Totoaba macdonaldi Farming Yields Big Gains, Minimal Risks

Illegal wildlife trade is a multi-billion dollar industry that drives biodiversity loss through unsustainable harvest¹, spreads zoonotic disease², and threatens animal welfare³. The Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) provides a regulatory framework that aims to ensure that international trade of CITES-listed wild animals and plants does not threaten their survival. Regulatory interventions such as trade bans and controls are associated with decreased poaching levels for some species, including elephants⁴ and wild cats^5,6. Yet, for many species, regulatory interventions have failed, and illegal trade in black markets continues to flourish^7,8. In such instances, supply-side interventions such as conservation farming can theoretically bolster conservation by lowering poaching incentives^9,10,11. Supply-side interventions have occasionally succeeded at reducing poaching and recovering wild populations – e.g. vicuña and crocodilians^12,13 – but they have also failed – e.g. green python and African elephant ivory stockpile sell-offs^14,15. Uncertainty around conservation outcomes from market-based approaches is driven by the complex nature of illegal market dynamics, which has led to continued reliance on trade bans and controls^8,16.

Determining whether farming will succeed or fail requires a holistic understanding of a specific illegal wildlife market¹, including the interplay between market conditions and species life history^9,17. Studies have pointed to a common set of farming pitfalls. Species with slow individual growth rates and low fecundity are often unable to grow supply quickly enough to displace illegal products. Further, if poaching costs and barriers to entry are low, it is impossible for farming to reduce poaching^9,11. Demand-side concerns are focused on substitutability between farmed and wild products. Consumers of wildlife for medicinal or conspicuous purposes have long been assumed to prefer wild products for greater perceived potency or associated social status^18,19,20, however a variety of factors have been shown to drive consumer purchase including legality, appearance of product, or even product availability²¹. Here, we develop a quantitative framework that comprehensively considers the complexity of wildlife trade markets while accounting for detailed species-specific and market information.

Another critical factor in driving the success or failure of farming is market structure: illegal markets are often characterized by imperfect competition – where an individual trader or a small number of traders (i.e., middlemen, cartels, gangs, or other criminal organizations) dominate illegal trade and exert significant control over market prices. This can typically arise with endemic species with no close substitutes.

A bioeconomic model that predicts how imperfectly competitive markets will respond to competition from farming was developed almost two decades ago^22,23. Predicted strategic responses depend on how a trader chooses to compete with farming. If a trader responds by price setting (an aggressive response where the trader tries to undercut farmed prices and take market shares), then poaching pressure will increase and can lead to the collapse of the wild population. On the other hand, if traders respond by quantity adjustment (a mutually beneficial response where the trader competes on the amount of output introduced to the market to supply the existing demand, letting market prices adjust), poaching pressure is reduced and wild populations have the possibility to increase. This model has been widely used to both justify^24,25 and discourage¹¹ prospective farming initiatives.

We explore the biological and economic performance of conservation farming for totoaba swim bladder in the context of poaching and trade under different strategic interactions that characterize market conditions²⁶. Specifically, we examine the evolution of poaching and wild totoaba biomass, as well as prices and profits for different economic actors. The lifecycle for totoaba has been successfully closed in aquaculture, and the species is currently farmed in Mexico for domestic meat production. Totoaba is endemic to Mexico’s Gulf of California and is threatened by a lucrative illegal international trade for its large swim bladder^27,28,29. A single totoaba swim bladder can sell for up to $80,000 USD per kilogram in Chinese end markets, where it is purchased for special occasions, gifting, and speculative investment^30,31,32. For nearly half a century, international trade for totoaba has been prohibited, and the legal totoaba commercial fishery has been closed. Historically, the damming of the Colorado River led to a sharp decline in both the totoaba population and the ecosystem’s carrying capacity. However, the primary threat to the species is ongoing poaching activities³³. Illegal fishing and trade continue and are controlled primarily by a single criminal organization (a cartel). This cartel has sought to monopolize illegal and legal fisheries along the northwestern states of Mexico, including other luxury seafood products such as geoduck (Panopea generosa and P. globosa) and abalone (Haliotis spp.)^34,35. The introduction of a farming sector capable of capturing market share will cause the cartel to respond strategically^23,34, as the non-cooperative interaction with the farming sector decision will determine its profit. Strategic responses may include market interactions but may also include deterrence through extortion or farm takeover.

