Winners and losers of a technical change: A case study of long-term management of the Northern European Hake

Fisheries Research 110 (2011) 98–110

Contents lists available at ScienceDirect

Fisheries Research journal homepage: www.elsevier.com/locate/fishres

Winners and losers of a technical change: A case study of long-term management of the Northern European Hake Dorleta Garcia ∗ , Raúl Prellezo 1 , Marina Santurtun 1 , Luis Arregi 1 Azti-Tecnalia, Txatxarramendi ugartea z/g, 48395 Sukarrieta, Spain

a r t i c l e

i n f o

Article history: Received 23 November 2010 Received in revised form 25 February 2011 Accepted 26 March 2011 Keywords: Discards Long term management Management strategy evaluation Northern European Hake Technical measures

a b s t r a c t Since 2004 management of the Northern Stock of European Hake has been focused on recovering the stock level up to a level consistent with the precautionary approach. After that, in 2007 and once this objective was on the track of being fulfilled a long term management plan was proposed. This plan has to be congruent with the maximum sustainable yield policy as well as producing stable yields and population levels. Thus, in that year, a bioeconomic impact assessment of long-term management plans for this stock was carried out. However the biological and economic assessments were not integrated and not fully congruent. On the basis of this assessment additional questions relating to the combination of harvest control rules with technical measures were raised by the managers and stakeholders. Here, the model used in the biological assessment is extended in order to integrate the economic part and to shed light on the effect of technical measures at stock and fleet level. Two scenarios are presented: a ‘base case’, where the model is parameterized from historical observations; and an ‘alternative case’ where an increase in the mesh size of some trawlers is simulated. In both scenarios the probability of falling below limit reference points is above 0, contrary to the result obtained in 2007. However, the relative trends of the median of population indicators are similar. While the biological performance of the base and alternative scenarios is also similar the trawlers are highly penalized when their mesh size is increased and the overall economic profit is lower. Furthermore, two fleets gain and the rest remain the same with the increase in the mesh size of trawlers. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Technical measures such as mesh size regulations are commonly used to strengthen catch quotas and effort management as part of recovery and long-term management plans in order to reduce growth overfishing and the discarding of undersized fish. This concept is based on the classical theory of fisheries management related to yield per recruit analysis (Armstrong et al., 1990; Quinn and Deriso, 1999; Scott and Sampson, 2011) or spawn at least once; policies (Myers and Mertz, 1998). In yield per recruit analysis an exploitation pattern is related to an optimum harvest rate which will maximize long-term yield and likewise a harvest rate is related to an optimum exploitation pattern. Within the context of Ecosystem Based Fisheries Management (EBFM) Froese et al. (2008) have recently defined the optimum size at first capture. Such approaches are based on stock dynamics and the overall exploitation and selection pattern, but they do not consider the dynamics

∗ Corresponding author. Tel.: +34 946029400; fax: +34 946870006. E-mail addresses: [email protected] (D. Garcia), [email protected] (R. Prellezo), [email protected] (M. Santurtun), [email protected] (L. Arregi). 1 Tel.: +34 946029400; fax: +34 946870006. 0165-7836/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2011.03.018

of the fleets involved in the exploitation of the stock. When the stock is exploited by a single homogeneous fleet, in terms of the selection pattern of individual vessels, the problem of increasing the selectivity and the benefit of this increase revert to the fleet itself. Usually stocks are exploited by several fleets operating with different fishing gears and mesh sizes. In general, each fleet is focused on a particular age range, even if there is some overlapping of ages between them. In some fleets like trawlers, when the mesh size is increased, the number of larger individuals retained increases and that of smaller ones decreases. Then, in a hypothetical situation where only the mesh size of the trawlers was increased, the level of exploitation of smaller individuals would decrease, increasing that of larger individuals. The result of this could be an increase in competition between fleets that catch larger individuals, if the future growth of smaller individuals, now underexploited, does not balance the exploitation increase of larger individuals. In the context of EBFM some authors have recently advocated a balanced exploitation of the individual stocks instead of selective fishing (Rochet et al., 2009; Zhou et al., 2010). Due to the nonlinear and complex nature of fishery systems, it is not straightforward to derive the effect of technical changes within the individual fleets. Suuronen and Sarda (2007) have anal-

