TL;DR: During a research stint at AIM the idea of a policy charity advocating for a ban or limitation of fish stocking in Canada was evaluated. Fish stocking is the practice of releasing cultured fish into natural water bodies to maintain or enhance fish populations. The idea was dropped as the probability of policy success was estimated to be merely 5% and the uncertainty is large, including doing harm with the intervention. These major uncertainties concerned key factors like fish suffering levels, re-catch rates of the stocked fish, and effects on wild populations, making it too risky to pursue.

Epistemic status

After researching this idea for about 100 hours, I am fairly certain that it is not worthwhile to pursue advocacy efforts against Salmon stocking in Canada. Do note that this is dependent on a bunch of subjective judgements, and one's risk appetite.  This research was done as part of the AIM Research Program, so keep that in mind. Whilst the findings were sense checked by the AIM research team, I have limited research experience of this type. Lastly, certain findings likely shift whenever different fish-species, fish stocking amounts and practises, geographical contexts or political contexts are considered^[1]. This might also change the overall conclusion, i.e. it might be a worthwhile pursuit in other contexts

Credits

Attention to this intervention came via a country report from Animal Ask on potential effective interventions in Canada. Noteworthy progress on this topic from an animal welfare perspective was first made by Rethink Priorities in 2019. I would like to thank those authors for bringing this topic to the forefront in the way they did. I would like to thank Koen van Pelt and George Bridgewater for their feedback on an earlier draft of this post. Special appreciation to Vicky Cox for excellent guidance, mentorship and frequent back and forth during the research program. Mistakes are my own.

Other introductory notes

The majority of this post was initially written for AIM, so it may look similar to a typical AIM report in both writing style and structure. This also means that in many instances I focused on decision relevant matters, instead of aiming to be fully complete. There remain many gaps in this write up, which could benefit from further research.  

Feel free to reach out for any additional explanation or background materials, whenever useful!

Figure 1. Fish stocked using a truck^[2]

Executive summary

To improve the amount of fish stocks in open waters in favor of commercial and recreational fishing, fish are being hatched and raised in hatchery environments. This practice is done around the world and affects tens of billions of fish.

The uncertainty in the CEA is very large, including a chance of having a net-negative effect. Additionally, we expect the probability of success^[3] of this idea to be low (5%), which is low even within the policy change reference class.

Main arguments in favour of this intervention

• It seems very likely that fish suffer during their time in the hatchery environment^[4], where densities are high and the environment is not natural.

• There are valid environmental and economic arguments against fish stocking which could be used to convince officials to ban the practice. Namely, fish stocking can reduce genetic diversity and fitness of wild populations while potentially harming native species and disrupting aquatic ecosystems. The economic argument hinges on the fact that the practice heavily relies on public funding that significantly exceeds commercial returns, effectively operating as an ineffective government subsidy. Being able to stay away from any animal welfare arguments by using these environmental and economic arguments instead likely improves tractability.

• One might worry about the substitution effect where consumers will increase their consumption of farmed fish rather than wild fish if the availability of wild fish decreases due to a ban on fish stocking. After careful consideration, this doesn’t weigh up against the downsides of fish stocking. This is mainly so as the recatch rates of stocked fish are so low and the survival rates of farmed fish are high. This means that one needs to stock many fishes (we used a point estimate of 16 in our model^[5]) to get one fish on a consumer's plate, whereas with farmed fish this number is relatively close to 1.^[6] 

Key concerns with this idea

• The main reason that we dropped this idea is because of its modelled cost-effectiveness. We estimate a cost-effectiveness^[7] of 18 - 35 SADs/$, which is around the bar of ~30 SADs/$. Meanwhile, the uncertainty of this model is very large. More specifically, after modeling the CEA in Guesstimate with current best guess assumptions, there is a chance that the intervention is net-negative. We have largely been unable to decrease our uncertainty for many model inputs:

  • Time spent in hatchery before release and how much they suffer from being in the hatchery. This feeds into the estimate of the amount of suffering during the hatchery phase.

  • Re-catch rates of stocked fish. This is used to estimate the substitution effect which is a key negative externality of this intervention.

  • As juveniles, the fish in hatcheries may have lower sentience or a lower likelihood of being sentient.

  • We also expect there to be some wild animal welfare effects of this intervention (as stocked fish impact wild fish and other wild animals). For example, studies have shown that these effects include predation, food and habitat competition and disease spread. Typical for wild animal welfare considerations, thinking through the effect gets tricky very quickly, and involves arguments both in favor and against fish stocking. We found these effects to be so uncertain that we didn’t even try to incorporate these in our model.^[8] 

There was limited information available for many of these key parameters.

We also note that we expect many of these parameters to change substantially between different hatcheries, species, years or stocking batches. This means there might be substantial variation in its effects on animal welfare between various fish stocking instances. Even if we were able to identify specific combinations of hatcheries/species/years/batches where this intervention looks uncontroversially net-positive from an animal welfare perspective, it doesn’t seem realistic that one is able to target one’s policy efforts. Therefore, one needs a robust case that such a ban would be net-positive overall.  

• Secondly, we estimated that the probability of success is about 5%. Even for a charity working on policy change, this is relatively low, i.e. we think that the problem is not very tractable. There are many reasons to believe that progress on this issue will be difficult:
  • There is a clear incentive for the fishing industry to keep the practice.
  • Various other non-industry stakeholders are in favor of the practice. This includes government officials, first nations, a big and relevant NGO working to improve wild salmon welfare in Canada with ties to the government.
  • There is just one example around the world of an NGO working on this issue and they do not seem to have made progress of the scale that we imagine here
  • A civil servant working at the relevant ministry in Canada that we spoke with said that our 5% chance of success was optimistic

As the number of fish involved is large and the CEA is very sensitive to some key parameters, it still seems possible that there might be countries or specific subcases where the CEA does turn out to be more robust and substantially above our bar, with a low (or no) chance of being net-negative. To be able to determine this, we would need to get hold of better information and data on some of the key parameters, like time spent in the hatcheries and the re-catch rates of stocked fish. 
 