There is an urgent need to reduce poaching for totoaba, as the vaquita (Phocoena sinus), a critically endangered porpoise also endemic to the Gulf of California, is caught as bycatch in gillnets used to catch totoaba. The vaquita is on the brink of extinction as there are now fewer than fifteen individuals remaining³⁶. Furthermore, illegal trade has had negative social welfare consequences, as cartels are increasingly extorting Mexican fishing communities^34,35. Despite Mexico’s attempts to stop totoaba poaching through various enforcement mechanisms, the country recently received wildlife trade sanctions (suspension of all Mexican commercial trade in CITES-listed species) for taking inadequate action^37,38. If successful, conservation farming presents a legal alternative to reduce illegal fishing with no detrimental impact on vaquita by manipulating market structure.

Of several possible biophysical, market, or regulatory roadblocks to conservation farming success¹⁰, we evaluate four key factors that could undermine conservation farming success for totoaba³⁹. First, the substitutability between farmed and wild products: the exceptionally high value of rare, wild-caught swim bladder has raised concerns that a parallel market will emerge where wild-caught swim bladders are considered a separate product from farmed swim bladders. The primary concern from this line of reasoning is that imperfect substitutability would not displace demand for wild-caught products. Second, even if substitutable with wild products, farmed products may exacerbate poaching by allowing for easier laundering of wild products through legal supply chains⁴⁰. Third, increased end-market demand: either by expanding existing consumer pathways, or by gaining new consumers who change their willingness to purchase when the stigma of illegality is removed⁴¹. Finally, the cost of farming production must remain lower relative to poaching to ensure the long-term economic viability of farming^9,10. As these roadblocks can also occur simultaneously, exacerbating the risks, we investigate their compounded effects.

We assemble and leverage a wealth of information on the totoaba stock, poaching sector, and farming sector to estimate the effects of market structure on poaching harvest and stock biomass. We focus on the market structure that best characterizes the current totoaba trade – a vertical monopoly where a single monopolist trader controls the entire supply chain – and evaluate how this trader will respond strategically to competition from farming. Additionally, we consider the case of multiple traders and farming operations, as well as a monopolist trader taking over farming. We also conduct sensitivity analyses for common pitfalls including substitutability between farmed and wild product, increased demand, and cost of the farming sector relative to the poaching sector. Our aim is to identify an effective policy space where all supply, demand, and market structure parameters align to ensure that conservation farming will reduce poaching.

We examine the effect of market structure and competition on poaching a population of wild animals using the logistic growth function (Fig. 1). The supply of poached product is driven by interactions between multiple poachers and a single trader, who acts as a ‘middleman’: an exclusive buyer from poachers and the sole supplier to the end market. Poachers determine their level of effort and catch by maximizing the difference between their revenue from selling to the trader and their fishing costs, which rise as the stock declines. As the sole seller in the end market, the trader sets prices and quantities that maximize its profit, factoring in demand characteristics. This, in turn, determines the quantities the trader purchases from the poachers.

Poached product supply intersects with population growth producing stable and unstable equilibria, depending on market and species characteristics. If poaching pressure is high relative to population growth (i.e., when demand is large and inelastic or poaching costs are low), a single stable equilibrium point is observed with a low wild abundance (an overharvested population). A combination of increased rarity and escalating value can enable extremely high poaching pressure which leads to population collapse, known as the anthropogenic Allee effect⁴². In the opposite scenario, where poaching pressure is low relative to population growth (i.e., when demand is small, or poaching costs are prohibitive), a single stable equilibrium point is observed with a high wild abundance (a healthy population). Between these extremes, two or three potential equilibria can emerge, with uncertain results that depend on the initial size of the population: a large initial population will result in a high abundance equilibrium point, and a small initial population will result in a low abundance equilibrium point.

To assess expectations for totoaba, we first calculate equilibrium points for the stock in the absence of conservation farming under vertical monopolistic conditions (hereafter referred to as monopolistic conditions for ease) (Fig. 2). A single trader exists in a single location where he is the sole buyer, typical of endemic species such as totoaba^43,44. The trader sells poached harvest on an end market where prices and quantities can be manipulated.

