Disentangling the socio-ecological drivers behind illegal fishing in a small-scale fishery managed by a TURF system
Silvia de Juan, Maria Dulce Subida, Andres Ospina-Alvarez, Ainara Aguilar, Miriam Fernandez
SSocio-ecological drivers of illegal fishing de Juan et al. 2020
Disentangling the socio-ecological drivers behind illegal fishing in asmall-scale fishery managed by a TURF system Silvia de Juan , Maria Dulce Subida , Andres Ospina-Alvarez , Ainara Aguilar ,Miriam Fernandez Spanish Scientific Research Council, Institute of Marine Sciences (ICM-CSIC),Passeig Maritim de la Barceloneta 37-49, Barcelona, Spain. Estacion Costera de Investigaciones Marinas, Pontificia Universidad Catoolicade Chile, Alameda 340, C.P. 6513677, Casilla 193, Correo 22, Santiago, Chile. Spanish Scientific Research Council, Mediterranean Institute for AdvancedStudies (IMEDEA-CSIC/UIB), C/ Miquel Marques 21, CP 07190 Esporles,Balearic Islands, Spain.*corresponding author: [email protected]
Highlights: • This study describes the development of a Bayesian network adapted tocomplex SSF.• A Bayesian network was designed to identify drivers of illegal fishing.• The network was based on the link between socio-economic drivers andresource state.• Scenario analysis explored effects of variables that are susceptible to bemanaged.
Abstract:
A substantial increase in illegal extraction of the benthic resourcesin central Chile is likely driven by an interplay of numerous socio-economiclocal factors that threatens the success of the fisheries’ management areas(MA) system. To assess this problem, the exploitation state of a commerciallyimportant benthic resource (i.e., keyhole limpet) in the MAs was related withsocio-economic drivers of the small-scale fisheries. The potential drivers of illegalextraction included rebound effect of fishing effort displacement by MAs, level ofenforcement, distance to surveillance authorities, wave exposure and land-basedaccess to the MA, and alternative economic activities in the fishing village. Theexploitation state of limpets was assessed by the proportion of the catch thatis below the minimum legal size, with high proportions indicating a poor state,and by the relative median size of limpets fished within the MAs in comparisonwith neighbouring OA areas, with larger relative sizes in the MA indicatinga good state. A Bayesian-Belief Network approach was adopted to assess theeffects of potential drivers of illegal fishing on the status of the benthic resourcein the MAs. Results evidenced the absence of a direct link between the levelof enforcement and the status of the resource, with other socio-economic (e.g., © 2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 licensehttp://creativecommons.org/licenses/by-nc-nd/4.0/ Preprint submitted to:
Marine Policy
Page 1 a r X i v : . [ ec on . GN ] D ec ocio-ecological drivers of illegal fishing de Juan et al. 2020alternative economic activities in the village) and context variables (e.g., fishingeffort or distance to surveillance authorities) playing important roles. Scenarioanalysis explored variables that are susceptible to be managed, evidencing thatBBN is a powerful approach to explore the role of multiple external drivers, andtheir impact on marine resources, in complex small-scale fisheries.Keywords: IUU, poaching, enforcement, fisheries management areas, artisanalfisheries, benthic resources.
INTRODUCTIONSmall-scale, artisanal, fisheries (SSF) are the pillar of wellbeing for many coastalcommunities, as it is estimated that they contribute to half of the global catch[1], while employing 90% of the world’s fisheries [2]. However, these fisheriesare largely unassessed [3] and often data poor [4]. SSF are becoming a priorityfor FAO, and efforts are on improving data gathering to guide monitoringand management protocols. Innovative approaches to maximize the availableinformation from SSF are critical, e.g., the incorporation of Traditional EcologicalKnowledge [5] or the use of fishers’ perception surveys [6]. While data gatheringfor the assessment of SSF is improving, there are several caveats driving fishstocks to overexploitation that need urgent assessment. These factors may beclassified into two interacting types: those related with failure of managementand conservation rules [7] and those related with the socio-economic context asthe loss of the culture of care and responsibility for marine resources promptedby strongly market oriented fishery policies [8].Scientific evidence suggests that non-compliance with fishing regulations is awidespread phenomenon in global fisheries [3,9,10] , and emerges among themost important factors contributing to the overexploitation of marine resources[11,12]. However, information on illegal fishing practices in SSF is still scarce.SSF are potentially more prone to violations to catch limits or minimum sizesdue to the nature of the operation [13]. Their usual spatially scattered natureimpose serious challenges to monitoring, surveillance and enforcement to detectany non-compliance activities, facilitating illegal fishing [3,9,10]. It is a complexproblem to solve, as numerous socio-economic factors are probably playing akey role, and it is likely to be highly conditioned by local context characteristics[8,14].The absence of incentives to comply with fisheries normative has been pointedout as a significant problem for SSF, particularly in scenarios where neoliberalfishing policies fostered severe extractivism in detriment of traditional and moresustainable fisheries ([8] and references therein). The common pool resourcesdilemma [15] has been in part solved using Territorial User Right in Fisheries(TURF). While these co-management systems aim to encourage trust, rulecompliance and fishers’ involvement in enforcement [16], they are also likely toallow fishers to sell fish at higher prices, reduce resource waste and increase fishers’incomes ([17] and references therein). Chile was pioneer on the implementationof a TURF system to the SSF at a national scale in 2003, known as ManagementPreprint submitted to:
Marine Policy
Page 2ocio-ecological drivers of illegal fishing de Juan et al. 2020and Exploitation Areas for Benthic Resources (AMERB, hereafter MA) [18]. TheTURF system in Chile has contributed in increasing abundance of commercialspecies [19] and has shown positive effects on biodiversity and trophic webstructure [20,21]. However, the fishing pressure outside these management areas[22] and a negative perception of fishers on the TURF system in aspects relatedwith economical revenues and government support [23,24] urge an upgrade inthis management paradigm. Recent studies suggest that illegal extraction inthe TURFs can be as high as 68% of the annual income obtained from thissystem in some regions of Chile [25], and recent biological surveys providedevidences of poaching in several management areas [26]. Research on fisheriesworking on a TURF basis, including some studies in central Chile, identified aseries of incentives promoting illegal fishing. For instance, the usually limitedhuman resources available for surveillance activities (both from the authoritiesand fishers) make it difficult to identify and punish poachers, and that there isa well-established black market in demand of fisheries products [27,28]. Thesefindings trigger the need of an integrative approach to identify which are themain drivers of illegal extraction in Chile TURF system.Usually, TURFs’ efficiency is assessed through the ecological state of biologicalcomponents or through the socio-economic benefits of the management [17]; theinteraction between these two aspects has seldom been addressed. Here, weprovide a quantitative assessment of the effects of socio-ecological drivers onthe illegal fishing of the traditional and culturally important keyhole limpetfishery in the central coast of Chile. A Bayesian Belief Network (BBN) modelwas developed linking key biological and socio-economic components of the SSF.BBNs have been applied to consider management systems governing artisanalfisheries where the effects of qualitative and quantitative factors are of concern[29–33], or when considering social, environmental and economic factors leadingto multi-objective management of coastal resources [34,35]. In the present work,we used a BBN model that integrates data from fisheries stakeholders andscientists to identify the key drivers that influence the proportion of illegalfishing in the TURF system, as well as the relationship between factors thathave generally been regarded as determining the state of the resource (e.g., levelof enforcement) and contextual factors (e.g., rebound effect of fishing effortdisplacement or distance to surveillance authorities). The BBN approach canhelp building a socio-ecological model of fisheries in an environment that ispartly data-poor. Moreover, the graphical outputs also facilitate communicatingthe results to stakeholders. The present paper addresses relevant issues drivingillegal extraction in the TURF system, including the identification of a) factorsthat determine the level of effective enforcement needed to reduce illegal fishing,and b) variables that can assist rapid assessment of effective enforcement andsuccess of the SSF system.METHODS
Small-scale fisheries in central Chile
Small-scale benthic fisheries in Chile are mostly organized around fishing covesPreprint submitted to:
Marine Policy
Page 3ocio-ecological drivers of illegal fishing de Juan et al. 2020locally known as caletas , which serve as operational bases for the local fleet andfisher organizations. The benthic fisheries operate around 17 km away fromthe coves [22] to a depth of approximately 20 m (official diving depth), andthe harvesting fishing grounds show two contrasting management regimes: (i)exclusive harvest rights assigned to fishers’ organizations (TURFs), locally knownas management areas (hereafter MAs) or (ii) historical fishing grounds withoutspatial entry restrictions, hereafter referred as open access areas (OAs). Themost common fishery resources that can be extracted in the MAs are the Chileanabalone or loco (
C. conholepas ), keyhole limpets (a set of species of the genus
Fissurella ), the red sea urchin (
Loxechinus albus ) and kelp (mainly
Lessonia spp.)that are exclusively exploited by fishermen pertaining to the fisheries association.Outside the MAs, both fishers belonging to the association and officially registeredun-associated fishers can extract fish and benthic resources (except loco). Whilesome species-specific regulations operate for both management regimes (e.g.temporal reproductive bans or minimum legal size) others apply exclusively toMAs (annual quotas) or OA (total ban of locos). However, due to differences inthe administration of the MAs, not all these areas exhibit a similar enforcementlevel [36]. A well-enforced MA implies that the maximum quotas are respected,and the area is surveyed to avoid illegal fishing. The control of catches andenforcement of regulations is usually performed by the National Fisheries Service,with about 25,000 enforcement actions during 2018 focusing on MLS regulationsin both artisanal and industrial fleet [24]. However, fishers’ associations mustcover surveillance costs, although it is not mandatory. In the present work,the term poaching refers to the illegal extraction of the resource, either due tononcompliance with the annual quotas by the associated fishermen, or to illegallyextracting the resource from MAs by un-associated fishermen.For the present study, data was gathered from 13 fishers’ associations thatmanage 24 areas in the central coast of Chile (Fig. 1). Each fishers’ associationpresented between 1 and 3 operational MAs at the time of the study. Differentsources of data were gathered for the study. In 2016-2020, face-to-face interviewsand telephonic questionnaires were carried out to the leaders of the fishers’associations to obtain information and perceptions on MA enforcement, poachingintensity and the organization of the fishers association. In 2017-2018, biologicalsurveys were conducted in MAs and neighbouring OAs to assess keyhole limpetsize structure.
Interviews with fishers
Interviews with the leaders of the fishers’ associations were carried out from2016 to 2020 in 40 associations along the central coast of Chile. Two sets ofinterviews were carried out: (a) face-to-face extended interviews focused on 16coves that covered ca. 250 km of the coast in central Chile; and (b) remote(telephonic) interviews focused on 24 coves covering the entire study area ( ca.
700 km). The two approaches used the same questionnaire that aimed to explorethe enforcement protocol, or lack of it, the evenness in enforcement across MAsand the fishers’ perception on the enforcement effectiveness and poaching recordsPreprint submitted to:
Marine Policy
Page 4ocio-ecological drivers of illegal fishing de Juan et al. 2020Figure 1: Map of the study area including the names of the 13 coves included inthe study (local names in red). Note that each association might have between1 and 2 management areas, resulting in 24 areas included in the analysis. Theranking corresponds to the enforcement level (5 being the highest enforcement).The ranking was performed with a wider set of 39 fishers’ associations (includedin the map).Preprint submitted to:
Marine Policy
Page 5ocio-ecological drivers of illegal fishing de Juan et al. 2020per year. The full questionnaire is provided in supplementary material.The information gathered during these interviews allowed defining the levelof enforcement endured by each fishers’ association. The enforcement rankingwas based on a dichotomy tree that represented all the options that a fishers’association could exhibit. As a result, an area could have a very high level ofenforcement (rank 5), when the fisher association hires an external person toperform 24h surveillance; on the other end, a very low enforcement (rank 1)occurs when there is no surveillance in the area (Fig. 2). Differences betweenhigh and very high enforcement, with 24 hours surveillance performed by eitherthe members of the fishermen association or by a payed person respectively, isjustified by an expected higher compromise by the paid person [23]. Fishersperception on enforcement effectiveness (i.e., fishermen perceived the enforcementin their MAs was either effective or non-effective) and surveillance differencesamong the MAs controlled by each fishers’ association (i.e., an association surveyall MA equally vs . surveillance efforts focus on a single MA) allowed the inclusionof an additional variable that considered the “effectiveness” of enforcement. Tocompute this variable, the enforcement level of a MA (ranked in 1 to 5) wasreduced by one score if the fishers’ association dedicated less time to surveythis MA (relative to other MAs controlled by the same association), and by anadditional score if the fishers perceived that enforcement in their MAs was noteffective. As a result, 20 out of the 24 MAs had an “effective enforcement” lowerthan the formal “enforcement level”. During the interviews, fishers also providedinformation on number of poaching events that had been reported in their MAson the past year: from low, with less than 20 events reported in the last year, tovery high, with more than 100 events reported in the year. All the variables perMA are included in the supplementary information (Table S1). Attributes of Management Areas
Additional context variables considered relevant to assess poaching intensity overkey-hole limpet were: 1) Availability of OA area per registered fisher in relationto the total area assigned to MA, in each cove (index IAOA); low availabilityof OA per fisher due to a high spatial density of MAs has been related with anincrease of illegal fishing in OAs due to an effort displacement [26]. Here wesuggest that such an increase in illegal fishing in OAs may yield, under certaincircumstances (like low enforcement, low compliance and/or lack of economicalternatives), a rebound effect of increased poaching in MAs. For each cove,the IAOA is a proxy of the proportion of OAs that corresponds to each diverofficially registered who may exert fishing effort around that cove. It considersfishing effort density and proportion of OA areas in relation to MA, in theaccessible fishing grounds of the cove. The lower the estimate of IAOA, the lowerthe proportion of OAs in relation to MAs (less areas open to fisheries availablenear a cove) and, therefore, higher poaching pressure over OAs and over MAsas a rebound effect. Details of the methodological approach are provided byFernandez et al. [26]. 2) Surface of the MA, as larger surfaces are expectedto be more difficult to guard. 3) Distance between the MAs and the officialPreprint submitted to:
Marine Policy
Page 6ocio-ecological drivers of illegal fishing de Juan et al. 2020Figure 2: Ranking of enforcement level based on the fishermen’s organization:payment of external personnel for the continuous enforcement (hired enforce-ment), the organization of the fishers to conduct the enforcement of their areas,or no enforcement of the areas. The duration of the enforcement (guard) couldvary from every day, 24hrs, to only occasional. The highest enforcement is 5(very high), lowest 1 (very low)Preprint submitted to:
Marine Policy
Page 7ocio-ecological drivers of illegal fishing de Juan et al. 2020surveillance base, as this official body is responsible for sanctioning the poachers,so fishermen contact them when a poacher is spotted in the area; the moredistant, the less effective is enforcement as poachers must be caught in act inorder to be sanctioned. 4) Access to the MA from land through main pavedroutes from the nearby location, considering also difficulty of sea access fromland, e.g., a cliff implies a difficult access; difficult access might act as a naturalprotection against poachers that often reach MAs from land. 5) Wave exposureof the fishing grounds; higher exposure acts as natural protection, as MAs withhigh wave exposure are less likely to be accessed by poachers (a high exposedarea in the central coast of Chile is an area facing south, while a protected areais facing north). 6) Existence of alternative economic activities in the cove (e.g.,tourism, construction, recreation); this, according to fishers’ leaders, resultsin less poaching pressure, as fishers find alternative sources of income. Thegeographical variables distance to surveillance, wave exposure and access fromland were estimated using Google Earth software. The presence of alternativeactivities was depicted from the fishers’ interviews. All the variables per MA areincluded in the supplementary information (Table S1).
