Citation: Romero Waldhorn, D., & Autric, E. (2022, December 21). Shrimp: The animals most commonly used and killed for food production. https://doi.org/10.31219/osf.io/b8n3t 

Summary

• Decapod crustaceans or, for short, decapods^[1] (e.g., crabs, shrimp, or crayfish) represent a major food source for humans across the globe. If these animals are sentient, the growing decapod production industry likely poses serious welfare concerns for these animals.
• Information about the number of decapods used for food is needed to better assess the scale of this problem and the expected value of helping these animals.
• In this work we estimated the number of shrimp and prawns farmed and killed in a year, given that they seem to be the vast majority of decapods used in the food system.
• We estimated that around:
  • 440 billion (90% subjective confidence interval [SCI]: 300 billion - 620 billion) farmed shrimp are killed per year, which vastly exceeds the figure of the most numerous farmed vertebrates used for food production–namely, fishes and chickens.
  • 230 billion (90% SCI: 150 billion - 370 billion) shrimp are alive on farms at any moment, which surpasses any farmed animal estimate known to date, including farmed insect numbers.
  • 25 trillion (90% SCI: 6.5 trillion - 66 trillion^[2]) wild shrimp are directly slaughtered annually, a figure that represents the vast majority of all animals directly killed by humans out of which food is produced.^[3]
• At this moment, the problem of shrimp production is greater in scale–i.e., number of individuals affected–than the problem of insect farming, fish captures, or the farming of any vertebrate for human consumption. Thus, while the case for shrimp sentience is weaker than that for vertebrates and other decapods, the expected value of helping shrimp and prawns might be higher than the expected value of helping other animals.

Introduction

Recently, Birch et al. (2021) and Crump et al. (2022a) reviewed the evidence of sentience in decapod crustaceans, with a focus on pain experience. Similar to findings previously reported by Waldhorn (2019; see also Waldhorn et al., 2020), these studies concluded that there is substantial, although limited, evidence that decapods might be sentient. Notably, the low strength of current evidence likely corresponds to the little scientific attention various decapod taxa have received–which is particularly the case for shrimp, especially for those belonging to the Penaeidae family (see also Comstock, 2022).

Decapods like crabs, lobsters, and shrimp serve as a major source of human food worldwide, having partly driven the global growth of the aquaculture sector in recent years (de Jong, 2018). Decapod production also represents the fastest-growing major fishery activity worldwide (Boenish et al., 2022). If these animals are sentient, current commercial practices pose serious welfare risks both when decapods are farmed and/or handled during capture, transport and sale, and when they are slaughtered (see Birch et al., 2021 and Crump et al., 2022b). These welfare issues might be particularly pressing given the high numbers of farmed decapods (see Mood & Brooke, 2019a), plus other uncounted individuals captured from the wild. Furthermore, increasing human population, changes in consumer preferences, technological advancements, and income growth suggest that decapod production stands to increase in the future (Boenish et al., 2022; FAO, 2022f), which may, in turn, augment the scale of the problem of decapod welfare.

Currently, only partial data exists about the scope of this issue. Mood & Brooke (2019a) calculated that between 255 billion and 605 billion decapods are farmed every year, the great majority of such individuals being shrimp and prawns–specifically, 213-530 billion, which amounts to 83.5%-87.6% of the total amount of farmed decapods killed annually. However, such estimates do not account for premature deaths, which seem to be a common issue in decapod farming. Indeed, this problem has shaped decapod aquaculture production over the last decades (Shinn et al., 2018; see also Benchmark Insights, 2019, p. 17). If so, the numbers of farmed individuals are likely higher than the estimates currently available. Similarly, no data is known about the number of decapods captured from the wild. While an initial interpretation based on total tonnages suggests that decapod production is dominated by aquaculture, it is unclear to what extent the number of captured individuals is significantly lower than the number of farmed decapods. Taken together, all that information would allow us to properly assess the scale of this animal welfare issue, estimate the expected value of helping these animals, and inform further research into how serious an ethical problem decapod production might pose.

Here, we present estimates of the numbers of shrimp and prawns slaughtered for food, covering both farmed and wild-caught individuals, and based on the most recent data released by the Food and Agriculture Organization of the United Nations (i.e., FAO, 2022c, 2022d). These estimates only focus on shrimp and prawns since current information, although incomplete, shows that these animals constitute the vast majority of farmed decapods. Likewise, statistics about decapod capture and the relatively small size of shrimp and prawns suggest that these animals are likely the majority of wild-caught decapods as well. We have quantified both farmed shrimp and prawns killed in a year and the number of these individuals alive at any time, accounting for mortality rates at different developmental stages. Numbers of individuals alive on farms at any given moment, rather than slaughter figures, are likely a more informative metric to assess the scope of potential welfare issues that standard shrimp farming conditions might pose. Regarding captures, we covered all the 92 decapod shrimp and prawn species caught in the wild across the globe.

All the shrimp and prawn figures we obtained were compared to equivalent available estimates for the most exploited vertebrates used in meat production (i.e., chickens and fishes), and also with available data on industrially farmed insects used to produce food and feed. Our findings suggest that, if these animals are sentient, shrimp production represents a massive animal welfare issue whose scale vastly surpasses all known figures of the animal species most exploited in the food system. Given the scale of shrimp production, the welfare of these animals deserves further attention.

A note on terminology

The terms ‘shrimp’ and ‘prawn’ are often used interchangeably. The two terms do not reliably track any phylogenetic differences between species. The preference for one term over another is geographically dependent (Gillett, 2008; Holthuis & Ng, 2009; New, 2002; Penn et al., 2001). Here, we use only the term “shrimp,” covering both shrimp and prawns. Note that members of the family Artemiidae are commonly referred to as "brine shrimp," but are not decapods and so are beyond the present scope.

Methods

FAO’s FishStatJ produces annual statistics of captured and farmed aquatic animals worldwide, measured in tonnes. To calculate the numbers of farmed shrimp annually slaughtered and those alive on farms at any time, we used FishStatJ’s Global Aquaculture Production dataset (FAO, 2022c). In the case of wild-caught individuals, the Global Capture Production dataset (FAO, 2022d) was consulted. In both cases, data is for 2020, the latest year available at the time the data for this work was recovered (September 2022).

Please note that this work only covers decapod shrimp (and prawn) species, mostly belonging to the Caridea and Dendrobranchiata groups. Non-decapod shrimp, like Artemia (family Artemiidae), are not decapods but belong to the Anostraca order and were thus excluded from this work.

Data for the species used to compare our shrimp estimates–i.e., farmed insects, fishes, and chickens–was obtained from the most recent sources available for each of them at the time of writing. These data were collected in November 2022.

Farmed shrimp estimates

To estimate the number of slaughtered farmed individuals and how many shrimp are alive on farms at any moment, all the countries and fishing areas of the corresponding FAO’s dataset (FAO, 2022c) were selected. Regarding the species included, all the species under the category “Shrimps, prawns” of the database were covered. Shrimp freshwater species under the category “Freshwater crustaceans” were also included.

Number of farmed shrimp slaughtered per year

The number of killed individuals was estimated by dividing tonnages by the average live weight of each species. These weights were taken from Mood and Brooke (2019a). An average live weight of 6-40 g for species for which Mood and Brooke (2019a) do not provide an estimate. This is a wide range that is a guess based on other species. Pre-slaughter mortality rates at larval, post-larval and juvenile-adult phases were also included. All these calculations were incorporated into a probabilistic model and performed in Guesstimate. The model is available here. This approach was chosen given that it uses Monte Carlo simulations, producing narrower and more meaningful value ranges than a deterministic model.

To check the validity of the model, it was used to calculate figures for 2018, an earlier year for which other data is available and, therefore, it might offer a more valid point of comparison than 2020 estimates. Our results converged with the numbers obtained using an alternative model designed with information from other sources for 2018 (see “Alternative estimation of the number of slaughtered shrimps” here), which suggests that the model used for 2020 estimates is roughly correct. Further details are provided in the Appendix.

Number of farmed shrimp alive at any time

To estimate the number of farmed shrimp alive at any time, the number of slaughtered individuals in 2020 was multiplied by the average total lifespan of individuals of that species. Then, the result was divided by 366 (the number of days in 2020).

The numbers of slaughtered shrimp correspond to the figures resulting from the previous probabilistic model for farmed shrimp (see here). Regarding farmed shrimp’s average lifespan, this input was based on FAO sources for the two most commonly farmed species–namely, FAO (2009) for Penaeus vannamei^[4] (family Penaeidae, suborder Dendrobranchiata), and FAO (2022e) for Penaeus monodon (belonging to the same family and suborder). For other species, we used figures from table 4.6 in Wickins and Lee (2002, p. 85).

These calculations were performed in a probabilistic model, using the same tool as before–Guesstimate. Adjustments to account for premature deaths were also incorporated. For further details, see the model here.

Wild-caught shrimp estimates

The method used is very similar to the one already adopted for calculating the number of farmed shrimp. In this case, all the countries and fishing areas of the Global Capture Production dataset (FAO, 2022d) for 2020 were selected. Regarding the species included, all the species under the category “Shrimps, prawns” of the database were covered. Shrimp freshwater species under the category “Freshwater crustaceans” were also included–i.e., Changallo shrimp, freshwater prawns, and shrimps nei^[5] (family Palaemonidae, suborder Pleocyemata), giant river prawn, oriental river prawn, river prawns nei, and Siberian prawn. Shrimp species of other orders were excluded.^[6] That way, 92 different species or taxon^[7] were finally considered (see here, column “B”).