Next, we add conservation farming to the monopolistic market structure, creating a duopolistic market (Fig. 2). We calculate equilibrium points for the totoaba stock if a monopolistic trader responds to conservation farming through quantity adjustment or price setting. From a policy assessment perspective, any scenario where poached harvest produces a single high stable equilibrium point, and the monopolist cartel loses income, presents clear conservation and social welfare benefits.

Totoaba stock under monopoly is sensitive to cost structure

A monopolistic market structure best characterizes the present consolidated totoaba trade^34,35. In this setting, a single monopolist trader restricts the supply of wildlife products to consumers, leading to increased prices and profits for the monopolist.

We initially calculate equilibrium points for totoaba assuming a quadratic cost structure, consistent with the original model (Fig. 3). Under the quadratic cost structure the totoaba wild stock biomass remains at a high steady-state equilibrium of 17,259 mt. However, we expand upon the quadratic cost structure, introducing a linear quadratic cost structure to account for energy costs associated with fishing. A linear quadratic cost structure more accurately represents new poachers being recruited to the fishery as fishing opportunities increase^45,46.

We find that under monopoly the linear quadratic cost structure is sensitive to cost parameter specifications, where relatively small changes in cost parameters can cause multiple steady states to emerge (Fig. 4). If an increase in poaching comes at a small cost increase compared to historical average costs, the aggregate cost is close to linear (e.g., W₂ = 0.47 and below, compared to baseline W₂ = 0.57). In this case, a low steady-state equilibrium of 1106 mt, an unstable intermediate equilibrium arises at 1842 mt and a high stable steady-state equilibrium of 17,277 mt in the vertical monopoly. Our model uses the best available information on the totoaba fishery, but uncertainty surrounding the projected evolution of fishery-wide poaching costs warrants a cautious assessment of monopoly performances: while it could maintain a healthy population, it can also lead to stock collapse.

Farming produces conservation benefits

While our results show that totoaba stock may remain healthy under the current monopolistic market conditions, these results are sensitive to changes in poaching costs (Fig. 4). We ask if conservation farming can improve upon the status quo by producing a robust single high stable equilibrium point and reduced cartel profits.

We add conservation farming to the monopolist model and now have two ‘firms’ – a trader and a farmer – competing on a duopolistic market. When farming supplies legal product to end-market consumers, the demand for illegal product will fall, assuming that wild and farmed products are substitutable (an assumption we explore later).

If illegal, imperfectly competitive markets arise, they are almost always characterized as competing through quantity adjustment^47,48. Under the assumption that products are substitutable, it is more profitable – and therefore more likely– for both firms to compete through quantity adjustment⁴⁹. When goods are substitutes, if both firms restrict the quantities supplied, they both enjoy higher prices. If they flood the market, prices and profits collapse. In the case of totoaba, we find that if traders respond through quantity adjustment under the linear quadratic cost structure, then the wild stock biomass increases by 5.46% (compared to a monopoly) to a steady state equilibrium of 18,220 mt, or to 90% of carrying capacity (Table 1). This represents a reduction in poaching harvest of 28.27% and $195.16 million USD of annual lost profit to the trader.

Even if traders respond aggressively through price setting, considered a less likely response⁴⁹, a single high equilibrium emerges (Fig. 3). Price setting is considered a much less likely response to competition because the trader would face steep profit losses. Under the high steady-state equilibrium with the linear quadratic cost structure, wild stock biomass decreases by 0.24% relative to monopoly, to a steady-state equilibrium of 17,235 mt, or to 85% of carrying capacity (Table 1). Although the high steady-state reflects a relatively small increase in poaching harvest by 5.85%, it would result in $313.84 million USD of annual lost profit to the cartel, making this strategy unlikely.

