Correspondence Brian R. MacKenzie, National Institute for Aquatic Resources (DTU-Aqua), Technical University of Denmark, Charlottenlund Castle, DK-2920 Charlottenlund, Denmark. Tel: +45-3396-3300; fax: +45-3396-3333. E-mail: email@example.com
The abundance of bluefin tuna, Thunnus thynnus, in the east Atlantic and Mediterranean has declined in recent decades. The International Commission for the Conservation of Atlantic Tunas (ICCAT), the regional bluefin tuna management authority, has developed a plan to promote recovery by 2022, while still permitting fishing to continue during the period 2008–2010. Here we predict that the adult population in 2011 will likely be 75% lower relative to 2005 and that quotas in some intervening years will allow the fishery to capture legally all of the adult fish. Population demographics (proportion of older fish and repeat spawners in population) indicate that buffering capacity against years of poor reproduction has been reduced. This population is at risk of collapse (90% decline in adult biomass within three generations, the criterion used by the IUCN for defining populations as Critically Endangered), even under the currently agreed recovery plan, unless new conservation measures are implemented in the next few years.
A key objective of fisheries management is to maintain populations at levels where the biomass of adults does not limit the production of new young fish (Myers & Barrowman 1996; Rosenberg 2003; Beddington et al. 2007). This objective has proven difficult to achieve in many of the world's commercially most important fisheries: 50% and 25% of all exploited marine fish populations in the world are considered to be fully- or over-exploited, respectively (Garcia et al. 2005). Because effective management is difficult to implement, populations often undergo large declines and, in some cases, entire fisheries have been closed to allow populations to recover. Examples of such declines of exploited marine populations are well documented (Jackson et al. 2001; Hutchings & Reynolds 2004; NRC 2006; Rosenberg et al. 2006). Recovery following fishing closures is often slow and sometimes impossible if the depleted population is still captured incidentally in fisheries targeting other species, or if changes to habitats, the environment or food web occur, which also affect offspring survival rates (Hutchings & Reynolds 2004). The difficulties associated with preventing declines in the first place or rebuilding depleted stocks are primarily due to the difficulty of reducing fishing pressure in the face of resistance by the fishing industry to strict management controls (Rosenberg 2003; Rosenberg et al. 2006; Beddington et al. 2007).
Here we demonstrate that bluefin tuna (Thunnus thynnus) in the northeast Atlantic and Mediterranean (east of 45° west longitude) may also be on the way to collapse (90% decline in adult biomass within three generations, the criterion used by the World Conservation Union for defining populations as Critically Endangered). This population has been declining for many years ( Fromentin & Powers 2005; ICCAT 2007) (Figure 1), and the biomass of adults (spawning stock biomass) is now (2006) at its lowest on record (approximately 40% of late 1950s’ biomasses; Figure 1), with the steepest decline in last 5–10 years (ICCAT 2008). At these population sizes, reproductive dynamics become increasingly uncertain and are likely limited by spawner biomass (Figure 1).
The spatial distribution of large, adult (> ∼30–35 kg and 4–5 years) bluefin tuna has also changed since the 1950s–1970s. These size-groups were common in the Bay of Biscay (Cort & Nøttestad 2007), the North Sea and Norwegian Sea (Mather et al. 1995; Fromentin & Powers 2005; MacKenzie & Myers 2007), but now are rare in these areas. Bluefin tuna disappeared from the Black Sea in 1987 (Karakulak 2004) and fishery-independent surveys shows that the species is now rare in parts of the Ionian Sea (northcentral Mediterranean; Bearzi et al. 2006). Their disappearance from these areas has coincided with or followed the expansion of fisheries in all these regions, and generally throughout the northeastern Atlantic and Mediterranean (Mather et al. 1995; Fromentin & Powers 2005; ICCAT 2006b). While the causes for these disappearances require further investigation, they are consistent with fishing-induced declines of abundance (FAO 1995, 1996; ICES 2005).
