Understanding fishing-induced extinctions in the Amazon · Understanding ﬁshing-induced extinctions in the Amazon LEANDRO CASTELLOa,*, CAROLINE CHAVES ARANTESb, DAVID GIBBS MCGRATHc,

Understanding fishing-induced extinctions in the Amazon

LEANDRO CASTELLOa,*, CAROLINE CHAVES ARANTESb, DAVID GIBBS MCGRATHc,DONALD JAMES STEWARTd and FABIO SARMENTO DE SOUSAe

aDepartment of Fish and Wildlife Conservation, Virginia Polytechnic Institute and State University, Blacksburg, VA, USAbDepartment of Wildlife and Fisheries Sciences, Texas A&M University, TX, USA

cEarth Innovations Institute, San Francisco, CA, USAdDepartment of Environmental and Forest Biology, State University of New York, Syracuse, NY, USA

eSociedade para a Pesquisa e Proteção do Meio Ambiente, Santarém, Pará, Brazil

ABSTRACT

1. Science and policy worldwide are influenced by predictions from bioeconomic theory that fishing cannot leadfish populations to extinction because fishing effort inevitably moves away from depleted resources. Yet suchpredictions contradict evidence of fishing-induced extinctions and in particular a model, called ‘fishing-down’, thatexplains historical reductions in mean size of harvested species in tropical multispecies fisheries through thegradual depletion and extinction of large-bodied species.

2. This study analysed data on fisheries for Arapaima spp., the most historically important and overexploited fishes ofthe Amazon Basin, to evaluate whether they supported bioeconomic or fishing-down predictions. The evaluation wasbased on census data on arapaima populations and interview data from 182 fishers with respect to fishing practices andmanagement regulations, which were collected in 81 fishing communities covering 1040km2 of Amazonian floodplains.

3. Arapaima populations were found to be ‘depleted’ in 76% of the fishing communities, ‘overexploited’ in 17%,‘well-managed’ in 5%, and ‘unfished’ in only 2%. Population densities were zero (i.e. locally extinct) in 19% of thecommunities. Twenty-three per cent of the fishers in each community harvested arapaima regardless of populationstatus. Similarly, the percentage of the catch in compliance with the size regulation did not vary with populationstatus, but compliance with the season regulation in communities with ‘overexploited’ or ‘depleted’ populations(72%) was lower than in communities with ‘well-managed’ or ‘unfished’ populations (97%).

4. These results support fishing-down predictions that fishing pressure continues to occur even when fishpopulations are depleted. The fishing-down process appeared to occur because of low gear selectivity and largerbody-size of target species as well as high species value and low fishing costs. These results and available dataelsewhere suggest that fishing-induced extinctions are more common than previously thought, endangeringbiodiversity and ecosystem functioning. Such extinctions are probably going unnoticed because high levels ofillegal fishing, geographic heterogeneity, and data scarcity make their identification difficult.Copyright # 2014 John Wiley & Sons, Ltd.

Received 9 February 2014; Revised 19 June 2014; Accepted 29 June 2014

KEY WORDS: Arapaima spp; biodiversity; conservation; food security; floodplains; Osteoglossidae; overfishing

*Correspondence to: Leandro Castello, Department of Fish and Wildlife Conservation, College of Natural Resources and Environment, VirginiaPolytechnic Institute and State University, 310 West Campus Drive, 148 Cheatham Hall, Blacksburg, VA, 24061, USA. Email: [email protected];http://fishwild.vt.edu/faculty/castello.htm.

Copyright # 2014 John Wiley & Sons, Ltd.

AQUATIC CONSERVATION: MARINE AND FRESHWATER ECOSYSTEMS

Aquatic Conserv: Mar. Freshw. Ecosyst. (2014)

Published online in Wiley Online Library(wileyonlinelibrary.com). DOI: 10.1002/aqc.2491

http://fishwild.vt.edu/faculty/castello.htm

INTRODUCTION

The notion that fishing cannot lead a species toextinction has influenced science and policy formany decades. This notion was founded on the oldidea that marine species are highly resilient tofishing (Roberts and Hawkins, 1999). It was foundedalso on predictions from bioeconomic theory thatfishing can overexploit and even deplete fishpopulations but cannot lead them to extinctionbecause the high costs of fishing-depleted populationsinevitably move effort away from them (Gordon,1954). Such optimistic predictions have profoundimplications for tropical fisheries. Not only aretropical fisheries embedded in the most biodiverseecosystems, playing key roles in the maintenance ofglobal biodiversity (Roberts et al., 2002), they alsoproduce one-third of global capture fish yields(Castello et al., 2007), sustaining key income- andfood-security services for the world’s poorestpopulations (Allison and Ellis, 2001; Pauly et al.,2005). Bioeconomic predictions that fishing does notcause extinction thus imply that tropical fisheries donot threaten biodiversity, food web structure andfunctioning, and income- and food-security services.

Evidence has been emerging, however, thatfishing causes extinctions. A literature reviewfound that fishing induced 55% of 133 documentedlocal, regional, and global extinctions of marinepopulations (Dulvy et al., 2003). The highlyfecund Bahaba taipingensis (Scienidae) was foundto be facing fishing-induced near extinction(Sadovy and Cheung, 2003). In another example, 22of 163 species of groupers may soon be at risk offishing-induced extinctions (de Mitcheson et al.,2013). Such evidence has been changing perceptions,although slowly. Whereas the resilience of marinespecies is no longer thought to protect them fromextinction (Dulvy et al., 2003), the notion thateconomic extinctions prevent species extinctionsremains prevalent in many policy and scientificarenas. For example, it was recently stated that‘where management is weak or nonexistent, multiplefishers compete to catch fish from a givenpopulation… An equilibrium… is reached onlywhen… catch rates are barely sufficient to cover thecosts of fishing. The population is then maintained atthis level through biological processes of natural

growth and reproduction’ (Beddington et al., 2007).Belief in this economic prediction is probably due to alack of studies assessing its validity, although italready has been argued that effort rarely can bedirected fully onto or away from any single speciesbecause most fisheries are multispecies, even intemperate regions (Dulvy et al., 2003). It has beenargued also that the prices of some fishes are inverselyproportional to their abundance (Pinnegar et al.,2002) such that the higher values of overexploitedspecies can promote extinction (Dulvy et al., 2003).