To estimate the number of individuals, capture tonnages were divided by each species' average live weight of captured shrimp. These weights–expressed in a range of possible values–were estimated based on relevant scientific literature or information provided by the industry. The search terms initially used were the species' scientific name(s),^[8] and the words "mean weight." If such an approach did not provide relevant results, "mean weight" was substituted by "weight."

In general, the mean weight ranges were based on mean weights found in the literature, minus and plus standard deviation. Typically, the lower bound of the range resulted from the lower average weight reported by the sources; the upper bound is the highest mean weight found in the literature plus the standard deviation. However, in several cases, the ranges relied primarily on specific sources and not equally on all the sources found. This is because some sources seemed more credible than others, mostly due to sample sizes and the capture methods employed. This approach was adopted for estimating mean weights for Acetes japonicus shrimp (family Sergestidae, suborder Dendrobranchiata), adjusting for sample sizes of the different sources (see here).

In several cases, no information was found on the typical weight of a species at the time of capture. Therefore, in those cases, we took as a reference the average live weights provided by scientific papers for farmed shrimp of that same species. Still, it should be noted that mean weights of farmed individuals are possibly different from those of individuals of the same species caught in the wild. Assuming this, the mean weight range was slightly adjusted accounting for available information for other wild-caught species phylogenetically close to the species at stake.

A similar approach was adopted for species for which no data at all was found. In these circumstances, possible mean values were estimated based on the average reported weights of shrimp species of the same genus. Whenever possible, more than one of the weights of species taken as references was reviewed. Then, mean lengths for the target species and the species used as references were also checked. Eventual differences were incorporated into the estimated mean weight for the species in question. Likewise, when no data at all was found for a whole genus, mean weights were estimated based on evidence for species with similar mean lengths, even if they belonged to a different genus.

In a few other cases, only data on mean shrimp weights when already dried was found (i.e., information provided by Chu et al., 1995). Kurun et al. (2007, p. 358) point out that the dry weight of the “edible” parts of shrimp is significantly lower than the weight of fresh parts. Therefore, dry mean weights were adjusted by 1.6. Further details about how this adjustment factor was calculated can be found in the Appendix.

The different approaches adopted are properly indicated on a case-to-case basis in the "Sources" column of the spreadsheet that compiles the mean weights for all the wild-caught shrimp species.

Initially, the number of captured shrimp per species was estimated by creating a simple deterministic model in a spreadsheet (see here). This first approach was primarily chosen due to the practicalities that a spreadsheet offers against other methods to input and review weight estimates and their sources for more than 90 species. This method allowed us to identify the most commonly wild-caught species–namely, A. japonicus, which accounts for at least ~70% of captured individuals. Next, the specific estimate for this species and for other non-A. japonicus shrimp species were combined into a probabilistic model to calculate the total number of wild-caught individuals using Guesstimate. This model is available here.

The total number of catches was therefore estimated with a probabilistic model, narrowing the ranges obtained with the initial deterministic approach.

Total shrimp estimates

All the shrimp estimates of slaughtered individuals, farmed and wild-caught, were combined in a single probabilistic model, using Guesstimate. This model is available here.

Results

Farmed shrimp

We estimated that, on average, between 300 billion and 620 billion farmed shrimp are killed annually (mean: 440 billion). Considering mortality rates for different species, at different life stages, between 150 billion and 370 billion shrimp are alive at farms at any time (mean: 230 billion). These figures are detailed in the following table. All estimates are available in this Guesstimate model.

Table 1. Table comparing estimated numbers of farmed shrimp of different species and genus annually killed (2020), average lifespan in days, and figures of individuals alive at any time on farms–adjusted and not adjusted for mortality. M stands for million (1,000,000), B stands for billion (1,000,000,000) and T for trillion (1,000,000,000,000). Numbers in parentheses are means.
Source: Own elaboration based on this probabilistic model.

In what follows, these findings are compared with equivalent data for farmed insects, farmed fishes, and chickens used for meat production.

Estimates of farmed shrimp killed every year

The number of farmed shrimp slaughtered per year is greater than available estimates of fishes killed in aquaculture or chickens killed for their meat. However, this number is lower than the total amount of farmed insects that die to be used as feed or food. This is illustrated in the table below:

Table 2. Table comparing estimated numbers of different farmed animals killed per year. B stands for billion (1,000,000,000) and T for trillion (1,000,000,000,000). Numbers in parentheses are means.
Source: Own elaboration based on:
Farmed insects: Source: Rowe (2020). Note that these figures include farmed insects sold live and others that die prior to processing. *: The most recent source of such an estimate is the year 2019, but older data was also considered.
Farmed fishes (2017): Source: Mood and Brooke (2019b). Mean estimated here.

Chickens (2020): Source: FAO (2022b).^[11]

Taking averages as a reference, the total number of farmed shrimp killed per year is equivalent to:

• 40% of the number of farmed insects slaughtered to produce food and feed and others that die prior to being processed;
• 4.5 times the number of farmed fishes killed per year;
• 6.2 times the number of chickens annually killed for their meat.

The following graph represents those differences:

Numbers of farmed shrimp and other farmed animals killed per year

Fig. 1. Estimated numbers (means) of different farmed animals annually slaughtered. Data for farmed shrimp is presented in darker blue. Note that the figure of “farmed insects” includes animals sold live and others that die prior to processing.

Estimates of shrimp alive on farms at any moment

Before being killed, farmed shrimp are likely to face several welfare concerns caused by a variety of factors (see Birch et al., 2021 and Crump et al., 2022b). If so, the number of farmed individuals alive at any time, rather than the figures of slaughtered animals, can better inform efforts to quantify the potential direct suffering posed by shrimp aquaculture.

We obtained that, on average, 150 billion to 370 billion farmed shrimp are alive at any moment (adjusting for mortality). While it was previously shown that more farmed insects annually die to produce food and feed than farmed shrimp for human consumption, this is not the case when considering the numbers of farmed individuals alive at any moment. The productive cycle of shrimp is typically longer than that of insects, so more farmed shrimp than insects are alive on farms at any time. Our estimate of shrimp alive on farms at any moment outnumber not only equivalent figures for insects but also for fishes and chickens. This is summarized in the following table and illustrated in the graph below.

Table 3. Table comparing estimated numbers of different farmed animals alive at any moment. B stands for billion (1,000,000,000) and T for trillion (1,000,000,000,000). Numbers in parentheses are averages. Source: Own elaboration based on:
Farmed insects: Source: Rowe (2020). : The most recent source of such an estimate is the year 2019, but older data was also considered.
Farmed fishes: Source: Šimčikas (2020). : The figure is an estimate of 2020 numbers, based on data from 2016.
Chickens (2020): Source: FAO (2022b).^[12]

Numbers of farmed shrimp and other farmed animals alive at any time

Fig. 2. Estimated numbers (means) of different farmed animals alive at any time. The figure for farmed shrimp is presented in darker blue.

Considering averages, the figures of farmed shrimp alive at any moment are equivalent to:

• 2.7 times the number of insects alive at any point in time on farms;
• 2.2 times the number of farmed fishes alive at any time;
• 6.9 times the number of chickens alive at any moment on farms.

Hence, the number of farmed shrimp alive at any moment surpasses equivalent estimates for the vertebrates most farmed by the food system, and also available figures for another farmed invertebrate group (i.e., insects) that are much smaller in size.

Wild shrimp

Using a deterministic model (spreadsheet), we calculated that between 4.1 trillion and 72 trillion shrimp are annually caught from the wild. Of these, the majority are individuals of a single species: A. japonicus. While in terms of total tonnage, this species accounts for 7.4% of shrimp captured in 2020, in terms of individuals, A. japonicus represents between 69.7% and 88.7% of wild-caught shrimp worldwide–that is, between 3.6 trillion and 50.2 trillion individuals.

Notably, in 2020, there was a sharp decrease in A. japonicus catches worldwide compared to 2019–namely, from 402,061 tonnes in 2019, similar to previous years’ catches, to 251,093 tonnes in 2020 (FAO, 2022d). It is unclear whether this drop obeys anomalous or longer-term factors, as discussed in the Appendix. Despite this decline, A. japonicus still dominates the number of wild-caught shrimp individuals, primarily because of their small size.

As other Acetes species, A. japonicus are tiny shrimp that usually do not reach more than 2 cm in length (Wong et al., 2015). They are found in the Indo-West Pacific, from the west coast of India to Korea, Japan, China, and Indonesia (Holthuis, 1980). As their common name indicates (“akiami paste shrimp”), they are used to produce “shrimp paste”–a salty and fermented paste based on crushed shrimp (Paterson, 2003, p. 5212).

After incorporating the deterministic model’s results for A. japonicus and non-A.japonicus species into a probabilistic model (see here), the following estimates of wild-caught shrimp captured in a year are obtained:

Table 4. Estimates of wild-caught shrimp, according to a probabilistic (Guesstimate) model. B stands for billion (1,000,000,000) and T for trillion (1,000,000,000,000). Numbers in parentheses are means.

