1. The Northern Gulf of California is an ecologically important marine area with a high degree of biodiversity, endemism and productivity. Mounting conservation concerns have prompted researchers to propose new management regulations, restricting fishing and protecting sensitive species. Compliance with existing regulations is poor. Rules that are currently in place, if followed, may go a long way towards achieving the ecological goals of management.
2. We conduct a review of existing fisheries regulations in this area. Then, using a spatially explicit marine ecosystem model (Atlantis), we estimate the benefits of compliance with existing fisheries regulations.
3. Under a full compliance scenario, we find large increases in protected species biomass within 25 years and a slowed rate of ecosystem degradation because of fishing. However, full compliance costs the fishing industry about 30% of its annual revenue.
4. We parse out the benefits offered by management instruments (including spatial management protections, seasonal fishery closures, gear restrictions, cessation of illegal fishing and vessel buy-out programmes) and conclude that a suite of measures is needed to address major conservation objectives.
5.Synthesis and applications. This exercise quantifies the benefits of improved fisheries enforcement and provides a benchmark by which the value of future regulatory amendments can be assessed. Where compliance with existing regulations is poor, conservation goals may be better served by strengthening enforcement than by enacting new rules and legislation.
There is little question that declining integrity of marine ecosystems is eroding the ability of these systems to provide the goods and services that humans need (Worm et al. 2006). It is less clear what to do to reverse this trend. In recent years, calls for the implementation of new approaches to marine resource management such as networks of marine protected areas (MPAs), marine zoning schemes, catch share programmes or systematic overhaul of governance structures have increased (Costello, Gaines & Lynham 2008; Halpern, Lester & McLeod 2010). In many cases, such calls follow an indictment of existing management, but often this condemnation of management lacks specificity. Management may fail for two very different reasons.
First, the management approach may be inadequate. For example, while both MPAs and traditional fisheries management can preserve fish biomass, MPAs may be superior at protecting the age structure of exploited populations (I. Kaplan, unpublished data). Thus, when a management goal is to preserve the age structure of stocks (e.g. Francis et al. 2007), a management scheme without MPAs is likely to be insufficient. Second, implementation of the approach may fail. Terrestrial management provides examples of where poor compliance with conservation rules can be easily identified as the primary factor limiting policy success (Hilborn et al. 2006). More often, the degree to which non-compliance impedes management is poorly understood because of difficulties in studying rule-breaking behaviour directly, and because of the complex and case-specific nature of the problem (Keane et al. 2008). Because of the social and fiscal costs of implementing new management schemes (e.g. Helvey 2004), it is helpful to distinguish between cases in which management has failed because the approach is fundamentally flawed and those in which management could succeed, but implementation has failed. We explore this issue in the Northern Gulf of California, Mexico.
This region is ecologically valuable for its biodiversity, its endemic species and for the critical breeding and feeding habitat it provides for birds, turtles and marine mammals (Enriquez-Andrade et al. 2005; Lluch-Cota et al. 2007). Fisheries are the chief source of economic productivity in the Northern Gulf, and they are critical for food security. Unfortunately, overfishing and environmental degradation during the last century have eliminated important fisheries and jeopardized others (e.g. Lercari & Chávez 2007). Growing conservation concerns regarding the vaquita porpoise Phocoena sinus, sea lions and sea turtles now further limit the economic options of coastal communities (e.g. Seminoff et al. 2003; Jaramillo-Legorreta et al. 2007).
Since the 1990s, Mexico has attempted to address these conservation concerns through the issuance of fisheries regulations (Hernandez & Kempton 2003), but most fisheries still operate as de facto open access systems (Cinti et al. 2010). New proposals attempt to plug some gaps in the regulatory framework. For example, the use of a catch shares programme has been suggested for the Northern Gulf, akin to the one implemented in the Sinaloan shrimp trawl fishery (J. Murrieta-Saldivar, personal communication). Other proposed measures would expand the existing network of MPAs to include additional juvenile fish habitat (Cudney-Bueno et al. 2009a) or to protect reefs (Sala et al. 2002). However, promising these proposals may be poor compliance threatens to scuttle benefits.