Few are aware that there is an alternative model,called the ‘fishing-down’ process, which explains thedepletion and extinction of species caused by fishingin tropical multispecies fisheries (Welcomme, 1999).The fishing-down process differs from the ‘fishingdown marine food webs’ concept (Pauly et al.,1998), which predicts declines in trophic levels.The fishing-down process stems from the notionthat body size largely determines extinction risk(Reynolds et al., 2001). Large-bodied animals tendto be more sought-after and possess life-historytraits associated with vulnerability, including latematurity, low intrinsic rates of populationincrease, behaviour that increases catchability, anddependence on vulnerable habitats (Reynoldset al., 2002). The fishing-down process predictsthat historical increases in fishing effort intropical multispecies fish communities reduce themean body size of harvested species through thegradual replacement of depleted large-bodiedspecies with small-bodied ones (Castello et al.,2013a). Bioeconomic and fishing-down predictionsthus differ mainly with respect to whether fishingcontinues once fish populations become depleted.

Support for the fishing-down process comes frommultispecies stock-production models, which arelike those for single species fisheries, showingincreasing yields with increasing effort up to amaximum, at which point they differ in showingconstant, not declining, yields with increasingeffort (Lae, 1997; Lorenzen et al., 2006). Suchconstancy in yields is maintained by the successivedepletion and replacement of target species. Aslarger-bodied species become depleted, they canbecome extinct because the harvest of smaller-bodiedspecies normally involves nets that often also selectjuveniles of the larger-bodied ones. The fishing-down

L. CASTELLO ET AL.

Copyright # 2014 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2014)

process shrunk the mean total length of harvestedspecies in the Oueme River in West Africa from78cm in the 1950s to 57 cm in the 1970s and then to22 cm in the 1990s, leading to the disappearance offour species from catches by 1965 (Welcomme, 1999).In the Amazon River in South America, fishing-downshrunk the mean maximum length of harvestedspecies from 206 cm in the early 1900s to 79 cmtoday, leaving several species under varying degreesof extinction risk, including manatees (Trichechusinunguis), three turtle species (Podocnemis spp.),two crocodilian species (Crocodilus crocodilus,Melanosuchus niger), and one or more species ofarapaima fishes (Arapaima spp.; Da Silveira andThorbjarnarson, 1999; Castello et al., 2013a).

Fishing-down and bioeconomic predictions aretwo competing hypotheses on the potential of fishingto cause extinction. To improve understanding offishing impacts in tropical fish communities, thisstudy evaluated whether the dynamics of fisheries

for arapaima supported bioeconomic or fishing-downpredictions. The evaluation was based on datacollected on the abundance, levels of fishingpressure, and sustainability of fishing practices forarapaima in floodplain communities of theAmazon river mainstem, near the municipality ofSantarém, State of Pará, Brazil (Figure 1). Thefishing-down prediction should be observablethrough locally extinct populations or continuedfishing of depleted populations. The bioeconomicprediction should be observable through absenceof local extinctions or presence of sustainablefishing practices if fishing continued after initialdepletions.

ARAPAIMA FISHERIES IN THE AMAZON

Biology, ecology, and population status of arapaima

The taxonomy of arapaima is poorly studied, so thegeographical distribution is known only for the

Figure 1. The study area in the lower Amazon region, showing censused arapaima densities in 41 communities. Population densities are measured asind km-2 of juveniles (1–1.5m TL) and adults (>1.5m TL). Inset shows location of study area in Brazil (black rectangle).

FISHING-INDUCED EXTINCTIONS IN THE AMAZON


genus and undescribed species may exist (Stewart,2013a, b; Castello et al., 2014). The genus waswrongly considered to be monotypic for 140 years(Günther, 1868) despite previous description of fourspecies by Schinz (in Cuvier, 1822; A. gigas) andValenciennes (in Cuvier and Valenciennes, 1847;Arapaima agasizzii, Arapaima mapae, andArapaima arapaima). Arapaima gigas, whose typespecimen came from the study area, has not beenseen in the field since the original specimen, whichwas obtained in about 1787 (Stewart, 2013a, b).Because of this taxonomic uncertainty, we refer toarapaima only at the genus level.

Historical overexploitation has led Arapaimagigas to be listed in Appendix II of the Conventionon International Trade of Endangered Species ofWild Fauna and Flora (Castello and Stewart,2010). This species is now listed also in the IUCNRed List of Threatened Species as ‘Data Deficient’,which means: ‘there is inadequate information tomake a direct, or indirect, assessment of its risk ofextinction based on its distribution and/orpopulation status’ (World Conservation MonitoringCentre, 1996). In Brazil, arapaima were notincluded in the national list of endangered speciesowing to lack of data.

The largest arapaima populations occur inwhitewater floodplains, a complex mosaic ofseasonally inundated forests, lakes, and channels thatare completely flooded annually (Irion et al., 1997).Arapaima are fished in floodplain lakes when lowwater levels (September–January) force all fish toseek refuge in remaining aquatic habitats (Castello,2008a). These fishes are vulnerable to fishing andhighly sought-after. Each individual is valuable, asthey grow up to 3m in length and 200kg in weightand attain high market prices (Arantes et al., 2010).They are obligate air-breathers that are exposed toharpoon-specialist fishers every 5–15min whenthey surface to breathe (Castello, 2004). They areparticularly vulnerable when they spawn in nestsbuilt on the margins of floodplain forestssurrounding lakes and channels. Males engaged inparental care are defenceless against fishers whouse such habitats as daily transport routes(Castello, 2008b).

