Study area

This study focuses on the main commercial fishing area in the Amazon Basin and the life history areas of the three most overexploited migratory species. The commercial fishing area is defined here as that which is fished for urban fish markets between the Andean Piedmont and the freshwater areas off the Amazon estuary in the east (Fig 1). The life history areas of the migratory species considered here are based on those already defined [8, 14].

Data sources

Fisheries data included annual fishery catches from different sources between 1976 and 2019, though with incomplete information for all years and geopolitical units. These data were used as an indication of the potential importance of each species for each geopolitical unit to estimate the Maximum Sustainable Yield (MSY) of the industrial fishery in the estuary, and to project the fishery trends for the other species. Fish length data from goliath catfish fisheries in the estuary and in the Andean piedmont were used to assess stock health. Further details on the fisheries data sources are provided in S1 Text.

Data analyses

The complex assemblage of data was analyzed with the objective to determine: (i) the importance of each geopolitical unit in the total production of the studied species, (ii) the stock status of the goliath catfishes exploited by the industrial fishery in the estuary and by the artisanal commercial fishery in the Andean piedmont, and (iii) for B. rousseauxii and C. macropomum the time series analyses of the fish stock status based on historical catches.

Brachyplatystoma vaillantii in Amazon River estuary

The historical analysis of trawl fisheries for B. vaillantii represents a 34 year dataset between 1972 and 2006 (Fig 5 and S1 Table). We estimated the maximum yield in three periods when the catch reached a plateau after the fishing effort increased over time (S1 Fig), a condition that validates the assumption of the relative rate of the catch increase (RRCI) method (S1 Text) [22]. The 1976–1980 period corresponded to the first expansion of the trawl fishery of B. vaillantii in the estuary. The second period included an increase in the number of trawler fishing vessels from 43 to 66 boats in the period 1986–1991. The third period, 1997–2000, corresponded to the second expansion of the fishery after the historic lowest catch in 1992, when many boats began trawling together despite a total estuary industrial fishing limit of 48 boats [4, 24]. After these three periods, fishing effort was restricted by limiting the number of boats that could work together and by the establishment of a closed season.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The goliath catfish fishery region and overfishing in the Amazon River estuary.

(A) The industrial trawl fishing zone and the restricted fishing area based on legislation in force. (B) Annual catches (tons) of the B. vaillantii bottom pair-trawler fishing fleet in the Amazon estuary (yellow bars) available for the 1972–2006 period, combined with the Maximum Sustainable Yield proxies (tons) with 95% confidence limits (black line) and the F/F_MSY ratio (triangle).

https://doi.org/10.1371/journal.pone.0264490.g005

The estimated MSY was very high for the first period (18,591 t/year) and similar for the last two years (14,602–14,915 t/year) (Fig 5 and Table 2). We consider the MSY estimate for the first period overestimated, because of the anomalous catch recorded in 1977 (22,486 t) after the fisheries were in operation for only five years. Our analyses indicate that B. vaillantii catches first exceeded MSY in the 1970s (Fig 5 and Table 2). Brachyplatystoma vaillantii has suffered high fishing mortality (F) for almost the entire industrial fisheries period since the 1970s and exploitation levels were at least twice F_MSY in 80% of the years (Fig 5). Considering all strategies to reduce fishing mortality, the effectiveness of goliath catfish management based on historical data showed a significant downward trend in fishing mortality just a few years after trawling began (Mann-MKT: tau = -0.393, 2-sided p value = 0.0382), suggesting that the measures applied had some effect on reducing fishing pressure.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Maximum Sustainable Yield (MSY) estimates for B. vaillantii.

RRCI’ smoothed relative rate of catch increase.

https://doi.org/10.1371/journal.pone.0264490.t002

Brachyplatystoma rousseauxii in the western Amazon region

Of the 5,482 B. rousseauxii specimens measured in the Western Amazon Region, 696 were from Madre de Dios River (Peru) in the period 2002–2004, and 4,786 from Ucayali-Urubamba River (Peru) in the period 2004–2005. Analyses and estimates of the population parameters of reproductive populations of B. rousseauxii in the Madeira (Brazil) and Ucayali-Urubamba basins, the two largest headwater regions, revealed different trends among these regions (Table 3 and S2–S5 Figs). The growth parameter of the fishes from the Madre de Dios River (department Madre de Dios) was slightly higher than that of the Ucayali-Urubamba River (L∞-_MD = 149.3 cm, and K_MD = 0.35 year^-1 against L∞-_UU = 141.8 cm, and K_UU = 0.29 year^-1). However, the current mortality rates influenced by fishing (Z and F) were quite different, with much higher values in Ucayali-Urubamba River (Z_MD = 1.18 year^-1, and F_MD = 0.81 year^-1 against Z_UU = 2.13 year^-1, and F_UU = 1.8 year^-1). The F/F_MSY relation indicates overfishing in the Ucayali-Urubamba River (F/F_MSY-UU = 2.73), and thus also in downstream fisheries exploiting stocks before they arrive near the Andes. An opposite situation occurred in the Madre de Dios River, where F/F_MSY indicates a “healthy” condition (F/F_MSY-MD = 0.91), at least before the Madeira dams blocked upstream migrations of B. rousseauxii.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Population parameters of B. rousseauxii from Madre de Dios and Ucayali rivers.