The declines in biomass and contractions of geographic range indicate that the population is overexploited (FAO 1996; ICES 2005), as has been recognized by ICCAT scientists (ICCAT 2008), the EU (EU 2007, 2008), and the conservation community (WWF 2006). A recovery plan adopted in January 2007 by the International Commission for the Conservation of Atlantic Tunas (ICCAT) was implemented during 2007 (ICCAT 2006a; EU 2007). It has been developed and adopted because of concerns that the highly efficient bluefin tuna fishery may lead to commercial and local extinction throughout much of its range. The overall goal of the recovery plan is to increase spawner biomass to a level (BMSY), which should support the largest long-term annual catch (i.e., maximum sustainable yield, MSY) with > 50% probability by 2022 (ICCAT 2006a). Management, which will have greatest impact on population dynamics, including the probability that biomass will increase to BMSY, will be that which most directly and immediately affects fishing mortality and abundance now and in the next few years. These measures include regulations to restrict catches during the period 2007–2010 (Table 1) and to increase the minimum size of captured fish from 10 to 30 kg. This latter measure is intended to increase survival of juvenile tuna so they can reproduce at least once before capture. Recently (12 June 2008), the EU implemented short-term (6 month) emergency conservation regulations in response to new evidence of damaging fishing practices by six EU countries during the period 2007–2008 (EU 2008).
Table 1. Settings for projection simulations of bluefin tuna population development during the period 2006–2022 in the northeast Atlantic and Mediterranean. Fi-j= age-specific fishing mortality for ages i–j; SQ = status quo, as represented by average for last 3 years of the most recent assessment (i.e., 2004–2006) and based on officially reported catch data (run 13 in ICCAT 2008); TAC = total allowable catch, as defined in the recovery plan (ICCAT 2006a), and is 29,500; 28,500; 27,500; and 25,500 t for 2007, 2008, 2009, and 2010, respectively. SSBlow (lowest observed spawner biomass) is the SSB below which recruitment is limited by SSB and above which recruitment varies independently of SSB (ICES 2003); SSBlow was estimated as the lowest observed spawner biomass (101,000 t) in the period of available data (1955–2006). See also Methods and Supporting Information for details
In addition to changes in abundance and range occupied, the size and age structure of exploited fish populations often changes systematically over time (Anderson et al. 2008). These changes, which have not yet been investigated for bluefin tuna in the northeast Atlantic and Mediterranean, include changes in age/size composition and spawning experience (first-time vs. repeat), and can have direct and indirect effects on reproductive and recruitment potential, including individual relative fecundity and offspring viability (Lambert 1990; Marshall et al. 2003). These reproductive demographic changes in turn can lead to higher recruitment variability in heavily exploited fish populations compared with populations experiencing lower exploitation (Hsieh et al. 2006; Anderson et al. 2008). We hypothesize that similar changes in age/size structure have occurred for this population and that they may be another indicator that the population is overexploited.
Our study has two objectives related to the exploitation status and sustainability of this population. We first evaluate the likelihood that the new recovery plan will result in conservation of this population and an increase in its size to levels where recruitment is no longer limited by spawner abundance. We predict using forward projection population dynamics models how the population will respond to the recovery plan, and in particular whether the plan, assuming full implementation and successful compliance by the fishing industry, will lead to recovery of this population within the expected time frame. Our second objective is to quantify changes in age structure and reproductive demographics over time.
We used an age-structured stochastic modeling approach similar to that used in many fisheries assessment working groups of the International Council for the Exploration of the Sea (e.g., ICES Baltic Fisheries Assessment Working Group (ICES 2006)). The model is used for predicting future development of fish populations (typically over 10–20 years) under different assumptions of fishing mortality and population biology (e.g., growth rates, maturity schedules, reproduction rate). Empirically derived estimates of the uncertainty of input variables is directly included in the model so that probabilistic outputs of population variables (e.g., biomasses) are derived. We used biological and fishing mortality rate information from assessment run 13 of ICCAT (ICCAT 2008) to parameterize the model.