The influence of fishing on arapaima populationdynamics appears to be strong. When Brazil’s

minimum size limit of 1.5m in total length (TL) isfollowed, arapaima grow fast, to 88 cm TL in oneyear, reaching maturity at 3–4 years of age whenthey measure 157–164 cm TL (Arantes et al., 2010).When the size limit is not followed, selectiveremoval of the faster-growing, under-sizedindividuals not only removes potential spawnersbut also lowers overall mean body growth rates. Insuch conditions, arapaima length at age is onaverage 27 cm shorter, and they reach maturity at5 years of age, which results in significantly lowerintrinsic rates of population increase (Arantes et al.,2010; Castello et al., 2011a).

Arapaima fisheries in the study area

In the study area, arapaima are exploited byfloodplain fishers living in geographically dispersedcommunities. For them, fish is the mosteconomically and nutritiously important resource.Some 40 different fish species are exploited overthe course of the year through daily fishing trips ofa few hours in wooden canoes (McGrath et al.,1998; Castello et al., 2013b). Gillnets dominate thecatch, but castnets, long-lines, fishing poles, andharpoons are also used. Arapaima are harvestedmostly using harpoons and gillnets, but manyyoung arapaima are caught as bycatch in gillnetsdirected to other smaller-bodied species (Batistaand Freitas, 2003).

Pressure over natural resources has beenincreasing in the study area because of newtechniques, human population growth, andeconomic development (Figure 1; McGrath et al.,1993; Isaac et al., 2008). Expansions of cattleranching and agriculture have led to deforestationof 56% of floodplain forests (Renó et al., 2011).Five of the nine most exploited species in thestudy area, including arapaima, are thought to beoverexploited, although there are few formal stockassessments (Isaac and Ruffino, 1996; Ruffino andIsaac, 1999; Castello et al., 2011b). Fisheriesmanagement in Brazil has relied on minimum sizeand closed season regulations, but compliance hasbeen dismal owing to lack of governmentresources and large geographical areas (Cramptonet al., 2004; Castello et al., 2013b). Also, fisheriesin Brazil are managed as open access resources,

L. CASTELLO ET AL.


which means that all citizens have equal rights ofuse (McGrath et al., 2008). Consequently, fishingfor arapaima is banned in the Brazilian states ofAcre and Amazonas where exception is made forcommunity-based management (CBM) schemes,but it is open in Pará State based on size andseason limits. Many floodplain fishers have soughtto curb overfishing by establishing CBM schemesvia implementation of gear, season, and arearegulations that are developed locally by eachcommunity (Castro and McGrath, 2003). Thenon-migratory behaviour of arapaima make themsuitable for, and hence a target of, CBM (Aranteset al., 2006; Castello et al., 2009, 2011c).

METHODS

Field data were collected based on censusingarapaima populations and through interviews withlocal fishers on levels of fishing pressure and thesustainability of fishing practices. To evaluatewhether the data supported bioeconomic orfishing-down predictions, the data were analysedfor each community to assess population statusper community and to evaluate how levels offishing pressure and the sustainability of fishingpractices varied with population status.

Data collection

Between July and September 2011, structuredinterviews were conducted with 182 fishers from 81communities, resulting in an average of 2.2 fishersbeing interviewed per community, with aminimum of two and a maximum of three fishersper community. The fishers interviewed wereselected following best-available methods forresearching local knowledge (Berkes et al., 2000;Davis and Wagner, 2003). Interviewed fishers wereselected by peers of their own communities asbeing ‘experts’ on fisheries matters. The fisherswere interviewed with respect to arapaimapopulations, management, and fishing practices inthe interviewees’ own communities. To assessarapaima populations, interviewees were askedto classify arapaima abundance (zero, low,medium, and high) and its trend in recent years(decreasing, stable, or increasing). To assess fishing

pressure, interviewees were asked to estimate totalnumber of fishers, and fishers targeting arapaima.To assess the sustainability of fishing practices,interviewees were asked to identify months ofthe year arapaima harvests occurred, if any,and typical TL of harvested individuals (in 20 cmsize-classes). Interviewees also were asked if therewere CBM rules for arapaima that were followedby the fishers. Field notes and interview resultswere transcribed and coded for quantitativeanalyses.

Arapaima populations were censused duringNovember and December 2011, in 41 of the 81communities included in the interviews, using themethod of Castello (2004) that allows expertfishers to count the arapaima at the moment ofaerial breathing, in two size classes: juvenile(1–1.5m TL) or adult (>1.5m TL). The countingmethod can be accurate and precise if properlyapplied. Arapaima were censused by eight trainedfishers whose individual counts possessederrors< 30%, as shown by the method of Aranteset al. (2007), which compares fishers’ counts ofarapaima with seine catches of the individuals inthe same closed lakes. The eight fishers who weretrained in the arapaima censusing method wereselected for this work first on the basis of theirinterest in learning the method, and later based onan informal assessment of their knowledge andskills on arapaima fishing, following guidelines foridentifying expert fishers provided in Castello(2004) and Castello et al. (2009). The same groupof eight fishers censused arapaima populations inlakes of all 41 surveyed communities. In eachcommunity, a minimum of one and a maximum offour lakes were censused. Arapaima density in theterritory of each censused community wasquantified per floodplain area (ind km-2), excludingriver channels, because such territories comprisedifferent sets of lakes that together host localarapaima populations (Castello et al., 2009, 2011b).

Data analyses

To identify arapaima population status, census datawere classified based on the following density rangeclasses: depleted (0–2.2 ind km-2), overexploited(2.3–17.7 ind km-2), well-managed (17.8–32.4 ind km-2),



and unfished (>32.5 ind km-2). These classes werederived by interpolating point density estimates fordifferent arapaima population statuses in equivalentvárzea floodplains from the Mamirauá Reserve inthe Solimões River: 4.4 ind km-2 is overexploited,31.1 ind km-2 is well-managed, and 33.8 ind km-2 is(close to) unfished conditions (Castello et al.,2011b). To assess historical trends in arapaimaabundance, the accuracy of interview responses onarapaima abundance (low, medium, or high) wasassessed by comparing them with censused data forcommunities for which such data were available.Then, interview responses on historical trends inarapaima abundance (decreasing, stable, orincreasing) were quantified. To determine whetherfishing pressure for arapaima has ceased orcontinued, the percentage of fishers targetingarapaima was calculated based on the total numberof fishers for each community, and this percentagewas compared across communities possessingdifferent arapaima population statuses (i.e. depleted,overexploited, well-managed, and unfished). Toquantify the sustainability of fishing practices,response data on months of the year when arapaimaharvests occurred, if any, and typical TL ofharvested individuals (in 20 cm size-classes) wereused to calculate the percentage of the catch that wasin compliance with size (>1.5m TL) and season(May–November) regulations. Such percentages ofcompliance with size and season regulations werecompared across communities possessing differentarapaima population statuses. Finally, interviewresponses on existence of CBM rules were quantified,and the effectiveness of these rules was assessed bycomparison with population census data in the samecommunities.