Biological surveys
Keyhole limpets (
Fissurella spp.) are targeted as primary resources in themanagement plans of the vast majority of MAs. This resource is currentlyclassified as fully exploited, with minimum landing size set at 6.5 cm for shelllength in the study area, which correspond to an age of 2 years of benthic life.In order to assess the influence of the management regime on the proportion ofundersize limpets in the catch, a field study was conducted between October2017 and July 2018 in MAs associated to the 13 fishers’ associations (Fig. 1), asdescribed in Fernández et al. [26]. Paired MA and OA sites were sampled ineach area by the local fishers with an observer onboard. In each area, at leasttwo sites were sampled since a minimum of one MA and one OA were required.The sampling procedure was identical for the two management regimes (MAand OA). Samples were directly obtained from the catch of a benthic fisher, i.e.,a semi-autonomous diver (hookah), and measured onboard the fishing boat orat the landing beach or small-scale port. Sample size is different among sitesbecause all individuals in the fishing bags were measured, until reaching theminimum sample size of 200 individuals, following the protocol established byAndreu-Cazenave et al. [37]. Size of keyhole limpets was measured as the totallength of the shell to the nearest mm. Non-parametric Wilcoxon signed-ranktests were used to compare medians of the size distributions of the catch betweenMA and OA within each cove (W will be used to indicate the Wilcoxon teststatistic). For further detail on the analysis of the data see Fernández et al. [26].
Bayesian Belief Network Development
Developed as essentially qualitative graphical models, BBNs are especially power-ful explaining the causal relationships between variables (nodes) via conditionalprobability distributions. In BBNs, the processes are not necessarily explicitlycaptured, on the contrary, the expected probabilities of the outcomes are basedPreprint submitted to:
Marine Policy
Page 8ocio-ecological drivers of illegal fishing de Juan et al. 2020on particular combinations of events. Moreover, the probabilities coming fromthe combination of qualitative and quantitative data can be assigned to the BBNnodes and could come from combinations of expert opinions, empirical field dataor previous experiences cited in the literature. Since BBNs provide a probabilityof an outcome rather than a discrete (deterministic) one, a mean (expected)outcome and a confidence interval can be determined. The way in which eachinput is combined to report the probability of an outcome is determined bya weighting combination rather than by a numerical estimation process. Inother words, it is not necessary to develop formal structural relationships linkingthe different components of the model, allowing for non-linear or discontinuousresults if deemed appropriately.BBNs consist of two structural models: 1) a conceptual model (directed acyclicgraph, DAG) that represents the best available knowledge about the functioningof the system and representing the links between the model variables (callednodes); and 2) conditional probability tables (CPTs) and conditional probabilitydistributions (CPDs), which determine the strength of the links in the DAG.Directed arrows representing cause-effect relationships between system variablesindicate the statistical dependence between different nodes. Each arrow startsin a parent node and ends in a child node. Feedback arrows from child nodes toparent nodes are allowed. The DAG can be developed by experts based on anunderstanding of the system and/or based on empirical observations. This initialstructure of the DAG is the basis for an operational BBN. The probabilisticrelationships between the model nodes are specified in the CPTs and CPDs. TheCPTs and CPDs can be parameterised based on expert opinion, derived frommathematical or logical equations, or learned from the relevant empirical datastructure. The nodes are restricted to a limited number of states that describethe probability distribution of system variables (e.g., a node can be discrete,with incremental or decremental states or levels, or be continuous type). Theprobability distribution of each node is contained in its CPT or CPD, and eachgiven state of one variable is associated with a probability between 0 and 1, sothat the sum of the state values adds up to 1 (100%).In our study the DAG was developed iteratively based on the system under-standing of the research team (see section 2.3 for the expected links betweenvariables). The first DAG was created by 12 initial nodes: 10 drivers, 1) MAsurface, 2) number of MA per fishermen association, 3) distance to surveillanceauthority, 4) access to MA from land, 5) wave exposure of the MA, 6) availabilityof OA (IAOA), 7) alternative activities in the cove, 8) enforcement level, 9)enforcement effectiveness, 10) perceived poaching level; and 2 response variables,1) the proportion of keyhole limpet below the minimum landing size (i.e., illegalproportion), as a high proportion is an indication of higher predisposition forpoaching; and 2) the difference in median size between the MA and adjacent OA(ê, as the normalized median MA/median OA). Values close to 1 indicate a goodstate of the resource with highest median sizes in MA compared to the pairedOA area. All possible relationships between the 12 initial nodes were consideredand quantified. The draft DAG structure included the main components relevantPreprint submitted to:
Marine Policy
Conditional probability queries or “What would happen if. . . ” scenarios
A query returns the probability of a specific event given some evidence. Forexample, a query could be of the type "If A occurs and B does not occur and C isgreater than X and less than Y, what is the probability that D is greater than Z?”.Based on this approach, a set of scenarios were explored to assess the influence ofthe external drivers on the state of the benthic resource. The scenarios consideredthat the management target is to improve the state of keyhole limpet stockand to reduce the proportion of limpets below minimum landing size. In ourcase study, the variables that are susceptible to be managed are those linked tothe fishers’ association and management bodies. The conditional queries firstconsider the probabilities of the response variable under conditional drivers basedon 2000 permutations. The conditional drivers were enforcement level, distanceto surveillance, availability of OA area and alternative economic activities in thePreprint submitted to:
Marine Policy
Page 10ocio-ecological drivers of illegal fishing de Juan et al. 2020cove. The management scenarios were also explored the other way around: whatwould be the probabilities of different values of the selected drivers for a specificbiological response? We considered as an optimal biological response a relativemedian size in the MA above 0.59 and an illegal proportion of limpet bellow0.31. These responses occurred in ca.
15% of the MAs included in the study.RESULTS
Size structure of keyhole limpet in Management Areas
Significant differences in median keyhole limpet sizes were found in 22 out 24OA-MA paired comparisons (11 out of 13 coves). Although in 19 of 22 pairedcomparisons median size was larger in MAs than in OAs, as expected for awell-enforced MA, in three cases the reverse pattern was observed, suggestingunwanted management outcomes (Algarrobo MA1, Chungungo MA1 and Chun-gungo MA2). Furthermore, we found two coves with illegality levels in MAsequal to or higher than OAs: Chigualoco, where the MA 3 showed a proportionof undersize individuals in the catch similar to the OA (13 and 11%, respectively)and Chungungo, where the MAs 1 and 2 showed significantly higher proportionsof undersize individuals (62 and 71%, respectively, see Fig. 6) in relation to the41% recorded in the OA (plots for median sizes in OA and MA and statisticaltest are included in supplementary information; Table S2).