In this article, we review existing rules and regulations governing marine fisheries and conservation in the Northern Gulf of California. We divide fisheries regulations into these categories: (i) spatial management protections; (ii) seasonal fishery closures; (iii) gear restrictions. Following regulations strictly, we would also see (iv) elimination of all illegal fishing; (v) by 2012, further reductions in fishing effort and changes in fleet structure as the Mexican fisheries authority carries out fishery license buy-out and gear switch-out programmes (SEMARNAT (Secretaría del Medio Ambiente y Recursos Naturales) 2008; CONAPESCA 2010). Each of these five management instruments represents potential limitations in where, when and how fishing can occur. All of these measures are implemented in current or impending regulations.
Using an ecosystem simulation model of the Northern Gulf of California (Fulton et al. 2004; Ainsworth et al. 2011), we modelled these five management measures independently to examine their effects on the ecosystem. We then compared the results to a status quo scenario, in which the current degree of compliance and fishing effort is represented. Combining all five management measures leads to a ‘full-enforcement’ scenario, in which all fishing regulations are adhered to, all illegal fishing is stopped and scheduled fleet reductions/changes are carried out. The ecological and economic benefits or costs achieved by this full-enforcement scenario provide a benchmark by which we can evaluate the costs of non-compliance and judge the value of proposed regulatory amendments. Policy success is framed in terms of species recovery, maintenance of biodiversity and fisheries value. With specific impacts of each policy described, we can explore the trade-off space inherent between economic and ecological costs and benefits.
Materials and methods
The Atlantis Ecosystem Model
Atlantis is a deterministic biogeochemical and biophysical modelling system that simulates the functioning of marine food webs and fisheries to serve as a policy exploration tool for ecosystem-based management (Fulton et al. 2004; Fulton, Smith & Smith 2007; Fulton 2010; Fulton et al. 2011). It is an ‘end-to-end’ model, in that it represents ecosystem components from marine bacteria to apex predators and human beings. We opted to use this framework because it allows testing of highly articulated management scenarios involving spatiotemporal fishing restrictions and it offers a wide range of economic and ecological outputs.
Fulton et al. (2004) provide a summary of the chief dynamics and process equations. Sub-models include consumption, biological production, waste production, reproduction, habitat dependency, age structure, mortality, decomposition and microbial cycles. The spatial domain is resolved in three dimensions using irregular polygons to represent biogeographic features. Exchange of biomass occurs between polygons according to seasonal migration and foraging behaviour, while water movement, heat and salinity flux across boundaries can be represented by a coupled hydrodynamic model. Plagányi (2007) and Rose et al. (2010) provide reviews of this and other marine ecosystem modelling approaches.
Northern Gulf of California Model
For simulations here, we utilize the model of Ainsworth et al. (2011). The spatial domain encompasses approximately 57 800 km2 of sea area and includes the Colorado River Delta in the Northern Gulf to areas as far south as the northern tip of Baja California Sur (Fig. 1). Functional groups used in the model are presented in Table S1 (Supporting Information).
Building an ecosystem model for this region required integrating diverse types of information. These included data emerging from the PANGAS project (http://www.pangas.arizona.edu), an effort by universities and non-governmental organizations to characterize small-scale fisheries and assess artisanal resources. Original field work included diet sampling (Ainsworth et al. 2010), trawl and underwater video surveys (Ainsworth et al. 2011), and fisher interviews to estimate species relative abundance trends (Ainsworth 2011). We forced salinity, water temperature and currents in Atlantis with output from a regional ocean modelling system (Lluch-Cota et al. 2010) for the years 1985–2008, looped into the future. Ainsworth et al. 2011 show how these data were used in the calibration and validation of the model.
All model simulations presented in this article run from 2008 to 2033 (25 years). The 2008 model represents current ecosystem structure and functioning. For these simulations, we assumed that targeting preference of fishers remains constant. Therefore, changes in fishing mortality rates result from effort differences between management scenarios. The scenarios tested are as follows: (i) status quo (partial compliance with regulations), (ii) full adherence to spatial management protections, (iii) full adherence to seasonal fishery closures, (iv) full adherence to physical gear restrictions, (v) no illegal artisanal fishing, (vi) planned fishery buy-outs and gear switch-outs are implemented and (vii) full enforcement (combined effects of scenarios ii–vi).