In brief, the modeling framework uses the starting population size at 1 January of each year and applies natural and fishing mortality rates to estimate numbers of survivors to 1 January of the following year. Annual estimates of adult biomass are derived from numbers of mature individuals multiplied by their weights-at-age. Production rate of new bluefin tuna each year (recruitment) is estimated from a functional relationship to spawner abundance. Full details of the modeling approach and the spawner biomass–recruitment relationship are described in Supporting Information.
We simulated population abundance according to five different scenarios to evaluate how the population would develop under full implementation of the new management plan (Table 1). The scenarios we considered investigated the uncertainty of catches in 2006 and 2007, the parameterization and uncertainty of the spawner biomass–recruitment relationship, the limits on catches for 2008–2010, and the consequences of a recent increase in minimum landing size from 10–30 kg for most, but not all, bluefin tuna fisheries in the northeast Atlantic and Mediterranean. The total allowable catches (TACs) in 2007, 2008, 2009, and 2010 are 29,500; 28,500; 27,500; and 25,500 t, respectively (ICCAT 2006a).
All scenarios started in 2006 (most recent year for which abundance estimates were available for all age groups) and end in 2022, when the current recovery plan should end (ICCAT 2007). We conducted five scenarios: two scenarios used the officially reported catches for 2006 (30,650 t), which ICCAT scientists (ICCAT 2008) also assumed applied for 2007 (their assessment run 13), and three scenarios assumed the real catches estimated by ICCAT for 2006 and 2007 (50,000 and 60,000 t, respectively). Additional details of the scenario settings are given in Table 1.
We present and interpret results both for overall trends in output variables (spawner biomass, yield, and recruitment) and for output values at the end of the initial TAC period (i.e., 2011) and end of the recovery period (2022). We compared biomass declines with those used by the World Conservation Union to define extinction risk. In particular, we evaluated the probability that a decline in adult biomass of 90% would occur within three generations, and consider this decline rate to be a measure of population collapse. This decline rate is used by IUCN to define populations as Critically Endangered (IUCN 2006).
Trends in population demographics
We quantified changes in age/size composition and reproductive demographics that have occurred for bluefin tuna in the eastern Atlantic and Mediterranean. We calculated the mean age of spawners, and the proportion of older bluefin tuna (i.e., age ≥ 8 years) using annual ICCAT population numbers-at-age data (ICCAT 2008). We also estimated the proportion of the population, which was a repeat spawner: because bluefin tuna mature at ages 4–5 years (ICCAT 2008), we assumed that half of all 5-year olds were repeat spawners (i.e., 50% of the 5-year olds were mature and spawned as 4-year olds) and that all fish aged ≥ 6 had spawned at least once.
We first consider scenarios 1 and 3, which assume officially reported catches for 2006–2007 and recovery plan catch during the period 2008–2010. In scenario 1, spawning stock biomass is estimated to be 99,000 in 2011 and increase to 210,000 t in 2022 (Figure 2A). Recruitment in 2012 and 2022 is expected to be 1.6 and 1.9 million individuals, respectively (Figure S2A). In comparison, geometric mean recruitment for 2001–2005 was 2 million (ICCAT 2008). Projected yield in 2022 is approximately 30,000 t (Figure S3A). In scenario 3, spawning stock biomass rises more sharply after the TAC period due to the lower assumed fishing mortality for ages ≥ 4 years. Spawner biomass could reach 335,000 t (Figure 2C), which would support yields of approximately 18,000 t even at these lower F-values (Figure S3C).