https://doi.org/10.1371/journal.pone.0264490.t003

Brachyplatystoma rousseauxii in the Brazilian Madeira River

Historical fishery data recorded at or near Porto Velho, Brazil, likely indicate temporal trends in the abundance of B. rousseauxii migrating upstream in the Madeira River. Initially, we tested the trend in the historical series of fish landings during the period 1977–2012. Next, we analyzed the variation of the annual fish production before and after the 1990 decade when stock depletion was first observed in other regions [2, 3, 25]. The first period considered only 1977–1989, the second period considered 1990–1998, and the third period 2004–2012 (Fig 6). There was a significant decrease in B. rousseauxii capture (MKT: tau = -0.039; p<0.05) during the entire period considered (Fig 6). The annual yield average of the three periods was 102, 80 and 16 tons, respectively, with significant differences between them (K-W H = 9,43, df = 2, p <0.05). Although catches in the first two periods were not significantly different, they were significantly higher than in the third period (Dunn post-test, p<0.05), which indicates a recent sharp depletion of the migratory stock in the Madeira Basin. Considering the large-scale migratory behavior of this species, a general decrease in the stock in the Madeira Basin could be due not only to overfishing but also be a consequence of the construction of the Santo Antônio dam that began in 2008 and was completed in 2011 [5, 26]. Overfishing and dams point to the deleterious synergy of these impacts on migratory fishes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Historical trends in the capture of C. macropomum and B. rousseauxii.

(A) Negative exponential trend adjusted to C. macropomum landings in Manaus (tau = -0.809; p<0.05); (B) Linear trend adjusted to C. macropomum landing in Tefé (tau = -0.032; p>0.05); (C) Linear trend adjusted to two periods of B. rousseauxii landings in Porto Velho (tau = -0.039; p<0.05).

https://doi.org/10.1371/journal.pone.0264490.g006

Colossoma macropomum in the Central Amazon

The maximum annual historical harvest of C. macropomum is concentrated in the state of Amazonas, Brazil, which accounted for three-fourths of total landings in the Amazon (Fig 7). Manaus, the capital of Amazonas, Brazil, is historically the most important fishing port for C. macropomum and alone accounted for up to 83% of the total monitored catch of the species at the city’s principal market where most of the catch is commercialized (Fig 7 and S2 and S3 Figs). The maximum recorded catches in Manaus occurred in 1976, when ~13,600 tons were registered [18]. Records for subsequent years in this city show a clear downward trend in catches (MKT: tau = -0.809; p<0.05), dropping to 41% of the maximum 10 years later in 1986, and to only 3%, 29 years later in 2004, the last year Manaus landings were recorded [16, 17, 27–29] (Fig 6). In the intermediate city of Tefé (Brazil) between Iquitos and Manaus, and which is associated with the Mamirauá Sustainable Development Reserve, C. macropomum catches have oscillated between 147 and 8 tons, with no clear temporal trend (MKT: tau = -0.032; p>0.05). The Tefé catches, however, are not representative of the overall high pressure on the species since its maximum annual landings represent only 1% of Manaus maximum annual landings (Fig 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Main states or departments and cities associated with overfishing.

Production in percentage categories. The highlighted cities are responsible for at least 50% of the production. Overfishing of the flagship species considered here needs to be controlled from the major urban ports that account for most of the landings.

https://doi.org/10.1371/journal.pone.0264490.g007

Goliath catfish overfishing

There are seven species of goliath catfishes, and all are known to be migratory to some extent [8, 40]. The goliath catfishes are major river channel and estuarine predators found in the Amazon and Orinoco basins and some species in the coastal rivers of the Guianas [41]. Several of these species in the Amazon make the longest freshwater fish migrations in the world [8]. The migratory life cycle of B. rousseauxii can be at least 11,600 km between the Amazon River estuary and the far western Amazon, including the Andean Piedmont to at least 250 m in elevation [8, 10, 38, 40–42]. The migratory life cycle of B. vaillantii is less known but reaches to at least the Andes in the Marañón River of Peru [8, 43].