1     Background

1.1        Introduction to fish stocking

Fish stocking^[9] is the practice of raising fish in hatcheries and releasing them into bodies of water in the wild. Fish stocking is a very common practice which is done around the world with 35-150 billion^[10] fishes stocked annually globally (Simcikas, 2019). It is done for three main purposes, listed from most to least common:

1. To support commercial fisheries and increase the catch of “wild” fish
2. To increase the number of fish available to be caught by recreational/sports/game anglers
3. To restore a population of threatened or endangered species, ie. conservation

Fish stocking is often implemented by the government (eg. in Canada), but in some countries the local angler clubs are the ones in charge (eg. in Germany). Undocumented and/or illegal stocking in the wild by individuals or stocking of fish in private waters falls outside of the scope of this report, and is expected to be much smaller in size.

What fish stocking looks like in practice may differ substantially from place to place, due to factors including the reasons for stocking, the type of species stocked, the amount of fish stocked, the lifecycle stage or age of the stocked fish, the environment in which the fish are hatched and released, and relevant economic circumstances^[11]. It is therefore hard to say anything definite about the practice, particularly as no database was found compiling information on fish stocking around the world. However, there are a few elements of stocking that generally seem to be prevalent in most cases:

1. Fish are raised in hatcheries before being released into the wild. They can be released anywhere from just a couple of weeks old up to over one year of age. Data on hatcheries in Canada show that this wide range is indeed happening even in the same hatchery and season.

2. Fish are transported from the hatcheries and released into the wild using a number of different methods, all of which seem stressful. Based on various grey literature (hatchery website, blogs, Youtube) these methods include designated trucks equipped for this practice which use tubes for the release, using buckets and simply throwing the fish into the water and dropping the fish from planes or helicopters^[12].

3. Many fish die after being stocked in the wild. This can be deduced from the fact that re-catch rates are low (in Section 5 we estimate this to be 0.7-20%) and the fact that the population is not increasing.

Moreover, some specific subtopics of fish stocking are surprisingly well-researched (eg. How size and timing of hatchery releases influence juvenile-to-adult survival rates, S. James) and well-documented (eg. detailed data on fish stocking in Canada). This availability of information might be due to the fact that it is a government funded program, which also funds research into the matter.

1.2        Fish welfare concerns

While there are some gaps in the Salmon welfare research field (L. Gaffney et al., 2022), we expect that fish raised in hatcheries have a net-negative life during their time in the hatcheries. We expect them to live this part of their life similar to fish that are hatched to be farmed instead of released. It involves high stocking densities, diseases, and barren environments where fish cannot show natural behavior (FWI, Norwegian Veterinary Institute). An impression of the stocking densities can be found on Youtube, although we didn’t find recent footage. While the claim “If salmon have less than a bathtub’s worth of water each, they begin to suffer due to the lack of available oxygen in the water” from FWI’s report seems to be about adults, it does point towards the problem of too high densities. The report also discusses that high stocking densities lead to higher levels of disease, and increased inter-fish aggression, which is confirmed by V. Braithwaite et al, 2011. The report from the Norwegian Veterinary Institute states that “Since mortality varies greatly between hatcheries, it indicates that many hatcheries have room for improvement.” It also describes the occurrence of various diseases, such as Haemorrhagic smolt syndrome (HSS - the bleeding of organs when salmon natures), wounds and Infectious pancreatic necrosis (IPN - virus disease). These specific datapoints substantiate the general intuition that life in hatcheries means suffering for (young) salmon.

Next to this, both the production of new fish in hatcheries by forced egg and sperm production, as well as the transport and release of stocked fish are very likely to be stressful events. However, these events seem to be short-lived and are therefore of limited overall concern. There may also be some “hidden” suffering in this hatchery stage as the feed of the fish might include other fish, thus involving further suffering. It is unclear what the size of this feed fish suffering is, and we omitted this from our evaluation.

One of the major concerns of fish stocking is that many fish die soon after being stocked. This means that many fish experience a net-negative life in the hatchery, after which they die in the wild after all. This means that fish stocking involves wasting the lives of many fish. We haven’t done research into the reasons why mortality is so high after stocking. It might be explained by the fact that natural mortality rates of wild fish might be somewhat similar, which is plausible looking at the amount of eggs per adult. Additionally, one could conjecture that stocked fish might not be as fit for the wild as wild-born fish. They spend the start of their lives in a contained and safe environment where they do not develop survival skills, e.g., not learning how to feed in the wild. 

For those fish that do not die soon after they are released, it is unclear to us whether they live net-positive or net-negative lives in the wild. However, our best guess is that they live better lives in the wild than they would if they had lived their whole lives on a farm. There is a high amount of uncertainty here which means that the overall welfare of stocked fish is unclear.

It is also worth noting that fish are still relatively young when they live in the hatcheries, which might make them less likely to be sentient (eg. as with insects). This adds another level of uncertainty to the overall welfare of stocked fish.

By banning or limiting fish stocking we can avert all of these fish welfare concerns and reduce suffering.

1.3        Substitution effect

Since the most common use of fish stocking is to support commercial fisheries that aim to catch wild fish, banning or limiting the practice may have unintended negative consequences: one expects consumers to switch to farmed fish. This is a common concern with animal interventions. It looks salient in this situation as well: whenever wild fish stocks are low, it seems very likely (85-95%) that banning or limiting fish stocking will decrease the amount of total catched wild fish^[13]. This will in turn drive the prices of wild fish up, which will then increase the consumption of relevant substitutes: most notably farmed fish.

In our best guess point estimates, we think that the increase in suffering of the farmed fish is smaller than the suffering of the stocked fish. The main driver of this is that the recatch rates of stocked fish are so low. Various papers mention re-catch rates of stocked fish between 0,7% and 10%(Hvidsten et al, E. Jokikokko et al), though this might vary even more depending on the details of implementation. In other words, tens or even more than one hundred fish need to be stocked to be able to catch just one of them. This means that many fish are “wasted”, i.e. many lives go into just one fish that will be caught and eaten. This contrasts to the farming of fish where there is a lot less “waste”. 

On the other hand, it is important to note that we expect farmed fish to live a worse life than stocked fish. In the case of farmed fish, the fish spend additional time in a farmed environment, in addition to the time in a hatchery. We also expect additional causes of suffering for farmed fish. The farmed fish might suffer from diseases like sea lice and they could be hatched as a triploid (three sets of chromosomes) which can cause cataracts and blindness, hunger due to jaw problems, and skeletal deformities. This argument points into the direction against a potential ban.