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

ysed the role of technical measures such as mesh size restrictions in European waters. They concluded that their effect, in biological terms, could not be as significant as expected and recognize that measures that imply cost or lost of earnings for the industry could be unsuccessful. So they recommend carrying out cost-benefit of technical measures before their implementation. Several studies have analysed the cost and benefits of an increase in mesh sizes, at fleet level, in particular cases (Heikinheimo et al., 2006; Macher et al., 2008; Pascoe and Revill, 2004; Thunberg et al., 1998; Tschernij et al., 2004). In the short term, almost all the scenarios in which an increase in selectivity at age was simulated resulted in short term losses for the fleet compared with status quo scenarios. Two scenarios in Macher et al. (2008) gave similar results to those obtained in the status quo scenario due to the increase in selectivity only affecting ages that were previously fully discarded, the landings thereby unaffected and with similar profits. In general, the performance of the overall fleet and the stock was better when selectivity was increased. In the study of Kuikka et al. (1996) the escapement mortality they modelled was considered to be high, so the increase in mesh size did not produce significant benefits in the stock and the yield in the long term. When the change in selectivity was not applied to all the fleets, the fleets not affected by the change gained more than those affected by the change (Heikinheimo et al., 2006; Pascoe and Revill, 2004). The variety of results obtained in these studies highlights the necessity of analysing the effect of technical changes, case by case, as pointed out by Suuronen and Sarda (2007). The Northern European Hake stock management is at the point of moving from a recovery plan to a long-term management plan (LTMP). In 2007, a bioeconomic impact assessment of LTMPs for this stock was carried out (SEC, 2007a,b). However, the biological and economic assessments were not integrated and not fully congruent.2 Based on the conclusions of these analyses the European Commission (EC) launched a consultation with the stakeholders about the LTMP. The consultation dealt with different aspects of the management of the stock. In particular, it considered the increment in mesh size on some fleets, such that the overall selection pattern improves and discards decrease. Thus, the need for an approach that could facilitate the assessment of such changes, within the framework of the long-term management of the stock, was identified. In May 2008, both North and South Western Water Regional Advisory Councils (NSWWRACs) stated a specific question in relation to what would be the effect in Northern Hake induced by the harmonisation of a unique mesh size of 100 or 120 mm for gillnetters and trawlers operating in Subarea VII and Divisions VIIIabd (Fig. 1). The modelling tool was developed and parameterized and various changes in mesh sizes were simulated under NSWWRACs premises. In this study, the results of the most data rich scenario are presented. This contribution aims to calculate the costs and benefits of a mesh size increase, within the framework of the LTMP of the Northern stock of Hake, analysing them at stock and fleet level. On the basis that the fleets involved in the Hake fishery used different gears and mesh sizes, the exploitation is divided into several fleets according to their technical characteristics. Furthermore, as discards are expected to be affected by the LTMP and mesh size increase, they have been included in the simulation. In 2007, the exploitation was simulated using a single fleet and discards were not included. Hence, to begin, the paper compares which are the differences between the impact assessment performed in 2007 (SEC, 2007b) with those undertaken here.

2 In the economic evaluation age structure was not considered and the production model used to generate the catch was not the same one used in the biological evaluation. Only the medians were considered and from 2015 onwards it was assumed that the system was stable.

99

Fig. 1. Northern European Hake distribution along ICES divisions and location of Fishing Units considered in the article and their mesh size.

The objective of this study is to perform an integrated bioeconomic analysis of the LTMP for Northern Hake stock in order to determine the robustness of the evaluations performed in 2007 (SEC, 2007a,b). The robustness is analysed in relation to the sustainability of the stock and to the fleets’ performance with and without the implementation of technical measures. First, the case study is presented including descriptions of the fishery and its management. Then the simulation model, data used and parameterization of simulations are described. Finally, the results and their discussion are presented within the framework of the management of this stock, together with (an analysis of) the usefulness of this approach. 2. The case study The Northern stock of Hake is considered to be a subpopulation of the European Hake (Merluccius merluccius), which is a demersal species distributed widely from Mauritania to Norway and the Mediterranean Sea (Casey and Pereiro, 1995). For management purposes, three different stocks are considered: the Mediterranean stock; and two stocks in the Northern East Atlantic, divided by the parallel 44.3◦ , the so called Northern and Southern stocks of Hake. The northern stock of European Hake, the only stock considered herein and referred to as Hake, is exploited principally by Spain and France, with 60% and 30% of the total international catch, respectively. The International Council for the Exploration of the Sea (ICES) divides the fishing activity of this stock into 15 different Fishery Units (FUs), characterised by the gear used and the fishing area (see Table 1). Five of these FUs account for 70% of the total international catch, FU01, FU03, FU04, FU13 and FU14. For a full review on Hake biology and of European Hake see Murua (2010).