RESULTS

Population status

The censusing surveys indicated that arapaimapopulations were ‘depleted’ in 76% of the fishingcommunities, ‘overexploited’ in 17%, ‘well-managed’in 5%, and ‘unfished’ in only 2%. Populationdensities were zero (i.e. locally extinct or extirpated)in 19% of the communities. In total, 1825 juvenilesand 1630 adults were censused in 1040km2 of

floodplain area, resulting in extremely low medianarapaima population densities around 0.55 ind km-2

for all size classes, with 0.34 ind km-2 for juvenilesand 0.11 ind km-2 for adults (Figures 1 and 2).

The foregoing assessment is supported by fishers’perceptions: 55% of the fishers classified arapaimaabundance as being ‘low’ in communities wheremedian censused arapaima abundance was 0.15ind km-2, 27% classified it as ‘medium’ where itwas 0.74 ind km-2, and 2% classified it as ‘high’where it was 2.39 ind km-2 (Mann–WhitneyU-tests, P< 0.01 for each pair-wise comparison).Such congruence between census and fishers’perception data supports the following inference ofdeclining abundance: 76% of the fishers said thatarapaima abundance in recent years has decreased,20% of them said it has increased, and 4% said ithas remained the same.

Fishing pressure

On average, 23% of the fishers per communityharvested the arapaima, and this level of fishingpressure was maintained regardless of populationstatus. Although the mean percentage of fishers ateach community targeting the arapaima was lowerin communities where arapaima populations were‘depleted’ or ‘overexploited’ (mean=15%) than incommunities where they were ‘well-managed’ or‘unfished’ (mean=33%), that difference was notstatistically significant (Kruskal–Wallis test, P=0.5).

Figure 2. Censused arapaima population densities in 41 communities ofthe lower Amazon region and associated population statuses classified

based on Castello et al. (2011a).

L. CASTELLO ET AL.


Sustainability of fishing practices

Compliance with season regulations varied withpopulation status, but compliance with sizeregulations did not. The percentage of the catchthat was in compliance with the season regulation(mean= 97%) was significantly higher incommunities where arapaima populations were‘well-managed’ or ‘unfished’ than in communitieswhere they were ‘depleted’ or ‘overexploited’(mean= 72%; Mann–Whitney U-tests, P< 0.05).The percentage of the catch in compliance withthe size regulation did not vary with populationstatus (Kruskal–Wallis test, P=0.5), and it wasextremely low on average, only 19%.

Only 27% of communities had management rulesfor arapaima harvests that were followed by thefishers, but those were effective at conservingarapaima and explained the high degree of spatialheterogeneity in arapaima density (Figure 1).Median arapaima density in communities withCBM rules (10.02 ind km-2) was two orders ofmagnitude higher than in communities without(0.55 ind km-2; Mann–Whitney U-test, P< 0.05).CBM areas contained 2506 individuals (or 75%) ofthe total censused.

DISCUSSION

Fishing-induced extinctions of arapaima

These results support fishing-down predictions thatfishing pressure continues to occur even whenfish populations are depleted. Contrary tobioeconomic predictions, arapaima populationswere found to be ‘depleted’ in 76% of thecommunities and locally extinct in 19% of them(Figure 1), yet fishers continued to harvest arapaimaregardless of population depletion, as indicated bylack of variation in the percentage of fisherstargeting arapaima (23%) across communities withdifferent population status. The sustainability ofsuch harvesting practices also contradictedbioeconomic predictions. Compliance with theclosed season regulation in communities with‘overexploited’ or ‘depleted’ populations (72%) waslower than in communities with ‘well-managed’ or‘unfished’ populations (97%). Such continued

unsustainable fishing practices that exploit depletedpopulations is probably the reason why 76% of thefishers believe that arapaima abundance in recentyears has decreased.

Could the trend of low and declining populationbe attributed to floodplain deforestation? Theavailable evidence suggests not. The ‘well-managed’or better (>17.8 ind km-2) arapaima densitiesobserved in three communities engaged in CBM(Figures 1 and 2) suggest that overfishing, notdeforestation (Renó et al., 2011), explains theobserved population patterns. These high arapaimadensities also endorse the reference data used hereto determine population status (e.g. ‘depleted’;Castello et al., 2011a).

The zero arapaima densities observed in eightcommunity territories reflect local extinctions(Figures 1 and 2), because the censused floodplainlakes in each territory were expected to host localarapaima populations (or at least, offer suitablehabitat) in the absence of fishing. Such lakes weredevoid of arapaima 2 years old and older, whichare the ages at which they are included in thecounts (i.e. > 1m TL; Arantes et al., 2010). If therewere young of the year individuals (i.e. < 1m) inthose lakes, the absence of adults (Figure 1) wouldindicate that they immigrated from surroundinglakes during the previous high water. Such lakeswithout adult arapaima may be population sinks,with no individuals surviving to reproduce.

The observed persistence of unsustainable fishingpractices over extremely low and decliningpopulations appears to have caused local extinctionsof arapaima, and it will probably continue to do so.Few communities (27%) have implemented rules forarapaima, and the widespread use of gillnetstargeting smaller-bodied species causes bycatch ofjuvenile arapaima, further undermining theirpopulations (Castello et al., 2011a). Such widespreadlack of management and incidental bycatch isexpected to cause additional local extinctions ofarapaima (Figure 1). Local extinctions usually signalthe potential for regional extinctions, which are thefirst steps towards global extinctions (Pitcher, 2001).