1.4        Wild animal welfare

As stocked fish are released into the wild, this practice has an impact on other animals already living in those environments. This impact can take various forms and gets complex rather quickly. It is therefore unclear what the overall net-welfare effects are of this practice to all other animals affected. Note that there is substantial uncertainty here, both about the direction and size of the welfare-effect relative to the welfare of the stocked fish.

Still, it is useful to briefly unpack what some of these impacts could look like. Impacts of stocked fish on the welfare of other animals can go via means of food, habitat and disease. First of all, note that stocked fish are often predators themselves. While often being released when they are still small, that means that the stocking has a direct impact on the food web. Stocked fish can both be eaten by larger fish, or eat smaller fish, either intra- or interspecies. Depending on what these smaller fish eat, the declined population of that smaller species might have second order effects as well. One could conceptualize fish stocking as being a shock to the food system, after which it needs to find its equilibrium again (Uusi-Heikkila, 2018), which might have a net-negative effect (MichaelStJules, 2020). Other impacts could be due to habitat competition (Pacific Salmon Foundation) or just altering habitats, eg. causing more sand to be stirred which decreases the available sunlight in the water (K. Lorenzen, 2010). Finally, one other way that fish stocking can impact the lives of wild fish is through the spread of diseases (Ingram and De Silva, 2015).

1.5        Ecological concerns

The impact that fish stocking has on the environment would be a central argument for the charity, as this is much more widely accepted as a valid argument than being concerned about the welfare of the stocked fish.

First of all, the aforementioned effects on wild animals in the previous section could all be considered environmental concerns. Furthermore, the fish stocking is driving down the genetic diversity of the stocked species, as a limited genetic pool of eggs and sperms is used to produce the fish. This in turn impacts the fitness of the species to the environment. A systematic review from 2010 (Araki et al) found that 23 out of 70 studies showed significantly negative effects of hatchery rearing on the fitness of stocked fish. This might in turn affect changes in the genetic composition of wild populations and their fitness (S. Kitada, 2018). Lowered fitness might afterall lead to lowered and less resistant populations.

Finally, there might be other effects on the non-animal environment. For example, effects on the types or amounts of plants in the water and the water quality. We do not consider this relevant in and of itself from a welfare perspective, though it might be a valid argument from an ecological perspective.

Overall, some of these ecological systems are hard to study, and often the consequences of fish stocking to the environment are unknown or not properly assessed. There are many accounts of scientists warning about these consequences (e.g. a meta-analysis by K. Korsu et al, 2010, another meta-analysis by S. Einum et al, 2001). Note that, on the other hand, it is less likely that positive environmental consequences of this practice is studied, which we also didn’t search for.

Relevant here is also that there is a so-called responsible and ecologically sustainable approach to fish stocking (eg. as introduced in 1995 and updated in 2010). This might mean that at least in some cases attention is given to the ecological effects, although it seems likely that this attention is not adequate or sufficient.

1.6        Economic viability of fish stocking

We have done limited research on the economic picture of fish stocking, though it seems likely that there is substantial variance^[14] in economic viability. Animal Ask’s country report on Canada considers economic viability and it concludes that it is unclear whether economic returns are larger than the investments made by the government. However, it is clear that public investments are substantial compared to the commercial returns, and thus one could think of the program as a - potentially inefficient - subsidy by the government to the fishing industry/recreational fishers.

2     Theories of change

The following Theory of Change (TOC) shows how by lobbying, researching and networking with other NGOs, a new organization could convince the government to ban or limit fish stocking, which would eventually lead to improved animal welfare.

2.1   Assumptions

Key uncertainty

• We think the probability of success is relatively low for this charity. Even when the relevant prior outputs are realized (steps 4, 5 and 6), it doesn’t seem likely that the relevant policy will be changed/introduced. That is, among other reasons, because one would expect the pushback to be substantial and potentially effective in preventing progress on the issue.

High uncertainty

• The effect of fewer fish stocked on the overall welfare of fish. As will be explained in more detail in the CEA in Section 5, there is substantial uncertainty around the effects and sizes of the substitution effect and the effect on wild animals already living in the stocked waters.  
• It is unclear that one is able to convince someone in the relevant governmental body up to such an extent that they are able and willing to push for the cause themselves as well, i.e. will there be a champion in the relevant governmental body. Elected officials might be tied to their parties’ stances and it might be hard to make this a priority for the politician.

Some uncertainty

• It is unclear whether one is able to influence the stance of other relevant NGOs. Potentially the topic is not as neglected as it might seem at first sight, and NGOs might have already formed strong opinions on the matter. For example, relatively big and relevant NGOs working to protect wild salmon in Canada (the Atlantic and Pacific Salmon Foundations) are in favor of fish stocking.

Low uncertainty

• If policies are introduced, the fish stocking will actually go down. As the practice is implemented by the government, it seems very likely that this is enforced as well.

3     Geographic assessment

3.1        Where existing organizations work

Work on this topic is very neglected. We have only been able to find one current and one historical example of work to ban fish stocking. Only one of these is motivated by animal welfare concerns:

• The Center for Biological Diversity in the US tried in 2010 to alter the practise of fish stocking via a court ruling, based on ecological reasons. Even after winning this court case, the governmental body pretty much dodged the ruling by not accurately acknowledging the ecological effects.
• An animal advocacy group in Paris called PAZ is working on the topic, trying to ban the practice in France. PAZ has made some progress since 2022: initiating National Assembly debate on fish farming VAT, meeting ecology officials, and analyzing fishing subsidies. However, with no dedicated staff, their campaign and results remain modest and not of the scale that we envision for this charity idea.

Based on this, we would rule out France as a potential target country and think that there is much to be done elsewhere.

3.2        Geographic assessment

Our attention came to this intervention via a country report from Animal Ask on potential effective asks in Canada, so Canada seemed like an obvious choice. However, we shallowly considered other options to ensure that we were not locking in this decision too quickly.

Fish stocking is prevalent all across the globe, with many countries raising billions of fish each year for the purpose. Estimates for the number of fish stocked by each country can be seen in Table 1 below.