Thus, the total number of officially wild-caught shrimp in 2020 is between 6.5 trillion and 66 trillion animals, and likely around 25 trillion individuals. As shown below, such figures outnumber all vertebrate animals farmed for food, and even current estimates of insects farmed for food and feed:

Table 5. Table comparing estimated numbers of different animals farmed or captured from the wild. B stands for billion (1,000,000,000) and T for trillion (1,000,000,000,000). Numbers in parentheses are means. Source: Own elaboration based on:
Farmed insects slaughtered annually: Source: Rowe (2020). Note that these figures include farmed insects sold live and others that die prior to processing. *: The most recent source of such an estimate is the year 2019, but older data was also considered.
Farmed fishes slaughtered (2017): Source: Mood and Brooke (2019b). Average estimated here.

Wild-caught fishes (2016): Source: Fishcount.org.uk (2018). Average estimated here.

Chickens slaughtered (2020): Source: FAO (2022b).^[13]

Taking averages as a reference, the total number of wild-caught shrimp is equivalent to:

• 22.7 times the number of farmed insects slaughtered to produce food and feed and others that die prior to being processed;
• 255.1 times the number of farmed fishes killed per year;
• 17.9 times current estimates of fishes captured from the wild;
• 353.3 times the number of chickens annually killed for their meat.

Such differences are illustrated in the following graph:

Numbers of wild-caught shrimp and other animals annually slaughtered

Fig. 3. Estimated numbers (means) of different animals slaughtered in a year. Data for wild-caught shrimp is presented in darker blue. Note that the figure of “farmed insects” include animals sold live and others that die prior to processing.

Hence, only figures of wild-caught shrimp account for the vast majority of all animals directly killed by humans for food.^[14]

Total shrimp estimates

In this probabilistic model, our previous estimates of slaughtered farmed and wild-caught shrimp were combined. Data for fishes was also included in order to make comparisons that might be useful for understanding the general scope of the problem. Such figures are summarized in the table below, along with estimates of farmed insects and chickens killed annually:

Table 6. Table comparing total estimates of animals annually slaughtered. Numbers in parentheses are means. Source: Own elaboration based on:
Farmed insects slaughtered annually: Source: Rowe (2020)*: The most recent source of such an estimate is the year 2019, but older data was also considered. Note that these figures include farmed insects sold live and others that die prior to processing.
Total number of fishes (farmed and wild-caught): Sources: Fishcount.org.uk (2018); Mood and Brooke (2019b).

Chickens (2020): Source: FAO (2022b).^[15]

Taking averages as a reference, the total number of shrimp annually killed is equivalent to:

• 24.5 times the number of farmed insects killed to produce food and feed, and others that die prior to being processed;
• 18 times the total number of fishes (farmed and captured) slaughtered per year;
• 381.5 times the number of chickens annually killed for their meat.

These differences are illustrated in the following graph:

Numbers of shrimp and other animals killed in a year

Fig. 4. Estimated numbers (means) of shrimp and other animals annually slaughtered. Data for farmed shrimp is presented in darker blue. Note that the categories “total shrimp” and “total fishes” include both farmed and wild-caught individuals. The “farmed insects” estimates include animals sold live and others that die prior to processing.

These figures outnumber any figure of animals killed for their meat known to date. Hence, as our previous estimates of wild-caught shrimp have already suggested, shrimp, in general, account for the majority of animals directly slaughtered for human consumption.^[16]

Discussion

We presented our estimates of the number of shrimp mostly intended for consumption, including farmed and wild-caught individuals. As shown, the number of farmed shrimp killed in a year (440 billion) exceeds several times the figure of the most numerous vertebrate species used in the food system–namely, fishes, and chickens. Certainly, more farmed insects than farmed shrimp are typically killed in a year. Nevertheless, significantly more shrimp than insects are alive on farms at any time. Furthermore, the number of farmed shrimp alive at any moment surpasses any farmed animal estimate known to date–even when compared to any other estimate of global captive vertebrate numbers as provided by Šimčikas (2020). Thus, the problem of shrimp farming is greater in scope–i.e., number of individuals affected–than the problem of insect farming or the farming of any vertebrate for human consumption.

Moreover, if shrimp are sentient, shrimp farming may not only be bigger in scope, but also likely to cause more severe harms than other animal farming industries. Unlike land vertebrates, shrimp and other decapods are often granted no protection by animal welfare legislation. Similarly, shrimp are typically not included in aquaculture certification schemes provisions, and “best-practice guidelines, where they exist at all, tend to prioritize product quality rather than animal welfare” (Crump et al., 2022b). As a result, shrimp and decapods, in general, can be farmed under conditions that otherwise would not be legally authorized for other industrially farmed animals. The exclusion of shrimp from certification programs also entails that producers lack incentives to adopt more humane practices. Moreover, the lack of standardized, welfare-oriented farming and slaughter guidelines means that even producers who might care about welfare may not have the resources to assess and improve their own practices. If negative experiences under typical farming conditions are severe, farming shrimp might pose one of the most serious animal welfare issues known to date, and one which is likely to worsen in the near future.

The problem of shrimp captured from the wild seems even bigger in scale. While the total tonnage of shrimp production from aquaculture typically exceeds the wild-capture output,^[17] this most likely isn’t the case in terms of the number of individuals affected. Indeed, this work has shown that, taking means as a reference, the total number of wild-caught shrimp (25 trillion) is equivalent to ~56.8 times the number of shrimp slaughtered annually in aquaculture (440 billion). Furthermore, to the best of our knowledge, the number of wild-caught shrimp is greater than that of any animal species killed for meat. Indeed, overall, wild-caught shrimp may account for the vast majority of all animals directly slaughtered by humans for food. This suggests that, even though shrimp captures may stagnate in the near future, the welfare risks caused by common capture methods will still cause, on aggregate, a very significant amount of avoidable suffering.

Nevertheless, our ability to produce estimates has been limited both by the quality of the available data and the nature of methods adopted to elaborate the models presented here. Regarding the first category of limitations, it should be noted that data about mortality rates of shrimp under different commercial farming conditions are scarce and not highly accurate (Boyd et al., 2017). Similarly, no specific information about mean weights was found for several shrimp species, or it was not possible to find more than one reliable source to guarantee the validity of a possible weight range.

Additionally, FAO data and other official figures may substantially underreport, or fail to report at all, data of non-commercial, small-scale, and illegal farms. Furthermore, FAO tonnages of farmed shrimp–and, by implication, our models–do not include data on shrimp used as breeders, or of young (postlarval^[18]) shrimp that die while being transported to growing farms.

Something similar happens with data regarding some shrimp capture operations. In this regard, countries typically report "landed weight," but not total catches. For example, Amani et al. (2011) warn that “the fishery statistics available is inadequate due to the fact that most Acetes shrimp caught are locally consumed and they are not been landed at fish landing jetties in the country [Malaysia].” Moreover, in some cases, some shrimp capture operations do not always appear in official records, so neither catches nor landed amounts are reported.

As to the second category of limitations, related to the methods adopted, our estimates admittedly could use further refining. In several cases, the ranges that feature in our models do not represent our subjective confidence intervals (lower than 90%), but simply ranges of possible values found in the literature–either for mean species weights or mortality rates in shrimp aquaculture. Ideally, it would have been desirable to aggregate data using the more common approach in meta-analyses, consisting in weighing the evidence by the inverse of the square of the standard errors of parameters that serve as inputs to the models, so as to arrive at weighted means and standard errors. Then, we would have been able to construct more accurate 90% confidence intervals for the mean weight of each species.^[19] However, the quality of the data available–particularly, for wild species’ mean weights–did not make such a method feasible.

Notably, the models are highly sensitive to slight variations in estimated mean weights for shrimp species. On the one hand, we are relatively confident in the weight ranges used for farmed species. On the other hand, there is greater uncertainty regarding the wild-caught species since reported mean weights vary significantly depending on capture practices, season, or the sex ratio of the individuals caught, among other factors. For these reasons, wide mean weight ranges were used. While we believe that the mean capture weight of A. japonicus very likely falls within the range here employed, minor weight changes–especially for small and commonly caught species–would result in wide variations in our estimates.

Additionally, although we confirmed that our farmed shrimp figures are roughly correct, we were not able to check the validity of our wild-caught shrimp calculations. Particularly, it would have been desirable to estimate the number of shrimp used to produce shrimp paste in order to see if it is similar to the number of A. japonicus here obtained. This would have also been helpful to possibly explain the sharp decrease in A. japonicus catches observed in 2020.^[20] Unfortunately, no reliable and updated sources for the estimation of the amount of shrimp paste produced or consumed per year were found. These and other previously discussed limitations are further addressed in the Methodological uncertainties and limitations section of the Appendix.

All things considered, while the lower and upper bounds of our estimates might be unlikely, we can provisionally say with high credence (> 90%)^[21] that the actual number of reported farmed and captured shrimp is somewhere in between the estimated ranges, and, conservatively, with mid credence ( ≈ 51-60%) that it is close to the averages (± 10%) here presented. In other words, we believe it is highly likely that currently, at least, 300 billion farmed shrimp are slaughtered per year (credence: ≈ 95%); that there are, at minimum, 150 billion shrimp alive on farms at any moment (credence: ≈ 95%); and that at least 4.1 trillion shrimp are captured from the wild annually (credence: > 90%^[22]).