The general migratory routes of the targeted goliath catfishes considered here include the Amazon River and its turbid tributaries with headwaters in the Andes or nearby headwater areas in the Purus and Juruá basins. Brachyplatystoma rousseauxii also enters low sediment tributaries, such as the blackwater Negro River and clearwater tributaries arising on the Brazilian and Guiana shields. Statistically distinct size classes of B. rousseauxii recorded in commercial and research fishing clearly indicate an east to west increase in size, thus the separation of nursery and spawning regions [41]. Genetic analyses of goliath catfish stocks are not yet conclusive, though hypotheses suggest geographical variability for each species. To date, a single stock of B. vaillantii has been identified in the Amazon Basin [11]. Genetic studies of B. rousseauxii have shown mixed results, ranging from no genetic segregation between the populations in the estuary and near the border of Brazil, Colombia, and Peru [9], to at least three clusters in five locations sampled in the upper Madeira Basin in the Bolivian Amazon [44]. Although genetic studies are insufficient alone to rule out homing, otolith isotope analyzes indicate natal homing of B. rousseauxii in the upper Amazon [26]. A combination of tagging experiments and genetic and otolith analyses will be needed to determine conclusively if homing occurs.

Industrial and artisanal fishing operations exploit goliath catfishes in the Amazon River estuary, whereas only artisanal fishing takes place in inland waters. The industrial fishing fleet in the estuary is composed of sea-worthy vessels that use bottom pair-trawlers that operate in fresh waters up to 50 km offshore of the Amazon River mouth [1, 45–48] (Fig 5). Artisanal fisheries for goliath catfishes across the Amazon sell fish to urban centers and refrigeration plants and employ a combination of drifting gillnets, seines, longlines in river channels, and gaffs at cataracts [49, 50].

Fishery subsidies and/or export markets were external drivers that historically contributed to goliath catfish overfishing. As early as the 1990s, B. vaillantii was reported overfished in the estuary [4], as also was B. filamentosum in the western Amazon [7]. Brachyplatystoma filamentosum spawns across the Amazon in great contrast to the long-distance migratory species, B. rousseauxii and B. vaillantii, whose nurseries and spawning areas are widely separated at the farthest eastern and western regions of the basin. Subsidies and export markets are common characteristics of goliath catfish fisheries in Brazil and Colombia, the two most important countries exploiting these species in the Amazon. In the early 1970s, industrial-scale fishing companies operating in the estuary, and subsidized by the Brazilian government, focused on B. vaillantii because of its export mainly to the United States [1, 45, 47]. In Colombia, between 1978 and 1982, increased coca production for the international cocaine market led to an economic boom that indirectly stimulated the catfish fisheries of the Amazon, Putumayo, and Caquetá rivers to supply fish to national markets [34, 50]. The Colombian port city of Leticia (department of Amazonas) served as the main base for these export fisheries based on artisanal operations. Subsidies and export markets also led to the establishment of fish processing companies in the Brazilian states of Pará and Amapá and the Colombian department of Amazonas. Other than the trawl fishery in the estuary, investments were made by national and regional (i.e., department or state) governments, without concomitant catch monitoring.

The Brazilian bottom pair-trawl operations in the Amazon Estuary represent a rare example of a single-species and industrial-scale freshwater inland fishery in the tropics, given that the main target is B. vaillantii, whose catch represents 76% of recent trawler production, followed by the 7% catch of B. rousseauxii [51]. Trawler bycatch has been reported but there are few details on how much is commercialized or discarded [52]. Initial government-based studies indicated MSY above 20,000 t/ year [24]. However, our data and recent estimates of MSY of B. vallantii indicate that the initial MSY was highly overestimated [53]. By 1992, the year with the lowest catch (6,299 tons), overfishing resulted in the collapse of B. vaillantii as an export species. It fell from the third most important marine or freshwater fish species exported from all of Brazil in 1986 and 1987 to no longer appearing in export lists [4, 24, 41]. The strategies to reduce fishing mortality have been focused on the minimum mesh size of the cod-end of the trawl net, the size of the fishing fleet, protected areas, and closed seasons [46]. Our data indicate that these strategies reduced mortality, although not enough to prevent overfishing.