Table 1: Estimates for number of fish stocked annually around the world (Šimčikas, 2019)

To narrow the scope, we consider the countries with a good fit for potential founders: this considers where the average participant of AIM’s Charity Entrepreneurship Incubation Program is from. We expect that it will be easier to lobby in your own country or a country that is culturally-similar to your own. This led us to rather quickly limit ourselves to the US, Canada, the UK and Europe.

• The US: ruled out for tractability concerns. Fish stocking programmes are directly funded by recreational fishing licenses creating strong buy-in from recreational anglers who we would expect to be a strong lobby against this work as recreational fishing is an important part of the US culture.
• Canada: Although the size of the industry is smaller, work here looks preliminarily promising based on Animal Ask’s report. Moreover, as we were scoping countries it became clear that the availability of information on the practice of fish stocking in Canada was unusually high. There is relatively detailed information and datasets available on some aspects of their fish stocking programs, including what species are stocked and the stage/age at the time of stocking.
• The UK: Ruled out based on scale. The size of the practice is very limited (< 0.1 million), not making it a worthwhile intervention.
• Europe: Ruled out based on concerns about scale, tractability, and availability of information.
  • Šimčikas 2019 estimates all of Europe to stock 14 billion fish, which we think could be an overestimate. This original estimate was based on a paper from 2000, as described in J. Cooke et al (pdf page 9). Based on our ~6 hours of desk research to break down this total number by country, we found information on some countries (Norway, Lithuania, Germany, Netherlands, Switzerland) that suggested that the numbers were much lower than one would expect solely based on this European grand total. We therefore expect the grand total of Europe to be more around the order of ~1-10 billion.
  • In Germany the practice is implemented by angler clubs and is thus out of scope for this version of the charity idea (and also likely to be not very tractable).      
  • The only other country where relevant information was found was Switzerland (see dataset here), which stocked about ~300 M fish in 2021, of which the biggest part was whitefish. Note that the majority of these stocked fish are fry (i.e. a very young life stage that starts when the fish can feed themselves as per Wikipedia). As further information on Switzerland wasn´t as readily available as Canada, we dropped it, though it might also be promising.

Based on this analysis, we think that Canada is the most promising target country for a new charity to advocate for a ban on fish stocking.

4   Quality of evidence

4.1   Evidence that a charity can effect change in this space

It has proven hard to find a suitable reference class for the type of change we are interested in. We have therefore relied on our subjective judgment, informed by arguments in favor or against possible change in the space of fish stocking policies. Overall, we think that it is hard to make change in this space, and set the probability of success at 5%.

Arguments in favour of policy change being possible

• There are environmental/ecological impacts of fish stocking, which might be an effective argument against fish stocking. In some cases these impacts can be severe. We expect environmental/ecological impacts to be much more compelling than animal welfare arguments.

• Depending on a few factors, such as the re-catch rates of stocked fish, you could make the argument that fish stocking is not economically feasible. It could be the case that the costs of fish stocking is larger than the profits from the additional fish catched by the industry, essentially wasting public resources. As in many instances the costs are at least larger than the related tax revenue, fish stocking is a de facto subsidy of the government to the fishing industry. The public might be against such financial support to the industry. Again, we expect that this argument would be much more compelling^[15] than animal welfare arguments.

• Other developed countries have dropped the practise due to environmental concerns (eg. it was common in the UK in the 60s and 70s, but not anymore in the 2000s, and in Germany it is also increasingly prohibited)

Arguments against policy change being possible

• There are clear and immediate financial incentives for the fishing industry to keep fish stocking, as it is a de facto subsidy for their industry. We expect them to be a strong lobby against this intervention.
• We also expect recreational anglers to be a strong lobby against this intervention as well, as they also benefit from the stocking practises.
• Although there no other organizations working on this issue from a welfare perspective, we expect there to be many organizations that have opinions on and stakes in this issue
  • In Canada, the government spends a lot of money on the Salmon Enhancement Program. Therefore, it seems to be something that they make an active choice in. Also, the hatcheries are being modernized, from which can be inferred that the government expects the practice to be around for the foreseeable future.
  • The Atlantic and Pacific Salmon Foundations (ASF and PSF) who are focused on protecting wild salmon are in favor of fish stocking, as described here. PSF has a ~CAD$20 million annual budget, and does regranting work as well. PSF seems to be pretty closely tied and working together with the relevant ministry in Canada.
• There are only very few examples of other NGOs working on fish stocking that could function as a case study or set a precedent.
• The FAO has meetings on the topic and seems to be in favor of implementing it in developing countries, as is shown for example here in 2015 in Asia.
• Given that there is something called the responsible and ecological approach, we make expect that the opposition (eg. the fishing industry) could easily dodge an out-right ban or limitation by coming up with promises that they will improve. Because of this, the environmental argument might not be the best argument to use. This could lead to alterations being made to the practice instead of an outright ban on the practice.
• Given the fact that the actual effect of fish stocking on the environment is quite hard to know, it seems relatively easy to use this uncertainty in favor of lobbying to defend the practice.

• Fish stocking has been done already for a long time^[16] so it is a fairly entrenched practice that may be hard to change.

4.2   Evidence that the change has the expected welfare effects

By banning fish stocking we will avert the suffering of stocked fish which we believe live net-negative lives as detailed in Section 1.2. More detailed information on what the lives of fishes in hatcheries is like might be found in a report on Salmon Welfare by Fish Welfare Initiative or in Section 1.4 of a previous AIM report on farmed fish welfare.

5     Cost-effectiveness analysis

The cost-effectiveness of this intervention was modeled using Guesstimate, as multiple inputs had substantial uncertainties. With the assumptions outlined below, we found a central estimate of the cost-effectiveness to be ~35 SADs/$. This is just above the aimed bar of ~30 SADs/$. The uncertainty is substantial though, as is seen in how far apart the percentiles lie:

Table 2. Cost effectiveness for various percentiles

There is one issue with interpreting these results naively: there is a correlation between key parameters, namely the time spent in the hatcheries and the recatch rates, which guesstimate does not allow us to model. The correlation works like this: the longer the fish are in the hatcheries for, the higher the chances they have of survival (eg. as mentioned in E. James et al) and thus there is a higher recatch rate. The way to think about correcting for this is to trim the outliers from both ends of the distribution and give more weight to the median values. We think that such a corrected distribution would have two effects that are relevant here. One, it skews the overall average up: as the positive values are even more extreme than the negative ones. Therefore, we may expect the actual cost-effectiveness to be closer to the median estimate of ~18 SADs/$, which is not anymore above our bar. Two, the 30% chance of making a negative impact that follows from table 1 is too pessimistic, this number must be lower. We are unsure what it should be instead, and didn’t prioritize formalizing this as the case already didn’t look good without such a formalization. This is also further explained this footnote^[17]. We therefore conclude the intervention to have a cost-effectiveness of 18-35 SADs/$ and that there is a chance, of unknown size but smaller than 30%, that - given the current best guess assumptions detailed below - this intervention is net-negative.  