Despite these sheer numbers, if the moral significance of shrimp sentience is negligible, then the total moral value of shrimp may not amount to much. In this regard, it should be noted that the evidence for sentience in penaeids (e.g., P. vannamei) and in other shrimp belonging to the Dendrobranchiata suborder (e.g., A. japonicus) is particularly scarce. Moreover, contrary to Birch et al.’s (2021) conclusions, Comstock (2022) has recently argued that Dendrobranchiates are probably not sentient. However, several open questions still remain, and lack of evidence of sentience should not be confused with evidence of lack of sentience. Thus, further research is needed to better understand the likelihood of sentience in these animals, focusing on the species most commonly farmed and killed for food production.

Assuming that there is a non-negligible chance that farmed and wild-caught shrimp can have valenced experiences–i.e., shrimp are at least 1% likely to be sentient–additional work about shrimp welfare needs under common farming conditions (particularly, for the P. vannamei and the P. monodon species) and the welfare risks posed by standard capture and slaughter methods is necessary. Given the exceptionally high numbers that wild-caught shrimp represent, it might be of special interest to investigate whether individuals of the most captured species–namely, A. japonicus–endure net negative or positive lives in the wild, to look into common natural deaths for individuals of these species, and to address whether the 2020 drop in this species’ catches is anomalous or part of a longer-term trend. Additionally, further resources are necessary to develop and promote science-based humane aquaculture and capture practices, taking into account possible species-specific needs.

Conclusion

This work presented the first known-to-date total estimates of shrimp slaughtered in our food system. It demonstrated that the number of farmed shrimp killed annually and alive in farms at any time vastly surpasses the amount of chickens, fishes, and insects farmed for their meat, or indeed, of any other farmed animal. Furthermore, the number of wild-caught shrimp are even higher, making the problem of shrimp used for food production greater in scale than the problem of insect farming or the farming of any vertebrate for human consumption.

Although shrimp sentience is not as strongly supported by existing evidence as that of vertebrates and other decapods, there are far more shrimp than farmed crabs, lobsters, chickens, or fishes slaughtered for food production. Thus, even if the case for shrimp sentience is weaker than the case for sentience in other decapods, chickens, or fishes, the expected value of researching and helping shrimp might be higher than the expected value of devoting resources to those other animals. Such expected value might become increasingly higher since shrimp production is likely to continue to grow in the future. Upcoming work by Rethink Priorities will provide a better understanding of this problem and possible ways to address it.

Acknowledgments

This research is a project of Rethink Priorities. It was written by Daniela R. Waldhorn, with contributions from Elisa Autric. Farmed shrimp estimates are indebted to a model previously developed by Saulius Šimčikas. Thanks to Hannah McKay for data visualizations, to Phil Brooke, Marcus A. Davis, Laura Duffy, Peter Singer, and Yip Fai Tse for helpful feedback on earlier versions of this report, and to Adam Papineau for copy-editing.

Daniela R. Waldhorn: Conceptualization, methodology, formal analysis, investigation, data curation, visualization, writing-original draft preparation.

Elisa Autric: Data curation, investigation.

Funding Statement: Rethink Priorities provided funding to the authors to conduct this work.

If you are interested in RP’s work, please visit our research database and subscribe to our newsletter.

References

Amani, A. A., Arshad, A. B., Amin, S. M. N., & Abd Aziz, N. (2011). Catch Composition of a Set Bag Net used for Acetes Trapping in the Estuarine Waters of Kedah, Peninsular Malaysia. Journal of Fisheries and Aquatic Science, 6, 279–284. https://doi.org/10.3923/jfas.2011.279.284

Amin, S. M. N., Arshad, A. B., Siraj, S., & Bujang, J. S. (2009). Population structure, growth, mortality and yield per recruit of segestid shrimp, Acetes japonicus (Decapoda: Sergestidae) from the coastal waters of Malacca, Peninsular Malaysia. Indian Journal of Marine Sciences, 38. https://perma.cc/T6N4-FT4V

Anderson, J. L., Valderrama, D., & Jory, D. E. (2019). GOAL 2019: Global shrimp production review. Global Seafood Alliance. https://perma.cc/PF5Q-2W6Z

Bauer, R. T. (2023). Shrimps: Their Diversity, Intriguing Adaptations and Varied Lifestyles (Fish & Fisheries Series Vol. 42). Springer. https://doi.org/10.1007/978-3-031-20966-6

Benchmark Insights. (2019). Technologies shaping the future of shrimp production. https://perma.cc/KA9P-GV2M

Birch, J., Burn, C., Browning, H., & Crump, A. (2021). Review of the Evidence of Sentience in Cephalopod Molluscs and Decapod Crustaceans. London School of Economics and Political Science. https://perma.cc/6FAT-3225

Boenish, R., Kritzer, J. P., Kleisner, K., Steneck, R. S., Werner, K. M., Zhu, W., Schram, F., Rader, D., Cheung, W., Ingles, J., Tian, Y., & Mimikakis, J. (2022). The global rise of crustacean fisheries. Frontiers in Ecology and the Environment, 20(2), 102–110. https://doi.org/10.1002/fee.2431

Boyd, C. E., McNevin, A. A., Racine, P., Tinh, H. Q., Minh, H. N., Viriyatum, R., Paungkaew, D., & Engle, C. (2017). Resource Use Assessment of Shrimp, Litopenaeus vannamei and Penaeus monodon, Production in Thailand and Vietnam. Journal of the World Aquaculture Society, 48(2), 201–226. https://doi.org/10.1111/jwas.12394

Bravo, E. (2020). Cuando el mar entra a la tierra. Producción camaronera en tierras altas. (Acción Ecológica). https://perma.cc/AW6U-NRCB

Carvalho, A. R., Pennino, M. G., Bellido, J. M., & Olavo, G. (2020). Small-scale shrimp fisheries bycatch: A multi-criteria approach for data-scarse situations. Marine Policy, 116, 103613. https://doi.org/10.1016/j.marpol.2019.103613

Chang, Y.-C., Zhang, X., & Khan, M. I. (2022). The impact of the COVID-19 on China’s fisheries sector and its countermeasures. Ocean & Coastal Management, 216, 105975. https://doi.org/10.1016/j.ocecoaman.2021.105975

Chu, K. H., Chen, Q. C., Huang, L. M., & Wong, C. K. (1995). Morphometric analysis of commercially important penaeid shrimps from the Zhujiang estuary, China. Fisheries Research, 23(1), 83–93. https://doi.org/10.1016/0165-7836(94)00342-T

Comstock, G. (2022). The Shrimp Hypothesis: Some, not all, decapods are sentient. Interspecies Comparisons of Welfare Conference, London, UK. https://perma.cc/7PTR-Q8QV

Continúa la discusión por la contaminación de aguas residuales, hay camaroneras que sí les dan tratamiento. (2020, July). El Universo. https://perma.cc/F6MW-5L8F

Crump, A., Browning, H., Schnell, A., Burn, C., & Birch, J. (2022a). Sentience in decapod crustaceans: A general framework and review of the evidence. Animal Sentience, 7(32). https://doi.org/10.51291/2377-7478.1691

Crump, A., Browning, H., Schnell, A. K., Burn, C., & Birch, J. (2022b). Invertebrate sentience and sustainable seafood. Nature Food, 3(11), Article 11. https://doi.org/10.1038/s43016-022-00632-6

de Jong, B. (2018). 100 Billion Dollar Baby: Aquaculture Keeps Growing. RaboResearch. https://perma.cc/8SRM-VMXZ

Demestre, M., & Abello, P. (1993). Growth and distribution of Solenocera membranacea (Risso, 1816) (Decapoda, Dendrobranchiata) in the northwestern Mediterranean Sea. Scientia Marina, 57, 161–166. https://perma.cc/PHX3-UD4N

Despalatović, M., Grubelić, I., & Šimunović, A. (2006). Distribution and Abundance of the Atlantic Mud Shrimp, Solenocera membranacea (Risso, 1816) (Decapoda, Solenoceridae) in the Northern and Central Adriatic Sea. Crustaceana, 79(9), 1025–1032. https://doi.org/10.1163/156854006778859678

Ding, Q., Shan, X., Jin, X., & Gorfine, H. (2021). A multidimensional analysis of marine capture fisheries in China’s coastal provinces. Fisheries Science, 87(3), 297–309. https://doi.org/10.1007/s12562-021-01514-9

FAO. (2009). Penaeus vannamei In Cultured aquatic species fact sheets. Text by Briggs, M. https://perma.cc/DT32-95S6

FAO. (2022a). Catch and landings | Coordinating Working Party on Fishery Statistics (CWP). https://perma.cc/ML2D-2XWP

FAO. (2022b). FAOSTAT—Crops and livestock products. License: CC BY-NC-SA 3.0 IGO. Extracted from: https://perma.cc/5E56-89F2. Date of access: 11-10-2022.

FAO. (2022c). Fisheries & Aquaculture—Global aquaculture production. License: CC BY-NC-SA 3.0 IGO. Extracted from: https://perma.cc/F3RX-8ZH9. Date of access: 16-09-2022.

FAO. (2022d). Fisheries & Aquaculture—Global capture production. License: CC BY-NC-SA 3.0 IGO. Extracted from: https://perma.cc/6ERV-W2F5. Date of access: 16-09-2022.