The status quo scenario represents our best understanding of present fishing patterns in the Northern Gulf. We represent the activity of 33 fishing fleets (Table S2, Supporting Information). Based on literature and discussions with local experts involved in the PANGAS project (P. Turk-Boyer, R. Cudney, W. Shaw, T. Pfister, P. Raimondi, L. Bourillon, J. Torre, M. Moreno, personal communication), we have assembled the set of fishery regulations that are followed by fishers in the Northern Gulf, partially or completely (Table S3, Supporting Information). These rules and level of compliance are implemented in the 2008 ecosystem model.
We assume that fisheries catch in the status quo scenario corresponds to the average catch from 2000 to 2007 (Table S4, Supporting Information). This series was assembled using official Mexican fishery statistics, port-level surveys with fishing cooperatives and various outputs of the PANGAS project. Complete references are provided in Appendix Table E2 of Ainsworth et al. (2011). Catch by functional group (in Table S4, Supporting Information) was allocated to each of the 33 fleets according to Table S5 (Supporting Information); references are in Table S6 (Supporting Information). Fishing areas are defined for artisanal fleets using Moreno-Báez et al. (2010), and for purse seine and squid fisheries using DOF (2006). Otherwise, fleet activity follows hydrographic contours: we assumed longlines operate to 150-m depth, shrimp trawls operate to 500 m and areas deeper than 500 m were exclusively used by pelagic fisheries.
Spatial Management Policies
Spatial management includes MPAs, concessions and regulations contained in Diario Oficial de la Federación and other governmental publications (Table S7, Supporting Information), as illustrated in Fig. S1 (Supporting Information). Major spatial restrictions are found within the Upper Gulf of California and Colorado River Delta Biosphere Reserve (hereafter called ‘Upper Gulf’), including a seasonal closure for artisanal fleets and a slated halving in the number of permitted shrimp trawlers (Table S7, Supporting Information), from 162 to 82 (SGPA (Subsecretaría de Gestión para la Protección Ambiental) 2009). This was simulated as a reduction in fishing mortality to 51% (82/162) of the 2008 levels. Spatial restrictions also consider the exclusion of gillnets from the vaquita distribution area, as is planned for 2012 (SEMARNAT (Secretaría del Medio Ambiente y Recursos Naturales) 2008). We considered this implemented as of 2008 for ease of modelling. This simplification has little effect on the long-term (25 year) simulations presented here.
Seasonal closures are active for Cortez oyster (Crassostrea corteziensis), spiny lobsters (Panulirus interruptus, Panulirus inflatus, Panulirus gracilis), Curvina golfina (Cynoscion othonopterus) and mullets (Mugil cephalus and Mugil curema) (Table S9, Supporting Information). The spatial closures were simulated as partial or complete spatial closures to specific model cells, with fishing effort reduced proportionately to area closed.
The gear restrictions scenario combined both catch age structure regulations and effort restriction regulations (Table S8, Supporting Information). Regulations that increase the minimum size of fish caught were simulated by increasing the age at first capture by one age class. Effort restrictions were simulated as either a total elimination of fishing mortality for protected species or minor or major reductions in fishing mortality (i.e. a 20% or 50% reduction relative to status quo on protected species). For sea turtle groups, we simulated the enforcement of turtle excluding devices as a 66% reduction in fishing mortality (Epperly 2003).
This scenario represents the possible effects of an upcoming trawl buy-out programme and an ongoing gear switch-out programme. The shrimp trawl buy-out aims to reduce trawl effort by 30% throughout Mexico by 2012 (CONAPESCA 2010). We therefore reduced fishing mortality for shrimp and fish trawlers to 70% of status quo for target and bycatch species in this scenario. The gillnet buy-out and switch-out programme aims to eliminate gill nets and driftnets from critical vaquita habitat or permanently substitute them for gears that avoid vaquita bycatch as part of the species recovery programme (SEMARNAT (Secretaría del Medio Ambiente y Recursos Naturales) 2008). Within this programme’s region, we assumed that 51% of gillnet vessels would opt for a buy-out, while 49% would shift to a light shrimp trawl (SEMARNAT (Secretaría del Medio Ambiente y Recursos Naturales) 2008; PRONATURA, CEC-CCA-CCE, CEDO Intercultural, & NOS 2010). We assumed that light shrimp trawls catch 80% less bycatch than the current trawl fleet and eliminate vaquita bycatch, but are less efficient at capturing the target species (catch is 56% of status quo levels; see Appendix S1 (Supporting Information) for details).