5.1   Costs

We think that one could run this policy advocacy campaign with a moderately lean team. We model fixed costs of $130,000 in Year 1 and $280,000 at scale. These fixed costs are held constant across all CEAs of AIM. 

Additionally, we expect one would need to hire a few lobbyists, which we modeled using 5 times the median Canadian salary of ~$52,000 (~$262,000 total) and an additional overhead of $50,000.

At scale this charity would cost a total of ~$590,000 per year.

We modeled the costs to last for 4 years in total, irrespective of whether the policy change was achieved or not. We used a total cost of $1.1 to 2.6 million for all 4 years of operating in our Guesstimate model, which is plus and minus 40% of our central estimate.

5.2   Effects

This CEA focuses specifically on salmon, as it was the most common fish species stocked^[18] and had the most information available^[19]. As a measure for the welfare of the Salmon, we used the SADs system from AIM. In this measurement system a SAD represents a day of intense pain felt by animals, with adjustments for pain intensity, sentience, and welfare range, where 1 SAD equals one day of disabling pain for a human. The only relevance for the effects here is that we estimate that a fish stocking salmon suffers 12.8 SADs per year in a hatchery environment and  farmed salmon suffers 31.6 SADs per year in a farmed environment. Part of the  difference here is due to the fact that we do not expect stocked fish to suffer from sea lice or being hatched as a triploid (cataracts and blindness, hunger due to jaw problems, and skeletal deformities) where farmed fish would. Farmed fish also spend longer in a farmed environment than stocked fish, which we assume is net-negative. For questions on the SADs system or any of the above estimates, Vicky Cox from AIM can be solicited.

While we can quantify suffering in the hatcheries, we excluded the welfare impacts on stocked fish after release into the wild from our analysis, as we are unsure whether this is positive or negative.

We also model the substitution effect by subtracting the increased suffering that may result from an increase in the number of farmed fish from the overall welfare impact of banning fish stocking.

We also include the probability of success, the expected amount of years of impact, and some relevant discounts.

In our model, we only focus on the release of salmon from a certain age upwards: smolts which are young-adult salmon, who can swim in salt water. This means that we excluded the release of eggs, alvea and fry (all stages before smolts) from our calculations. Although the numbers of fish released at this stage are higher, we think that the overall impact is lower because of the shorter stay in the hatchery and an additional discount for potential lower sentience capacity. We also have less information on this type of release.  

Table 3. Amount of fish stocked for various categories

We have made the following assumptions and choices in modeling this intervention. If ranges are given, they represent a 90% confidence interval:

• Welfare effects on stocked salmon
  • 12.8 SADs per year for salmon in a hatchery, as explained before. This amount was fixed in the Guesstimate model.
  • The time that a stocked fish is reared in the hatchery varies greatly. Smolts stay in the hatcheries for a few months up to two years, we modeled this to be between 0.2 to 2 years. This is based on data from the Canadian Salmon Enhancement Programme and information on the time spent at various life stages.
  • We have applied a discount of 0 to 60% for the fact that the fish are juveniles when they are living in the hatcheries, and they might have less capacity for sentience at this stage. This is informed by our conversation with Wild Animal Initiative and priors based on research on insect juveniles.
• Substitution effect
  • We assume a 1:1 substitution effect, i.e. for every kg fewer wild salmon produced, there will be a kg increase in the number of farmed salmon. This includes simple assumptions on preferences and consumer behavior, although there is already a price difference between farmed and wild salmon (the former being cheaper). We also assume that the weights of farmed and wild caught salmon are the same, i.e. they produce the same amount of kg per animal lived.
  • We assumed a re-catch rate of 0.7% to 20%, with a point estimate of 6%. This is a relatively large range, mostly because information on this was limited. In Japan there is a re-catch rate of 2.7% for Chum Salmon (Kitada), in Sweden there are re-catch rates of 2.5% for Atlantic salmon (Peterson), and re-catch rates in a river in Finland are between <1% and 10% (Jokikokko).
  • Mortality rates of farmed fish are between 8% and 20%, with a point estimate of 13%. This is mostly based on a website on farmed salmon die off in Canada.
  • Farmed fish spend around 1.1 years in the hatcheries, and 1.8 years in the grow out phase in sea pens. During their time in the hatcheries , we modeled the same amount of SADs as the stocked fish, but during their time in the sea pen there is a higher welfare burden of 31.6 SADs per year.
• Inputs and discounts
  • We modeled the amount of released fish based on the Salmon Enhancement Programme (SEP) of Canada (data from 2023 here). One finds that there are 45 million smolts released for harvesting purposes.
  • We have given the charity a probability of success of 5%. See Section 4.1 for arguments behind this estimate.  We put the range between 1% and 13%.
  • We modeled that we might not have a complete ban, but a partial ban that only limits the fish stocking. We decided to model this to be a reduction between 70% and 100%, with a point estimate of 85%
  • We have used a yearly 4% time-discount for both the costs side and the animal welfare side. We assumed that it would take 4 years to accomplish the policy change, and that counterfactual benefits would last for 21 years thereafter (until 2045). We assume that one would stop after 4 years if unsuccessful, and continue until year 6 if one is successful.  With some quick calculations on net present value, this translates to a grand total of 0.75 time discount for the SAD’s and a grand total of 0.93 time discount for the costs.
  • As fish stocking has been around for a long time, and there is no sign of any change, we didn’t apply a discount for the fact that the policy change might happen anyway before 2045, i.e. the counterfactual.
• Other assumptions
  • We haven’t discounted for enforcement, as we expect that a new policy would be effectively enforced as the hatcheries are government led/owned.
  • As mentioned before, we haven’t modeled the effects of fish stocking on wild animals whatsoever. This is a large unknown with unknown direction, and should be kept in mind when judging the overall CEA of this charity.
  • We did not incorporate the welfare implications of fish feed. Note that both fish in the hatcheries as well as farmed fish are being fed other fish and potentially other animal products. The farmed fish may need to be fed for a longer period of time.