FAO. (2022e). Penaeus monodon. Cultured Aquatic Species Information Programme. Text by Kongkeo, H.. Fisheries and Aquaculture Division [online]. https://perma.cc/5ST4-8DED

FAO. (2022f). The State of World Fisheries and Aquaculture 2022. Towards Blue Transformation. FAO. https://doi.org/10.4060/cc0461en

Feng, E., & Cheng, A. (2020, February 21). Restrictions And Rewards: How China Is Locking Down Half A Billion Citizens. NPR. https://perma.cc/8BB3-TB9Y

Fishcount. (n.d.). Estimated numbers of individuals in annual global capture tonnage (FAO) of fish species (2007—2016). https://perma.cc/3C6D-BGAU

Garcia, S. (1988). Tropical penaeid prawns. In Gulland, J. A, (ed.) Fish Population Dynamics (2nd Ed., pp. 219-249). John Wiley & Sons Ltd. https://perma.cc/3C29-X485

Gillett, R. (2008). Global study of shrimp fisheries. FAO. https://perma.cc/Y6TC-Y772

Graça, G. N. S. N. D. (2008). Aspectos da biologia do Camarão Plesionika narval (Fabricius, 1787) nos Açores (Doctoral dissertation). https://perma.cc/DA79-E553

Heu, M.-S., Kim, J.-S., & Shahidi, F. (2003). Components and nutritional quality of shrimp processing by-products. Food Chemistry, 82(2), 235–242. https://doi.org/10.1016/S0308-8146(02)00519-8

Holthuis, L. B. (1980). Shrimps and prawns of the world: An annotated catalogue of species of interest to fisheries. Food and Agriculture Organization of the United Nations. https://perma.cc/7K27-WVMB

Holthuis, L. B., & Ng, P. K. L. (2009). Nomenclature and Taxonomy. In New, M. B., Valenti, W. C., Tidwell, J. H., D'Abramo, L. R., & Kutty, M. N. (Eds.), Freshwater Prawns: Biology and Farming (pp. 12-17). John Wiley & Sons. https://doi.org/10.1002/9781444314649.ch2

Jones, B. L., Unsworth, R. K. F., Udagedara, S., & Cullen-Unsworth, L. C. (2018). Conservation Concerns of Small-Scale Fisheries: By-Catch Impacts of a Shrimp and Finfish Fishery in a Sri Lankan Lagoon. Frontiers in Marine Science, 5. https://doi.org/10.3389/fmars.2018.00052

Kurun, A., Balkıs, H., & Balkıs, N. (2007). Accumulations of total metal in dominant shrimp species (Palaemon adspersus, Palaemon serratus, Parapenaeus longirostris) and bottom surface sediments obtained from the Northern Inner Shelf of the Sea of Marmara. Environmental Monitoring and Assessment, 135(1), 353–367. https://doi.org/10.1007/s10661-007-9655-9

Louisiana Direct Seafood. (n.d.). Seafood Handbook. https://perma.cc/NS3R-UNHJ

Mood, A., & Brooke, P. (2019a). Estimated numbers of individuals in global aquaculture production (FAO) of decapod species (2017). Fishcount. https://perma.cc/2EDY-UY63

Mood, A., & Brooke, P. (2019b). Estimated numbers of individuals in global aquaculture production (FAO) of fish species (2017). Fishcount. https://perma.cc/EE9P-DAZP

New, M. B. (2002). Farming freshwater prawns. A manual for the culture of the giant river prawn (Macrobrachium rosenbergii). FAO Fisheries Technical Paper, (No. 428). https://perma.cc/8KSC-SHWG

Otto, R. S., & Jamieson, G. S. (2001). Commercially Important Crabs, Shrimps and Lobsters of the North Pacific Ocean (No. 19; PICES Scientific Report). North Pacific Marine Science Organization (PICES). https://perma.cc/B9PS-AX9W

Paramo, J., & Saint-Paul, U. (2012). Deep-sea shrimps Aristaeomorpha foliacea and Pleoticus robustus (Crustacea: Penaeoidea) in the Colombian Caribbean Sea as a new potential fishing resource. Journal of the Marine Biological Association of the United Kingdom, 92(4), 811-818. https://doi.org/10.1017/S0025315411001202

Paterson, B. D. (2003). Commercially Important Crustacea. In B. Caballero (Ed.), Encyclopedia of Food Sciences and Nutrition (Second Edition) (pp. 5210–5216). Academic Press. https://doi.org/10.1016/B0-12-227055-X/01077-4

Pauly, D., & Zeller, D. (2016). Catch reconstructions reveal that global marine fisheries catches are higher than reported and declining. Nature Communications, 7(1), Article 1. https://doi.org/10.1038/ncomms10244

Penn, J. W., Caputi, N., & Melville-Smith, R. (2001). Crustacean Fisheries. In J. H. Steele (Ed.), Encyclopedia of Ocean Sciences (Second Edition) (pp. 699–707). Academic Press. https://doi.org/10.1016/B978-012374473-9.00453-7

Rowe, A. (2020). Insects raised for food and feed—Global scale, practices, and policy. Rethink Priorities. https://perma.cc/S4G3-8PU7

Shinn, A., Pratoomyot, J., Griffiths, D., Trọng, T., Nguyen, V., Jiravanichpaisal, P., & Briggs, M. (2018). Asian Shrimp Production and the Economic Costs of Disease. Asian Fisheries Science, 31S. https://doi.org/10.33997/j.afs.2018.31.S1.003

Šimčikas, S. (2020). Estimates of global captive vertebrate numbers. Rethink Priorities. https://perma.cc/9FZP-7A7S

Singh, G. (2021). How a shrimp farming "mafia” is displacing Indian fishing communities. https://perma.cc/3RKE-WHAH

Surat: Razing of illegal shrimp farms begins. (2021, January 28). The Times of India. https://perma.cc/4S4Y-QYA4

Suresh, A. (2015). Selective Breeding in Shrimp Farming. Fishing Chimes. https://perma.cc/7YJB-LJVN

Torry Research Station. (2001). Handling and processing shrimp (No. 54; Torry Advisory Note). Torry Research Station [Electronic version by SIFAR-FAO, 2000]. https://perma.cc/CH2K-PL63

Turnbull, C. T., & Atfield, J. C. (2007). Fisheries Long Term Monitoring Program—Summary of tiger and endeavour prawn survey results:1998-2006. 34. https://perma.cc/3TZH-FBG7

Urbina, I. (2020, August 17). How China’s Expanding Fishing Fleet Is Depleting the World’s Oceans. Yale Environment 360. https://perma.cc/MJ8C-RTWE

USDA (Foreign Agricultural Service). (2021). 2021 China’s Fishery Report (p. 17). https://perma.cc/V6JZ-YY4W

Waldhorn, D. (2019). Invertebrate sentience: Summary of findings, part 2. Rethink Priorities. https://perma.cc/L6E9-F4WJ

Waldhorn, D., Schukraft, J., Carpendale, M., Wildeford, P., & Davis, M. A. (2020). Invertebrate Sentience Table. Rethink Priorities. https://perma.cc/2RXZ-6TDR

Watson, R. A. (2017). A database of global marine commercial, small-scale, illegal and unreported fisheries catch 1950–2014. Scientific Data, 4(1), Article 1. https://doi.org/10.1038/sdata.2017.39

Wickins, J. F., & Lee, D. O. (2002). Crustacean Farming: Ranching and Culture, 2nd Edition. Wiley-Blackwell. http://doi.org/10.1002/9780470995082

Wong, B. Y., Ong, H. K., & Khoo, G. (2015). Length-weight relationships of Acetes spp. Sampled along the west coast of Peninsular Malaysia. Sains Malaysiana, 44(3), 379–386 https://perma.cc/N4XD-BJNT

Zeller, D., Harper, S., Zylich, K., & Pauly, D. (2015). Synthesis of underreported small-scale fisheries catch in Pacific island waters. Coral Reefs, 34(1), 25–39. https://doi.org/10.1007/s00338-014-1219-1

Appendix

Farmed shrimp estimates

Number of farmed shrimp slaughtered per year

In order to check whether the model built to estimate the number of farmed shrimp killed in a year produced reliable results, we used it to calculate 2018 figures. This year was chosen given that (i) it’s a closer point of comparison for existing 2017 estimates by Mood and Brooke (2019a), and (ii) other information for that same year would allow us to produce an alternative model and thus, compare its results with those obtained with the model initially built.

We calculated that between 270 and 550 billion farmed shrimp were slaughtered in 2018 (mean: 390 billion, see here). These results are very similar to those provided by Mood and Brooke (2019a)–namely, between 213 and 530 billion. Then, we performed an alternative estimate based on data offered by the Global Aquaculture Alliance’s GOAL (Global Outlook for Aquaculture Leadership) 2019 survey (Anderson et al., 2019) for the year 2018. That source describes what percentage of farmed shrimp produced in 2018 belong to different weight categories (see this figure). Using this information, it was calculated that the average farmed shrimp weight in 2018 was about 10.3 grams. According to an email exchange with one of the authors, the survey asked for weights of shrimp with their shells on but with their heads off. According to an article by Louisiana Direct Seafood (n.d.),^[23] “removing the heads [of shrimp] will decrease total weight by 35%.” Hence, the mean live weight of farmed shrimp in 2018 was about 10.3 grams / (1 - 0.35) = 15.8 grams.

