by Brian Tomasik
First written: 14 Dec. 2015; last update: 24 Feb. 2016


Unlike farmed animals, wild-caught fish are a finite resource whose populations decrease in response to overconsumption by humans. It's plausible that more fishing of an overexploited species now reduces the net amount of human killing of that species over the long term. However, this argument is counteracted by other considerations -- such as substitution toward consumption of other kinds of fish, including farmed fish and more smaller wild fish. On balance, it's unclear whether an animal-rights activist whose goal is to reduce total human killing of animals for food should be happy or dismayed about overfishing.

Given anticipated increases in fish farming to substitute for wild fish, it's plausible to me that overfishing is somewhat bad on balance.


if we don't stop overfishing, there will be no fish, and personally, I want to keep eating fish. So it's a strictly selfish business for me. I want to eat sushi in 30 or 40 years' time, and in order to do that, we have to behave sensibly right now.
--George Duffield, filmmaker

Consider Fred, an animal-rights activist who opposes humans killing other animals. In particular, Fred wants to minimize the expected number of (possibly brain-complexity-weighted) animals that humans kill over the long-term future. Fred is not concerned about natural suffering of animals in the wild. (For an analysis of overfishing that focuses on wild-animal suffering, see this article.)

Naively it seems obvious that Fred should oppose commercial fishing in the wild -- after all, this increases human killing of fish in the short run. But given that many fisheries are overexploited, it could be that eating more fish now reduces long-run catches of fish, thereby reducing the total number of fish killed by humans. This piece analyzes that argument and enumerates a number of complicating factors.

Which fish are overexploited?

As will be elaborated below, how Fred analyzes increased fishing depends a lot on whether the species being eaten is currently overexploited. It's difficult for ordinary consumers to assess precisely whether a given fish in the supermarket comes from an overexploited fishery or not, but there are "good fish guides" that advise at a high level about the exploitation status of fish species. If Fred thinks it's better to overfish exploited stocks than to fish sustainably, then Fred would want to essentially reverse the advice of these guides.

Ray Hilborn believes that claims of overfishing are exaggerated and that many fisheries are actually recovering, especially in Western countries. Indeed, the UN FAO estimates that only 29.9% of global fish stocks are fished beyond a sustainable yield. So it's important not to assume that any given fish that a person might buy comes from overexploited fisheries.

Considering only a single species

MSY model

In this section I discuss the maximum sustainable yield (MSY) model of fisheries. For background, read the Wikipedia article first. In this piece, I use the same notation as that article.

Theragra chalcogramma fishingThe MSY model assumes that fisherpeople harvest a constant amount of fish per unit time to achieve a sustainable fishery population. The harvested fish are "interest" that accumulates on their "bank account". Like with a regular bank account, the amount of interest increases as the fish population increases from 0 to K/2 (half the fishery's carrying capacity).a But unlike a bank account, interest decreases as the number of fish in the bank increases above K/2 the because population growth rate is slower closer to carrying capacity. Hence, to maximize long-term "revenue", the fisherpeople should aim to keep the fishery's population at K/2, since this gives the maximal interest payment per unit time for all future time periods.

Since Fred only cares about preventing human-caused fish deaths, Fred aims to minimize the same quantity that the fishery managers aim to maximize: number of fish caught over the long run -- which, for a sustainable fishery with a constant yield rate, is proportional to number of fish caught per unit time. Hence, Fred should want the fishery population to be as far from the MSY population (N = K/2) as possible.b

For example, if the fishery population is large (N is close to K), Fred should oppose fishing, since doing so will kill fish in the short run and move N closer to K/2. On the other hand, if the fishery is overexploited, with N closer to 0, Fred should (within the context of this simplified model) favor more fishing now, because of the following result.


Within the basic MSY model, if N ≤ K/2, then harvesting an extra fish never increases and may decrease long-run fish catches regardless of the fishery-management policy used.

Proof sketch:

Let the current fishery population be N0, where 3 ≤ N0 ≤ K/2. Due to increased demand for fish, the fisherpeople harvest one more fish, leaving a population of only N0-1 fish.

If fishery managers want to maximize total long-term fish harvesting (ignoring the benefit of harvesting sooner in light of the time value of money, etc.), then at a population size of N0-1, they should optimally catch one less fish than they would have had the population been N0. This is because they want to leave an extra fish in order to restore the population level back to N0. They should replenish the population right away rather than waiting because dN/dt increases with N for N ≤ N0, so the fishery should try to spend as little time as possible at a population below N0. The one less fish caught that serves to replenish the population compensates for the one extra fish caught earlier, so catching an extra fish now doesn't increase long-run fish catches.