Enforcement level and effectiveness
From the 39 coves where enforcement was assessed (Fig. 1), 13% had noenforcement, 33% only occasional enforcement, 23% a daily 8-hours enforcement.The rest had daily enforcement (24 h), which is performed either by fishers(8%) or hired personnel (23%). The subset of 13 coves included in the studyexhibited variable enforcement, from low to very high, with no example of noenforcement (very low); therefore, the enforcement variable included in the BBNhas 4 levels. Three fishermen associations had low enforcement, 1 moderate, 2high and 7 very high enforcement. However, by considering the effectivenessof the enforcement level endured by the fishermen association, in 20 MAs theenforcement level was reduced by 1 or 2 scores. In 11 MAs from associationswith very high enforcement level, the effective enforcement was lower than theformal enforcement: 1 MA changed from very high to low; 5 from very high tomoderate; 5 from very high to high (supplementary information; Table S2).There was no linear relationship between the level of enforcement and thenumber of poaching events per year reported by the fishermen leaders during theinterviews. Generally, sites with high enforcement have moderate-low perceivedpoaching; and low enforcement is related with high or very high perceivedpoaching. However, in some cases, high enforcement is related with very highperceived poaching (Quintay and Hornos) or low enforcement is related with lowperceived poaching (Chungungo) (Fig. 4).Similarly, there is no evident relationship between the state of the resource,measured either as the illegal proportion of keyhole limpets or as the relative sizePreprint submitted to:
Marine Policy
Page 11ocio-ecological drivers of illegal fishing de Juan et al. 2020of individuals in the catch in MAs, and the enforcement level and enforcementeffectiveness in each MA (Fig. 5). The absence of a direct link suggests othervariables are playing a role in the state of the benthic resource.Figure 4: Cleveland dot plot showing the overlap between the ranking of enforce-ment endured by the fishers’ association (green dots, 1, low, to 4, high) and thelevel of poaching perceived by the fishermen (red dots, 1, low, to 4, high)
Bayesian-belief network
A BBN approach was adopted to explore the influence of a set of externalpotential drivers of poaching intensity over keyhole limpet state within theMAs (Fig. 3). Keyhole limpet state was assessed through the relative sizeof limpets fished in MAs and the illegal proportion of limpets in the catch(see methods for further details). The limpet state is considered a proxy forthe effects of poaching on benthic resources. The level of poaching was alsoassessed relying on fishermen perceptions (supplementary information; TableS1); however, the perceived poaching exhibited an unexpected link with thePreprint submitted to:
Marine Policy
Page 12ocio-ecological drivers of illegal fishing de Juan et al. 2020Figure 5: Relationship between the relative median size of individual limpetsfished in the MA (ê), left panels, and the proportion of illegal limpets in thecatch, right panels, and the enforcement level (upper panels) and enforcementeffectiveness (bottom panels) in the case studies (24 MAs, indicated by theirlocal names).Preprint submitted to:
Marine Policy
Page 13ocio-ecological drivers of illegal fishing de Juan et al. 2020condition of the resource, as lower reported poaching events were linked withlower average sizes in MA and higher proportions of illegal size. Therefore, thisvariable was excluded from the final BBN. Access and number of MAs werealso excluded from the optimal network, as these variables had no link with thebiological variables in our case study.The most parsimonious BBN included most links that were predicted by expertknowledge in the draft DAG (Fig. 3 vs . Fig. 6). However, some links wererather unexpected (Fig. 6). For example, the existence of alternative activitiesplays a major role in the network, influencing both the magnitude of the effectsof availability of OA area (with less fishing pressure rebounding on the MAs)and enforcement level, as alternative activities might reduce the time dedicatedto survey the MAs. On the other hand, availability of OA area is linked to theillegal proportion of limpet through the distance to the surveillance authorityand, therefore, poaching reports are not effective. Enforcement has effectsover the relative size in the MA (ê) through the effectiveness of enforcement;therefore, it highlights the relevance of uneven surveillance efforts across severalMAs controlled by a single association. Some of the links were probably casual,considering the limited number of coves and MAs included in the study (Fig.6): availability of OAs and distance to surveillance is linked to the surface ofthe MA and to wave exposure. Whereas the availability of OAs is controlledby both the number and extension of the MA available per fisher in an area ofthe coast, probably the sections of the coast most exposed to waves, and/or lessaccessible from the coves, tend to concentrate less MAs.Figure 6: Bayesian-belief Network obtained for the case study. The thickness ofthe arrow represents the strength of the link. Scenarios to achieve a good state of the fishery resource
Preprint submitted to:
Marine Policy
Page 14ocio-ecological drivers of illegal fishing de Juan et al. 2020In order to observe a low proportion of illegal limpets in the catch, the idealscenario includes a high availability of OA areas per fisher and short distance toa surveillance authority base (scenario 2 in Table 1). The combination of anyof these variables with alternative economic activities for the fishers and highenforcement increases the probability of a lower proportion of limpets of illegalsize in the catch (scenario 1, 3 and in Table 1). In particular, the combinationof high availability of OA areas with any of the other three variables yielded ahigh probability (>0.88) of an illegal proportion below 30% of the catch. Onthe other hand, the external drivers had a weak effect on the relative mediansize of limpets fished in MAs, with values closer to 1 indicating larger mediansizes in the MA compared to neighbouring OA. Either high enforcement orhigh effectiveness combined with the presence of other activities did not yieldprobabilities higher than 39% (Table 1).Table 1. Queries performed over the BBN (Fig. 6). Probabilities of the responsevariable under conditional drivers based on 2000 permutations
Scenarios Variable Variable 2 Response Probability
Sce. 1 Available OA= very highor high Otheractivities =Y Illegalproportion<= 0.3 0.881Sce. 2 Available OA=very highor high Distance tosurveillance= close Illegalproportion<= 0.3 0.935Sce. 3 Available OA=very highor high Enforcement= very highor high Illegalproportion<= 0.3 0.961Sce. 4 Enforcement= very highor high Otheractivities =Y Relative sizeat MA >=0.4 0.378Sce. 5 Effectiveness= moderateto very high Otheractivities =Y Relative sizeat MA >=0.6 0.394Sce. 6 Otheractivities =Y Distance tosurveillance= close Illegalproportion<= 0.3 0.991Sce. 7 Effectiveness= very highor high Distance tosurveillance= close Illegalproportion<= 0.3 0.987In order to attain a good state of the fishery resource, with an illegal proportionof limpet catch below 30% and the relative median size in the MA over 60%(i.e., larger limpets in the MA in reference to neighbouring OA), the optimalcombination of variables indicate that 1) availability of OA area should be highto very high (probability of 0.53), 2) the enforcement should be high to veryPreprint submitted to:
Marine Policy
Page 15ocio-ecological drivers of illegal fishing de Juan et al. 2020high (probability of 0.80), 3) the effectiveness can be maintained at moderate(probability of 0.77), 4) the distance to surveillance should be short (probabilityof 0.59), 4) while the presence or absence of other activities does not have amajor effect (Table 2).DISCUSSIONTo disentangle SSF complex dynamics, innovative approaches are necessary todeal with the scarcity of data and the need to take advantage of both scientificexpertise and fishers’ knowledge [4]. Bayesian-belief networks are useful forSSF research due to flexibility in incorporating data of different nature, whereexpert knowledge plays a key role. The BBN exercise on a TURF system incentral Chile allowed identifying the role of multiple external socio-economic andgeographical context drivers on the sustainability of the fishery resource. OurBBN provided evidence of the absence of a direct link between the level of MAenforcement and the state of the benthic resource, with other socio-economic(e.g., alternative economic activities) and context variables (e.g., fishing effortpressure or distance to surveillance authorities) playing important roles. Theinterviews with fishers evidenced high variability in enforcement protocols andalso that a concern frequently shared by the fishers’ leaders was the high cost ofallocating human resources to survey the area, as these costs must be covered bythe fishers’ association. Often, one association is in charge of more than one MA,which prompts the allocation of the limited enforcement resources only to onearea, either the most accessible or the most productive one. This variabilitywas considered in the conception of the variable “effective enforcement” and themodel evidenced the importance of this variable in the state of the resource, asin some cases the level of enforcement achieved by the fishermen associationwas lowered by an uneven and non-effective enforcement across MAs. Theseobservations are aligned with previous results in the same study area wherefishers reported that their organizations often decide not to monitor the MAsthat are furthest away (less accessible) from the cove [23].The relationship between the enforcement level and the number of poachingPreprint submitted to:
Marine Policy
Page 16ocio-ecological drivers of illegal fishing de Juan et al. 2020events per year reported by the fishers was not consistent, although an oppositerelationship was observed in several coves. The observed lack of consistency inthese results might be showing a bias due to the negative fishers’ perception on theeffectiveness of the poaching sanctioning system [23]. Lab-in-field experimentsfocusing on the effect of co-enforcement on limiting the access to commonpool resources in the study area showed that the existence of external poachingsanctions (like those imposed by a government authority) improved the willingnessof resource users to invest in enforcement, ending up with a reduction in poaching[40]. Compliance-monitoring agencies in Chile concentrate on access controlwith much less emphasis on compliance to quotas, bans or minimum legalsize. According to the national fisheries law, the National Fisheries ServiceSERNAPESCA has to publicly deliver an annual report of enforcement actions,which, in its last edition reveals the low percentage of enforcement work achievedin the central coast of Chile (8.8% of the total field enforcement actions in thecountry), where the present study focuses [41]. This might be related with thegeographic location of coves, several of them very distant from the closest city,where enforcement agencies are usually based. The implementation of localenforcement tools must be necessarily considered to address this difficulty. Thesustainability of SSF is a complex socio-economic and ecological problem [42],with many local variables playing an important role and no bullet-proof solution.In fact, illegal fishing of Chilean SSF has been termed as a wicked problem byNahuelhual et al. [43], as it is characterised by its complexity, uncertainty andinterdependence of factors. Therefore, a multi-dimensional assessment, suchas the one presented here, is necessary to address the illegal fishing of benthicresources in Chile.A Bayesian-belief network is a strong tool to address multi-dimensional problems,as it nourishes from expert knowledge and it is flexible to be used for scenarioanalysis by local actors. In our study, the optimal model included most of thelinks predicted by expert knowledge (Fig. 3 vs . Fig. 7). However, some links wererather unexpected. For example, the existence of alternative economic activitiesis probably alleviating fishing pressure around the coves, and, in the case of areaswith low availability of OAs, this translates into lower poaching in the MAs.The scenario analysis explores variables that are susceptible to be managed,including distance to surveillance authorities and a spatial planning approachto fisheries restrictions. It is legitimate to state that, according to our model,lowering illegality in benthic artisanal fisheries in central Chile depends mostlyon two government administrative matters: i) improving authority surveillancemechanisms and rules (in order to increase effectiveness of anti-poaching controlsand reduce the negative consequences of higher distances to surveillance authoritybases); ii) specific actions in marine spatial planning of the TURF system (inorder to increase the availability of OAs around each MAs).An additional administrative issue is the closure of the delivery of artisanalfishing permits for more than 10 years, that has let young generations of fisherswith no choice but to fish illegally [43]. Fishers leaders’ perceptions gatheredduring this study support this statement, as leaders often pointed out thatPreprint submitted to: Marine Policy
Page 17ocio-ecological drivers of illegal fishing de Juan et al. 2020there is an increasingly large proportion of young, un-associated, fishers thatare discouraged by the current co-management system that does not fulfil thecost-benefit balance. The recurrent idea among different stakeholders (i.e.,fishers, government and intermediaries of resource value chain) is that illegalfishing is a consequence of the lack of opportunities and economic needs [43,44],reflected by the variable “alternative economic activities” in our model. Thus,managers must ensure fishers do not have “a reason to poach”, by puttingefforts on the socio-economic well-being of the cove. But, bearing in mind thelater as a necessary long-term task, immediate actions on preventing poachingmust be considered. Unmonitored and unsanctioned poaching in TURFs mayimpede fishers of sustaining their resources and benefit from harvesting surplus.Contrastingly, higher sanctions to poaching may both help fishers to defendtheir TURF resources and improve un-associated fishers’ willingness to bind toa TURF system [40]. Nevertheless, a strengthening in the sanctions to poachingmust be framed in the artisanal SSF perspective [12], where non-compliance isstrongly conditioned by local socio-economic contexts, as mentioned above [14].Currently, fishers must cover the (increasingly high) cost of surveillance of theirTURFs but rely on the administration to punish poachers [24]. Although thelegislation considers formal sanctions to poaching and illegal practices, these areinformally allowed by the enforcement agencies. This suggests a vicious circlearound the problem of poaching in TURFs: economic needs prompt poaching anddeter fishers to invest in surveillance, in response the government administrationstrengthens formal sanctions but does not improve enforcement mechanisms,letting the SSF stuck in an illegality trap [8].From our results, we can suggest that illegal fishing on key-hole limpet in centralChile arises from a complex network of drivers, from methodological to logisticproblems, which appear to be linked to monitoring of compliance, lack of supportof government agencies in the co-management process, socio-economic contextof fishers, and an unsuitable spatial planning of the TURFs. These results are inline with observations focusing in other benthic resources like the highly valuableloco or Chilean abalone [27] and the king crab [43]. A change in paradigm to alocal perspective in the implementation of mechanisms to overcome enforcementand planning weaknesses is needed. This change in paradigm should be aided bysuitable diagnosis of local socio-economic contexts of fishers, and the improvementof local and regional governance tools (e.g., Management Committees, which arealready considered in the national regulation).In the current TURF system in Chile, fishers select an area, and it is assignedunder request, but there is no advice from decision makers and scientists onwhere to allocate the area (only on quotas). There is an inherent problem on theplanification of the system that might lead, for example, to high fishing pressurein adjacent open access areas [22,26]. Ospina-Alvarez et al. [45] showed thebenefits of a planned network of MAs in the central coast of chile that wouldhave positive effects on both fisheries and biodiversity conservation. However,this spatial modelling exercise also evidenced the importance of enforcementfor the effective functioning of the restricted areas network and the need toPreprint submitted to:
Marine Policy
Page 18ocio-ecological drivers of illegal fishing de Juan et al. 2020consider all environmental and human factors in the optimization criteria. Highfishing pressure on the OA areas, with overexploited benthic resources [37],has a negative feedback on the MAs that are exposed to illegal extraction ofa benthic resource that is generally in a better condition than neighbouringOA areas [46]. A systematic and science-based spatial planning of the fisheryrestricted areas would avoid high concentrations of MAs in sections of the coast,generally those more densely populated, while identifying the most productiveareas for the placement of fishery restricted areas [45]. This spatial approachto the problem would alleviate fishing pressure on the restricted MAs; however,additional actions, like increasing surveillance efforts, are necessary for theeffective performance of a network of well-enforced MA.While TURFs systems aim to incentive fishers’ care of their fishing grounds,these need to be well planned and supported financially and logistically by theadministration. Property rights schemes could benefit from a re-formulationthat considers the social rights and benefits (beyond the individual rights) andfishers’ traditional ecological knowledge [6], so the conservation of the ecosystemis a priority over the individual interests to exploit the resources [47]. However,in the case study, the top-down component of the co-management system seemsto fail as there is poor planning and financing. The decision-making powergranted to fishers in the current co-management process is very much limitedto a MA monitoring and surveillance, with insufficient resources and supportto carry out this task. In Chile, and in SSF in general, the balance of thefishermen-administration role must be planned to ensure a fully democratisedsystem that is able to cope with illegal fishing and overexploitation. Addressingthese complex problems inherent to the functioning of SSF are central, as SSFcan play a key role in the sustainable development of global fisheries.