If the total 2018 shrimp aquaculture production from FAO’s FishStatJ (~6.6 million tonnes; FAO, 2022c) is divided by 15.8 grams, it is obtained that about 417 billion farmed shrimp and prawns were produced in 2018. We performed the same estimation using ranges for the mean number of shrimp per pound in each size category (see here). As observed in the model, it is estimated that between 360 billion and 470 billion farmed shrimp were slaughtered in 2018. This finding is similar to the figures initially calculated for that very same year. This convergence of results suggests that the model built for 2020 estimates is roughly correct.

Methodological uncertainties and limitations

There are limitations of our work that result both from the quality of the data available and from the method used. These are briefly discussed in what follows.

Limitations associated with the data available

Unlike the FAO data of the tonnage of animals captured, aquaculture outputs refer exclusively to amounts of aquatic animals harvested for human consumption. That is, the data provided do not include shrimp reared for other purposes. Nevertheless, other limitations should be taken into account:

In some cases, aquaculture production data are not fully recorded: Similar to what happens with some shrimp capture operations (see below), FAO and official figures may substantially underreport, or not report at all, data of non-commercial small-scale and illegal farms. In India’s Odisha state, some testimonials of local residents claim that this might be a rising phenomenon (Singh, 2021). In the Indian city of Surat, it is estimated that around 1,000 illegal shrimp ponds have been constructed on government land (“Surat: Razing of illegal shrimp farms begins,” 2021). Informants interviewed for this report confirmed the existence of illegal farms in the country, which is also one of the major shrimp producers worldwide. Similarly, a report states that the shrimp business in Ecuador–another main producing country–operates largely illegally (Bravo, 2020, p. 8; see also “Continúa la discusión por,” 2020^[24]). No additional, reliable and updated sources were found to assess the scope of this issue.

Furthermore, good data for some countries may not be available. In those cases, FAO estimates the amount of farmed shrimp production. However, if those countries have not reported official figures repeatedly over the years, FAO statistics may underestimate the actual amounts. For example, according to FAO’s Global Aquaculture Production dataset reviewed in June 2021, some important producing countries–e.g., India, Ecuador, and Honduras–had not reported shrimp aquaculture production numbers for 2018 yet. At the time we are drafting this report (December 2022), 2020 data for China is partially incomplete and Indonesia–one of the world's largest shrimp exporters–has not fully reported its shrimp captures either, so some of their figures are estimates (see here, extracted from FAO, 2022d).

The total tonnage of farmed shrimp do not account for the total amount of shrimp that die in aquaculture: In particular, FAO figures–and hence, our models–do not include:

• Numbers of shrimp used as broodstock or as “stock”
• Postlarvae that die while being transported to growing farms
• Other shrimp that die or are slaughtered prematurely–e.g., due to disease outbreaks.

However, we attempted to account for premature mortality in our second model–i.e., numbers of farmed shrimp alive at any time.

Limited updated information about mortality rates and other variables in commercial conditions: One crucial concern about these estimates is the limited up-to-date information on lifespan, mean weights, and mortality figures found in the literature. Importantly, accurate mortality rates for different species under different production systems are particularly scarce. According to Boyd et al. (2017), farmers typically fail to properly estimate the proportion of shrimp that survive the whole production cycle, and “failed crops” are not always included in reported mean mortality rates. Additionally, it may be that such mortality rates do not account for current diseases that are threatening shrimp production.

Limitations associated with the methodology adopted

Some methodological constraints are likely to lead to an underestimate of the actual numbers of farmed shrimp. In this sense, for our first model on the number of shrimp slaughtered in aquaculture, at least three limitations should be acknowledged:

In several cases, our subjective confidence level is lower than 90%: Guesstimate is made to be used with subjective 90% confidence intervals. Similarly to our calculations for wild-caught shrimp, in many cases, we entered ranges in the model that are not our subjective confidence intervals, but simply ranges of possible values that we found in the literature. Future work might refine our estimates by better accounting for confidence intervals.

Estimated mean weights per several species may not be accurate, and this would change the total number of killed animals: According to Anderson et al. (2019; see fig. 4), there has been a shift towards farming smaller shrimp since 2011. If so, some average weights used in our model may be outdated–that is, they are likely to be higher than the actual mean weights of farmed shrimp. As a result, the total number of farmed shrimp slaughtered per year would be greater than what we calculated. However, Suresh (2015) claims that bigger individuals can be produced nowadays: “Production of 15 g vannamei in 120-135 days at a stocking density of 20 animals/sq.m was the norm twenty years ago. Today 28 g shrimp can be produced in the same duration….” It might also be the case that there are trends to farm smaller or bigger shrimp in each country due to demand changes.

Lifespan and mortality figures may not be accurate, and this would change the number of farmed shrimp alive at any time: It could be that the sources we used are outdated, or that these numbers are not sufficiently valid across all producing countries or farming systems. If the aquaculture industry is nowadays shifting to produce smaller shrimp–as Anderson et al. (2019) suggests–then, life cycles in captivity are likely to be shorter than what was estimated for these models. Similar concerns apply to the mortality rates used.

The model’s convergent validity may not be as strong as it may seem: The results of our model are similar to those obtained by other authors and with an alternative model. Those other calculations are, to the best of our knowledge, the only available sources of comparison to check the sanity of our estimates. However, those other models may not be the best sources for ensuring the validity of our calculations:

• First, the mean weights used in our model are based on data provided by Mood and Brooke (2019a)–the other model we compared our results with. It is reasonable to obtain similar figures if several inputs of both models are exactly the same. Still, it should be noted that Mood and Brooke (2019a) developed a deterministic model while we used a probabilistic one.
• Second, the alternative estimate provided here is mostly based on one survey, but it is unclear how representative it is and whether its results are accurate.

With respect to the models designed to estimate the numbers of farmed shrimp alive at any time, the following limitations should be considered:

The limitations of the first model on the number of farmed shrimp killed were incorporated in the calculations of the number of shrimp alive on farms at any time: Since the later estimate required as an input the number of farmed shrimp slaughtered in a year, any inaccuracy in the first calculations was also accumulated in our estimates of the number of shrimp alive at any moment on farms.

Wild-caught shrimp estimates

Adjustment factor to convert dry weights into wet weights

When only dried weight data was found, an adjustment factor was used. It was calculated as follows:

1. According to figures provided by Kurun et al. (2007) (see table 4 on page 358), when shrimp meat is dried, it loses approximately 75% of its initial weight.
2. It is hypothesized that it is unlikely that similar weight losses affect non “edible” shrimp parts–i.e., shrimp exoskeleton and their heads. Therefore, we assume that when shrimp are blotted dry, there is no such significant weight loss of such non-edible parts.
3. According to Louisiana Direct Seafood (n.d.),^[25] a shrimp’s head and exoskeleton together represent 50% of the animal’s total weight.^[26]
4. Thus, if a shrimp is dried, the animal will lose 75% of the 50% weight of her total weight. Therefore, the total weight lost upon drying is 37.5% of the animal’s initial weight (0.75x0.5). In other words, a dried shrimp preserves 62.5% of her initial weight (0.5 + 0.5x0.25 = 0.5 + 0.125 = 0.625).
5. This 62.5% retained weight must be multiplied by 1.6 to obtain the live animal weight (100/62.5 = 1.6).

Nevertheless, assumption (2), although possible, may be disputed.^[27] Whether shrimp’s non “edible” parts would lose weight in any significant way depends on how and to what extent the animals are dried out. Unfortunately, not enough details are provided in this regard.^[28] In order to account for the possibility that this hypothesis might be wrong, all the values resulting from the above calculations were also compared with the already available mean weights of shrimp species of similar size, and when possible, of the same genus. When necessary, some small adjustments were made. Those cases are indicated in the “Type of weight” column of the spreadsheet as “WW, adjusted & estimated.”

Methodological uncertainties and limitations

Limitations associated with the data available

In what follows, the limitations of this work that result from the quality of the data currently available are discussed. Such constraints are likely to lead to an underestimate of the actual number of wild-caught shrimp.

Data provided by FAO refer to shrimp captured for various purposes: FAO statistics cover shrimp captured for all commercial, industrial, recreational, and subsistence purposes–that is, not exclusively for direct human consumption. Unfortunately, the database does not allow a breakdown of the quantities of animals landed (see below) for different purposes. Nevertheless, the majority of global catches are likely industrial fisheries that commercialize shrimp for human consumption (FAO-Fisheries Division, personal communication, May 27, 2021; Gillett, 2008, p. 15; see also Pauly & Zeller, 2016).

Countries typically report “landed weight,” not total catches: The FAO global capture database refers to the total retained catches, which includes “landed” and “non-landed” catches (FAO-Fisheries Division, personal communication, May 27, 2021). That is, it includes the mass of animals that are captured, whether they arrive at the port (“landed”) or not.

However, in most cases, countries are only able to report the quantities of landed catches (FAO-Fisheries Division, personal communication, May 27, 2021). That is, it is likely that FAO’s Fishery Statistical Collections on Global Capture Production does not contain data about shrimp (and other animals) that were captured, killed, and used prior to landing, either because they were consumed on board, or used as bait. Similarly, captured shrimp that got spoiled onboard are typically dumped before landing, and therefore, are not reported either. Lastly, landed weight data do not account for shrimp that are unintentionally lost when handling at sea or when landing.