We challenge a key conclusion from the original bioeconomic model by arguing that price setting does not universally lead to increased poaching pressure^22,23. Farming puts an upper bound on the price traders can pay to poachers in order to remain competitive. When the cost of farming becomes lower than the combined cost of poaching and trading, price-setting competition does not inevitably result in the overexploitation of the wild stock. This is because when farming costs are low, traders have an incentive to maintain large stocks by poaching less to remain competitive with farmers. This limits the price paid to poachers. On the other hand, when farming costs are large, traders have an incentive to poach more, as they can afford to pay the increase in fishing costs corresponding to lower stocks, while remaining competitive with the farming sector. In the case of totoaba, species specific traits and market characteristics result in a slight increase in poaching in the price setting scenario. However, if the carrying capacity were smaller, or demand larger, the price-setting equilibrium would result in conservation benefits (Supplementary Fig. 8).

Farming is beneficial under alternative market structures

While we focus on the effect of conservation farming on a monopolistic market structure, given that this scenario best represents our best understanding of the totoaba fishery, we explore how alternative market structures affect poaching intensity and wild totoaba population size. First, we revise our status quo scenario to capture multiple traders (cartels) controlling the totoaba trade such that the status quo is best represented by a duopoly or oligopoly market structure, rather than a monopoly. Next, we add competition from a single farmer or multiple farmers under this alternative duopoly or oligopoly market structure. Organized crime groups in Mexico have been shown to extort otherwise legal fish processing operations³⁵, and we explore the possibility of a trader taking over farming operations.

We find that in the absence of farming, a status quo with a trader duopoly or oligopoly has less incentive to restrict supply of poached product and therefore poaching will increase. Under this scenario, multiple population equilibria emerge, all of which generate a smaller wild population than in the monopoly case, with a low stable equilibrium point that can lead to population collapse (Supplementary Fig. 1 and Table 1). When we add farming, the competition among traders and farmers unambiguously leads to lower poaching levels as long as the number of farmers exceeds (or equals) the number of traders (Supplementary Fig. 2) in the more likely quantity adjustment scenario. The price setting scenario is insensitive to the number of players. Lastly, we examine the implications if the cartel seized control of farming. We find that poaching is reduced and the wild totoaba population increases by 6.57%, due to relatively low farming costs (Supplementary Fig. 3 and Supplementary Table 3). Additionally, profits increase by 17 million USD ( + 4.4%), making this scenario plausible. These alternative scenarios strengthen the case for conservation farming, given that farming presents lower poaching and higher wild totoaba populations relative to the alternative status quo scenarios.

A subsidy can keep farming competitive

We find that the cost of conservation farming for totoaba can be high and still competitive with poaching, but this is contingent on the cost for traders also being high (Supplementary Fig. 6). When farming costs are large, if traders compete with poachers under the more likely quantity adjustment response, both parties limit the quantity supplied to maintain high prices and the population remains healthy, even increasing by nearly 6% from the monopoly steady state. However, if traders compete with farmers by price setting, high prices of farmed substitutes give leeway to traders to increase fishing pressure and afford the increase in fishing costs associated with declining population stocks while undercutting farmed substitute prices. This can lead to a decrease in the wild population biomass modestly by 0.24% from the monopoly steady state. Policymakers can support farming success by subsidizing farming to keep the cost low while maintaining enforcement to keep the cost of poaching high (for totoaba this includes marine patrols, fisheries closures, and gillnet bans). To mitigate the possibility of stock decline under the less-likely price-setting response, we identify that maintaining conservation farming below $77,339 USD per mt of totoaba (amounting to a 14% subsidy on unit production cost and $5.57 million USD per year) will prevent any increase in poaching pressure under either competitive response, assuming no effect of law enforcement in our baseline model. Additionally, operational costs may rise due to security threats posed by the cartel, leading to higher protection expenditures. To ensure farming remains competitive, efficient law enforcement is paramount.

Opening the market to farmed totoaba risks lowering the costs of transaction for poached products, thus creating a “parallel market”³⁹. We analyze the potential laundering effect by assuming large illegal transactions costs that vanish once farming is introduced (Supplementary Fig. 4 and Data Table 2). We find that conservation farming still leads to substantial poaching reductions in the quantity adjustment scenario, while the unlikely, low profit price-setting scenario risks increasing poaching.