100

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

Table 1 Fishery Unit description, name, gear used, area, average contribution to the total landings and average contribution to total discards. FU

Description

Sub-area

Landings

Discards

FU01 FU02 FU03 FU04 FU05 FU06 FU08 FU09 FU10 FU12 FU13 FU14 FU15 FU16 FU00

Long-line in medium to deep water Long-line in shallow water Gillnets Non-Nephrops trawling in medium to deep water Non-Nephrops trawling in shallow to deep water Beam trawling in shallow water Nephrops trawling in medium to deep water Nephrops trawling in shallow to deep water Trawling in shallow to medium water Longline in medium to deep water Gillnets in shallow to medium water Trawling in medium to deep water Miscellaneous Outsiders French unknown

VII VII VII VII VII VII VII VIIIabd VIIIabd VIIIabd VIIIabd VIIIabd VII and VIIIabd IIIa, IV, V and VI –

20.5% 0.9% 10.1% 16.3% 0.7% 0.0% 0.3% 2.0% 3.5% 1.7% 9.6% 12.2% 1.8% 20.0% 0.4%

0% 0% 0% 54% 0% 0% 13% 0% 0% 0% 0% 33% 0% 0% 0%

Hake is managed by the EC, on the basis of the scientific advice provided by ICES. Annually, a Total Allowable Catch (TAC) is set by the EC and is divided among European States, according to a fixed quota share. Furthermore, several technical measures have been adopted throughout the years in order to protect juveniles. Since 2004, the TAC has been set according to the recovery plan established within that year (EC No. 811/2004) with the aim to achieve, in two consecutive years, Spawning Stock Biomass (SSB) levels above the precautionary limit set by ICES at BPA 3 = 140,000 tonnes. The TAC is set limiting the fishing mortality to the precautionary level FPA = 0.25, at the same time a maximum 15% variation in TAC is allowed between consecutive years. If the SSB falls below BLIM 4 = 100,000 tonnes, the recovery plan allows a higher reduction in TAC. In order to ensure fulfilment of the plan, additional control measures were set in place. In 1998, technical measures were established to protect juvenile fish in European waters (EC No. 850/98). The minimum landing size for Hake was set at 27 cm, except for Kattegat and Skagerrak, where it was set at 30 cm. Moreover; minimum mesh sizes were determined, dependent upon gear, area and targeted stock. The assessment carried out by ICES, in 2000, indicated that the stock was in a serious risk of collapse. Thus in 2001, an emergency plan was put into force (EC No. 1162/2001) with the aim of reducing the catch of juveniles. Additional minimum mesh size to those set in 1998 was established. At present, the mesh size configuration map for the main FUs in Area VII is 120 mm for gillnetters and nonNephrops trawlers (FU03 and FU04) with the minimum mesh size of 100 mm for those segments in Subarea VIIIabd (FU13 and FU14) (see Fig. 1). The ICES assessment carried out in 2007 indicated that Hake SSB was close to BPA in 2006 and exceeded it in 2007 (ICES, 2007) thereby recommending, according to the recovery plan, its replacement with a management plan. In 2007, two assessments were carried out by the EC in order to evaluate the biological and economic impact of possible LTMPs for Hake (SEC, 2007a,b). The Hake LTMP should be in accordance with the plan of implementation of the World Summit on Sustainable Development (WSSD, 2002), ratified by the European Community and Member States, as stated in Council regulation EC No. 2006/360. One of the recommendations

3 BPA and FPA represent the precautionary reference points of spawning stock biomass and fishing mortality, respectively as defined by ICES. They represent a buffer zone that overcomes the uncertainty associated with the estimation of limit reference points BLIM and FLIM . 4 BLIM and FLIM are the limit reference points defined within ICES precautionary approach framework. In this case, BLIM corresponds with the Lowest Observed Spawning Stock Biomass (BLOSS ) in the assessment of 2003 and FLIM with FLOSS . See Hauge et al. (2007) for details on the definition of ICES Precautionary Approach framework.