Fishing-down may be exerting impacts on one ormore arapaima species, depending on their life-historytraits. For example, it has been estimated that if aslower-growing arapaima species with maturity at



6–7 years ever existed, it would have been eliminatedby fishingmortality rates that are today considered tobe sustainable for arapaima, even if current size andseason regulations were followed (Castello et al.,2011a). Similar fishing effects on sympatric skatespecies have been observed in the North Atlantic,where fishing pressure on skates nearly caused theextinction of a skate species that was unrecognizedtaxonomically, primarily because of its larger bodysize and later age of reproduction (Dulvy andReynolds, 2009).

Fishing-induced extinctions

These findings indicate that fishing causes moreimpact on tropical fish communities thanpreviously recognized (Dulvy et al., 2003).Fishing-induced extinctions affect tropical aquaticecosystems and biodiversity. Tropical fisheriesbased on the use of gillnets generally lead to anincrease in diversity of target species, because of theprogressive elimination of larger-bodied species andthe naturally higher abundances of smaller-bodiedspecies (Welcomme, 1999). In the study area, thediversity of target species in communities wherearapaima have become locally extinct is higher thanin communities where arapaima are ‘well-managed’or better (Castello et al., 2013b). However, tropicalmultigear fisheries generally decrease the diversity oftarget species by targeting, and hence progressivelydepleting, the entire fish assemblage (Jenningset al., 1995; Welcomme, 1999). Species extinctionsundermine diversity in functional groups and canproduce cascading effects. In the Amazon,ecological extinction of manatees, turtles, andcapybaras (Hydrochaerus hydrochaerus) havebeen linked to historical growth of aquatic andsemi-aquatic macrophyte beds (Junk, 2000). Lossesof apex predatory species and primary consumersalter whole ecosystems through modification ofenergy flows with severe consequences forbiodiversity (Estes et al., 2011).

Do fishing-induced extinctions undermine fisheryyields and associated income- and food-securityservices? Two meta-analyses based on datasetscovering a diversity of tropical fisheries worldwidefound that multispecies catch–effort responsesfollowed a negative parabolic curve, indicating

that yields increase with effort up to a maximumand then decrease with increasing effort (Bayley,1988; Halls et al., 2006). However, two othercomprehensive meta-analyses found thatmultispecies catch–effort responses conformed tothe asymptotic model in which catch levelsincrease up to a maximum and remain stableindefinitely with increasing effort (Laë, 1997;Lorenzen et al., 2006). Thus, it remains uncertainwhether fishing-induced extinctions underminetropical multispecies fisheries yields. However, lossof target species adversely affects the livelihoods offishers by making gillnet-based fisheries dependenton the natural productive cycles of fewer species(Jennings and Polunin, 1996). Fishers may beforced also to adapt to new species and gears andto find new fishing grounds. In addition,differences between species availability and marketpreferences may affect fishers’ economy(Hoeinghaus et al., 2009), at least until peopleadjust to the less desirable species.

Lessons from the Amazon

Examination of the conditions in which arapaimaextinctions occurred suggests that many fishing-induced extinctions in the tropics are going unnoticedbecause three characteristics of the fishing-downprocess make it difficult to identify:

1. Lack of data: Most tropical countries possesslimited human and financial resources tostudy the biology, taxonomy, and ecologicalinteractions of fishery species as well as tomonitor and assess fisheries statuses (Mahon,1997; Johannes, 1998). Such data scarcityprecludes identification of population declines.

2. Illegal fishing: Where fishery regulations existin the tropics, compliance is low owing topoor enforcement. Not only does illegalfishing targeting intensely exploitedresources occur at the margins of limitedfishery monitoring systems, it also limits theusefulness of catch statistics and underminesthe sustainability of such resources. Lack ofcompliance with size and season regulationsdegrades the capacity of fish populations tosustain fishing pressure and recover fromoverexploitation (Myers and Mertz, 1998).

L. CASTELLO ET AL.


3. Geographic heterogeneity: The tropics possesslarge numbers of small-scale fishery stocksbecause of high species diversity (Stevens,1989). Inter-community differences in fishingpractices, ecological conditions, and economicactivities, which also are high in the tropics(Castello et al., 2013b), create heterogeneousmosaics of resource abundance that are notcaptured by deficient monitoring schemesbased in a few cities (Figure 1).

Scientists interested in documenting fishing-inducedextinctions should consider alternative data andanalytical approaches to overcome the inadequacy ofconventional approaches. Sophisticated analysis ofcomprehensive fisheries statistics, as is done inEurope and North America for example, is seldoman option in the tropics. Rather, scientists in theseregions will benefit from historical records (Hardt,2009) as well as fishers’ knowledge to document pastand current fishery patterns, among other datasources (Berkes et al., 2000; Johannes et al., 2000;Sadovy and Cheung, 2003). In the present study,fishers’ ecological knowledge was used as the primarybasis for assessing arapaima populations anddocument fishing practices and trends. The usefulnessof local ecological knowledge to overcome datascarcity and promote user participation inmanagement and conservation schemes has been thetopic of increased interest in recent years (Berkeset al., 2000), but little has been done to apply it(Castello et al., 2011c). The time has come to applylocal ecological knowledge for problem solving.Combining alternative research methods withconventional methods such as those based on catchstatistics, even if limited in availability, can fosteraquatic conservation research in regions of the worldthat need it the most.