Overfishing of B. rousseauxii, the fish species undertaking the longest migrations in the Amazon Basin, was reported for the Caquetá River, Colombia, in the period 1995–1998 [2] and Amazon River, Brazil, in the period 2002–2003 [3, 25]. To reach the western Amazon, the migratory catfish populations must first avoid capture along a several thousand-kilometer-long route where commercial fisheries operate in the Amazon River and its tributaries in Brazil, Colombia, and Peru [1, 25, 34, 47]. In contrast to B. vaillantii, after migrating upstream from the estuary, B. rousseauxii does not return to it, though the juveniles of both species born in the western Amazon migrate back to the estuary. Brachyplatystoma rousseauxii is heavily exploited during its long 2–3-year upstream migration to the western Amazon, thus making it vulnerable to both growth overfishing of young fish by the trawler fleet in its estuary nursery and recruitment overfishing of mature stocks upstream [2, 3, 25].

Colossoma macropomum overfishing

Colossoma macropomum is the second largest scaled fish in the Amazon and one of its most highly prized food species and cultural-culinary biodiversity symbols. Historically, C. macropomum reached 100 cm TL and lived up to 65 years of age [16, 19]. The species is exploited by fishers in nine states or departments, ranging from the eastern Amazon near the estuary in the states of Amapá and Pará in Brazil to the far west in the departments of Cochabamba and Beni in Bolivia and Madre de Dios and Ucayali in Peru. In the 1970s, it was the most important species commercialized in Manaus, the Amazon’s largest inland fish market [17, 18]. It is the iconic symbol of the flooded forests because of its fruit- and seed-eating habits and its role in dispersing many seed species that pass through its intestinal system and remain viable for germination [16, 54–56]. The species is widely distributed in the Amazon Basin and found in all river types. Its nurseries, however, are found mostly in floodplain lakes of turbid rivers with headwaters in the Andes, and include the Amazon, Madeira, Purus, and Ucayali [57]. When water levels begin to fall rapidly, adult fish largely move from the floodplains to the river channels where migrations take place at various intervals until the fish spawn with rapidly rising river levels. After spawning, the fish migrate to flooded forests of all river water types. In addition to fishers’ and our field observations, C. macropomum migrations have been verified by ichthyoplankton fishing in river channels where the newborn can be carried 400–1,300 km downstream before entering a floodplain nursery [16, 58, 59]. Thus, it can be assumed that adults, on average, migrate a similar distance upstream during the six-month period or longer that they reside in Andes-Amazon River channels before spawning.

The widespread use of gillnets beginning in the 1960s made it possible to catch large quantities of C. macropomum in floodplain lakes and flooded forests [35]. The increased use of purse seines also greatly increased fishing pressure on adults found in river channels during the low water periods when these fish are migrating or seeking refuge along woody shore areas. Immature, 2-3-year-old C. macropomum are among the largest fishes exploited in the open waters of floodplains [1]. Although principally a fruit- and seed eater after about 2 years of age, young C. macropomum are highly zooplanktivorous and feed in open waters of floodplain lakes where they are vulnerable to both gillnets and seines [16, 18, 60]. As adult C. macropomum became rarer in the 1980s, local communities increased their catches of young fish, an occurrence that was and still is easily observable in urban markets in Brazil and Peru (Fig 4). Once C. macropomum adults became relatively rare, an adult fish of 12 kg or larger sold for $75–100, thus compensating for the low catch per unit of effort required to capture it. Additionally, as passenger-freight boats began to be used to transport ice to and fish from far-flung communities along the rivers, catches could be delivered from hundreds of km away on a weekly basis [61]. This new type of fish transportation not only replaced urban-based large fishing boats that previously accounted for most of the C. macropomum catch, but also increased the maximum distance from urban markets where captures were made. In the case of Manaus, the distance expanded from perhaps 500 km in the 1970s [18] to at least 1,000 km after the year 2000, a phenomenon that has been referred to as the ‘rainforest metropolis defaunation shadow’ [62].

The most robust datasets for C. macropomum stock assessment are from commercial fisheries, though with serious hiatuses. Empirical abundance data independent of fisheries data is available for a few locations and measured by ichthyoplankton densities [63]. The life history of the species is reasonably understood, including life history areas [14, 16], fish size range [17, 64], age [16, 17, 65, 66], general migration patterns [57], and reproduction [16, 19, 58, 67, 68]. Our analyses rely mostly on historical fisheries data to detect overfishing trends, especially as indicated by data for the large Manaus market that accounts for most of C. macropomum catches.

Due to its flooded large forests and floodplain lakes on which C. macropomum depends [14], the central Amazon is the center of C. macropomum fisheries. The first yield-per-recruit (Y/R) study of C. macropomum covered the period 1976–1978, and that included a large area of the Central Amazon, concluded that there was no growth overfishing for this period [17, 69]. Later studies, however, indicated a 50% reduction in catch per unit of effort between 1979 and 1986 and that the C. macropomum stock had been overfished in the Central Amazon [29]. This coincided with a period when the Manaus fishing fleet expanded its fishing river distance from 1,700 km to 3,700 km [70]. Subsequent Y/R studies in the 1980s and 1990s also indicated C. macropomum overfishing in the lower Amazon centered on the city of Santarém [6, 71, 72], in the Manacapuru tributary just west of Manaus in the Central Amazon [73], and farther west centered on the city of Tefé near the Amazon (Solimões) River [74]. A large study of length frequency distribution of C. macropomum in the Manaus market between1993 and 2006 further indicated growth overfishing of the species [64].