If the fishery isn't managed optimally (perhaps because of competition among fisherpeople, short-term greed, lack of real-time population monitoring, etc.), then catching one extra fish will keep N below N0 for some time. Even if the fisherpeople respond by reducing catch rate to the sustainable yield for a population size of N0-1, dN/dt will be smaller than if the population were N0, which will reduce total future catch. If the fisherpeople don't decrease catch, they'll drive the population to extinction in the long run, which definitely decreases long-run catch.

In general, any fishery policy other than an optimal policy will decrease long-run fish catch, because the fishery's objective is to maximize long-term catch, and anything suboptimal doesn't achieve the maximum long-term catch level. ■

Empirical considerations

The idea behind the MSY model seems to have some real-world validity. For example, Brian Tissot presented the results of adding marine protected areas in Hawaii (which is tantamount to moving N from close to 0 toward K/2): Both fish populations and fish catches increased.

This makes sense when we see fisherpeople as agents in a tragedy of the commons: The sum total of fish caught by fisherpeople decreases when the commons are overfished. For those who oppose the exploitation of fish, this might be called a victory of the commons, with the addition of marine protected areas being the tragedy.

That said, the MSY model can also be unrealistic. It "ignores the size and age of the animal being taken, its reproductive status, and it focuses solely on the species in question, ignoring the damage to the ecosystem caused by the designated level of exploitation and the issue of bycatch. Among conservation biologists it is widely regarded as dangerous and misused.[2][3]" This article notes that

Whereas species can be identified and counted, species interactions might be referred to as the invisible fabric of nature because the strength of their influence is detectable only after a perturbation to one or more of the interacting species.

Still, MSY remains an important concept in fisheries management, especially in its less naive forms.

Due to imperfect knowledge, measurement noise, and random variations in fishery populations, a naive application of MSY policy can lead to fishery depletion:

An example of this was the New Zealand Orange roughy fishery. Early quotas were based on an assumption that the orange roughy had a fairly short lifespan and bred relatively quickly. However, it was later discovered that the orange roughy lived a long time and had bred slowly (~30 years). By this stage stocks had been largely depleted.

If these errors in applying MSY tend to overexploit more than they underexploit, they'll tend to lead to more declines in fishery stocks, which further accentuates the argument that fishing now reduces fishing later. When the fish population is reduced to a very small level, even if it's later allowed to grow back to a size of K/2, potential catches will have been lost in the meanwhile because the growth rate is slower at lower population sizes. (As an analogy, if you take money out of your bank account and then later return it, you will have foregone interest that you could have been earning in the intervening period.)

Possible permanent collapse

Fishing beyond a sustainable yield could lead to a complete fishery collapse. Daniel Pauly's "Aquacalypse Now: The End of Fish" explains that without intervention, "just as a Ponzi scheme will collapse once the pool of potential investors has been drained, so too will the fishing industry collapse as the oceans are drained of life."

Opinions differ on how permanent fish-population collapses will be. Pauly himself said of the effects of overfishing:

Strangely enough, these effects are all reversible, all the animals that have disappeared would reappear, all the animals that were small would grow, all the relationships that you can't see any more would re-establish themselves, and the system would re-emerge. So that's one thing to be optimistic about. The oceans, much more so than the land, are reversible [...].

On the other hand, many reports suggest irreversible changes. For instance, this study reports (p. 271):

Realization is growing that overfishing may result in possibly irreversible changes to the structure and functioning of marine ecosystems. [...] We have shown that the collapse of the low-trophic-level planktivorous species has resulted in drastic changes in a productive upwelling ecosystem, leading to a possible alternate stable state that has affected the entire food web and is characterized by a high biomass of jellyfishes and loss of productivity of the higher trophic levels, including endangered top predators and commercially important predatory fish stocks.


When considering only a single fish species harvested in the wild, it seems that if the fish is currently overexploited, then increasing demand for it will probably reduce total catches of this fish species over the long term. This is the case if the MSY model is pretty accurate and if the MSY model is inaccurate in such a way that management strategies tend to diminish fishery populations.

Substitution across fish types

The analysis becomes more complex when considering other wild fish species and fish farming. A decrease in the wild population of a given fish species may not reduce total fish consumption but might merely shift consumption to other fish.

Fishing down the food web

Big fish tend to be overexploited first, since they're easier to catch and take longer to mature. As populations of big fish decline, the result is fishing down the food web, i.e., harvesting smaller fish species. To produce the same amount of fish meat, more total fish will need to be harvested. Even if Fred weighs the badness of killing fish by the brain complexity of the fish killed, as long as Fred's brain-complexity function is less than linear in the mass of the fish being caught, then catching a given mass of smaller fish will be worse in Fred's eyes.c Assuming the total mass of fish harvested stays roughly the same now and into the long run, then eating more big fish and thereby reducing their long-term yields will have made things worse for Fred.