The full-enforcement scenario assumes perfect compliance with each of these five classes of regulations: spatial management policies, seasonal closures, gear restrictions, elimination of illegal fishing and reduction in fishing effort through buy-out and switch-out.
Policy Success Metrics
We discriminate the outcomes of these scenarios by considering species-level, fishery and ecosystem-based metrics. Species-level metrics include changes in biomass, body size and the ratio of reproductively mature individuals to juveniles. Fishery metrics include changes in catch (total amount and distribution among fleets), landed value and average trophic level (a proxy for ecosystem maturity, Cury et al. 2004). Ecosystem-based metrics include spatial distribution of biomass and species biodiversity measured by Kempton’s Q index (adapted for ecosystem models, Q90: Ainsworth & Pitcher 2006). This metric represents the interdecile slope of the cumulative species log-abundance curve, where functional group biomass, sorted into bins, substitutes for abundance. The metric increases with species richness and evenness and behaves consistently across a wide range of model structures.
Our model projects that species of conservation concern, oceanic turtles, pinnipeds, vaquita and totoaba (Totoaba macdonaldi) (see Ainsworth et al. 2011 for group definitions), experience population increases once full compliance with fisheries regulations is achieved (Fig. 2a). The full-enforcement scenario results in the highest biomass of threatened species in all cases, and the status quo scenario results in the lowest. Under the full-enforcement option, vaquita achieve an increase in biomass 3·5 times over present-day (2008) levels; though, they do not reach their population maximum within the span of the simulations. Under the status quo scenario, they decrease to 1/3 of their current biomass and stabilize. Totoaba are predicted to recover under all policy simulations including the status quo. Of the five regulatory instruments tested, spatial management allows the largest increases in turtles, vaquita and pinnipeds, while totoaba respond most positively to gear restrictions (i.e. larger gillnet mesh sizes reduce juvenile mortality). The initial decrease in the biomass of the long-lived groups, oceanic turtles and pinnipeds, reflects the lag time between new recruitment and individual body growth.
Abundance of commercially exploited species shows a steady increase throughout the 25-year simulation (Fig. 2b). Full enforcement yields the greatest increase in biomass overall, while the status quo scenario projects a modest increase for the first 15 years followed by stabilization. The increase in commercial biomass under the status quo regime mainly occurs in groups below trophic level (TL) 2·5; these increase 22% on average, while groups above TL 2·5 decrease by 49%. This shunting of secondary production towards the lower food web occurs in all scenarios to some degree, but it is moderated by restrictive fishing policies. For example, under full enforcement, the majority (64%) of predator groups above TL 2·5 increase in biomass, while under status quo the majority (54%) decrease.
High trophic level groups benefit the most from full enforcement (Fig. 3). These tend to be large-bodied exploited species, whose population abundance lies far below ecosystem carrying capacity. The positive relationship between trophic level and species recovery is pronounced under the full-enforcement scenario (mean regression slope in Fig. 3 is 0·36), but spatial management alone also favours predator recovery to some extent (slope = 0·19, not shown). Gear restrictions (0·07) and vessel buy-outs (0·04) marginally increase the relative biomass of predators.
Spatial management alone allows an increase in exploitable biomass to between 125% and 200% of the status quo (Fig. 4a,b). The increase occurs broadly over the study area, even though spatial management areas are restricted to the coast and Upper Gulf (Fig. S1, Supporting Information). There are smaller biomass increases in protected areas than in unprotected areas, but this is an artefact of the simplifying assumption regarding static spatial fishing effort. Biomass accumulates in unprotected areas because fisheries do not re-allocate effort appropriately, while biomass in protected areas is kept low by the presence of fisheries. Factoring in all of the current management regulations, the full-enforcement scenario (Fig. 4c) achieves a dramatic increase in exploitable biomass with the greatest improvements occurring in the delta region and along coastlines.
