Summary

1. Where there’s overfishing, reducing fishing pressure or harvest rates — roughly the share of the population or biomass caught in a fishery per fishing period — actually allows more animals to be caught in the long run.
2. Sustainable fishery management policies are generally aimed at maximizing or maintaining high levels of catch — the biomass of wild aquatic animals caught — in the long run. More restrictive policies that would actually reduce long-run catch generally seem politically infeasible, and less restrictive policies that increase long-run catch don’t seem like a stepping stone to more restrictive ones that decrease it.
3. Demand reductions for wild-caught aquatic animals may increase or decrease actual catch, and it’s very unclear which. My highly uncertain tentative best guesses are that
   1. they seem slightly more likely to increase than decrease catch in the near term but bioeconomic “long run”, e.g. over the next 10-20 years, and
   2. persistent demand reductions and cumulative work towards them seem slightly more likely to decrease than increase catch over longer timelines with more sustainable fishery management and eventual population decline, but it’s not clear if and when catch would actually be consistently lower on average than otherwise.

Acknowledgements

Thanks to Brian Tomasik, Ren Ryba and Tori for their feedback on an earlier draft, and Saulius Šimčikas for his supervision on an earlier unpublished project. All errors are my own.

Basic terminology

1. I use ‘fishing’ to include the capture of any wild aquatic animal, including crustaceans, not just fish.
2. I refer to the long run (and long-run) in fishing as long enough for all production factors, including the number of boats/vessels, amount of fishing equipment, employment and the number of fishing companies or businesses to increase or decrease and approximately reach a new equilibrium in response to a permanent shift in prices, supply or demand. I’d expect this to typically be less than a decade. This is a standard term in economics.

Introduction

Reductions in fishing pressure or harvest rates — roughly the share of the population or biomass caught in a fishery per fishing period — can result from reductions in demand or from improvements in fishery management, like the use of quotas, smaller fishing net mesh sizes, seasonal closures or restrictions on fishing vessels or their numbers. However, these reductions can also lead to increases in catch where there’s overfishing, by allowing stocks to recover, resulting in more fish to catch. Fishery management policies that preserve or increase stocks are also typically aimed at increasing long-run catch.

If we’re concerned with reducing total exploitation, injustice or harm caused by humans or moral/rational agents, then this would count against this kind of work (see also Tomasik, 2015, who made this point earlier). The same could hold for a (weighted) average level of exploitation, injustice or harm by humans across all animals or moral patients.^[1][2] I don’t personally take exploitation or harm by humans in particular to be worse than other types of harms, independently of their effects on subjective welfare,^[3] but this seems to be a common position. Furthermore, harm caused by humans might matter more for indirect reasons related to subjective welfare in practice, tracking our willingness to help or make other sacrifices for other animals.^[4]

Or, if fishing deaths are particularly bad compared to natural deaths and will continue to be so (e.g. humane capture and humane slaughter won’t become widespread), and bad enough to be worth preventing regardless of the population effects and effects on natural deaths, then this, too, would count against improving fishery management and possibly against demand reductions.

I start with an illustration and then discuss potential intervention effects in more detail.

Fishing like cattle ranching

In some ways, fishing is like farming animals on pasture. In both, the animals are allowed to roam relatively freely most of their lives. Then, convincing fishers, whether via price or demand shifts, policy or outreach, to reduce their fishing pressure or harvest rates — roughly the share of the population or biomass caught in a fishery per fishing period — or improve fishery management in a way that allows them to increase yields would be like convincing a cattle rancher to keep more of the cattle longer for breeding, allowing them to increase the population available for slaughter overall. That seems morally questionable if we’re concerned with minimizing these deaths or exploitation, injustice or harm by humans.

There are of course important differences. Some of them are:

1. Wild-caught aquatic animals are more free from direct human interference and harm, and almost entirely free, except for capture. Pasture-raised beef cattle are less free, and male calves are often castrated without any pain management.^[5]
   1. Still, wild-caught aquatic animals may also be harmed (or benefitted) indirectly by us, through our environmental impacts, e.g. climate change and pollution, and fishing of other species.
2. Wild animals of a given species in a given region would have existed whether or not we fished them, but farmed animals wouldn’t really exist if not for our selective breeding and other interference. More specifically and to pick out the morally relevant differences,^[6] there would probably have been fewer descendents of the progenitors of the farmed species, if not for farming them. On the other hand, there would have been more members of the fished species, if not for fishing them.
3. Fishing deaths seem likely to cause more suffering on average than slaughter with stunning, as is more common for cattle.
4. Plausibly average welfare. Compared to the lives of beef cattle on pasture, I’d guess fishes’ and crustaceans’ lives are worse on average (relative to how good or bad their experiences can be) and more likely than cattle’s to be bad overall. Increasing their populations seems therefore more likely to be bad for them than doing the same for beef cattle, in terms of their subjective welfare.
   1. Cattle are practically guaranteed access to enough food and generally protected from predators. They may also receive some medical treatment.
   2. Fish and crustaceans typically have far more offspring and much higher natural mortality rates. Fewer than 10% of beef cattle die before slaughter,^[7] but often >90% of larval fish and crustaceans die before reaching the juvenile stage, and mortality rates in juveniles remain high, as well.^[8]
   3. That being said, humans aren’t actively responsible for the natural harms to wild aquatic animals, but if it weren’t for us breeding farmed animals, they wouldn’t be harmed at all, directly by us or by more natural harms, so it seems we are at least partially responsible for all harms to farmed animals.

Reducing fishing rates can of course sometimes decrease catch even in the long run, too, when it gets low enough.

Would we increase or decrease catch?

Sustainable fishing policy reform probably increases catch overall. I’d guess reducing demand increases catch in the near term but could decrease it over the longer term, with recent trends reversing.

Fishery management policies

Many fishery management policies are enforced to limit catch, like quotas, fishing net mesh size restrictions, restrictions on the number of vessels, and seasonal closures. However, they are generally designed to increase long-run catch or prevent it from decreasing, by preventing overfishing. If our primary goal were to decrease long-run catch, and there were no other important indirect effects counting in favour, we shouldn’t support policies that would increase it. That’s the easy case. And it could even be worth it to oppose such policies, although I expect that to be morally ambiguous for other reasons discussed in this article.

On the other hand, policies that are so limiting that they would reduce long-run catch wouldn’t be very politically feasible, at least not in the near term, perhaps other than in special cases where maximally sustainably fishing one species endangers another, via indiscriminate catch, by-catch, or for already endangered predators that depend on them, directly or indirectly.

But maybe there are other important indirect effects that could help reduce catch. Perhaps it comes through coalition-building with conservationists or environmentalists or association with those movements, gaining allies to reduce catch overall. Many people worried about overfishing are worried in part because they care about conservation for its own sake or the sake of the animals. Maybe we can build on that. Or, maybe we can make fishery policies more and more limiting over time, to the point of eventually decreasing long-run catch.

However, policies that maintain or increase long-run catch are also in the economic interests of countries and industries, and, if properly implemented and enforced, can prevent overfishing, population collapse and species loss, and maintain populations at significant shares of their natural levels (around 30% or 50% according to standard models, Ritchie & Roser, 2021–2024). We might expect such policies to be eventually implemented without our intervention, too, precisely because of economic benefits. On the other hand, reducing long-run catch is against their economic interests and isn’t necessary to meet these basic conservation goals, so is far less politically tractable.

Policies that maintain or increase long-run catch can also politically and socially empower the industry by increasing their profits and employment, so supporting those first to later support stricter policies reducing long-run catch would empower the very opponents of those stricter policies.

So, policy work to limit catch, if it has any effect at all, seems reasonably likely to just increase long-run catch by reducing overfishing, and bring us no closer politically to actually reducing long-run catch, in case that was our goal.

Reducing demand

On the other hand, reducing demand, more precisely negative demand shifts — like getting people (and farmed animals) to consume fewer wild-caught aquatic animals, via diet advocacy or support for plant-based or cultured substitutes — have more complicated effects.

Formally, fishing pressure, also called the harvest rate, is the ratio, $U=C/B$, of catch, $C$, over the fishing period, in mass, to the biomass at the start of the fishing period, $B$. Or, it's the ratio, $U=C/N$, of catch, $C$, over the fishing period, in number of individuals, to the population at the start of the fishing period, $N$.^[9] Some fishery models use biomass and others use the number of individuals.