Fish farming

In addition to causing consumption of more small wild fish, overfishing of big species in the wild creates pressure to produce farmed fish. Some farmed fish are fed smaller wild-caught fish, which means fish farming can significantly increase total fishing. This would be prima facie bad in Fred's eyes.

However, there might be cases where fishing pressure on smaller wild-caught fish who are fed to bigger fish is sufficiently intense that populations of the smaller fish also decline. One example of this was the Peruvian anchovy fishery, which had supplied fish meal for livestock but which collapsed, leading to increased demand for soy protein instead.

Unfortunately, other fish-feed sources might be even worse than wild-caught fish, such as insects. Farming insects to feed farmed fish would significantly multiply the number of animals killed by humans. That said, fish feed may also include plant-based ingredients.

How much would substitution happen?

In rich countries, I would expect that demand for fish would remain pretty stable in the face of further fishery declines. As there become fewer big wild fish, prices of fish increase slightly, which slightly reduces the quantity demanded. And some eco-conscious consumers might cut back on fish in response to overfishing. But on the whole I expect the effect to be modest for affluent omnivores.

However, the effect might be more pronounced in poor countries, which may lack the resources to undertake aquaculture. According to the UN FAO:

In many areas of sub-Saharan Africa and South Asia, [...] fish consumption levels remain too low and they are failing to benefit from the contributions that fisheries and aquaculture are increasingly making elsewhere in terms of sustainable food security and income.

Given that up to 90% of US seafood comes from outside the US and that most rich countries import a lot of seafood from developing countries, increased consumption of fish in rich countries might indeed reduce long-term fishing by poor countries (to the detrmiment of indigent people in those countries).

Total fish production

One macroscopic perspective from which to assess the total impact of fishing is the following graph (compiled by Earth Policy Institute). I think the trends in the graph might be overstated if these figures include China, which is a major fish producer that's widely believed to overreport its fish numbers.d So the increase in fishing is possibly less dramatic than what's shown. But assuming the overall trend of wild + farmed fish is still an increasing one, then this is a bad sign from Fred's perspective, even if Fred only cares about fish in proportion to their mass (so that Fred wants to minimize the total biomass of fish caught). This is weak evidence that further demand for fish would make things worse for Fred, since most of the trend in the graph was probably driven by growing fish demand.

Of course, some fishery doomsayers might claim that the upward trend in the graph can't last forever (especially since fish farming often relies on wild-caught fish), and if they're correct, then drawing conclusions from the portion of the curve we can see now would give the wrong impression. But given that some fisheries are recovering, especially in the US, I'm personally skeptical about "end of fish" scenarios.


When considering substitution of fish consumers away from a declining fish species toward other species and fish farming, the effect of marginal big-fish consumption in the short run on total long-term fish harvesting becomes unclear. It's not obvious how much a reduction in populations of big wild fish reduces total fish consumption especially by the global poor (which could be good from Fred's perspective) versus how much it merely shifts consumption to other fish types (which could be quite bad from Fred's perspective).


Fishing kills not just the target organisms for harvest but other marine life that gets caught in nets. Fred views bycatch as wrong, because it represents another form of humans directly killing animals.

Reasons more fishing now might increase bycatch:

  • In the short term, more demand for fish means more fishing and hence more bycatch.
  • Overfishing means more total fishing effort for the same or lower fish catches, which may mean more total dragging of nets through the oceans.

Reasons more fishing now might decrease bycatch:

  • If a fishery becomes sufficiently overexploited that it's no longer fished, then bycatch will no longer happen there.
  • Bycatch is lower on fish farms (I assume), and fish farms increase as overfishing increases.

Anti-overfishing policies

Perhaps unfortunately for Fred, as overfishing pressures increase, there will be stronger pushes for marine protected areas and fishing rights: "Rights-based fishing has already transformed fisheries in the U.S., where it’s known as catch shares. Today, more than two-thirds of the seafood supply in the U.S. is sustainable, and more than 100 species are now on the path to recovery."

Of course, it seems unlikely (though possible) that more overfishing by some amount X now will actually reduce overfishing by some amount greater than X in the near future by creating pressure for anti-overfishing policies. But at least anti-overfishing policies will dampen whatever effects overfishing would be expected to have in the business-as-usual scenario -- e.g., overfishing by X amount now only implies a net amount of overfishing of, say, X/2 in the long term given reactions to overfishing.

That said, developing nations seem unlikely to implement strong anti-overfishing measures any time soon, and a lot of US fish comes from other countries, so maybe the contribution of more overfishing now toward triggering protection measures is relatively small?