Female wild C. macropomum mature as small as 45 cm, though most are larger than 55 cm [16]. The 55 cm maturity baseline, which includes fish about four years of age [65, 66], has also been used as the legal limit for commercial catches since the 1970s. By the 1990s, over 90% of C. macropomum sold in urban centers along the Amazon River were fish less than 55 cm, which was the minimum catch length legally allowed in the 1977–1978 period. By the 2007–2008 period nearly all C. macropomum sold in urban markets were illegal (Table 4). Recent larval studies in the Ucayali and Marañón rivers in Peru [39] and in the Solimões River in Brazil [75] revealed alarmingly low densities of C. macropomum. The absence of C. macropomum breeding adults is now so pronounced that fishers report they no longer detect reproductive adults of the species in the lower Amazon [76].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. C. macropomum size composition in three fishing ports of the Amazon.

https://doi.org/10.1371/journal.pone.0264490.t004

Colossoma macropomum contains a single and large panmitic population along the Amazon River and genetically differentiated populations in tributary headwaters, especially in the Madeira Basin above the cataract stretch in Rondônia, Brazil [78]. The low inter-population genetic differentiation in the Amazon River and lower and middle courses of its tributaries is probably related to the large-scale migratory behavior of the species that acts as a homogenizing agent [79]. Most of the commercial catch consists of this genetic stock. The widespread evidence of C. macropomum overfishing indicates that the large homogenous population is under a severe threat of collapse. Above the Madeira Rapids in Bolivia, earlier data indicated a relatively unexploited but minor stock compared to that in the Central Amazon [19, 80], but the status of the species in Bolivia is unclear now.

As a result of overfishing and the popularity of C. macropomum in aquaculture, it is either the most important or second most farmed species in the Amazon [16, 81]. Though data are minimal on farmed fish species in the Amazon, the total C. macropomum production sold in Manaus is probably greater than maximum wild catches recorded in the mid-1970s. It should not be assumed, however, that aquaculture has reduced fishing pressure on wild C. macropomum. Farmed C. macropomum are sold at about 1.5–3.0 kg, that is, immature fish and, as with wild fish of the species, are destined to the middle and upper economic classes. Large adults are only available from wild fisheries and have commanded the highest prices of any fish species since the 1990s, thus recompensing the effort invested in catching them despite their rarity [16]. Finally, despite the contribution from aquaculture, young wild C. macropomum also continue to be exploited across the Amazon, including in protected areas.

Impacts other than overfishing

The potential major impacts on fisheries other than overfishing are wetland and upland deforestation, dams, and water quality degradation. Though upland deforestation and water quality degradation could eventually surpass overfishing, wetland deforestation, and dams as the principal impacts on fisheries, at present these impacts are less limiting ecologically. Climate change is also expected to cause major hydrological changes [e.g., 42, 82]. In addition to overfishing discussed above, here we address dams and wetland deforestation as the two impacts that require immediate mitigation for robust fisheries management.

Reinstate government responsibility to collect urban fishery market data

The importance of collecting fisheries data in the Amazon dates to early Portuguese colonies in the 17^th century and was emphasized during the rubber boom at the end of the nineteenth century in the first major overview of the region’s fisheries [101]. Nevertheless, in general, data collecting has not been continuous, even after the exponential growth of the fisheries beginning in the 1960s. Early modern efforts to collect fisheries data in the 1970s and early 1980s were largely led by individuals with the support of various governmental or non-governmental institutions, and give a snapshot of the geographical scale, species exploited, catches, and fishing effort before overfishing became a widespread problem in the Amazon [18, 49, 102, 103]. These studies emphasized and demonstrated the importance of urban market and refrigeration company data collected in major cities for managing fisheries based on migratory species. In the last two decades some continuous data has only been collected by Peruvian and Colombian fisheries agencies and locally by the Mamirauá Institute located in Tefé, Brazil (S1 Text).