High substitutability is important to farming success

Laundering of wild product through legal supply channels is a key concern when farming is introduced to an otherwise illegal market^14,40. We highlight that high – but not perfect – substitutability is key to the success of farming, given that it leads to larger conservation benefits in the quantity-setting equilibrium, under the assumption that demand remains stable (Fig. 5). Swim bladders have a wide variety of uses and values, and it is possible that farmed totoaba swim bladders may enter into these different product streams, or a product stream unique from wild totoaba swim bladders³¹. In the case of no substitutability, two separate, non-competitive markets emerge. In this scenario the status quo is maintained, both firms set high prices, and traders continue to operate as a monopoly because farmed product does not compete with wild product. For low to medium substitutability (i.e., 10-50%), traders and farmers operate on relatively independent markets. Both farmers and traders are still likely to set prices high and limit quantities, as the impact of their competitors on their respective market segments is limited. As such, both price setting and quantity adjustment yield modest increases in steady state population, and slightly curb poaching. For high substitutability (i.e. 75% and 90%), competitors have significant impacts on each other’s market shares, fueling stronger responses in either the price setting or the quantity adjustment equilibria. This yields more marked responses in terms of poaching and steady state population, as reflected in our main results. At the other extreme, in the case of perfect substitutability, consumers prefer the cheaper option without any preference of source. This increases the intensity of potential price-setting competition between firms and further depletes the stock, but also the incentive for traders and farmers to both restrict quantities, further reducing poaching and increasing steady state population. To comply with CITES captive breeding guidelines totoaba must be identified as farmed⁵⁰, and distinguishing between products to meet regulatory obligations can artificially lower substitutability. Given that opening of a legal market may create a parallel market, we show that maintaining large substitutability is key to achieving poaching reductions of up to 13% (Supplementary Fig. 5).

Interaction between substitutability and demand

The value of totoaba swim bladder is tied to rarity, and while demand evolution is an open empirical question, we test the sensitivity of our results to simultaneous changes in demand and substitutability (Fig. 5). Totoaba swim bladder purchases are ‘conspicuous consumption,’ luxury products commonly purchased for social status and speculative investing by wealthy consumers^31,51. A decrease in swim bladder price resulting from conservation farming may actually undermine the desirability of totoaba swim bladders in Chinese end markets, given that the high monetary value is linked to high social status⁵². However, some increase in demand may be expected if a legal product becomes available, as law-abiding consumers will be more likely to purchase wildlife products when those products are traded and purchased legally¹⁰. Under our high substitutability assumption (75%), competition through quantity adjustment can withstand a 40% increase in demand, whereas competition through price setting is not robust to demand increases. For price setting, a demand increase of 40% would cause the equilibrium population to decrease by 10% from the monopoly status quo, increasing poaching by 216 mt.

There is a much higher threat to the wild population if demand increases under low to medium substitutability (i.e. 10-50%), given that this additional demand cannot be fully met by farmed product (Fig. 5). In the best-case and most likely scenario, medium substitutability (50%) can meet a 20% increase in demand if competition occurs through quantity adjustment, although uncertain outcomes (e.g., high and low steady states) start to emerge if demand increases by 60% or more. In the worst-case scenario, if competition occurs through price setting and products have medium substitutability (50%), any increase in demand reduces the wild population from the status quo. While increases in demand of 20–40% still produce a single high equilibrium point, the population size is lower than under monopoly. Furthermore, if demand increases beyond 80%, uncertain outcomes emerge, with the wild population either stabilizing at a high equilibrium point (14,322 mt in the price setting scenario; 15,886 mt in the quantity adjustment scenario) or being pushed to a low equilibrium point (ranging from 763 mt in the quantity adjustment scenario; 909 mt in the price setting scenario). We recommend that stated preference investigations on wild versus farmed product should be undertaken in Chinese end-markets and that these investigations include questions focused on perceived social status benefit and legality²¹.

Conservation benefits depend on the interplay of demand, costs, and substitutability

Our analysis provides a quantitative framework that can identify an effective policy space where all supply, demand, and market structure parameters align to ensure that conservation farming will reduce poaching, improving greatly on the original bioeconomic model and the limitations of binary qualitative approaches^{10,11,22,23,53}. This bioeconomic model allows researchers to quantify: (a) how much cheaper farming must be relative to poaching to be competitive; (b) how much of a demand increase can be absorbed by farming; and (c) how substitutable must wildlife products be for farmed products to displace wild products. Critically, we also explore how the interaction between these factors may affect outcomes. Our results offer general and totoaba-specific policy solutions to help ensure that conservation farming remains in the effective policy space.