If demand and resulting catch prices are reduced enough in a fishery, then fishing pressure and catch should be lower. Large enough reductions in demand and prices should also reduce global catch. You might think any negative demand shift, even a small one, should decrease catch, or at least reduce expected catch. But things are more complicated in wild fisheries, and the catch can instead increase in response to reduced demand and reduced fishing pressure if there’s overfishing, i.e. fishing pressure $U>U_{MSY}$, as illustrated in the graph below, which is common:

For more background and illustrations of wild fishery models, see Haddon, 2023, chapter 3, Maximum sustainable yield - Wikipedia, Sustainable Fisheries UW, 2019, and/or pages 9-16 from the FAO report Melnychuk et al., 2020. For bioeconomic models, catch supply functions of price based on these fishery models, see Eide, 2012, Eide, 2011, Copes, 1970 and/or Pham & Flaaten, 2013. These models make some simplifications and have limitations, but they’re still useful for capturing general features.^[10]

In the relatively near term (but bioeconomic long run), e.g. the next 10 years or so, it looks like catch would increase, given little effect on maximally sustainably fished stocks, increased catch in overfished stocks, decreased catch in underfished stocks, and, based on FAO, 2022, Figure 23, more overfished than underfished stocks^[11]:

Of course, the actual answer depends on the relative magnitudes of effects per stock, which depends on the number of fish and number of fish caught per stock, as well as supply elasticities. So, that catch would increase is a first rough guess, but subject to possible revision.

However, over the longer term, the trends may reverse and facilitate low enough fishing pressure for a marginal but persistent reduction in demand to reduce catch later. Reasons for reversal include:

1. The United Nations projects the global human population to peak around 2090 and then start to decline, with globally falling birth rates (e.g. Ritchie et al., 2023a, Ritchie et al., 2023b and Cillufo and Ruiz, 2019, graphing projections by UN DESA, Population Division, 2022).
2. It seems per capita animal product consumption has also already roughly peaked in high-income countries, and others should eventually peak, too, as they become richer, even if they will increase per capita consumption first (Whitton et al., 2019, Ólafsson, 2023). Then future advocacy, higher incomes to afford to be more concerned with animals or other impacts of diet, and advances in substitutes could facilitate eventual declines.
3. Economic interests support fishery policies that maximize sustainable yields and eliminate overfishing, or at least prevent short-run increases in catch, like well-designed and well-enforced quotas. Overfishing is also rarer among high-income or developed countries than it is among low- and middle-income or developing countries (Ritchie & Roser, 2021–2024). Total allowable catches and quotas have been increasingly implemented, especially in wealthier countries (Lawson & Smith, 2023, European Commission, accessed 2024), but also notably, as accounting for large shares of global catch, for Peruvian anchoveta^[12] (Molinari, 2022, Molinari, 2024, Fishbase a, Fishbase b, Fishbase c, Fishbase d) and increasingly in China^[13] (Ding et al., 2023). So we may expect overfishing to be mostly eliminated in the longer term, as countries become wealthier and come to manage their resources more rationally for the long term. With little overfishing in the longer term, lower demand then would be unlikely to substantially increase global catch.

However, we should be careful about extrapolating trends, especially the population trends in 1.

I can imagine the expected future declining population to reverse again. Increases in individual wealth and advances in technology — like automation freeing up our time and artificial wombs — and government incentives may make having and raising children much easier or relatively attractive, so it’s conceivable fertility rates will surpass and eventually remain above replacement. Or, subgroups of humans with stably higher fertility rates than replacement will come to outnumber the rest and then the global human population could increase again.

Plus, the projections account only for modest reductions in mortality rates and modest increases in life expectancies, so that deaths will eventually outnumber births each year. But we could have radical life extension, curing cancer and heart disease, replacing aging cells, and far lower mortality rates than they project. If people live far longer and in better health and energy throughout their lives, then maybe they will have enough children to replace themselves, even if it means spreading births out more over time.

On the other hand, our efforts today could be made mostly redundant, if there’s a very large drop in demand that would have happened either way. For example, artificial intelligence and automation may facilitate research on substitutes, but the solution that will actually replace enough animal products won’t arrive until we have sufficiently advanced such technologies. It may arrive regardless of our work on substitutes before then, so our work on substitutes before then won’t help. Or, there could be a human population collapse in a global catastrophe, and our efforts today never pay off, because demand will reach 0 anyway or our work will be undone by the catastrophe. Or, there could be species collapse practically eliminating fishing independent of our efforts to reduce fishing.

In these cases, our impacts will be more limited to the nearer term effects, which seem slightly more likely to increase catch.