Unfortunately, the lack of continuous monitoring of fisheries across the Amazon over the years compromised the assessment of stocks and the establishment of adequate management measures to detect and prevent overfishing. Despite the Amazon’s large basin size and ever-increasing infrastructure development, academics, NGOs, and financial institutions still favor small-scale fisheries initiatives, usually under the social umbrella of community-based management, to guarantee at least the livelihoods of the human populations that live along floodplain lakes and river channels [99, 104–108]. It is unclear to what extent these initiatives have been successful at regional scales for fisheries management of large and widespread stocks. Their actions have had well documented success in the recovery of relatively sedentary species, such as Arapaima gigas [104], and socially for the organization and transfer of fishing rights from outside users to rural communities. Although community-based management can protect sedentary stocks and favor production for floodplain lake fishers, this strategy alone is not adequate for managing long-distance migratory species at the large scales of migratory fish life histories. To manage the migratory species at sustainable fishing levels, and thus the most important commercial species, it is necessary to maintain continuous data collecting in urban markets for at least the most important fisheries. These data can then be used for routine stock assessments, to detect overfishing, and support the agencies responsible for managing the fishery resources to make the most appropriate decisions to ensure the sustainability of the fisheries.

Successful fisheries management worldwide is associated with solid monitoring and robust stock assessment, and especially for larger stocks that receive far more management attention than smaller and unassessed stocks [109]. The most basic data for fishery management are historical and contemporary catch and effort, size composition, and sexual maturity of the catch [110]. Despite the large-scale increase in the exploitation of fish stocks, Brazil, the largest country with the largest area in the Amazon Basin, has not collected fishery data in most urban centers, including the largest, Manaus, since 2004. Furthermore, the use of statistics for setting science-based management regulations for Amazon fisheries continues to be ambiguous due to lack of confidence in governmental regulatory implementation and, as mentioned above, the relinquishment of governmental fishery responsibilities to local community-based management alone. Proper assessment of stocks could be reinforced if researchers, fishers, NGOs, consumers, and other stakeholders insist on the need for reinstating governmental responsibility to collect data and assess urban markets, ultimately providing the international framework required to understand overfishing at migratory fish life history scales.

Reduce fishing effort through enforced market regulations

Given the large areas involved in the Amazon and the precarious environmental infrastructure to patrol the interior, the most effective option to manage fishing effort is through enforced market regulations, such as size limits and closed seasons of fish sold in urban markets or to refrigeration companies. This obviously assumes that if the fishers cannot sell prohibited species, they will not capture them in large quantities. An argument against targeting species might be that in multi-species and multi-gear fisheries it is extremely difficult to target single species. The three species discussed here, and most of the other important migratory species, however, are highly targeted because of some combination of their large size, market and cultural values, and local fishers’ intimate knowledge of their life history locations and seasonal migratory behavior.

Although the results of B. vaillantii and B. rousseauxii overfishing are worrying, the fishery collapse for these species can be prevented and stocks rebuilt within a decade if fishing mortality is rapidly reduced below the fishing pressure that gives the MSY [111]. The most effective way to reduce fishing pressure on goliath catfishes is to reduce the fishing effort of the industrial fishing fleet in the estuary [24, 53, 102] and to adopt a capture quota or total allowable catches (TACs) for the Amazon Basin. The TACs are one of the most positive measures for rebuilding overfished stocks [21], and can be established at the main landing sites, such as at fish processing plants and markets in the estuary and along the inland rivers in the Amazon [47]. Local managers have had some success in reducing fishing pressure in the estuary, but not enough to control catches by the trawler fleet. Unfortunately, the necessary recovery time before expected collapse is running out. Finally, the lack of monitoring for more than a decade by Brazilian fishery agencies is ominous for the management of estuarine trawl and other fisheries. The most daunting challenge facing the management of B. rousseauxii is the exploitation of mature migratory fish in Colombia and Peru in the western Amazon that has led to recruitment overfishing. The loss of major spawning areas in the Madeira Basin because of the construction of dams blocking upstream migrations has added further to declining B. rousseauxii stocks in the Amazon Basin. A limited fishing period during the main spawning period of the species near the Andes would benefit the fisheries of Andean countries and Brazil [8, 25].

Colossoma macropomum has been so heavily overfished that a fish market moratorium will be required to rebuild its stock [112]. This will also require collecting separate data for wild-caught fish and those from farming operations to detect to what extent young wild fish are being exploited. Rather the moratorium is spatially and/or temporally implemented would not necessarily require a complete closure of the fishery [113]. How long the moratorium should be in place would depend on abundance data other than fisheries since prohibiting its sale also implies that its regeneration could not be detected in urban markets where the species was prohibited. The easiest, cheapest, and most quantitative method to do this would be through ichthyoplankton surveys in river channels during the spawning season [67, 68].