The next two scenarios (2 and 4) employ the more realistic and higher estimates of catches in 2006 and 2007 (ICCAT 2008) than those officially reported to ICCAT; otherwise settings are the same as for scenarios 1 and 3, respectively. Scenario 2 predicts a reduction of spawning stock biomass to 25,000 t in 2010 and 28,000 t in 2011 (Figure 2B). In 2010 (final year of existing recovery plan when TAC = 25,500 t), the fishery is legally able to capture all spawners, and there is a 5% chance that spawning stock biomass will fall to 2,000 t (Figure 2B). Recruitment will fall as spawning stock biomass declines in the early 2010s (Figure S2B). Low fishing mortality starting in 2011 allows spawning stock biomass to increase nearly four-fold to approximately 100,000 t in 2022 (Figure 2B).
Scenario 4 settings were similar to scenario 2 except that fishing mortality of age groups ≥ 4 during 2011–2022 was reduced from 50% of status quo (2004–2006) to 15% of status quo. A similar decline in spawning stock biomass until end of 2010 is therefore seen in both scenarios 2 and 4, but as expected, it increases more in the reduced fishing scenario (4) to approximately 177,000 t by 2022 (Figure 2D).
Under scenario 2, there is only approximately 50% probability that the population will recover to its size before implementation of the recovery plan, and that the population will exceed the threshold that no longer limits reproduction (defined as SSBlow; Figure 3). Moreover, there is < 5% probability that spawning stock biomass in 2022 will exceed the long-term mean biomass (Figure 3). In contrast, under scenario 4, assuming low fishing mortality for all age groups starting in 2008, there is approximately 80% chance that spawning stock biomass in 2022 will exceed SSBlow, and approximately 40% chance that it will exceed the long-term mean (Figure 3).
Scenarios 2 and 5 compare projected stock development assuming different spawner biomass–recruitment models (Figure S1); all other model settings were identical. Mean ± 95% confidence limits for spawner biomass in 2011 were 28,000 t (1,000–62,000 t) and 48,000 t (16,000–85,000 t) for the two scenarios, respectively (Figures 2B and 2E). Based on both models, it is likely that recruitment in next few years will be substantially reduced due to low spawner abundance (Figure S2B and E). In 2022, projected spawner biomass was 100,000 and 254,000 t, respectively (Figures 2B and 2E).
Several population-level indicators of age structure and reproductive demographics have declined since the start of available datasets in 1955. The mean age of mature bluefin tuna has declined since the mid-1980s and the proportion of age ≥ 8 years has declined, especially since the late 1970s (Figure 4A). The share of repeat spawners in the population has declined and remained generally low since the mid- to late 1980s (Figure 4B).
The recent decline of this bluefin tuna population is primarily a result of high exploitation for too many years (ICCAT 2008). Scientific advice by ICCAT scientists and some conservation organizations has not been fully implemented by nations that exploit this population; in cases where the advice has been partly followed (e.g., quotas or fishing capacity reduced but not sufficiently to prevent decline), it has not been effectively enforced. Evidence of weak implementation and enforcement of regulations are reports of illegal fishing (e.g., capture of undersized specimens, failure to report catches, fishing outside closed seasons or areas) summarized by ICCAT scientists and others (ICCAT 2008; WWF 2006), even during the recovery plan period itself (EU 2008). Given the history of fish population collapses elsewhere, and their ecological and socio-economic consequences (Hutchings & Reynolds 2004; Beddington et al. 2007), it is perhaps surprising that authorities responsible for managing this population have not been more rigorous in supporting effective conservation measures. We hypothesize that authorities have been unable or unwilling to resist political pressure by the bluefin tuna fishing industry to implement recommended measures, as has been the case in many other jurisdictions for other species (Rosenberg 2003; Rosenberg et al. 2006; Beddington et al. 2007). These difficulties are particularly acute for economically valuable species such as bluefin tuna whose long-distance and high-seas migratory behavior exposes the species to fishing fleets of a large number of nations inside and beyond the territorial waters of single nations. Hence, enforcement of regulations is difficult and compliance is incomplete (ICCAT 2008).