Total catch decreases under the full-enforcement scenario for all gear types relative to the status quo (Fig. 5). However, for all fleets except diving, the reduced fishing effort mandated by the full-enforcement scenario permits a steady increase in exploitable biomass, so catch increases after an initial dip. Catches by the gillnet fleets for fish and driftnet fleet for shrimp are reduced immediately to 21% of the status quo amount, mainly because of spatial management rules that restrict fishing in the Upper Gulf and the vaquita refuge. Those controls reduce total effort of pelagic gillnets by 66% inshore and 78% offshore. The initial drop in fishing effort reduces annual catch value (see Table S10, Supporting Information for prices) for pelagic gillnet fisheries from $0·9 million (US dollars) under the status quo to around $0·25 million (US dollars) under the full enforcement.
Trap fisheries are impacted the least: total catch is reduced to 74% of the status quo initially but recovers almost completely. Total value from fisheries in 2009 (i.e. 1 year after full enforcement is effected) is reduced for most fleets relative to status quo. The biggest losers are the following: shrimp driftnets −95%, pelagic gillnets −74%, demersal gillnets −61%, diving −57% and seine −47%. This lost revenue is likely to be an overestimate as we have not factored in adaptive behaviour from fishermen. The only winning fleets are fish trawl +2·6% and sport fisheries +15%.
Trade-Offs Between Fisheries and Conservation
Scenarios that take a large amount of catch tend to result in lower end-state species biodiversity (Fig. 6a). Spatial management regulations appear apt at preserving biodiversity. In fact, no additional benefit is realized by the full-enforcement scenario in our simulations. The shrimp trawl buy-out programme, which reduces incidental mortality on a broad range of species, also improves ecosystem biodiversity as measured by the Q90 metric from 5·42 (initial 2008 value) to 5·93. Average fish body size and average trophic level of artisanal catch are at their maximum values under the full-enforcement scenario (Fig. 6b,c). The trade-off frontiers suggest there may be convex relationships between catch and some ecological proxies, indicating that joint benefits are maximized by moderate fishing solutions. Translating catch into landed value by the price matrix in Table S10 (Supporting Information), the full-enforcement scenario could cost fishers $230 million (US dollars) annually from lost revenue (about 30% of gross revenue).
Each of the regulatory instruments shows high performance in at least one economic or ecological policy criterion (Fig. 7). For example, seasonal closures maximize shrimp landings, gear restrictions preserve elasmobranch biomass and spatial management protects biodiversity and vaquita biomass. No one regulatory instrument achieves exceptional results across all criteria, but layering these regulations together in the full-enforcement scenario yields outstanding conservation benefits to the detriment of fisheries revenue.
The Northern Gulf of California demonstrates a case where management regulations are likely to be sufficient to achieve a broad range of conservation objectives, but management falls short because implementation is lacking. The full-enforcement scenario offers significant ecological benefits including increased ecosystem biodiversity and recovery prospects for depleted species. We did not attempt to estimate the costs of achieving compliance, but they are likely to be substantial. In addition to enforcement costs, we know that fisheries would suffer economic losses under strict implementation of existing rules, perhaps losing 30% of revenue. Moreover, we recognize that it would require an organizational feat to improve compliance significantly, addressing issues like coordination between agencies, stakeholder education, regulatory oversight, legality and corruption (Moreno, Recio-Blanco & Michel 2010). Nevertheless, conserving this ecologically rich area may offer a cost-effective means to protect global biodiversity, as has been suggested by cost-benefit analysis of conservation in tropical terrestrial systems (Balmford et al. 2003).
A Toolbox Approach
Regulatory instruments excel at different aspects of resource management. Seasonal closures increase shrimp landings as they restrict harvests during breeding periods. Gear restrictions, aimed at reducing incidental capture of juvenile fish and non-target species, help increase commercial biomass. Cessation of illegal fishing has only minor ecological and economic effects at the ecosystem scale, although benefits to legitimate operators could be substantial; for instance, in the south-east of the study area where unregulated trap operations are common (Bóurillon-Moreno 2002). The buy-out programme for shrimp trawlers has a positive effect on biodiversity and reduces bycatch, but the benefits of buy-outs and spatial management are not additive because the main areas of effect closely overlap. The buy-out may help to reduce benthic habitat damage, but this was not modelled. Spatial management protections assist in the recovery of predatory fish and vaquita, and they preserve biodiversity because of the wide umbrella of protection offered across ecological communities. However, no single instrument can match the combined protections offered by the full-enforcement scenario.