The findings of this study suggest that fishing-downis caused not only by the impacts of low gearselectivity on the larger-bodied species, as originallyproposed (Welcomme, 1999), but also by theeconomics of the small-scale fisheries that dominatefish yields in the tropics (McManus et al., 1992;Mahon, 1997). Because most small-scale fisheries arelow-cost, fishers can exploit high-value, large-bodiedresources even at very low abundances. In thisstudy, fishers continued to target the arapaima

despite depletion, because they use home-madecanoes and harpoons and can locate and harvesteven a single arapaima when it surfaces tobreathe. Even if fishers were to try to move effortaway from large-bodied overexploited species, asbioeconomic theory predicts, incidental harvest ofjuveniles caused by gillnets would furtherundermine the survival of the large-bodied species.This inability of tropical small-scale fishers to curboverfishing in the face of declining fishpopulations is partly explained by the literature onpoverty traps, which suggests that lack ofeconomic alternatives makes fishers unable toovercome market- or environment-related shockssuch as declining resources (Allison and Ellis,2001). In small-scale fisheries in Kenya, fishersfrom poorer households were less likely to stopfishing than fishers from wealthier households(Cinner et al., 2009).

Curbing the fishing-down process requiresmanagement policies that are multi-pronged toaddress issues related not only to size and seasonof target species but also to gear and poverty. Thestudied community of Ilha de São Miguel offers alesson in this regard: it banned the use of gillnetsand seines two decades ago and today it possessesthe highest arapaima densities in the region(35 ind km-2; Figures 1 and 2). Although banninggillnets can be expected to disrupt food andincome security because of their disproportionatecontribution to catches, multispecies catch per uniteffort in Ilha de São Miguel is the highest in thestudy area (Castello et al., 2011a; 2013b). Castnetsare allowed, however, and they yield abundantfishes for local consumption, so food security isnot compromised. Similarly, in Papua NewGuinea, selective restrictions on the use of gillnets,line fishing, and spearguns improved theconservation of fish communities associated withcoral reefs (McClanahan and Cinner, 2008).Addressing the poverty of the fishers is moredifficult, however, but it can be done by providingtargeted assistance in the form of education,insurance, and institutional support (Costanza,1987) as well as alternative-livelihood opportunities(McClanahan et al., 2005). Although historicallymany such assistance programmes have failed(Allison and Ellis, 2001), the need to address the



socioeconomic context of tropical fisheries for theirconservation demands further work on the topic.This is particularly important in river conservationschemes where biodiversity protection often conflictswith maintaining ecosystem services (e.g. food security)to an extent much greater than in other ecosystems.The conservation of species exploited by tropicalsmall-scale fisheries in river ecosystems thus requiresintegrative approaches to fostering environmentalconservation and human wellbeing (Castello et al.,2013a; Ormerod, 2014).

ACKNOWLEDGEMENTS

W. Rocha helped conduct the interviewquestionnaires; D. Gurdak helped censusarapaima populations; R. Welcomme clarified ourunderstanding of the fishing-down process. Thisresearch was supported by funding from theBrazilian Conselho Nacional de Pesquisa (CNPq)and the Gordon and Betty Moore Foundation.The authors declare no conflict of interest.

REFERENCES

Allison EH, Ellis F. 2001. The livelihoods approach andmanagement of small-scale fisheries. Marine Policy25: 377–388.

Arantes C, Garcez DS, Castello L. 2006. Densidades depirarucu (Arapaima gigas, Teleostei, Osteoglossidae) emlagos das Reservas de Desenvolvimento SustentávelMamirauá e Amanã, Amazonas, Brasil. Uakari 2: 37–43.

Arantes CC, Castello L, Garcez DS. 2007. Variações entrecontagens de Arapaima gigas (Schinz) (Osteoglossomorpha,Osteoglossidae) feitas por pescadores individualmente emMamirauá, Brasil. Pan-American Journal of AquaticSciences 2: 263–269.

Arantes CC, Castello L, Stewart DJ, Cetra M, Queiroz HL.2010. Population density, growth and reproduction ofarapaima in an Amazonian river-floodplain. Ecology ofFreshwater Fish 19: 455–465.

Batista VS, Freitas VS. 2003. O descarte de pescado na pescacom rede de cerco no baixo rio Solimões, AmazôniaCentral. Acta Amazonica 33: 127–143.

Bayley PB. 1988. Accounting for effort when comparingtropical fisheries in lakes, river-floodplains, and lagoons.Limnology and Oceanography 33: 963–972.

Beddington JR, Agnew DJ, Clark CW. 2007. Currentproblems in the management of marine fisheries. Science316: 1713–1716.

Berkes F, Colding J, Folke C. 2000. Rediscovery of traditionalecological knowledge as adaptive management. EcologicalApplications 10: 251–262.

Castello L. 2004. A method to count pirarucu Arapaima gigas:fishers, assessment and management. North AmericanJournal of Fisheries Management 24: 379–389.

Castello L. 2008a. Lateral migration of Arapaima gigas infloodplains of the Amazon. Ecology of Freshwater Fish17: 38–46.

Castello L. 2008b. Nesting habitat of pirarucu Arapaima gigasin floodplains of the Amazon. Journal of Fish Biology72: 1520–1528.

Castello L, Stewart DJ. 2010. Assessing CITES non-detrimentfindings procedures for Arapaima in Brazil. Journal ofApplied Ichthyology 26: 49–56.

Castello L, Castello JP, Hall CAS. 2007. Problemas en el manejode las pesquerias tropicales. Gaceta Ecológica 18: 65–73.

Castello L, Viana JP, Watkins G, Pinedo-Vasquez M, LuzadisVA. 2009. Lessons from integrating fishers of arapaima insmall-scale fisheries management at the Mamirauá Reserve,Amazon. Environmental Management 43: 197–209.

Castello L, McGrath DG, Beck PSA. 2011a. Resourcesustainability in small-scale fisheries in the Lower Amazonfloodplains. Fisheries Research 110: 356–364.

Castello L, Stewart DJ, Arantes CC. 2011b. Modelingpopulation dynamics and conservation of arapaima in theAmazon. Reviews in Fish Biology and Fisheries 21: 623–640.

Castello L, Viana JP, Pinedo-Vasquez M. 2011c. Participatoryconservation and local knowledge in the Amazon várzea: thepirarucu management scheme in Mamirauá. In The AmazonVárzea: The Past Decade and the Decade Ahead, Pinedo-Vasquez M, Ruffino M, Padoch CJ, Brondízio ES (eds).Springer: New York; 259–273.