In some cases, neither catches nor landed amounts are recorded: Some shrimp capture operations do not always appear in official records. In general, FAO figures may substantially underreport data of commercial and non-commercial small-scale fisheries and illegal or non-regulated catches (Pauly & Zeller, 2016; Watson, 2017; Zeller et al., 2015). This underestimation is known and acknowledged by FAO–e.g., the organism states, “non-reporting of landings is a major concern in some fisheries” (FAO, 2022a).

Still, the magnitude of these underestimations for the shrimp fishing industry is unclear. However, there are reasons to think that official data significantly underestimates the total catch. First, illegal and non-regulated shrimp trawling is a common issue in major producing countries, like Indonesia (Gillett, 2008, pp. 191-213) and China (Urbina, 2020). That is why, in 2001, the North Pacific Marine Science Organization (PICES) stated that “world landings of Acetes [shrimp]^[29] are likely to be grossly underestimated” (Otto & Jamieson, 2001, p. 34). If that continues to be the case, given the small size of these animals–who resemble krill–these estimates would fall short by several orders of magnitude.

Second, good data for some countries is not always available (Gillett, 2008, p. 33). This might be the case of currently available A. japonicus figures, as discussed below. Some countries catch important amounts of shrimp and have not reported official data to FAO for years (FAO-Fisheries Division, personal communication, May 27, 2021).^[30]

Additionally, due to practical and logistical constraints, lack of reliable data is a common issue for subsistence and small-scale shrimp catches, particularly in low- and middle-income countries (FAO-Fisheries Division, personal communication, May 27, 2021; Gillett, 2008, p. 33). One may believe that such catches do not represent significant amounts, but Gillett (2008, p. 9) states that “small-scale fishing is responsible for a surprisingly large proportion of the world’s shrimp catch.” More recent sources confirm the above (Carvalho et al., 2020; Jones et al., 2018). Additionally, it should be noted that small-scale operators typically employ gear suitable for estuarine waters and lagoons. Therefore, the shrimp captured will preferably be small individuals–i.e., larvae, juveniles, and migrating subadults (Gillett, 2008, p. 22; see also Garcia, 1988, p. 220). That is, even if the non-reported amounts by small-scale operators are relatively reduced, this deficit is likely more important in terms of the number of individuals involved.

Also note that recreational catches are likely to be under-estimated in many countries (FAO-Fisheries Division, personal communication, May 27, 2021), but that those sorts of captures are beyond the scope of this work.

A large drop in A. japonicus catches due to unclear factors: In 2020, there was a sharp decrease in A. japonicus catches worldwide compared to 2019. While in 2020 251,093 tonnes of A. japonicus were captured from the wild (data viewed: September 2022), captures in 2019 for the same species reached 402,061 tonnes–similar to previous years. The 2020 dip came from China, whose A. japonicus captures decreased by 157,546 tonnes (-40%) from 2019 to 2020 (FAO, 2022d). No generalized drop was observed for China’s total capture numbers for other marine species over that same year, and according to the United States Department of Agriculture (USDA, 2021), China’s wild catches only declined by 5.4% in 2020 compared to 2019.

Given A. japonicus’ small size, even little variations in tonnages lead to significant variations in numbers of individuals. Initially, it was assessed whether the 2019-20 difference in A. japonicus catches was due to late data reporting–that is, available information for 2020 might still be incomplete–rather than an actual dip in catches. In this regard, FAO-Fisheries Division experts stated that “catches for the latest year (currently 2020) should generally be treated as preliminary and subject to revision at the next release of the data, scheduled for March 2023. In 2021, catches for 2019 were revised by China, so it is reasonable to expect that catches for 2020 may also be revised as new or more complete data becomes available” (Statistics team at FAO-Fisheries and Aquaculture Division, personal communication, September 9, 2022). If 2020 catches are actually similar to 2019 figures, we may be underestimating the numbers of wild-caught shrimp by 2.2 trillion to 30.2 trillion individuals.

However, while catches are generally revised upwards by countries, in the case of China, the revisions to 2019 data resulted in a marginal (just over 1%) increase in total catches. Therefore, it is reasonable to assume that 2020 revised data for A. japonicus will unlikely change the already-2020 reported figures by any large magnitude. It is even more unlikely that complete 2020 data will reach 2019 levels. If so, late data reporting does not likely explain the decrease in A. japonicus captures, and other factors thus probably offer more plausible reasons.

After further research and also consulting with FAO-Fisheries Division experts, this sharp decrease may be multifactorial, possibly including the following:

• Circumstantial effects associated with the COVID-19 pandemic: The pandemic, including lockdowns and closure of markets, caused major disruption to fishing activities and quantities of landed catches, and slowed down trade (FAO, 2022f, pp. 195-200). In China, A. japonicus is mostly caught in the Zhejiang province (Ding et al., 2021), which was one of the most affected areas in terms of COVID-19 circulation in 2020 (Feng & Cheng, 2020). The province was on lockdown from mid-February 2020. It therefore seems like a plausible hypothesis, which was also mentioned by the FAO’s statistics team in their response (Statistics team at FAO-Fisheries and Aquaculture Division, personal communication, September 9, 2022). This hypothesis was deemed “meaningful” and “able to develop as a very good research” by Yen-Chiang Chang, Professor of International Law at Shandong University and co-author of “The impact of the COVID-19 on China’s fisheries sector and its countermeasures” (2022) (personal communication, October 5, 2022). However:
  • It is unclear why A. japonicus captures would be much more affected by the pandemic than catches of other species–which, in some cases, even increased, while China’s total shrimp captures are generally consistent with previous years. It could be argued that A. japonicus production, commercialization and/or demand were particularly vulnerable to pandemic restrictions, unlike other shrimp products. In this regard, Gillett (2008, p. 28) claims that “akiami paste shrimp is distinct from most other [shrimp] species (magnitude of production, fishing technique, product form, end market).” If so, assuming that the pandemic impacted aquatic food production differently (FAO, 2022f, p. 196), it is yet to be investigated whether A. japonicus production, trade, or consumption were actually particularly affected by the COVID-19 pandemic.
  • So far, no direct and more specific proof linking the pandemic restrictions with this dip in A. japonicus catches was found.
• A decrease in catches caused by structural factors: While China remains the world’s top producer of marine captures, according to the FAO experts consulted (Statistics team at FAO-Fisheries and Aquaculture Division, personal communication, September 9, 2022) and USDA (2021), China is voluntarily reducing wild catches. Over the last years (2016-2020), this has been happening “due to fewer resources, reduced fishing fleets, and [new] regulations” put in place by the Chinese government (USDA, 2021, p. 3-4). However, assuming that China did indeed voluntarily reduce its A. japonicus captures, it should be noted that the country’s catches of other shrimp species in 2020 grew, while China’s total shrimp captures are generally consistent with previous years. It is unknown whether and why A. japonicus catches would have been particularly affected by recent fishing policies of the Chinese administration.
• In general, without 2021 data, it is difficult to speculate whether the decrease in A. japonicus captures was anomalous or is part of a longer-term trend and possible change in targeting by Chinese vessels. Forthcoming data for 2021 and onwards will likely shed light on this matter.

Total retained captures do not account for the total amount of animals killed by shrimp fishing: In general, FAO data of total retained catches of aquatic animals do not account for the total amount of shrimp and other animals that are actually killed, like:

• Shrimp (and other animals) that died as a result of capture operations but were not caught.
• Shrimp and other animals of non-target species that were captured and killed but were considered undersized,^[31] unsaleable, or undesirable for other reasons, and were discarded. This includes what is commonly called “bycatches.” It is widely known that global discarded bycatches are mainly caused by shrimp trawling (Gillett, 2008, pp. 45-67; Jones et al., 2018; Pauly & Zeller, 2016).

Estimating the number of these animals indirectly killed by shrimp fishing is beyond the scope of this work.

These estimates do not account for all the shrimp and other non-human individuals harmed by shrimp capture operations: Capture operations can harm shrimp and other animals living in the wild, without being caught. Similarly, captured shrimp and other animals may escape alive–but physically hurt–from the fishing gear. Other animals are captured, but may be considered commercially undesirable, and are discarded while still alive (“bycatches”). It is likely that those individuals, if sentient, also suffered from the intensive handling that capture operations pose. There is no specific data on these harms caused by capturing shrimp.

Note that these limitations are not exclusive to this work. They derive from the quality of the available data and affect any estimate of the number of captured aquatic animals based on figures provided by FAO.

Limited data of the mean weight of all the captured shrimp species exist: Shrimp of more than 92 different species are captured from the wild. For just a few species, the mean commercial weight at which those individuals are typically caught is readily available. However, for several species, we were unable to find such specific information. In many cases, it was not possible to find more than one reliable source, or we had to rely on factsheets where no additional information was provided to directly assess the accuracy of the data provided.

Limitations associated with the methodology adopted

There are other constraints related to the method used to calculate the number of shrimp captured from the wild. These are described below.

In several cases, our subjective confidence level is lower than 90%: In many cases, mean species weights that are not our subjective confidence intervals were used, but simply ranges of possible values that were found in the literature, or values that were inferred based on the literature for other species of the same genus.