Remove subsidies

Fisheries strongly driven by export markets are usually difficult to control because economic pressure contributes to removing the last remnants of traditional local management [114]. Export markets from the Amazon are now restricted to each country and most important in Brazil and Colombia. Fishing subsidies directly or indirectly contribute to fleet overcapacity and stock overfishing [115]. Subsidies and export markets in Brazil and Colombia were the main incentives to the exploitation of goliath catfishes in the Amazon, which is reflected by the maximum annual fishery production of B. rousseauxii and B. vaillantii in each state or department (S2 Table). The Brazilian states of Pará and Amazonas and the Colombian department of Amazonas represent 84% and 95%, respectively, of total production of B. rousseauxii and B. vaillantii in the Amazon Basin. The cities of Belém in Brazil and Leticia in Colombia are responsible for at least 50% of all yields of these two species (Fig 7).

Currently, Brazil still maintains fishery subsidies in the form of unemployment benefits to artisanal fishers during the closed fishing season [116], and a tax exemption on the purchase of diesel oil for fishing vessels [117]. Fishing subsidies have had negative effects in the Amazon in both the industrial and artisanal fisheries. The subsides weakened market forces that might have curbed fishing pressure. Most ominous were federal fishery subsidies associated with the Closed Fishing Season Law in effect since 2003 in Brazil, a supposedly legal means to decrease fishing pressure that had the opposite effect. The law encouraged fishers to participate in the fisheries solely for government subsidies without any meaningful monitoring of catches such as for the highly threatened C. macropomum [118].

Unite urban and rural fishery management

The synergies between overfishing, hydroelectric dams, and floodplain deforestation present immediate major challenges for fisheries management in the Amazon, and especially for migratory species [10]. Hydrological changes due to climate change present further challenges to overall impacts in the long term [64, 119]. The migratory species are especially vulnerable to overfishing in their floodplain or estuarine nurseries and in river channels when they are schooling and migrating. The three migratory species considered here represent only a trio among dozens of commercial fish species that migrate in river channels, but their decades-long overexploitation is leading to the potential tipping points of stock collapse. We believe it would be a mistake to accept the collapse of these species as an unavoidable consequence of the fishing-down concept where larger species, often predators, are overexploited and the commercial fisheries turn to smaller species at lower trophic levels. If these species are not preserved, we believe that incentives to protect other targeted migratory species and the wetlands on which they depend, will be further weakened, and exacerbate the need for fisheries management at the ecosystem scale. Such iconic species have high cultural and other values that need to be factored into management strategies [120].

Fisheries managers have various options to control overfishing, though no one option is ideal for all fishing operations and species in the Amazon [69, 121]. Due to limited governmental intervention in the fisheries, NGOs and academics launched community-based management strategies in the 1990s. The major oversight of community-based management was the failure to recognize the overwhelming importance of migratory species (80+% of total catches) in commercial fisheries, and thus the necessity to concomitantly enhance fisheries data collecting in urban centers that are necessary to detect regional fishery patterns, such as overfishing and stock collapses. Furthermore, the lack of governmental fisheries management at urban levels resulted in a pessimistic scenario of the ability to manage fisheries through enforced market regulations based on annual statistics of catches and fishing effort.

Fisheries management in the Amazon should not be binary as either directed solely by governmental authorities or solely by hundreds of local communities largely operating independently along the rivers. There are few examples where local communities have attempted to manage C. macropomum overfishing. Controlled fishing zones seemed to have some positive effect on local C. macropomum populations where no-take floodplain lakes provided recruits of larger fish in the area of the Piagaçu-Purus Sustainable Development Reserve in the lower Purus River of Brazil [122]. It is unlikely that such an approach would be successful across the large commercial fishing area as it would be highly challenging to implement and monitor. A more dynamic approach is required where the two complementary components of urban regulated fisheries and community-based management are developed and interact together through regional fishery associations or the like. Some progress to this end has been made at small regional scales, such as at the Mamirauá Sustainable Development Reserve in the central Amazon that accompanies fishery catches in the small nearby city of Tefé to inform the Reserve’s management. A similar situation occurs in the Pacaya-Samiria National Reserve near the confluence of Marañón and Ucayali Rivers in Peru, where the governmental agencies of production (DIREPRO) and conservation (Servicio Nacional de Áreas Naturales Protegidas- SENAMP) share the registers of fish production in the protected area. This coupling of urban and rural initiatives has less continuity outside of extractive reserves and indigenous territories. Although encompassing large areas, even together, the ongoing local efforts are insufficient to manage long-distance migratory species within their large life history areas to supply the high demand for fish in urban centers.