Castello L, McGrath DG, Hess LL, Coe MT, Lefebvre PA,Petry P, Macedo MN, Reno V, Arantes CC. 2013a. Thevulnerability of Amazon freshwater ecosystems.Conservation Letters 6: 217–229.

Castello L, McGrath DG, Arantes CC, Almeida OT. 2013b.Accounting for heterogeneity in small-scale fisheriesmanagement: the Amazon case. Marine Policy 38: 557–565.

Castello L, Stewart DJ, Arantes CC. 2014. O que sabemos eprecisamos fazer a respeito da conservação do pirarucu(Arapaima spp.) na Amazônia. In Biologia, Conservação eManejo Participativo de Pirarucus na Pan-Amazônia,Amaral E (ed.). Instituto de Desenvolvimento SustentávelMamirauá: Tefé, Brasil; 17–31.

Castro F, McGrath DG. 2003. Moving toward sustainability inthe local management of floodplain lake fisheries in theBrazilian Amazon. Human Organization 62: 123–133.

Cinner JE, Daw T, McClanahan TR. 2009. Socioeconomicfactors that affect artisanal fishers’ readiness to exit adeclining fishery. Conservation Biology 23: 124–130.

Costanza R. 1987. Social traps and environmental policy.BioScience 37: 407–412.

Crampton WGR, Castello L, Viana JP. 2004. Fisheries in theAmazon várzea: historical trends, current status, and factorsaffecting sustainability. In People in Nature: Wildlife Conservationin South and Central America, Silvius K, Bodmer R, FragosoJMV (eds). Columbia University Press: New York; 76–95.

Cuvier G. 1822. Das Thierreich eingetheilt nach dem bau derThiere als Grundlage ihrer Naturgeschichte und dervergleichenden Anatomie. Mit vielen Zusätzenversehen vonH.R. Schinz. Vol. 2. Cotta: Stuttgart and Tübingen.

Cuvier G, Valenciennes A. 1847.Histoire naturelle des poissons.Tome dix-neuvième. Suite du livre dix-neuvième. Brochets ou

L. CASTELLO ET AL.


Lucioïdes. Livre vingtième. De quelques familles deMalacoptérygiens, intermédiaires entre les Brochets et lesClupes. Libraire de la Société Géologique de France: Paris.

Da Silveira R., Thorbjarnarson JB. 1999. Conservationimplications of commercial hunting of black and spectacledcaiman in the Mamirauá Sustainable Development Reserve,Brazil. Biological Conservation 88: 103–109.

Davis A, Wagner JR. 2003. Who knows? On the importance ofidentifying “experts” when researching local ecologicalknowledge. Human Ecology 31: 463–489.

de Mitcheson YS, Craig MT, Bertoncini AA, Carpenter AA,Cheung WWL, Choat JH, Cornish AS, Fennessy ST, FerreiraBP, Heemstra PC, et al. 2013. Fishing groupers towardsextinction: a global assessment of threats and extinction risksin a billion dollar fishery. Fish and Fisheries 14: 119–136.

Dulvy NK, Reynolds JD. 2009. Biodiversity: skates on thin ice.Nature 462: 417.

Dulvy NK, Sadovy Y, Reynolds JD. 2003. Extinctionvulnerability in marine populations. Fish and Fisheries 4: 25–64.

Estes JA, Terborgh J, Brashares JS, Power ME, Berger J, BondWJ, Carpenter SR, Essington TE, Holt RD, Jackson JBC,et al. 2011. Trophic downgrading of planet Earth. Science333: 301–306.

Gordon HS. 1954. The economic theory of a common propertyresource: the fishery. Journal of Political Economy 62: 124–142.

Günther A. 1868. Catalogue of the Physostomi, containing theFamilies Heteroptygii, Cyprinidae, Gonorynchidae,Hyodontidae, Osteoglossidae, Clupeidae, Chirocentridae,Alepocephalidae, Notopteridae, Halosauridae, in theCollection of the British Museum. Catalogue of the Fishes inthe British Museum, Vol. 7. British Museum Trustees: London.

Halls AS, Welcomme RL, Burn, RW. 2006. The relationshipbetween multi-species catch and effort: among fisherycomparisons. Fisheries Research 77: 78–83.

Hardt MJ. 2009. Lessons from the past: the collapse ofJamaican coral reefs. Fish and Fisheries 10: 143–158.

Hoeinghaus DJ, Agostinho AA, Gomes LC, Okada EK,Pelicice FM, Kashiwaqui EAL, Latini JD, Winemiller KO.2009. Effects of river impoundment on ecosystem servicesof large tropical rivers: embodied energy and market valueof artisanal fisheries. Conservation Biology 23: 1222–1231.

Irion G, Junk WJ, Mello JASN. 1997. The large centralAmazonian river floodplains near Manaus: geological,climatological, hydrological and geomorphologicalaspects. In The Central Amazon Floodplain: Ecology of aPulsing System, Junk WJ (ed.), Springer Verlag: Berlin;23–46.

Isaac V, Ruffino M. 1996. Population dynamics of tambaqui,Colossoma macropomum Cuvier, in the Lower Amazon,Brazil. Fisheries Management and Ecology 3: 315–333.

Isaac VJ, Da Silva CO, RuffinoML. 2008. The artisanal fisheryfleet of the lower Amazon. Fisheries Management andEcology 15: 179–187.

Jennings S, Polunin N, 1996. Impacts of fishing on tropical reefecosystems. Ambio 25: 44–49.

Jennings S, Grandcourt SMl, Polunin NVC. 1995. Theeffects of fishing on the diversity, biomass and trophicstructure of Seychelles’ reef fish communities. Coral Reefs14: 225–235.

Johannes RE. 1998. The case for data-less marine resourcemanagement: examples from tropical nearshore finfisheries.Trends in Ecology & Evolution 13: 243–246.

Johannes RE, Freeman MMR, Hamilton RJ. 2000. Ignorefishers’ knowledge and miss the boat. Fish and Fisheries1: 257–271.

