weight sentience in proportion to sqrt(# of neurons), the comparison becomes sqrt(250 million) = 16,000 for chickens vs. sqrt(100,000) = 320 for insects, or a ratio of 50:1. In this case, we might think effects on insects are more important than those on chickens.

Given that the sign of crop cultivation for insects isn't even clearly positive or negative, if the variance in the insect effects of chicken farming are only of a comparable magnitude with direct chicken suffering, then we might incline toward opposing chicken farming to be on the safe side, pending more thorough investigation. That said, I haven't considered other effects of chicken farming, including energy use. And if any of the soy fed to chickens is rainforest-grown, this might be significant as well. So the uncertainty remains large. But at least for now, I'd encourage people to pass on the chicken nuggets.

The following figure illustrates my current state of belief regarding this question. The blue error bars might be smaller than what's shown, but given the error bars on my error bars, I kept them reasonably large.

Footnotes

1. My current best guess is that grazing on pasture often reduces expected invertebrate suffering, while crop cultivation has less clear net impact on invertebrate suffering. On average, it seems that the diets of dairy cattle include a somewhat higher fraction of cultivated crops than the diets of beef cattle. This is why I expressed a bit more uncertainty about the rating of dairy cattle than beef cattle in my table here.

   However, this beef vs. dairy distinction is crude, and it ignores other considerations. For example, both crops and pastures that are heavily irrigated may increase invertebrate suffering in cases where, absent irrigation, much less vegetation would have existed. So, for example, grazing beef cattle on irrigated pastures in California might be worse than feeding dairy cattle rainfed corn in the Eastern USA. This is because the rainfed crops are displacing lots of native vegetation, while irrigated alfalfa pastures in Calfornia may not be.  (back)