Our flexible framework can be applied to other species in other illegal markets to assess the potential effects of farming on poaching intensity and population size, however we do not explore the impact of our intervention on the vaquita. As vaquita has experienced a 98.6% population decline since 2011⁵⁴, a possible extension of our model is examining how a totoaba farming intervention a decade ago may have ameliorated such rapid declines. Lessons learned from the vaquita may extend to other species that are caught incidentally in multispecies exploitation systems. Additionally, our model can be extended beyond endemic species like totoaba to species with cosmopolitan distributions for which trade to an end market is sourced from multiple geographic locations. This cosmopolitan model extension builds on the different market structure scenarios we examine. Our approach provides a rigorous alternative to existing qualitative frameworks that are unable to analyze the interaction between multiple variables. While our analysis focuses on totoaba, the bioeconomic model is flexible and can be applied more broadly to other species and contexts to examine the effect of conservation farming on a wild population.

Here we briefly discuss our methods with an emphasis on the empirical application. Information on our theoretical conclusions from the bioeconomic model we revisited, lemmas and proofs can be found in the Supplementary Information Section 1 and 2. Detailed methods and additional sensitivity analysis can also be found in the Supplementary Information Section 2 and 3.

The poaching model

The growth of the fish stock follows a logistic curve and the stock is poached following a Gordon-Schaefer production model. Totoaba population growth parameters were obtained from the 2017 stock assessment, where the carrying capacity (K) was 20,226 mt, and the stock biomass in 2017 was 14,844 mt³³. The intrinsic rate of population increase (r) was predicted using the FishLife package in R, which estimates growth parameters using totoaba-specific life history data from FishBase⁵⁵. The growth equation is g(x) = rx(1-x/K) using a predicted r of 0.20. We do not consider potential effects of hyperstability of the stock resulting from poaching on seasonal spawning aggregations⁵⁶, or age structure.

Poachers optimally determine their effort to maximize their profit, with constant catchability, σ, and stock biomass, x, obtained from the 2017 stock assessment³³, and a linear quadratic cost of effort function, E. The poaching equation is q = σxE where σ = 0.00002.

Poachers are faced with a linear quadratic cost function C(E) = W₁E + W₂E². We calculate two poaching cost parameters W₁ (the linear coefficient of the cost function) and W₂ (the quadratic coefficient of the cost function) by (a) estimating total and average annual operating costs of the fishing fleet using semi-structured interviews conducted by the authors of this study; and (b) calibrating a linear quadratic cost function that matches historical data and predicts future cost evolution.

We conducted semi-structured interviews in the upper Gulf of California with two fishing cooperative leaders and four fishers in July and August 2018. These interviews informed annual poaching costs: food and fuel, labor, gear replacement, and bribes paid to fisheries officials. The fishery operates over six months with a variable number of active vessels, monthly fishing days, and sets per day³³. Poaching costs also include annual fleet-wide costs related to gear confiscations, vessel replacement, and fines, adapted and extrapolated from a summary of law enforcement actions provided by Mexico⁵⁷. The cost per fishing trip was estimated to be $5,051.26 during the low season (January and June), $8,385.34 during the mid-season (February and May), and $14,386.7 during the high season (March and April) (Supplementary Table 5). In our analysis we reconstructed a linear quadratic cost function with cumulative effort. We considered effort in each season cumulative with effort in less intense seasons. We used a low-season average cost for effort levels between 0 and low-season effort; for effort levels between low-season effort and cumulated low and mid-season efforts, we used a mid-season average cost.

We estimated the corresponding poaching cost parameters to match the observed average cost and modeled marginal costs at historical levels (resulting in cost parameters W₁ = 12,200 & W₂ = 0.57). Our low sample size precludes a robust statistical estimation of these cost parameters, e.g., of the historical cost function and of the evolution of costs if the fishery were to increase. To account for this uncertainty, we run a sensitivity analysis on two dimensions of costs. First, we use different estimates for the average cost and reconstructed total costs, ranging from −10% to +30% of our high season average cost estimates. Second, we test weights for the linear and quadratic costs, ranging from a purely linear cost (W1 = 14,386,7, W2 = 0) to a purely quadratic cost (W1 = 0; W2 = 3,74).