The expected biomass trajectory, together with the declines in indicators of stock status (discussed below), is alarming: a management plan, which has been developed to promote recovery of a population, will cause it to fall to record lows and could result in both fish population and fisheries collapse. The current ICCAT and EU recovery plan as presently implemented (similar to scenario 2) is predicted to reduce the adult population close to small sizes (i.e., 50% and 5% probabilities that spawning stock biomass = 25,000 and 2000 t in 2010) within one generation of the stock, given realistic fishing mortality rates and biological uncertainty, including that related to recruitment. Even if a near-complete ban on all bluefin tuna fishing in the northeast Atlantic and Mediterranean were implemented immediately in 2008 and enforced until 2022 (scenario 4), the population will probably fall to record lows in the next few years, unless environmental conditions promote exceptionally high recruitment. In the coming years, the existing quota would allow capture of most, or all, adults; this situation makes population sustainability sensitive to the success of individual year classes and reduces buffering capacity against a series of years when environmental conditions reduce offspring survival.
At the projected low biomasses, recruitment will become severely impaired due to (1) low numbers of spawners (Myers & Barrowman 1996; ICCAT 2007), (2) potential Allee effects on reproductive success (Frank & Brickman 2000), and (3) increased risk of recruitment failures due to adverse environmental conditions (Brander 2005). In particular, there is a 25% chance that the expected decline (scenario 2) in biomass between 1999 (spawning stock biomass = 142,000 t) and 2010 (25th percentile = 14,000 t) will be 90% under the agreed recovery plan management measures. Given that a decline > 90% within three generations (= 12–15 years for this population) is one of the criteria for a population to be listed as Critically Endangered on the World Conservation Union's Red List (IUCN 2006), the current recovery plan (which allows high catches during the period 2008–2010) ironically could justify a species-at-risk listing, rather than a sustainably managed population. Furthermore, our calculations demonstrate that reproduction will be limited by spawner abundance (i.e., SSB < SSBlow= 101,000 t) for many years to come. The status of the eastern bluefin tuna population may therefore reach that of the population in the West Atlantic, which is now extremely low and also in danger of collapse (Safina & Klinger 2008).
Alternative scenarios could be constructed to evaluate various fishing mortalities and compliance levels on recovery, but our overall results and conclusions regarding the decline of the population would not change; in general, higher fishing mortality or implementation failure will prolong population decline and subsequent recovery. In particular, the trajectory after 2010 is uncertain because there presently are no recovery-specific TAC regulations in place following the expiration of the current TAC schedule.
Our projections of biomass development depend, as do all fish population projections, on the recruitment–spawner biomass relationship. We used two such models; both demonstrate that the population will decline to record lows in the next few years under the recovery plan. Our analyses indicate that the hockey–stick model follows the recent recruitment pattern more closely than the Ricker model (see Supporting Information). In addition, the hockey stick model has superior management implications: in particular, meta-analyses based on 100s of fish populations show that it generally does not overestimate maximum reproductive rates at low population sizes, as do alternative models ( Barrowman & Myers 2000), and therefore does not overestimate the resilience of the population to declining biomass. These characteristics are particularly important for the specific case of bluefin tuna because the population is in rapid decline, the spawner biomass–recruitment relationship is uncertain, a recovery plan is in place, yet implementation of the management measures and compliance by the fishing industry is difficult (ICCAT 2006b; EU 2007, 2008). Using the hockey stick model for management decisions under these circumstances is therefore more consistent with the precautionary approach to fisheries management (FAO 1995, 1996; ICES 2003).