Appendix 1 - Interview with Canadian civil servant from the relevant ministry

To get a better understanding of what the work for such a charity would look like, we spoke with someone working in the Department of Fisheries and Oceans (DFO) of Canada, i.e. the relevant ministry. This person was knowledgeable about effects and interactions between farmed and wild salmon, and was decently informed about fish stocking specifically.

He confirmed that the DFO is the relevant political body if one would want to bring about change in this space. This country-level body is the one in power, rather than the state level^[20]. He described the department to be rather divided between various incentives, most prominently divided over being mostly concerned with wild fish and conservation versus being mostly concerned with the needs and wants of commercial fisheries. This division was tangible within the department, and the consensus generally favoured the commercial fisheries.

The person we spoke with thinks that the welfare and ecological issues raised are real indeed. For example, he agrees with the ecological arguments such as the increase in the amount of disease, the loss of genetic diversity and the increased competition for food/habitat.

Still, this person thinks that the practice of fish stocking, e.g. the Salmon Enhancement Programme(SEP), is unlikely to stop or change significantly anytime soon. Moreover, he thinks that it would be very unlikely that lobbying efforts would bring about a ban or substantial limitation to the project. This person considers our estimate of the probability of success of 5% to be high. The following arguments argue for this unlikelihood of change in this space:

• This person expects the opposition to be very strong and diverse.
  • The industry is well organized and has substantial financial means.
  • Many government officials would be against a change

  • Other NGO’s like Pacific Salmon Foundation (PSF) are doing work in favor of wild salmon and trying to increase its population. They would likely be against a ban^[21].

  • First nations are in favor of stocking.
  • There would be very few allies on this front, potentially only some conservationists.
• This person expects that the susceptibility of the department for the environmental argument would be very limited.
  • Financial and economic interests are very strong. Canada is in many ways an extracting economy: logging and mining are practices of taking from nature. Fishing is an extracting industry as well. Such types of economic activities are substantial parts of the economy and the paradigm of the foreseeable future is that the government encourages these practices.
  • Canada is relatively less conservation focussed than the US.
  • The only signs that the department is indeed caring about conservation are the various research projects that they fund in the conservation space. Though, the results of these research efforts often are ignored: although the effects of diseases spread because of fish stocking is well known within the department, it does not lead to any change in practice.
• This person thinks that animal welfare arguments would not be effective at all in any political or lobbying pursuits.

One way to interpret this, is to conclude that the space is actually not very neglected. Indeed, there is already substantial discussion and money flowing in and around the fish stocking program and quite some stakeholders are engaged with the matter.

--

We conducted a couple other interviews on the matter, though decided to incorporate these within the main text rather than laying out a separate interview summary here.

Reference list

Animal ask (2025), Animal Advocacy in Canada, https://www.animalask.org/post/animal-advocacy-in-canada

S. Simcikas, Rethink Priorities (2019), 35-150 billion fish are raised in captivity to be released into the wild every year https://rethinkpriorities.org/research-area/fish-raised-in-capacity-and-released-wild/ including it’s Guesstimate https://www.getguesstimate.com/models/12854 and forum post with the same text https://forum.effectivealtruism.org/posts/4FSANaX3GvKHnTgbw/35-150-billion-fish-are-raised-in-captivity-to-be-released#References

J. Cooke, I. Cowx (2005), Contrasting recreational and commercial fishing: Searching for common issues to promote unified conservation of fisheries resources and aquatic environments, https://www3.carleton.ca/fecpl/pdfs/Cooke%20and%20Cowx%20Biol%20Cons%202006.pdf

S. James et al (2003), Size and timing of hatchery releases influence juvenile-to-adult survival rates of British Columbia Chinook (Oncorhynchus tshawytscha) and coho (Oncorhynchus kisutch) salmon https://cdnsciencepub.com/doi/10.1139/cjfas-2022-0121

Salmonid Enhancement Program (SEP) Production Plan Archive - 2023 Production Plan (English), (cited: october/november 2024) https://open.canada.ca/data/en/dataset/7cb82f30-3bdf-4abf-9d1b-63be6a6f9f7a/resource/c7990626-84c1-4dc5-865a-6b79329716b2

L. Gaffney et al., (2022), Research Before Policy: Identifying Gaps in Salmonid Welfare Research That Require Further Study to Inform Evidence-Based Aquaculture Guidelines in Canada https://www.frontiersin.org/journals/veterinary-science/articles/10.3389/fvets.2021.768558/full

Mark Borthwick, Fish Welfare Initiative (2020), Welfare Issues in Farmed Atlantic Salmon https://www.fishwelfareinitiative.org/salmon-welfare

Norwegian Veterinary institute, (2019) Animal welfare in fish hatcheries, https://www.vetinst.no/rapporter-og-publikasjoner/rapporter/2019/animal-welfare-in-fish-hatcheries-smafiskvel/_/attachment/download/d95b1229-378e-4d51-b25d-5a27785bc80a:50a3f437d19609cee0bfc4323b1fbfbaba10e4bd/Animal-welfare-in-fish-hatcheries-SMAFISKVEL.pdf

Youtube (2013), Salmon And Trout Hatchery Documentary https://youtu.be/8x810mszCfU?si=cU4h8d512KJaEq3I&t=373

V. Braithwaite et al. (2011), Welfare Issues Related to Fish Stocking https://www.researchgate.net/publication/267894881_Welfare_Issues_Related_to_Fish_Stocking

Bob Fischer (2023), If Adult Insects Matter, How Much Do Juveniles Matter?, https://forum.effectivealtruism.org/posts/E7xdBbxqPNLjhrnz6/if-adult-insects-matter-how-much-do-juveniles-matter