Consumption of land animals

Especially for wealthy consumers who already eat a lot of animal flesh, increased consumption of fish by the typical non-vegetarian person would probably imply somewhat reduced consumption of land farm animals. Fred should be a bit relieved by this, but given that many fish (especially sardines and such) are smaller than even chickens, a trade of killing fewer land farm animals in return for killing many more fish doesn't seem worth it, ignoring the other complicating factors in this piece.

For those who, unlike Fred, are also concerned with suffering in the wild, the sign of reduced consumption of land farm animals is less clear.

Memetic considerations

It's typically assumed that eating fish implies disregard for fish suffering. Hence, eating more fish conduces toward less social concern for animals in general, which is probably bad from Fred's perspective when taking a long-term view. This is especially so given that Fred only cares about human killing of animals and hence isn't worried about potential downsides of animal rights for wild animals.

Eating any fish vs. switching to eating overexploited fish

In this piece, I've largely conflated two distinct questions:

  1. Should Fred be glad about additional consumption of overexploited fish rather than not eating any fish?
  2. Should Fred be glad when a person switches from eating 1 sustainable fish to eating 1 equally sized overexploited fish?

The answer to question #2 seems more likely to be "yes" than the answer to question #1, but both questions are uncertain.

Let's say the sustainable fish comes from a fishery where the fish population is above the MSY population. When someone eats such fish, demand increases slightly, which slightly increases fishing effort, which slightly reduces the population and hence actually increases sustainable yield. Imagine that eating 1 sustainable fish leads to a long-term increase of, say, 2 total fish killed (since sustainable yield is at least temporarily higher). Assume for simplicity that this increase in fish killed in the future has no substitution impact on other types of fish (although my point in this section would still work even if this assumption were relaxed).

Now compare this with the effect of eating overexploited fish. An overexploited fish population is less than the MSY population, so eating a bit more reduces the population further and thus reduces harvest rates of that species into the future. Let's say for the sake of this example that eating 1 extra overexploited fish reduces future fish catch of this species by 2. However, because this leaves fewer fish in the future, people also substitute toward eating other kinds of fish. In particular, people substitute toward eating a number F of extra fish of other kinds. Assume the substituted fish matter equally as the overexploited fish. Then the total change in fish killed by eating the 1 overexploited fish is 1 (original fish eaten) - 2 (future fish of that species prevented from being killed) + F (other types of fish eaten) = F-1.

The options are

  • eat no fish: net fish killed = 0
  • eat 1 sustainable fish: net fish killed = 2
  • eat 1 overexploited fish: net fish killed = F-1.

So if F < 1, eating the overexploited fish is actually better than not eating any fish. If 1 < F < 3, eating overexploited fish is worse than eating no fish but better than eating sustainable fish. And if F > 3, eating overexploited fish is worse even than eating sustainable fish.

Recreational fishing

The ideas in this piece don't necessarily translate to recreational fishing. Lakes where people fish for fun may be restocked, which counteracts the overfishing benefits of fish consumption:

Hundreds of Minnesota walleye lakes would today offer little or no walleye fishing were it not for regular stocking. Stocking is responsible for the tiger muskies hooked each year in Twin Cities Metro Region lakes. Stocking has helped restore the native lake trout population on Lake Superior.

And the ideas in this piece especially don't apply to catch-and-release fishing, which causes trauma and injury without necessarily changing fish populations.

Reader feedback

This Facebook thread includes some good comments on this piece.


  1. In practice, the MSY population is often around 30% of K.  (back)
  2. Purely from the perspective of how many fish humans catch, a constant catch rate of a fish population above the MSY population would be equivalent to a constant catch rate the same amount below the MSY population. But the two states of affairs are not equivalent from the perspective of wild-animal suffering. A main reason growth rates are low when N is close to K is that most offspring that are born die due to lack of resources. So total death rates for N close to K are vastly higher than for N close to 0, even though sustainable harvest rates may be similar between the two cases. So ignoring effects on other species, a wild-animal advocate would prefer to keep the fish population below rather than above the MSY level.  (back)
  3. To see this, consider a total mass M of fish being harvested, consisting of N equally sized individuals. Hence, each individual has mass M/N. Let c(m) be the brain complexity of a fish of mass m. Since c is less than linear, c(αm) < αc(m) for any m > 0 and α > 1. The total badness B of killing the N fish is B(N) = N * c(M/N). For β < 1, B(βN) = βN * c(M/[βN]) < βN * (1/β) * c(M/N) = N * c(M/N) = B(N). So total badness decreases as the fish caught are larger.  (back)
  4. That said, China also underreports its foreign fish catches by about 5 million tons.  (back)