Other indicators of population status have changed. Age structure and reproductive demographics for the population have shifted to configurations that likely reduce reproductive potential and increase vulnerability of the remaining population to additional stressors, such as ecosystem variability. Although the contribution of different ages to recruit production, parent–offspring relationships and frequency of skipped (nonannual) spawning for bluefin tuna remain to be fully investigated, changes in reproductive demographics like those documented here can lead to a reduction in reproductive and recruitment potential and increased recruitment variability in many other fish species (Marshall et al. 2003; Anderson et al. 2008). The narrowing and displacement of age structure toward younger individuals with less spawning experience is a common feature in exploited fish populations (Anderson et al. 2008). When the population is eventually allowed to recover, recruit production per spawner will likely be lower, and population recovery will probably take longer, than for a population of the same biomass composed of older, more experienced spawners.
The declines in abundance and age structure may also be factors responsible for the disappearance of bluefin tuna from formerly occupied areas in the northeast Atlantic and Mediterranean. For example, many other fish species expand (or contract) their geographic ranges when abundant (or rare) (MacCall 1990; Garrison 2001; Bakun 2005), probably as a response to density-dependent feedbacks as local carrying capacity is reached (Matsukawa 2006). A rebuilding population could reoccupy former areas of the distributional range.
The existing recovery plan, whose progress is scheduled to be assessed in 2008 by ICCAT (ICCAT 2006a), therefore needs rapid adjustment to minimize the rate of further decline especially in 2008–2010. Some other evaluations of the recovery plan using different fishing and biological assumptions also show that the recovery plan objective may not be met (BFT species group 2007) and that implementation success of the new fishery regulations will be critical for achieving the plan objective (Fromentin 2007). Experience with collapses of other fish populations and their recoveries shows that (1) cessation of fishing is the single most effective measure available for promoting the rebuilding of most collapsed fish populations (Rosenberg 2003; Beddington et al. 2007), (2) recovery sometimes requires many decades even when exploitation has been greatly reduced, and (3) recovery tends to occur fastest when conservation plans are implemented soon after declines become evident and over short periods (Hutchings & Reynolds 2004; Shertzer & Prager 2007).
A modified recovery plan (i.e., no or little bluefin tuna fishing of any kind for several years) in the northeast Atlantic and Mediterranean urgently needs to be implemented to reduce the risk of population collapse. Implementing major reductions in fishing mortality for this population is difficult (EU 2008); however, past experience with recovery successes shows that fishing mortalities need to be very low in order to promote recovery (Hutchings & Reynolds 2004; Rosenberg et al. 2006; Beddington et al. 2007). A retrospective analysis by ICCAT showed that fishing mortality since 2003–2004 was three-fold higher than that which would lead to MSY (ICCAT 2007, 2008), and similar difficulties with implementation of fishing restrictions for species (cod Gadus morhua, herring Clupea harengus, etc.) in other jurisdictions are common (Rosenberg 2003; Beddington et al. 2007). Nevertheless, delaying implementation for bluefin tuna will mean that this species in the NE Atlantic and Mediterranean will take longer to recover and become more susceptible to collapse, possibly within one generation. Such a fishing-induced collapse would be an ecological disaster for the population, a fisheries management failure by and for the fishing industry, managers and regulatory agencies, and socio-economic hardship for those depending on the population for livelihoods. Given the slow recovery rate of other collapsed fish populations, many years or even decades may be necessary before a sustainably exploitable population could be reestablished (Hutchings & Reynolds 2004). Moreover, a fishing-induced collapse would contradict the intentions of international agreements such as the FAO Code Of Conduct for Responsible Fisheries Including the Precautionary Approach (FAO 1995, 1996) and the Conventions on Biodiversity (UN 1992), and Sustainable Development (UN 2002), which many signatories of ICCAT and the EU have otherwise adopted.
Editor : Corey Bradshaw
We thank Eskild Kirkegaard, Irene Mantzouni, Jerry Scott, Morten Vinther, and the ICCAT Secretariat for assistance, editors and anonymous reviewers for comments and suggestions, and our late colleague and friend, Ransom A. Myers, for inspiration. This work is a contribution to the History of Marine Animal Populations project of the Census of Marine Life and to two EU Networks of Excellence (MarBEF [publication MPS-08053]; Eur-oceans).