1. ^{^}

    However, averagist views seem to imply ethical egoism due to the possibility of solipsism (Oesterheld, 2017, Tarsney, 2023), and more generally, we should give far more weight to outcomes with much smaller total populations of moral patients, e.g. the unconsciousness of nonhuman animals, total extinction soon, a generally unpopulated universe. This is because it’s easier in practice to affect the average in smaller populations, without larger and larger populations you can’t or can hardly affect at all. This seems to me to count very strongly against averagist views.

2. ^{^}

    Roughly, this is because the extra animals will be more exploited than the average of all animals — as only a small share of all animals (wild, farmed or otherwise) are or have been exploited — and so their addition increases the average level of exploitation. It’s less clear if we include far future moral patients.

   Then, from reducing fishing pressure where there’s overfishing, we have some additional animals who will be exploited and some additional animals who won’t be exploited, and the average rate of the exploitation of these extra animals is greater than that of the average animal in the background population. Therefore, average exploitation would increase.

   Plus, it’s not even clear the number of unexploited individuals will increase, in case the additional unfished animals further suppress the populations of their prey, although this could have further effects to lower trophic levels, too. Or, we could be underestimating the number of unexploited animals added, in case many more lower down the food chain are added than are prevented.

   If we’re concerned with other injustices or harms caused by humans besides direct exploitation, like from climate change, crop agriculture and plastic pollution, then reducing overfishing would increase the average level of human-caused harm for the background population and the extra individuals, so it’s not clear this would affect the conclusion.

3. ^{^}

    Animals themselves don’t care that they are harmed by humans in particular, rather than by predators or other causes. They will care about how they’re killed indirectly, depending on how it affects their experiences, and they would otherwise care about their experiences if not killed earlier. As such, exploitation by humans doesn’t seem to matter much in itself.

4. ^{^}

    In this case, the distribution of willingness could track the distribution of attitudes and individual consumption. Total consumption could be less informative of what people would be willing to do for the interests of other animals, because total consumption tracks the total human population. That being said, some people may decide not to have children in part for the benefit of other animals.

5. ^{^}

    Beef cattle are generally fenced in, even if on large plots of land.

   They are mostly allowed to breed naturally, but with farmers intentionally exposing cows (and heifers) to specific bulls (APHIS USDA, 2009). Calves are weaned and separated from their mothers at around 4–8 or 6–8 months of age (APHIS USDA, 2017, p.7, Oregon State University, Rasby, 2007).

   In the US, 79% of male beef calves born in 2017 were or would be castrated before sale (APHIS, USDA, 2017, p.29), with only 30% of male calf castration procedures and 20% of veterinarians performing the procedure on male calves using any pain management (anaesthesia or analgesia) (Fajt et al., 2011 and Coetzee et al., 2010, respectively). In the UK, 67% of respondents who performed castration on calves used local anaesthesia (Remnant et al., 2017).

   In the US, only 7.8% of beef cattle would grow horns, given widespread use of polled breeds, so beef cattle generally need not be dehorned in the US (APHIS USDA, 2017, p.iii).

6. ^{^}

    Some other differences don’t actually hold or don’t seem morally relevant.

   “Fished animals” wouldn’t exist if not for our interference; because they wouldn’t be fished. “Farmed” and “fished” are just properties of individuals. “Farmed animals” have or would otherwise have had natural wild counterparts without farming, from their common ancestors. We can just not fish or stop farming any specific individual.

   Fishing also affects which specific individuals will exist, by changing which individuals have offspring at all, which animals fertilize which eggs or even conditional on the same paired individuals, which specific eggs get fertilized by which specific sperm through the precise timings of events and so which animals. I’d guess individual genetic identities have little overlap between the current fished populations and what they would have been if unfished.

   Selective breeding is common in farming, and it changes the distribution of genes in the population. However, fishing can also exert some specific adaptive pressures, also changing the distribution of genes in (in principle) predictable directions, too. If we care about the specific distribution of genes, or the specific genetic shifts in farmed animals but not those in wild-caught aquatic animals, or that the shifts in farmed animals are larger, then this seems essentially speciesist, or a kind of discrimination on the basis of mere group genetic distribution.

7. ^{^}

    Cow pregnancies last around 283 days or around 9 months (UNL Beef, 2015), and generally result in one calf per pregnancy. Around 6-7% of beef calves were born dead or die before weaning in the US in 2007 (APHIS USDA, 2010) and around 8% in Canada (BCRC, 2023). In the US after weaning, according to APHIS USDA, 2010, “1.5 percent of beef breeding cattle, weaned or older, died or were lost to other causes (such as theft) in 2007.” So, around 7% to 10% of beef cattle die before slaughter.