Hvidsten et al (2011) Increased Recapture Rate of Adult Atlantic Salmon Released as Smolts into Large Shoals of Wild Smolts in the River Orkla, Norway https://www.tandfonline.com/doi/abs/10.1577/1548-8675%281993%29013%3C0272%3AIRROAA%3E2.3.CO%3B2

E. Jokikokko et al (2006), Effect of origin, sex and sea age of Atlantic salmon on their recapture rate after river ascent, https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1439-0426.2006.00747.x

Uusi-Heikkila (2018), Species’ ecological functionality alters the outcome of fish stocking success predicted by a food-web model https://royalsocietypublishing.org/doi/pdf/10.1098/rsos.180465

MichaelStJules (2020), Changes in conditions are a priori bad for average animal welfare https://forum.effectivealtruism.org/posts/5bmpoXjZKcy74w4ss/changes-in-conditions-are-a-priori-bad-for-average-animal

Pacific Salmon Foundation (cited: october/november 2024)  https://psf.ca/salmon/#hatcheries

K. Lorenzen et al. (2010). Responsible Approach to Marine Stock Enhancement : An Update https://www.stockenhancement.org/pdf/Responsible_Approach_Update_2010.pdf

Ingram and De Silva (2015). General aspects of stock enhancement in fisheries developments https://enaca.org/?id=289

Araki et al. (2010), Is hatchery stocking a help or harm? Evidence, limitations and future directions in ecological and genetic surveys https://www.researchgate.net/publication/222532279_Is_hatchery_stocking_a_help_or_harm_Evidence_limitations_and_future_directions_in_ecological_and_genetic_surveys

S. Kitada (2018), Economic, ecological and genetic impacts of marine stock enhancement and sea ranching: A systematic review https://onlinelibrary.wiley.com/doi/10.1111/faf.12271 and it’s appendix https://onlinelibrary.wiley.com/action/downloadSupplement?doi=10.1111/faf.12271&file=faf12271-sup-0002-TableS1-S8.pdf

K. Korsu et al (2010) Impacts of invasive stream salmonids on native fish: using meta-analysis to summarize four decades of research, https://www.semanticscholar.org/paper/Impacts-of-invasive-stream-salmonids-on-native-to-Korsu-Huusko/9e00c0a02fb4cee7e8ce0064c9e553c6cef30a7b

 S. Einum et al (2001), Implications of Stocking: Ecological Interactions Between Wild and Released Salmonids https://www.semanticscholar.org/paper/Implications-of-Stocking%3A-Ecological-Interactions-Einum-Fleming/6914ebcae29cfbf69750bd87faabe383075bce00

H. Blankenship and K. Leber (1995), A responsible approach to Marine stock enhancement https://www.stockenhancement.org/pdf/responsible_approach_1995.pdf

Center for biological diversity (cited: october/november 2024)  https://www.biologicaldiversity.org/campaigns/fish-stocking_reform/index.html

Zoopolis (cited: october/november 2024)   https://zoopolis.fr/nos-campagnes/les-animaux-ne-sont-pas-des-jouets/enquete-sur-lempoissonnement-dans-le-but-de-pecher/

New Hampshire bulletin (2022) State’s fish hatcheries are about to get a boost; some environmentalists say that’s a problem https://newhampshirebulletin.com/2022/05/11/states-fish-hatcheries-are-about-to-get-a-boost-some-environmentalists-say-thats-a-problem/

UK 2023 fish stocking figures https://www.nidirect.gov.uk/articles/2023-fish-stocking-figures

IGB, (cited: october/november 2024) Hand in Hand für eine nachhaltige Angelfi scherei https://www.igb-berlin.de/sites/default/files/media-files/download-files/IGB-Bericht-28.pdf

Bundesamt für Umwelt (cited: october/november 2024) https://www.fischereistatistik.ch/de/statistics?tt=%204&dt=0&at=0&st=0&dp=0&ar=CH&wt=0&th=0&un=0&in=0&yr%5Bfrom%5D=2000&yr%5Bto%5D=2021&sp=100102

Wikipedia, Juvenile fish, (cited: october/november 2024) https://www.wikiwand.com/en/articles/Juvenile_fish

Institut für Binnenfischerei e.V (2019) Jahresbericht zur Deutschen Binnenfischerei und Binnenaquakultur https://www.bmel-statistik.de/fileadmin/daten/0510100-2019.pdf

P. Pearse (1994), An Assessment of the Salmon Stock Development Program on Canada's Pacific Coast, https://waves-vagues.dfo-mpo.gc.ca/Library/179181.pdf

Hatchery modernization, Fisheries and Oceans Canada (cited: october/november 2024) https://www.pac.dfo-mpo.gc.ca/sep-pmvs/hatcheries-ecloseries/modernization-modernisation-eng.html

FAO, EIFAAC/CACFish Fish Stocking Guidelines Consultation (cited: october/november 2024), https://www.fao.org/fishery/en/meeting/41461

FAO (2015), Responsible stocking and enhancement of inland waters in Asia, https://openknowledge.fao.org/server/api/core/bitstreams/1baeac4d-463a-4b09-956c-3d1d84f1ed52/content

Ministry of Natural Resources, Fish management history (cited: october/november 2024) https://www.ontario.ca/page/fish-management-history

SADs (Suffering-Adjusted Days): An improved animal welfare measurement tool  (cited: october/november 2024) https://www.charityentrepreneurship.com/post/sads-suffering-adjusted-days-an-improved-animal-welfare-measurement-tool

Fisheries and Oceans Canada , Salmonid Enhancement Program (cited: october/november 2024)  https://www.pac.dfo-mpo.gc.ca/sep-pmvs/index-eng.html

Ontario’s fish stocking program, (cited: october/november 2024)  https://www.ontario.ca/page/ontarios-fish-stocking-program

Fish stocking report - 2021 fish stocking report, (cited: october/november 2024)   https://open.canada.ca/data/en/dataset/1bd8313d-afc9-472e-94e7-e2dc73ab82cd/resource/fbe83862-b78c-4319-a946-e5faffcdb4db

Government of Canada and Tŝilhqot’in National Government partner to support Fraser River Chinook salmon with new hatchery, (cited: october/november 2024)

https://www.canada.ca/en/fisheries-oceans/news/2024/12/government-of-canada-and-tsilhqotin-national-government-partner-to-support-fraser-river-chinook-salmon-with-new-hatchery.html

Guardian (2023), Sea-lice outbreak on Icelandic salmon farm a ‘welfare disaster’, footage shows

https://www.theguardian.com/environment/2023/nov/03/sea-lice-outbreak-icelandic-salmon-farm-welfare-disaster-footage-shows

Institute of Marine Research (2024), Challenges and Opportunities with Sterile Salmon https://www.hi.no/en/hi/news/2024/november/challenges-and-opportunities-with-sterile-salmon#:~:text=%E2%80%9CTriploid%20salmon%20are%20more%20prone,2023%20due%20to%20welfare%20challenges.