The model is highly sensitive to slight variations in estimated mean weights of shrimp species: Some of the species most commonly captured are particularly small–e.g., A. japonicus, Acetes spp, or Sergestidae shrimp weighing less than 1 g. Therefore, any tiny change in shrimp estimated mean weights–particularly, of the mentioned species–results in wide variations in the total number of wild-caught individuals. While we are confident that the mean weight of A. japonicus very likely falls within the range provided, mean weights for some other species may be further refined given that:

• Mean weights are based on relatively small samples: While farmed species' mean weights might be more homogeneous, that is unlikely to be true for the 92 species of wild-caught individuals. Here we are, at best, relying on sample sizes of up to a few thousand shrimp to extrapolate the mean weights of trillions of individuals.
  • No data on the mean weight of some species were found: When no average weight information for a given species seemed to exist, it was inferred from scientific descriptions of the species or the mean weight of other comparable species. Those cases are flagged in the spreadsheet as "WW [wet weight], estimated." When the live weight of a species had to be calculated based on data about its dry weight, it is indicated as “WW, adjusted.” Still, it should be noted that, in many cases, it was not possible to find more than one reliable source to guarantee the validity of the estimated average weight.
• The average weight of adult individuals of a given species can vary greatly: For most shrimp species, there are significant weight differences between adult males and females of the same species (see e.g., Chu et al., 1995), and it is unknown whether most catches are females or males for any given species. For example, based on Amin et al., (2009), it is reasonable to think that a higher proportion of females in catches is likely for A. japonicus (Amin et al., 2009), which represents between 69.7% and 88.7% of individual shrimp caught globally. However, evidence by Amani et al. (2011) does not support such an assumption, and none of these studies provides strong enough evidence to generalize across A. japonicus capture practices worldwide. Due to these uncertainties, on some occasions, the range of estimated mean weights is particularly wide.
• Commercially desirable weights also vary greatly: Different markets require different species and sizes. In some countries, the industry has developed standards to classify commercially valuable shrimp and discard smaller individuals (Louisiana Direct Seafood, n.d.; see also Gillett, 2008, p. 37-38). In general, large individuals are sold at higher prices than small shrimp (Gillett, 2008, pp. 37-38; Louisiana Direct Seafood, n.d.; Torry Research Station, 2001), and are likely to be preferred for export (Turnbull & Atfield, 2007). However, while specimens of a given species can be considered too small for some markets (i.e., juveniles), they can indeed meet commercial expectations in other locations. Indeed, small individuals that are discarded in some countries, are commercially valuable in other locations, especially in lower-income countries–e.g., tiny shrimp are ground and used as a condiment (e.g., in Indonesia or Nigeria), directly consumed, or used for other purposes (Gillett, 2008, p. 33; Garcia, 1988, p. 222; see also Torry Research Station, 2001). According to observations in markets in China, smaller shrimp are preferred over large ones (Tse Yip Fai, personal communication, June 16, 2021) while that is not typically the case in Western countries (see e.g., Gillett, 2008, p. 37). These preferences inform capture practices that affect mean weights of shrimp catches, as discussed in the next point.
• Mean weights of a given species can also vary due to capture practices: Wild-caught shrimp sizes might vary greatly depending on the season of the year they are captured (see e.g., Amani et al., 2011; Demestre & Abelló, 1993), and mesh size selection (Graça, 2008; Oh et al., 2003). Due to behavioral patterns, sizes also change depending on the depth at which shrimp are collected (see e.g., Demestre & Abelló, 1993; Paramo, 2012), and, for some species, even the time of day capture efforts take place (see e.g., Despalatović et al., 2006). Similarly, it might be the case that a species' common sizes are slightly different in different geographies. Unfortunately, it was not always possible to obtain data on the average weight of a species in the area where it is most frequently captured–for example, for A. japonicus caught by China. All these considerations might affect the reliability of the sources used to calculate wild-caught shrimp mean weights–that is, some weight-reporting sources rely on catching methods and/or locations that may not be representative of global shrimp capture practices for that same species. This is an additional reason why we used relatively wide mean weight ranges.

Notes

1. ^{^}

   Decapods (a term derived from the Greek, meaning "ten footed") are an order of crustaceans within the class Malacostraca. They are characterized by having the first three pairs of legs being functionally associated with feeding, while the other five pairs of legs (“ten feet”) are used for various purposes ranging from locomotion to grooming. Decapods can have many appendages, and vary morphologically significantly. Some well known decapods are crabs, lobsters, shrimp, and crayfish (Bauer, 2023, p. 9).

2. ^{^}

   The wild shrimp estimates depended on some inputs for which we have less than 90% subjective confidence. Therefore, our estimated 90% SCI for our aggregate wild shrimp estimates is somewhat narrower than a true 90% SCI.

3. ^{^}

   Note that this is different from all animals killed directly for food production, since this category could be construed to include invertebrates killed during pesticide application on crops.

4. ^{^}

   We opt for the use of Penaeus vannamei over Litopenaeus vannamei (to which this species is often referred), due to recognition of this nomenclature by ITIS, WorMS, and FAO’s ASFIS List of Species for Fishery Statistics Purposes at time of writing. 

5. ^{^}

   In FAO statistics, 'nei' means 'not elsewhere included.'

6. ^{^}

   I.e., Artemia salina (brine shrimp) and Erugosquilla massavensis (red sea mantis shrimp), which can be found under the category “Miscellaneous marine crustaceans” in FAO’s dataset.

7. ^{^}

   In some cases, FAO groups different species under a broader category, like a genus or family.

8. ^{^}

   Some shrimp species names have changed over the years. One of the most well-known examples is that of Penaeus vannamei, also known as Litopenaues vannamei. 

9. ^{^}

   Family Penaeidae, suborder Dendrobranchiata. 

10. ^{^}

    Family Palaemonidae, suborder Pleocyemata. 

11. ^{^}

    Parameters: Livestock Primary; World + (Total); Producing animals/Slaughtered; Meat of chickens, fresh or chilled; 2020.

12. ^{^}

    Parameters: Crops and livestock products; World + (Total); Stocks; Live Animals (> List): Chickens; 2020. 

13. ^{^}

    Parameters: Livestock Primary; World + (Total); Producing animals/Slaughtered; Meat, chicken; 2020. 

14. ^{^}

    Note that figures for farmed insects include some individuals indirectly killed for human consumption–i.e., insects to be used as animal feed.

15. ^{^}

    Parameters: Crops and livestock products; World + (Total); Stocks; Live Animals (> List): Chickens; 2020.

16. ^{^}

    Note that figures for farmed insects include individuals indirectly killed for human consumption–i.e., insects to be used as animal feed.

17. ^{^}

    For example, according to FAO’s FishStatJ, 7,426,401.19 tonnes of farmed shrimp were produced in 2020, while 3,386,207.69 tonnes of wild shrimp were captured that same year (FAO, 2022d). 

18. ^{^}

    Postlarvae are 15-20 days-old shrimp (depending on the species) that distinguish themselves from larval individuals because they resemble the adult form, including the presence of functional setae (tiny hairs) on the pleopods (swimming legs).

19. ^{^}

    Credit to Laura Duffy for this suggestion.

20. ^{^}

    While in 2020 251,093 tonnes of A. japonicus were captured from the wild (data viewed: September 2022), captures in 2019 for the same species reached 402,061 tonnes–similar to previous years. Given A. japonicus’ small size, even little variations in tonnages lead to significant variations in numbers of individuals, so such an important difference in tonnages represents a difference of 2.2 to 30.2 trillion individuals between 2019 and 2020. See the Appendix for further discussion of this decrease in catches.

21. ^{^}

    Credence levels cited in this paragraph reflect both authors’ subjective credences.

22. ^{^}

    If it were proven that 2020 was an anomalous year for Acetes japonicus landings, our credence level would be higher. 

23. ^{^}

    This source may not be accessible if the reader is retrieving it from outside the US. 

24. ^{^}

    Note that these two sources are in Spanish. 

25. ^{^}

    Last accessed: July 2021. 

26. ^{^}

    It is probable that these average weight losses vary by species, but I was not able to find such specific information. Therefore, I used this data for all species to which the conversion factor was applied. 

27. ^{^}

    We thank Phil Brooke for raising this point.

28. ^{^}

    Heu et al. (2003), for example, demonstrate that moisture in shrimp’s muscle is higher than that in shrimp’s by-products–i.e., non “edible” parts. Still, those differences are just around 5%. These findings suggest that hypothesis (2) is somewhat plausible, but also that drying non “edible” parts could affect the total dry weight of the animals, in particular, if they are oven-dried–as Heu et al. (2003) did. However, Chu et al. (1995) just left the animals to dry in the open air for an unspecified period of time.

29. ^{^}

    I.e., akiami paste shrimp (Acetes chinensis and Acetes japonicus).

30. ^{^}

    When exploring fishery production data by country, FAO indicates with an "E" those figures that are estimates as opposed to officially reported data. Thus, data for these countries have a higher level of uncertainty, and therefore they are more likely to be inaccurate. In September 2022, that was the case of Bahrain, Benin, Brazil, Cambodia, Cameroon, China, El Salvador, Eritrea, Fiji, France, Gabon, Greece, Guinea, Haiti, Indonesia, Iran, Israel, Ivory Coast, Kuwait, Lebanon, Libya, Mauritania, Myanmar, Nicaragua, Papua New Guinea, Sierra Leone, Solomon Islands, Tunisia, Uruguay, Venezuela, and Yemen for 2020 data of some species. 

31. ^{^}

    Mechanical peelers require a supply of shrimp within a fixed size range. If shrimp are too small, they are considered uneconomic to peel, and therefore, are discarded (Torry Research Station, 2001).