Junk WJ. 2000. The central Amazon river floodplain: conceptsfor the sustainable use of its resources. In The CentralAmazon Floodplain: Actual Use and Options for aSustainable Management, Junk WJ, Ohly JJ, Piedade MTF,Soares MGM (eds). Backhuys Publishers: Lieben; 75–94.

Lae R. 1997. Does overfishing lead to a decrease in catches andyields? An example of two West African coastal lagoons.Fisheries Management and Ecology 4: 149–164.

Lorenzen K, Almeida O, Arthur R, Garaway C, Khoa SN.2006. Aggregated yield and fishing effort in multispeciesfisheries: an empirical analysis. Canadian Journal ofFisheries and Aquatic Sciences 63: 1334–1343.

Mahon R. 1997. Does fisheries science serve the needs ofmanagers of small stocks in developing countries? CanadianJournal of Fisheries and Aquatic Sciences 54: 2207–2213.

McClanahan TR, Cinner JE. 2008. A framework for adaptivegear and ecosystem-based management in the artisanal coralreef fishery of Papua New Guinea. Aquatic Conservation:Marine and Freshwater Ecosystems 18: 493–507.

McClanahan, TR, Maina J, Davies J. 2005. Perceptions ofresource users and managers towards fisheries managementoptions in Kenyan coral reefs. Fisheries Management andEcology 12: 105–112.

McGrath DG, Castro F, Futemma C, Amaral BD, Calabria J.1993. Fisheries and evolution of resource management on theLower Amazon floodplain. Human Ecology 21: 167–195.

McGrath DG, Silva UL, Martinelli MC. 1998. A traditionalfloodplain fishery of the lower Amazon River, Brazil. Naga,the ICLARM Quarterly 1: 4–11.

McGrath DG, Cardoso A, Almeida OT, Pezzuti J. 2008.Constructing a policy and institutional framework for anecosystem-based approach to managing the LowerAmazon floodplain. Environment, Development andSustainability 10: 677–695.

McManus JW, Nañola JCL, Reyes RB, Kesner KN. 1992.Resource Ecology of the Bolinao Coral Reef System,ICLARM Studies and Reviews, ICLARM: Manila.

Myers RA, Mertz G. 1998. The limits of exploitation: aprecautionary approach. Ecological Applications 8: 165–169.

Ormerod, SJ. 2014. Rebalancing the philosophy of riverconservation. Aquatic Conservation: Marine and FreshwaterEcosystems 24: 147–152.

Pauly D, Christensen V, Dalsgaard J, Froese R, Francisco T.1998. Fishing down marine food webs. Science 279: 860–863.

Pauly D, Watson R, Alder J. 2005. Global trends in worldfisheries: impacts on marine ecosystems and food security.Philosophical Transactions of the Royal Society B –BiologicalSciences 360: 5–12.

Pinnegar JK, Jennings S, O’Brien C, Polunin NVC. 2002.Long-term changes in the trophic level of the Celtic Sea fishcommunity and fish market price distribution. Journal ofApplied Ecology 39: 377–390.

Pitcher TJ. 2001. Fisheries managed to rebuild ecosystems?Reconstructing the past to salvage the future. EcologicalApplications 11: 601–617.

Renó VF, Novo EMLM, Suemitsu C, Rennó CD, Silva TSF.2011. Assessment of deforestation in the Lower Amazonfloodplain using historical Landsat MSS/TM imagery.Remote Sensing of Environment 115: 3446–3456.



Reynolds JD, Jennings S, Dulvy NK. 2001. Life histories offishes and population responses to exploitation. InConservation of Exploited Species, Reynolds JD, Mace GM,Redford KH, Robinson JG (eds). Cambridge UniversityPress: Cambridge; 147–168.

Reynolds JD, Dulvy NK, Roberts CM. 2002. Exploitation andother threats to fish conservation. In Handbook of FishBiology and Fisheries, Hart PJ, Reynolds JD (eds).Blackwell: Cornwall; 319–341.

Roberts CM, Hawkins JP. 1999. Extinction risk in the sea.Trends in Ecology & Evolution 14: 241–246.

Roberts CM, McClean CJ, Veron JEN, Hawkins JP, Allen GR,McAllister DE, Mittermeier CG, Schueler FW, Spalding M,Wells F, et al. 2002. Marine biodiversity hotspots andconservation priorities for tropical reefs. Science 295: 1280–1284.

Ruffino ML, Isaac VJ. 1999. Dinâmica populacional dosurubim-tigre, Pseudoplatystoma tigrinum (Valenciennes,1840) no médio Amazonas (Siluriformes, Pimelodidae).Acta Amazonica 29: 463–476.

Sadovy Y, Cheung WL. 2003. Near extinction of a highlyfecund fish: the one that nearly got away. Fish and Fisheries4: 86–99.

Stevens GC. 1989. The latitudinal gradient in geographicalrange: how so many species coexist in the tropics. AmericanNaturalist 33: 240–256.

Stewart DJ. 2013a. A new species of Arapaima(Osteoglossomorpha: Osteoglossidae) from the SolimõesRiver, Amazonas State, Brazil. Copeia 2013: 470–476.

Stewart DJ. 2013b. Re-description of Arapaima agassizii(Valenciennes), a rare fish from Brazil (Osteoglossomorpha:Osteoglossidae). Copeia 2013: 38–51.

Welcomme RL. 1999. A review of a model for qualitativeevaluation of exploitation levels in multi-species fisheries.Fisheries Management and Ecology 6: 1–19.

World Conservation Monitoring Centre. 1996. Arapaimagigas. In IUCN Red List of Threatened Species. Version2011.2. <www.iucnredlist.org>. Downloaded on 11November 2013.

L. CASTELLO ET AL.


http://www.iucnredlist.org

Understanding fishing-induced extinctions in the Amazon · Understanding ﬁshing-induced extinctions in the Amazon LEANDRO CASTELLOa,*, CAROLINE CHAVES ARANTESb, DAVID GIBBS MCGRATHc,

Documents