Acknowledgements
This work was partly funded by a Fondecyt Project 1171603 grant to MDSand MF and the Iniciativa Científica Milenio (Project CCM RC 1300004) fromMinisterio de Economía, Fomento y Turismo de Chile. The authors want tothank all participants of the surveys for their essential contribution to this study.
Author contribution
Silvia de Juan: conceptualization, investigation, methodology, formal analysis,writing-original draft. Dulce Subida: conceptualization, investigation, formalanalysis, writing-original draft. Andres Ospina-Alvarez: methodology, software,formal analysis. Ainara Aguilar: investigation. Miriam Fernandez: conceptual-ization, writing-review and editing, funding acquisition.
References [1] J. Jacquet, D. Pauly, Funding Priorities: Big Barriers to Small-ScaleFisheries:
Funding for Fisheries , Conservation Biology. 22 (2008) 832–835.https://doi.org/10.1111/j.1523-1739.2008.00978.x.Preprint submitted to:
Marine Policy
Page 19ocio-ecological drivers of illegal fishing de Juan et al. 2020[2] FAO, Securing sustainable small-scale fisheries: sharing good practices fromaround the world, FAO, 2019.[3] C. Costello, D. Ovando, R. Hilborn, S.D. Gaines, O. Deschenes, S.E. Lester,Status and Solutions for the World’s Unassessed Fisheries, Science. 338 (2012)517–520. https://doi.org/10.1126/science.1223389.[4] C. Pita, S. Villasante, J.J. Pascual-Fernández, Managing small-scale fisheriesunder data poor scenarios: lessons from around the world, Marine Policy. 101(2019) 154–157. https://doi.org/10.1016/j.marpol.2019.02.008.[5] M.-J. Roux, R.F. Tallman, Z.A. Martin, Small-scale fisheries in Canada’s Arc-tic: Combining science and fishers knowledge towards sustainable management,Marine Policy. 101 (2019) 177–186. https://doi.org/10.1016/j.marpol.2018.01.016.[6] C. Ruano-Chamorro, M.D. Subida, M. Fernandez, Fishers’ perception: Analternative source of information to assess the data-poor benthic small-scaleartisanal fisheries of central Chile, Ocean and Coastal Management. 146 (2017)67–76. https://doi.org/10.1016/j.ocecoaman.2017.06.007.[7] J. Aburto, W. Stotz, Learning about TURFs and natural variability: Failureof surf clam management in Chile, Ocean & Coastal Management. 71 (2013)88–98. https://doi.org/10.1016/j.ocecoaman.2012.10.013.[8] L. Nahuelhual, G. Saavedra, M.A. Mellado, X.V. Vergara, T. Vallejos, Asocial-ecological trap perspective to explain the emergence and persistence ofillegal fishing in small-scale fisheries, Maritime Studies. 19 (2020) 105–117.https://doi.org/10.1007/s40152-019-00154-1.[9] B. Worm, R. Hilborn, J.K. Baum, T.A. Branch, J.S. Collie, C. Costello, M.J.Fogarty, E.A. Fulton, J.A. Hutchings, S.R. Jennings, O.P. Jensen, H.K. Lotze,P.M. Mace, T.R. McClanahan, C. Minto, S.R. Palumbi, A.M. Parma, D. Ricard,A.A. Rosenberg, R. Watson, D. Zeller, Rebuilding global fisheries, Science. 325(2009) 578–585.[10] C. Mora, R.A. Myers, M. Coll, S. Libralato, T.J. Pitcher, R. Sumaila,D. Zeller, R. Watson, K.J. Gaston, B. Worm, Management effective-ness of the world’s marine fisheries, PLoS Biology. 7 (2009) e1000131.https://doi.org/10.1371/journal.pbio.1000131.[11] D.J. Agnew, J. Pearce, G. Pramod, T. Peatman, R. Watson, J.R. Beddington,T.J. Pitcher, Estimating the Worldwide Extent of Illegal Fishing, PLoS ONE. 4(2009) e4570. https://doi.org/10.1371/journal.pone.0004570.[12] C.J. Donlan, C. Wilcox, G.M. Luque, S. Gelcich, Estimating il-legal fishing from enforcement officers, Sci Rep. 10 (2020) 12478.https://doi.org/10.1038/s41598-020-69311-5.[13] P. Veiga, C. Pita, M. Rangel, J.M.S. Gonçalves, A. Campos, P.G. Fernandes,A. Sala, M. Virgili, A. Lucchetti, J. Brčić, S. Villasante, M.A. Ballesteros, R.Preprint submitted to:
Marine Policy
Page 20ocio-ecological drivers of illegal fishing de Juan et al. 2020Chapela, J.L. Santiago, S. Agnarsson, Ó. Ögmundarson, K. Erzini, The EU land-ing obligation and European small-scale fisheries: What are the odds for success?,Marine Policy. 64 (2016) 64–71. https://doi.org/10.1016/j.marpol.2015.11.008.[14] R. Oyanedel, S. Gelcich, E.J. Milner-Gulland, Motivations for (non-)compliance with conservation rules by small-scale resource users, CONSERVA-TION LETTERS. (2020). https://doi.org/10.1111/conl.12725.[15] E. Ostrom, V. Ostrom, Public economy organization and service delivery,Dearborn University of Michigan, 1977. http://dlc.dlib.indiana.edu/dlc/handle/10535/732.[16] W. Battista, R. Romero-Canyas, S.L. Smith, J. Fraire, M. Effron, D. Larson-Konar, R. Fujita, Behavior Change Interventions to Reduce Illegal Fishing, Front.Mar. Sci. 5 (2018) 403. https://doi.org/10.3389/fmars.2018.00403.[17] C. Nguyen Thi Quynh, S. Schilizzi, A. Hailu, S. Iftekhar, Territorial UseRights for Fisheries (TURFs): State of the art and the road ahead, Marine Policy.75 (2017) 41–52. https://doi.org/10.1016/j.marpol.2016.10.004.[18] M. Fernández, J.C. Castilla, Marine Conservation in Chile: HistoricalPerspective, Lessons, and Challenges, Conservation Biology. 19 (2005) 1752–1762. https://doi.org/10.1111/j.1523-1739.2005.00277.x.[19] J. Aburto, G. Gallardo, W. Stotz, C. Cerda, C. Mondaca-Schachermayer,K. Vera, Territorial user rights for artisanal fisheries in Chile – intended andunintended outcomes, Ocean & Coastal Management. 71 (2013) 284–295.https://doi.org/10.1016/j.ocecoaman.2012.09.015.[20] S. Gelcich, M. Fernandez, N. Godoy, A. Canepa, L. Prado, J.C. Castilla,Territorial user rights for fisheries as ancillary instruments for marinecoastal conservation in Chile., Conservation Biology. 26 (2012) 1005–1015.https://doi.org/10.1111/j.1523-1739.2012.01928.x.[21] A. Pérez-Matus, A. Ospina-Alvarez, P.A. Camus, S.A. Carrasco, M. Fer-nandez, S. Gelcich, N. Godoy, F.P. Ojeda, L.M. Pardo, N. Rozbaczylo, M.D.Subida, M. Thiel, E.A. Wieters, S.A. Navarrete, Temperate rocky subtidal reefcommunity reveals human impacts across the entire food web, Marine EcologyProgress Series. 567 (2017) 1–16. https://doi.org/10.3354/meps12057.[22] J. Beckensteiner, A.M. Scheld, M. Fernández, D.M. Kaplan, Driversand trends in catch of benthic resources in Chilean TURFs and surroundingopen access areas, Ocean & Coastal Management. 183 (2020) 104961.https://doi.org/10.1016/j.ocecoaman.2019.104961.[23] K.J. Davis, M.E. Kragt, S. Gelcich, M. Burton, S. Schilizzi, D.J. Pannell,Why are Fishers not Enforcing Their Marine User Rights?, Environ ResourceEcon. 67 (2017) 661–681. https://doi.org/10.1007/s10640-015-9992-z.[24] S. Gelcich, J. Cinner, C. Donlan, S. Tapia-Lewin, N. Godoy, J.Castilla, Fishers’ perceptions on the Chilean coastal TURF system after twoPreprint submitted to:
Marine Policy
Page 21ocio-ecological drivers of illegal fishing de Juan et al. 2020decades: problems, benefits, and emerging needs, Bms. 93 (2017) 53–67.https://doi.org/10.5343/bms.2015.1082.[25] R.M. Bandin Llanos, Impacto de la captura ilegal en la pesquería del recurso’loco’ (Concholepas concholepas) En áreas de manejo y explotación de recursosbentónicos: El caso de la Isla Mocha, Thesis, Universidad de Concepcion, 2013.http://repositorio.udec.cl/jspui/handle/11594/763.[26] M. Fernández, M. Kriegl, V. Garmendia, A. Aguilar, M.D. Subida, Evidenceof illegal catch in the benthic artisanal fisheries of central Chile: patterns acrossspecies and management regimes, Lat. Am. J. Aquat. Res. 48 (2020) 287–303.https://doi.org/10.3856/vol48-issue2-fulltext-2475.[27] R. Oyanedel, A. Keim, J.C. Castilla, S. Gelcich, Illegal fishing and territorialuser rights in Chile: Illegal Fishing, Conservation Biology. 32 (2018) 619–627.https://doi.org/10.1111/cobi.13048.[28] H.M. Ballesteros, G. Rodríguez-Rodríguez, “Acceptable” and “unacceptable”poachers: Lessons in managing poaching from the Galician shellfish sector, MarinePolicy. 87 (2018) 104–110. https://doi.org/10.1016/j.marpol.2017.10.015.[29] L. Little, S. Kuikka, A. Punt, F. Pantus, C. Davies, B. Mapstone, Infor-mation flow among fishing vessels modelled using a Bayesian network, Environ-mental Modelling & Software. 19 (2004) 27–34. https://doi.org/10.1016/S1364-8152(03)00100-2.[30] P. Haapasaari, C.G.J. Michielsens, T.P. Karjalainen, K. Reinikainen, S.Kuikka, Management measures and fishers’ commitment to sustainable exploita-tion: a case study of Atlantic salmon fisheries in the Baltic Sea, ICES Journal ofMarine Science. 64 (2007) 825–833. https://doi.org/10.1093/icesjms/fsm002.[31] P. Levontin, S. Kulmala, P. Haapasaari, S. Kuikka, Integration of biological,economic, and sociological knowledge by Bayesian belief networks: the interdisci-plinary evaluation of potential management plans for Baltic salmon, ICES Journalof Marine Science. 68 (2011) 632–638. https://doi.org/10.1093/icesjms/fsr004.[32] I. van Putten, A. Lalancette, P. Bayliss, D. Dennis, T. Hutton, A. Norman-López, S. Pascoe, E. Plagányi, T. Skewes, A Bayesian model of factors influencingindigenous participation in the Torres Strait tropical rocklobster fishery, MarinePolicy. 37 (2013) 96–105. https://doi.org/10.1016/j.marpol.2012.04.001.[33] C.A. Pollino, O. Woodberry, A. Nicholson, K. Korb, B.T. Hart, Pa-rameterisation and evaluation of a Bayesian network for use in an ecologicalrisk assessment, Environmental Modelling & Software. 22 (2007) 1140–1152.https://doi.org/10.1016/j.envsoft.2006.03.006.[34] E. Hoshino, I. van Putten, W. Girsang, B.P. Resosudarmo, S. Yamazaki, ABayesian belief network model for community-based coastal resource managementin the Kei Islands, Indonesia, E&S. 21 (2016) art16. https://doi.org/10.5751/ES-08285-210216.Preprint submitted to:
Marine Policy
Page 22ocio-ecological drivers of illegal fishing de Juan et al. 2020[35] M.J. Slater, Y.D. Mgaya, A.C. Mill, S.P. Rushton, S.M. Stead,Effect of social and economic drivers on choosing aquaculture as acoastal livelihood, Ocean & Coastal Management. 73 (2013) 22–30.https://doi.org/10.1016/j.ocecoaman.2012.12.002.[36] S. Gelcich, R. Guzman, C. Rodríguez-Sickert, J.C. Castilla, J.C. Cárdenas,Exploring external validity of common pool resource experiments: insightsfrom artisanal benthic fisheries in Chile, Ecology and Society. 18 (2013) art2.https://doi.org/10.5751/ES-05598-180302.[37] M. Andreu-Cazenave, M.D. Subida, M. Fernandez, Exploitation rates oftwo benthic resources across management regimes in central Chile: Evidence ofillegal fishing in artisanal fisheries operating in open access areas, PloS ONE. 12(2017) e0180012. https://doi.org/10.1371/journal.pone.0180012.[38] M. Scutari, C. Vitolo, A. Tucker, Learning Bayesian networks from big datawith greedy search: computational complexity and efficient implementation, StatComput. 29 (2019) 1095–1108. https://doi.org/10.1007/s11222-019-09857-1.[39] M. Scutari, Learning Bayesian Networks with the bnlearn R Package,arXiv:0908.3817 [Stat]. (2010). http://arxiv.org/abs/0908.3817 (accessed De-cember 10, 2020).[40] C.A. Chávez, J.J. Murphy, J.K. Stranlund, Managing and defend-ing the commons: Experimental evidence from TURFs in Chile, Jour-nal of Environmental Economics and Management. 91 (2018) 229–246.https://doi.org/10.1016/j.jeem.2018.07.004.[41] SERNAPESCA, Fiscalización en Pesca y Acuicultura. Informe de actividadesdel 2019, Ministerio de Economía Fomento y Turismo, Gobierno de Chile, 2019.[42] X. Basurto, S. Gelcich, E. Ostrom, The social–ecological systemframework as a knowledge classificatory system for benthic small-scale fisheries, Global Environmental Change. 23 (2013) 1366–1380.https://doi.org/10.1016/j.gloenvcha.2013.08.001.[43] L. Nahuelhual, G. Saavedra, G. Blanco, E. Wesselink, G. Campos,X. Vergara, On super fishers and black capture: Images of illegal fishingin artisanal fisheries of southern Chile, Marine Policy. 95 (2018) 36–45.https://doi.org/10.1016/j.marpol.2018.06.020.[44] AU-IBAR, Economic, Social and Environmental impact of Illegal, Unreportedand Unregulated Fishing in Africa, 2016.[45] A. Ospina-Alvarez, S. de Juan, K.J. Davis, C. González, M. Fernández, S.A.Navarrete, Integration of biophysical connectivity in the spatial optimization ofcoastal ecosystem services, Science of the Total Environment. 733 (2020) 139367.https://doi.org/10.1016/j.scitotenv.2020.139367.[46] S. de Juan, S. Gelcich, A. Ospina-Alvarez, A. Pérez-Matus, M. Fernandez,Applying an ecosystem service approach to unravel links between ecosystemsPreprint submitted to:
Marine Policy
Page 23ocio-ecological drivers of illegal fishing de Juan et al. 2020and society in the coast of central Chile, Science of The Total Environment. 533(2015) 122–132. https://doi.org/10.1016/j.scitotenv.2015.06.094.[47] K. Moon, D. Marsh, C. Cvitanovic, Coupling property rights with responsi-bilities to improve conservation outcomes across land and seascapes, CONSER-VATION LETTERS. (2020). https://doi.org/10.1111/conl.12767.Preprint submitted to:
Marine Policy
Page 24
Figure S1 - Length-frequency distributions (in %) of keyhole limpets for the OA and different MAs and of each cove, sorted from north to south. Solid grey line indicates the MLS (=6.5 cm) for this benthic resource. Dashed black lines indicate the median size of the catches for each sampling site able S1 – Data gathered in the Management Areas and considered as potential drivers for illegal fishing in SSF in central Chile.