of the plan of implementation is to maintain or restore fish stocks to Maximum Sustainable Yield (MSY) level by no later than 2015. At the biological impact assessment meeting, different management strategies were evaluated by means of Monte Carlo simulations, under a Management Strategy (MSE) framework (Butterworth, 2007; Butterworth and Punt, 1999; De la Mare, 1998; Punt and Donovan, 2007; Rademeyer et al., 2007). Several Harvest Control Rules (HCRs) were tested against different assumptions made in relation to the stock dynamics (different stock-recruitment relationships and different growth patterns). All the HCRs were based on achieving a fixed target fishing mortality while allowing a fixed maximum annual variation in it. However, advice was provided in terms of TAC, after transforming output fishing mortality into catch. At the economical impact assessment meeting, a socio-economic evaluation was carried out based on the biological assessment. However, the link between the biological and economic assessments was not undertaken in an integrated way. Furthermore, the assumptions and the formulae used in both evaluations differed. Thus, the results obtained in both evaluations were not fully congruent. In 2008, on the basis of the findings of the impact assessments, the EC proposed a HCR based on achieving a fishing mortality equal to FMAX 5 = 0.17, reducing it annually by no more than 10%. Moreover it launched a consultation with the stakeholders about the suitability of this HCR and other possible management actions, such as increases in the mesh size of some fleets. 3. Materials and methods Two different simulations were undertaken using an extended version of the simulation model used in the biological evaluation carried out in 2007 (SEC, 2007b). This followed a MSE approach and was coded in R (R Development Core Team, 2009), using FLR packages (Kell et al., 2007). Kell et al. (2005) and Murua et al. (2010) analysed Hake management under a MSE approach. Kell et al. (2005) analysed the management of the stock based on the Precautionary Approach framework used by ICES until 2009 instead of a long-term management framework. Murua et al. (2010) used the same algorithm used in 2007 (SEC, 2007b) but they did not disaggregate the exploitation by fleet and the work was focused on the robustness of the LTMP to uncertainty in reproductive dynamics of the stock instead of fleet performance.

5 FMAX : The rate of fishing mortality that produces the maximum level of yield per recruit. It was selected as a proxy of FMSY because it is independent of the stockrecruitment relationship and well defined for this stock, and was estimated to be equal to 0.17.

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

101

Fig. 2. Conceptual diagram of the simulation algorithm.

The simulation model took an initial random population and projected it into the future, under pre-specified conditions. The model parameterization was based on historical data and parameters, together with an initial fit of an XSA assessment model (Shepherd, 1999) to a catch-at-age data matrix and a set of abundance indices. This catch-at-age matrix was equal to the sum of the landings-at-age used in the assessment working group of the stock and an estimated discards at age matrix (ICES, 2007, SEC, 2007b). Initial population spanned from 1978 to the first of January of 2007 and the projection from there to 2040.

ity of the fleet which, apart from Hake, catches other species. As such, costs were weighted by the proportion of Hake in the total catch. Estimates of historical catchability-at-age for each FU were obtained dividing partial fishing mortality-at-age by observed effort. In turn, the fleet’s partial fishing mortality was calculated as the overall fishing mortality multiplied by the fleet’s catch-at-age proportions.

3.1. The initial random population

The main components of the simulation model and main assumptions made in each scenario are summarised in a conceptual diagram in Fig. 2. The model was composed of several modules which interacted among themselves, the two main modules being the Operating Model (OM) and the Management Procedure Model (MPM). The OM simulated the real system and the MPM the whole management process, from data collection to management advice.