Foster interstate and international agreements

The definition of migratory species is somewhat subjective, as nearly all species migrate to some extent, and depending on the species, ranging from a few kilometers to thousands of kilometers during their life histories in the Amazon [123]. One category is long-distance migratory species that for the Amazon have been defined as total life cycle movements greater than 1,000 km in river channels [8, 14]. The Amazon Basin includes five countries relevant to long-distance migratory fish conservation and seven states or departments in four countries that exploit these species (Fig 1). The transnational migratory behavior of goliath catfishes with nurseries and reproductive areas in different states or departments and countries presents major political challenges to the management and conservation of these species. The most important secondary administrative units where overfishing takes place are the Brazilian states of Pará, Amazonas, Rondônia, and Amapá, the Colombian department of Amazonas, and the Peruvian departments of Loreto and Ucayali (S3 Table). These jurisdictions are largely responsible for allowing excessive fishing pressure that has resulted in overfished stocks of the long-distance migratory B. vaillantii, B. rousseauxii, and C. macropomum and the local migratory B. filamentosum, the last which was reported seriously overfished in the early 1990s [7].

To date, fishery policies across jurisdictions are not coordinated, though the importance of migratory fish species is now recognized, and studies are increasing [10]. Brazil occupies the largest area of the Amazon Basin, but hydrographically it is downstream of all other countries. At the country level, the western headwater region associated with the Andes embraces four countries: Bolivia, Colombia, Ecuador, and Peru. With 15 secondary administrative levels (states, departments, or provinces), the Andean piedmont region of the Amazon Basin is far more fragmented geopolitically than the lowlands. Although all Amazonian countries except Colombia are signatories to the Convention on the Conservation of Migratory Species of Wild Animals, no Amazonian freshwater migratory fish species are listed in the Convention. Other possible international organizations that could help stimulate fishery management across countries include the Food & Agriculture Organization (FAO), the Organization Cooperation Treaty Organization (OTCA), the Ramsar Convention, and the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) [15].

Mitigate the Madeira dams

The Madeira dams represent the first major infrastructure mitigation challenge to fisheries management linked from the Andes to the Amazon estuary. A fish bypass was constructed at the first downstream dam, Santo Antonio, though it does not function as planned [124]. Because legal contracts require migratory fish passage through the Madeira dams, the bypass options need to be investigated further. We believe it would be a mistake to forgo further mitigation of goliath catfish passage through or around the Madeira dams as legal agreements require it until successful, and to not do so would set an ominous precedent for the future.

Monitor floodplain deforestation

Floodplain deforestation directly impacts species that use the floodplains as nurseries, as spawning habitats for many species, and for growth of older fish, such as feeding seasonally in flooded forests [97, 122]. The historical trend of large-scale floodplain deforestation in the Amazon is east to west along the Amazon River, and if it continues at a similar rate in the next 3–5 decades then it will reach as far west as the Andean countries.

Floodable forests are the most common natural wetland type of floodplains of large Amazon rivers and the estuary [125]. Floodplain deforestation analyses have often been overlooked or subsumed into upland deforestation, yet the impacts on wetlands are far more drastic per area because they affect directly aquatic, arboreal, and terrestrial biodiversity. The amount of flooded forest correlates with fishery production and fish biodiversity because the floodplains are nurseries and feeding areas for most species that migrate in the river channels [14, 98, 99]. More than 1,000 km of the lower Amazon River floodplain has been largely deforested and is now in relatively species-poor secondary forest or open herbaceous wetlands where forest existed before. Although beyond the purview of this paper to suggest specific interventions to decrease wetland deforestation, direct governmental intervention is necessary. Especially helpful would be a clear separation of upland and wetland deforestation to call more attention to the importance of both types of forest in the aquatic ecosystems of the Amazon. The impacts of upland deforestation, especially in headwater areas near or in the Andes, are also highly relevant to the life history of long-distance migratory fishes because they protect watersheds, but this relationship is poorly understood [38]. Upland deforestation is further complicated geopolitically because Andean western headwaters are shared by four countries (Bolivia, Peru, Ecuador, and Colombia), but with the largest country, Brazil, downstream of all of them.

Include overfishing as part of ecosystem level impacts

Though little investigated, overfishing of freshwater migratory fish species can have ecosystem-level consequences on species other than just the exploited species [126]. For example, C. macropomum is an important seed disperser in flooded forests of the Amazon, and not only because of the large number of plant species whose fruits it eats [127], but also the long distances that it disperses seeds [56, 128]. Thus, overexploitation probably disrupts an ancient coevolutionary relationship between C. macropomum and Amazonian plants, and especially because of the large size of the fish that is associated with a higher level of seed dispersal [127, 129, 130] The decrease in C. macropomum biomass through overfishing also eliminates one of the most important seed dispersal agents of the Amazon’s flooded forests. Many other migratory species exploited in fisheries are also seed dispersers [127].