   For Angus beef in the US, cattle are slaughtered between 9 and 30 months old (Angus Media, 2019). In Great Britain, beef cattle are typically slaughtered between 14 and 30 months old (Clarke, 2022).

   However, probably more than half of successful fertilizations result in pregnancy loss (Reese et al., 2020), and some of these embryos or fetuses may matter morally, too, and with relative weight to adult cattle similar to larval fish and crustaceans to their adults. On the other hand, they may be capable of morally relevant conscious experiences, e.g. conscious pain, but actually almost entirely unconscious, e.g. sleeping, and die painlessly in their sleep.

8. ^{^}

    About Peruvian anchoveta, one of the most wild-caught fish, representing around 28% of the number of fish caught annually on average (Mood & Brooke, 2024), Molina-Valdivia et al. (2020) write:

   Early larval growth and mortality of anchoveta is highly variable at intra-seasonal and latitudinal scales along the Chilean coast. In northern Chile (23 °S), larval growth varies between 0.50- and 0.85-mm day−1, with daily losses of 16–23% (Contreras et al., 2017); meanwhile, in central Chile (36 °S), larvae grew at 0.40-0.57 mm day−1, with daily losses of 4–7% (Castro and Hernández, 2000; Hernández and Castro, 2000).

   See also Fig. 4. H (and G for another species) for the drop in abundance over time as larvae, leaving less than 1% after 40 days according to their fitted model.

   Dall et al., (1991, p.357, Table 11.1) report that for the family of penaeid shrimp, >70% of larvae die per week over a period of 2–3 weeks, 10–25% of juveniles die per week over a period of 2–3 months, and 2–10% die per week as adults thereafter.

9. ^{^}

    This is not identical to the share of the biomass or population caught per period, and can actually be >1, because more animals will be born over the fishing period, the denominator is the biomass at the start of the period. We can also take fishing periods to be arbitrarily long, so that U>1.

10. ^{^}

     The standard model assumes a single-species fishery, so no predators of the fished species being caught (the influence of predators may otherwise be captured in other model parameters) and ignores changes to age at capture, e.g. smaller net meshes can catch juveniles and reduce long run yields. Furthermore, some parameters may be treated as constant but can change over time, e.g. the unfished biomass should really be a counterfactual in the same environmental conditions, not the biomass before humans started fishing the population, but may be subject to variation with temperature and weather changes, like the El Niño-Southern Oscillation.

    For criticism of this model and its use, misuse and abuse in policy, see Maunder, 2003, Mesnil, 2012, Finley and Oreskes, 2013 and Ramesh and Namboothri, 2018. For a response defending the original Schaefer models and against more recent variants, see Pauly and Froese, 2020. For variants, see Haddon, 2023, Chapter 7 and Cadima, 2003.

11. ^{^}

    Whether or not at equilibrium, I expect a reduction in fishing pressure to reduce catch if underfishing and if we can expect underfishing to continue. Similarly, I expect a reduction in fishing pressure to increase catch if overfishing and if we can expect overfishing to continue, without fishery management policies limiting catch or fish stocking.

    Continued underfishing leads to the stock becoming underfished, i.e. $B>B_{MSY}$ (Melnychuk et al., 2020, p.13), and at equilibrium, a stock is underfished if and only if it’s subject to underfishing. Similarly, whether or not at equilibrium, continued overfishing leads to the stock becoming overfished, i.e. $B<B_{MSY}$ (Melnychuk et al., 2020, p.13), and at equilibrium, a stock is overfished if and only if it’s subject to overfishing. I also expect reductions in fishing pressure to increase catch if overfished and overfishing, even if not at equilibrium. This seems relatively likely in open-access fisheries, i.e. fisheries without current policies limiting fishing pressure or future policies that would be put in place in response to changing stock status. Such policies tend to reduce the responsiveness of catch to price and demand shifts, anyway (St. Jules, 2024).

12. ^{^}

     Peruvian anchoveta are the most wild-caught fish species by tonnage and numbers, representing around 28% of the number of fish wild-caught annually on average (Mood & Brooke, 2024).

13. ^{^}

     China is the country that catches the most by tonnage, representing around 10% of global catch (Pauly et al./ Sea Around Us).