All of the tested scenarios, including the status quo scenario, projected an increase in commercially exploitable biomass over the next 25 years, but this does not indicate species recovery. When fisheries deplete piscivore populations, trophic chains in the ecosystem are shortened overall and this can lead to biomass increases in the lower food web (Casini et al. 2009). Indeed, there are economic reasons to allow this type of reorganization (Pauly et al. 2000). However, the decision to do so should be made with care as simplification of the ecosystem is akin to a ratcheting effect (in that it is one-way process) and may be difficult to reverse for ecological and socioeconomic reasons (Pitcher 2005). Also, such ecosystems can be prone to collapse (Nyström, Folke & Moberg 2000). Policies with increased enforcement slow this degradation process.
The Atlantis model predicts that vaquita are not yet doomed to extinction as some policies can elicit an increase in numbers. This is the tacit consensus among researchers who call for immediate management protections (e.g. Jaramillo-Legorreta et al. 2007; Aragón-Noriega et al. 2010). However, our model has not considered the effects of genetic bottlenecks, which could become a significant obstacle to recovery after prolonged population depression (Rojas-Bracho & Taylor 1999). The driftnet switch-out and gillnet buy-out initiatives included as part of the species recovery programme (SEMARNAT (Secretaría del Medio Ambiente y Recursos Naturales) 2008) appear to be insufficient to permit recovery even under perfect compliance, which is not to be expected (Aragón-Noriega et al. 2010). Recovery is achieved only under the spatial management and full-enforcement scenarios, both of which assume gillnets are excluded entirely from the vaquita distribution area (Fig. S1, Supporting Information). A coordinated approach involving spatial protections seems necessary. This will surely impact the profitability of fisheries in nearby communities like San Felipe, Golfo de Santa Clara and Puerto Peñasco (Aragón-Noriega et al. 2010).
Totoaba are predicted to recover under all policy variations including the status quo. However, as there has been little change in stock status since the fishery was banned in 1975, it is likely that we have missed or underestimated an important source of mortality. The impacts of poaching and environmental stress were not included in the model but they may be significant (Lercari & Chávez 2007). Juvenile bycatch seems to be an important influence judging from the success of simulated gear restriction policies; however, severity may be underestimated.
There has been amazing progress made in the last 15 years in the use of end-to-end and ecosystem modelling approaches. These tools can play an important role in providing the quantitative evaluation and synthesis needed to support ecosystem-based management. In particular, end-to-end models are useful for evaluating the impacts of management policies that affect numerous species and ecosystem processes, like those tested in this study. However, the inclusivity of end-to-end models comes at a price as they incorporate processes that are poorly understood. We must therefore use caution in interpreting predictive results (Rose et al. 2010). Moreover, the slow run-times of complex end-to-end models affect our ability to explore uncertainty surrounding projections through sensitivity analysis. Current efforts to parallelize code aim to overcome this limitation (e.g. in Atlantis: E. Fulton, personal communication, and Ecospace: C. Walters, personal communication).
Several general conclusions can be drawn from this study. First, improving compliance with existing fisheries regulations (e.g. through improved education, monitoring and enforcement) might offer great conservation benefits. As new management policies are developed for this region, there must be a concurrent attempt to improve compliance. Second, no one regulatory instrument is able to address all major conservation and fishery concerns; an effective regulatory framework must include a suite of measures. Third, conservation costs money, and this cost will be borne unevenly among fishing communities in the Northern Gulf of California.
This project was funded by the David and Lucile Packard Foundation. PANGAS members provided useful input on scenario development: J. Torre (COBI A.C.), W. Shaw, M. Moreno Baez and T. Pfister (The University of Arizona), R. Cudney-Bueno (Packard Foundation), P. Turk-Boyer, R. Loaiza-Villanueva and I. Martinez-Tovar (CEDO Intercultural), A. Cinti (The University of Arizona), P. Raimondi (University of California-Santa Cruz), E. Torreblanca and G. Danemann (PRONATURA A.C.). M. Moreno-Baez and C. Moreno provided data on fishers’ interviews and logbooks. The Gordon and Betty Moore Foundation supported improvements to the Atlantis code base.