Fishex, The Fascinating Life of an Alaska Salmon (cited: october/november 2024) https://fishex.com/salmon/alaska/salmon-life-cycles

E. Petersson et al (2013), Fin erosion and injuries in relation to adult recapture rates in cultured smolts of Atlantic salmon and brown trout, https://www.researchgate.net/publication/242019002_Fin_erosion_and_injuries_in_relation_to_adult_recapture_rates_in_cultured_smolts_of_Atlantic_salmon_and_brown_trout

E.Jokikokko et al. (2003) The survival of stocked Atlantic salmon smolts during sea run and the timing of migration in the river Simojoki, northern Finland https://www.sciencedirect.com/science/article/abs/pii/S0044848602005823

Swimdrinkfish.ca  (cited: october/november 2024) 600,000 farmed salmon die of disease in BC annually — with what effect on wild fish and human health?, https://www.swimdrinkfish.ca/fraser-riverkeeper/blog/600000-farmed-salmon-die-of-disease-in-bc-annually-with-what-effect-on-wild-fish-and-human-health#:~:text=Approximately%203%20million%20fish%20died,potentially%20died%20due%20to%20disease%E2%80%A6.

1. ^{^}

    Though I expect that substantial uncertainty will remain in the foreseeable future due to meaningful interaction with ecosystems and the wild animals it inhabits

2. ^{^}

    Source: Lasco Press

3. ^{^}

    Here, success is considered to be substantial policy progress on the matter. See section 5 on the CEA for more details.

^{^}

 We largely base this on our belief that farmed fish in the grow-out phase (the final phase of their lives before they are slaughtered that is typically the target of other animal welfare interventions) have a net-negative life. It seems like this would also be the case for hatcheries, especially as there is less incentive to reduce mortality in hatchery systems as each individual fish is less economically valuable at this stage.

^{^}

 The 90% confidence interval suggests 5 - 143 fishes, more on this in section 5.

^{^}

 Note that this does not consider any fish that may be fed to farmed fish in their feed.

^{^}

 AIM uses Suffering Adjusted Days (SADs) to compare the suffering of animal welfare. This system calculates the number of days of intense pain experienced by animals, adjusted for factors like pain intensity, sentience, and welfare range.

^{^}

 Do note that we even weren’t able to find a meaningful upper bound to these effects. At least in theory these could be bigger than the direct effects of this practice as modeled in the CEA. Though this is a somewhat risk-averse framing of the fact that we just don’t know.  

^{^}

 Other names for the same practice include fish restocking, fish stock enhancement, marine enhancement and wild sea ranching. It is one tool in the toolbox of the so-called fisheries management.

^{^}

 We haven’t been able to properly assess or update the numbers underlying this grand total, as there is still no global database on fish stocking available (by e.g. FAO) and we didn’t cover all countries in our geographical assessment. The only relevant update that we can report is that Šimčikas 2019 estimates all of Europe to stock 14 billion fish, which seems like it could be an overestimate. This original estimate was based on a paper from 2000, as described in J. Cooke et al (pdf page 9). Based on our ~6 hours of desk research to break down this total number by country, we found information on some countries (eg. Norway, Lithuania, Germany, Netherlands, Switzerland) that suggested that the numbers were much lower than one would expect solely based on this European grand total. We therefore expect the grand total of Europe to be more around the order of ~1-10 billion.

^{^}

 Eg. costs of stocking and size of the fishing industry.

^{^}

 We do not know which of these practices are most common, though we can imagine the truck option to  be the cheapest way to handle large amounts of fish.

^{^}

 Some papers argue that in some instances the fish stocking merely replaces wild populations rather than improving overall fish stock, due to competition for habitat and food. This seems very unlikely though whenever fish stocks are particularly low.

^{^}

 Table 3 in appendix of S. Kitada shows the ratio of net income to the release cost (excluding personnel expenses and expenditure for hatchery facilities) ranging between 0.5 to 9.6. Note that this concerns different species though.

^{^}

 On the other hand, exactly this economic analysis of the program was made in an internal report by the relevant ministry stating that the benefit-cost ratio of 0.6 is disappointing, but this hasn’t led to stopping the program.

^{^}

 With long meaning really long: this source on Ontario suggests it started before 1900

^{^}

 The most optimistic CEA values come from a situation where fish spend a very long time in the hatchery (which means that they have more suffering to avert as we believe the time spent in hatcheries is net-negative) and low recatch rates (which means that you need to stock many fish to get the required output in number of fish caught), which seems to be very unlikely to happen simultaneously due to the aforementioned correlation. This also holds for the pessimistic CEA. As the 5% percentile CEA is closer to the median/average than the 95% percentile CEA (ie. the extremes on the positive side affect the average more than the extremes on the negative sides) we expect that this makes the naive average CEA too optimistic. Clearly also trimming the extreme negatives would give overall lower mass to being negative overall, i.e. the 30% chance of being net-negative is too pessimistic.

^{^}

 Salmon has its own so-called Salmon Enhancement Program (SEP) in Canada. Other stocked fish include trout and whitefish, which turn out to be ~4 million in 2021, which is much smaller than the >300 million from the SEP.

^{^}

 We had an existing SADs estimate for salmon that could be adjusted and the Canadian government published data on the number of salmon being released and at what age/life stage they were released.

^{^}

 As opposed to how it works in the US where states do have more power

^{^}

 Though note that they might also be against the (size of) current wild and farmed fishing. One could imagine them to be in favour of a ban/limitations of fish stocking after wild fishing is greatly reduced and wild stocks are at a substantial level.

Show all footnotes