Caleta MA name Nº MA per caleta Access from land Wave exposure Surface MA (ha) Other activities Availability OA Enforcement Effectivenness Distance from surveillance Nº of poaching events Chungungo Chungungo E 3 Moderate Protected 21,45 No 0,02572 low low far <20 (low) Chungungo D 3 Difficult Exposed 18,82 No 0,02572 low very low far <20 (low) Temblador 3 Difficult Moderate 20,42 No 0,02572 low very low far <20 (low) San Pedro – Los Vilos Ñague 2 Moderate Moderate 280,77 Yes 0,03251 very high high close 50-100 (high) Ñague B 2 Difficult Exposed 56,49 Yes 0,03251 very high high close 50-100 (high) Las Conchas La Cachina 2 Moderate Moderate 116,2 No 0,03287 high moderate close 20-50 (moderate) Los Vilos Sector C 2 Easy Moderate 60,21 No 0,03287 high moderate close 20-50 (moderate) Hornos Hornos 2 Moderate Moderate 370,76 No 0,03496 very high moderate average >100 (very high) Hornos B 2 Difficult Exposed 84,32 No 0,03496 very high moderate average 50-100 (high) Chigualoco Caleta Boca del Barco 3 Moderate Exposed 112,17 No 0,03789 moderate low close 50-100 (high) Chepiquilla 3 Moderate Exposed 81,8 No 0,03789 moderate very low close >100 (very high) Chigualoco 3 Easy Exposed 343,11 No 0,03789 moderate low close >100 (very high) Huentelauquén Huentelauquén 1 Difficult Exposed 309,74 No 0,04815 high high average <20 (low) Los Molles Los Molles 2 Easy Exposed 95 Yes 0,10035 very high high average 20-50 (moderate) Playa Los Molles 2 Moderate Exposed 63,18 Yes 0,10035 very high high average 20-50 (moderate) Algarrobo Algarrobo A 3 Easy Protected 34,8 No 0,11637 very high moderate close 20-50 (moderate) lgarrobo C 3 Easy Exposed 62,87 No 0,11637 very high high close 20-50 (moderate) Pichidangui Pichidangui 2 Easy Exposed 93,6 Yes 0,120033 low very low close >100 (very high) Horcón Horcón 1 Easy Exposed 98,71 Yes 0,13074 low very low close 50-100 (high) Pichicuy Pichicuy 1 Moderate Exposed 189,16 Yes 0,14445 very high moderate close 20-50 (moderate) Papudo Papudo 2 Moderate Moderate 187,26 Yes 0,1891 very high moderate close 50-100 (high) Punta Pite 2 Difficult Exposed 112,72 Yes 0,1891 very high low close 50-100 (high) Quintay Quintay A 2 Moderate Protected 53,42 Yes 0,34513 very high very high average >100 (very high) Quintay B 2 Moderate Exposed 105 Yes 0,34513 very high very high average >100 (very high) able S2. Results from the Wilcoxon rank tests (W) comparing medians of the length of the catch of keyhole limpets (
Fissurella spp.) between management regimes (MA indicates management areas and OA open access areas) at each cove (from south to north). N indicates sample size and includes the percentage of individuals below the minimum legal size (rounded). The index of availability of open areas is shown between parentheses for each cove. P- values are shown between parenthesis and values <0.05 in bold. Sites where the % of illegal catch (below MLS) was ≥
49% are marked with a *.
Cove Site / Regime Lapa (
Fissurella spp .) N_total (N_undersize; % undersize) W statistic for the catch (P value in parenthesis) Algarrobo MA 1- Sector A 303 (20; 7%)
MA 2- Sector C 204 (5; 2%)
OA 284 (22; 8%) - Quintay MA 1- Sector A 215 (0; 0%)
MA 2- Sector B 245 (3; 1%)
OA 220 (9; 4%) - Horcón MA 1- Horcón 220 (27; 13%)
OA * 222 (131; 59%) - Papudo MA 1 – Papudo 255 (2; 1%)
MA 2 – Pta. Pite 217 (8; 4%)
OA 200 (28; 14%) - Pichicuy MA 1 – Pichicuy 250 (28; 11%)
OA 240 (59; 25%) - Los Molles MA 1 – Los Molles 272 (51; 19%)
MA 2 – Playa Los Molles 322 (43; 13%)
OA * 212 (104; 49%) - Pichidangui MA 1 – Pichidangui 224 (0; 0%)
OA 250 (0; 0%) - Las Conchas MA 1 – La Cachina 241 (0; 0%)
MA 2 – Sector C 207 (71; 34%)
OA * 256 (137; 53%) - San Pedro MA 1- Ñague 284 (27; 9%)
MA 2 – Ñague B 204 (12; 6%)
OA 205 (41; 20%) - higualoco MA 1 – Chigualoco 255 (21; 8%)
MA 2 – Boca del Barco 261 (13; 5%)
MA 3 – Chepiquilla 181 (23; 13%) 20818 (0.4811) OA 221 (25; 11%) - Huentelauquen MA 1 – Huentelauquen 225 (17; 8%) 22904 (0.288) OA 192 (23; 12%) - Hornos MA 1 – Hornos 211 (0; 0%)
MA 2 – Hornos B 263 (21; 8%)
OA * 201 (136; 68%) - Chungungo MA 1 – Chungungo E * 271 (167; 62%)
MA 2 – Sector D * 210 (149; 71%)
MA 3 – Temblador 257 (66; 26%)
OA 210 (87; 41%) -
UESTIONNAIRE: TURFs enforecement: implementation level and associated costs
Date: Cove:
Fisher’s organization:
President’s name:
TURF’s name: Do you have hired a surveillance service? a.
Yes b.
No. Who does the surveillance? 2.
Duration and frequency of surveillance a.
Only during the day b.
Day and night c.
Only in weekdays d.
Only with calm seas e.
Other 3.
This situation was different 10 years ago? 4.
There is any protocol to carry out the surveillance? (e.g. patrol route, person who contact with NAVY) a.
Yes. Who decided the protocol? b.
No 5.
Who pays the surveillance? Is the fisher’s organization or another external entity? a. Fisher’s organization b. External entity (Who?) c.
Both 6.
If the fisher’s organizations must pay the surveillance. How much it cost? a. More than 75% of the organization budget b.
More than 50% of the organization budget c.
More than 25% of the organization budget d.
Less than 10% of the organization budget 7.
If the fisher’s organizations have more than one TURF. The surveillance is the same for all? a. Yes b.
No i.
What TURF has more surveillance? ii.
Why those differences? .
In the last year. How many cases of poaching did the organization register in each of the TURFS? 9.
In a scale 1 to 7. What is the effectivity of the surveillance? 10.
In a scale 1 to 7. What is the importance of the surveillance in a TURF for its good working? 11.
What need your organization to have a good surveillance? a.
More money b.
More cooperation between members c.
Less necessity for people that need to poach d.