The resulting poaching profit function is calculated as follows: Π = pσxE − W₁E + W₂E²

Traders operate on the end market, taking prices as given (competitive scenario) or determining prices (monopolistic scenario) to maximize profits. Traders face a linear demand function. We estimate a linear demand function by regressing price data on estimated catch from 2014 to 2017³³, yielding the equation p(q) = α – βq where the intercept, α, is $1,625,837 USD and the slope coefficient, β, is $1,563.75 USD (see Supplementary Table 6). Price data were obtained from available literature that provided estimated weight and value of totoaba maw seizures^27,29,58,59. In addition to the literature review, valuable insights were obtained through personal communication with Wild Aid Investigators (pers. comm. Anonymous Wild Aid Investigators, 2018) as well as with local fishers and cooperative leaders in the upper Gulf of California, as previously described. The information shared by investigators and stakeholders was aggregated with the existing data from the literature. To ensure consistency and comparability, we standardized the weight measurements to grams and the currency values to US dollars. We assume that annual catch reaches the market during the same year, i.e, there is no stockpiling. As data are notoriously difficult to acquire for illegal trade, we pool observations and estimate a stationary demand function (Supplementary Table 6).

Traders buy totoaba from poachers at price s (USD/metric ton). The price paid to poachers balances demand from traders and supply to poachers. It decreases as the population increases, as fishing becomes less demanding. Traders also pay a unit transaction cost c (USD/metric ton), which we conservatively estimated to be zero. At a minimum this unit transaction cost includes transport (land and air travel), and payment to two or three ‘runners’ who carry up to ten swim bladders each (pers. comm. Anonymous Wild Aid Investigators, 2018). We know through anecdotal evidence that unit transaction costs are likely large³⁰.

The farming model

We use a linear profit model with fixed costs for aquaculture and estimate a unit farming production cost parameter v (USD/metric ton) using annual operational costs (labor, feed, vessel fuel, facility and administrative fees), as well as annual maintenance of pens (including cleaning) and vessels, using information provided by existing aquaculture facilities. The fixed set-up costs corresponds to the cost associated with the aquaculture pens (Supplementary Table 6). Growth rates differ in the wild and in captivity. Using captive growth rates obtained from personal communication with totoaba aquaculture producers, we consider harvestable size to be between 4.5 and 5 years old (an adult weight of 21.43–27.2 kg), associated with a swim bladder size between 417 and 529 g (Supplementary Fig. 7). A minimum farmed harvestable size of 4.5 years closely corresponds to the mean swim bladder size (500 g) and estimated adult totoaba size (25.7 kg), as reported in surveys of individuals harvested in the wild³³. We considered this to be the size at which farmed totoaba would be competitive with the average wild-caught totoaba. We assume that aquaculture operates on a homogenous rotation⁶⁰. The implications of this assumption are discussed (Supplementary Materials – Section 2.4.1). We compute the farming cost per metric ton as the capitalized sum of annual costs over 4.5 years at a 10% interest rate.

We include a substitutability parameter, which measures the imperfect substitutability between farmed and wild products in the linear demand functions. When farmed products are introduced, the linear demand function is modified such that pⁱ(qⁱ, q^k) = αⁱ–βⁱqⁱ–γq^k where qⁱ and q^f indicate the supply from the wild (w) and the farmed supply (f). This demand system emerges from a linear quadratic utility function in Supplementary Text (section 1.3.2). When demand intercept αⁱ are equal, and own price effect βⁱ = β^j = γ are equal, products are perfect substitutes. When demand intercepts are equal, but own price effects differ (βⁱ ≠ β^j), then γ²/(βⁱβ^j) denotes the degree of product substitutability.

At present, there has been no stated preference investigation for wild and farmed totoaba swim bladders in Chinese end-markets, although we know that the end-market economic value for fish maw is determined by taxon, size, and thickness of swim bladder³¹. Investigative work in Mexico reports that it is challenging to distinguish between wild and farmed specimens³⁰. Therefore, we assume high substitutability (75% product substitutability) and check for smaller substitutability values in our sensitivity analysis (Fig. 5) (see Supplementary Table 8 for a list of parameters).