The initial fit of XSA provided annual estimates of stock numbers, fishing mortalities and catchabilities of the abundances indices used. In order to obtain an initial random population, it was assumed that the main source of uncertainty in the estimation was the uncertainty in the abundance indices. Thus, 500 parametric bootstrap samples of the abundance indices were generated (Efron and Tibshirani, 1993). Subsequently, XSA was fitted to each sample, together with the original catch-at-age matrix. Thus, 500 sets of historical estimates of numbers-at-age and fishing mortality-atage were generated, which were then used to parameterize the historical stock and the fleets. 3.1.1. The stock Historical stock numbers-at-age obtained through the replication of XSA and biological parameters, such as maturity, weight and natural mortality-at-age, was equal to those used in the assessment Working Group (ICES, 2007). 3.1.2. The fleet Overall exploitation was divided into 5 fleets: gillnetters and demersal trawlers operating in ICES Subarea VII (FU03 and FU04); gillnetters and demersal trawlers operating in ICES Divisions VIIIabd (FU13 and FU14); and an aggregated fleet, which accounted for the remaining catch, FUXX. The first four fleets account for 60% of the total catch of the stock. Such segmentation was selected based on the data availability and purpose of the study, the simulation of mesh size changes. Available data for these fleets were timeseries of effort and time-series of landings-at-length for each of the 4 fleets and discards-at-length for the trawlers, FU04 and FU14, and the aggregated fleet FUXX. The time-series were transformed into age, using an overall Age Length Key (ALK) taken from the assessment Working Group. Historical data on variable and fixed costs were available for FU14. Costs were associated to the whole activ-

3.2. The simulation model

3.2.1. The operating model The OM was the part of the model that simulated the ‘real world’, consisting of the stock, the fleet and their interaction. Evolution of the biological population was described using the exponential survival model for the existing year classes, together with a stock recruitment relationship to simulate births or recruits. Using the historical stock number estimates and SSB as the stock’s proxy, 500 different stock recruitment relationships were estimated using a Ricker stock-recruitment model,6 one for each of the iterations of the historical random population. Three parameters were obtained in each fit: the two parameters that describe the functional form of the model and an estimate of the variance around it. In the projection, recruitment was simulated by multiplying, in each year and iteration, a lognormal random error to the point estimate obtained with the stock-recruitment model. The lognormal distributions used had mean equal to 1 and the coefficient of variation (CV) varied from iteration to iteration according to the variance estimate, but in all the iterations it was close to 20%. The number of age classes was considered finite and in the plus group (age 8), all the individuals older than 8 were clustered. The instantaneous mortality rate was divided in natural and fishing mortalities. In turn, the fishing mortality was divided into the fishing mortalities exerted by each individual fleet. For all fleets, a

6 Beverton and Holt, Segmented regression and Ricker stock-recruitment relationships were initially fitted to the data and based on Akaike Information Criteria (AIC), Ricker stock-recruitment relationship was chosen to carry out the simulation.

1.0

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

qayf .Eyf

(1)

f

where Z, M, F represent total, natural and fishing mortalities, respectively, E denotes effort, q catchability and finally a, f and y subscripts indicate age, fleet and year. The fishing mortality of each individual fleet was further divided into fishing mortality associated to discards (FD ) and fishing mortality associated to landings (FL ), using a discard selectivity parameter d: Fay,f = FL,ayf + FD,ayf = (1 − dayf )qayf Eyf + dayf qayf Eyf

(2)

The catch of each fleet, by age and year, was given by Baranov catch equation, as derived from the growth equation used. Price was modelled following the approach used by (Kraak et al., 2004). It was assumed that price was formed from total international landings, hence it was the same for all the fleets:

0.6



proportion

f

Fayf = May +

0.2



0.0

Zay = May + Fay = May +

0.8

linear relationship between fishing mortality-at-age and effort was assumed. Thus:

0.4

102

2

4

age

6

8

Fig. 3. FU14’s selectivity-at-age in BC (black) and FU14130 (red) scenarios.

where 0 stands for the historic period, e for flexibility,7 P for price and L for landings. FU14’s profit was calculated by subtracting fixed and variable costs from gross revenue:

from 2004 to 2006. An exception was FU14’s catchability when the mesh size change was simulated. In the alternative case FU14130 scenario, an increase in FU14’s mesh size was simulated from 100 mm to 130 mm mesh size. Catchability can be discomposed as a product of the selectivity of the gear used, together with a parameter which incorporates the vulnerability, accessibility and availability of the fish to the fleet (Arreguín-Sánchez, 1996), i.e.:

a=A 

qa,f,ms = Sa,f,ms ra,f



Pay = Pa,0

La,y La,0

e

(3)

Pa,y La,y,f − FCf − VCf Eyf

(4)

a=0

Fixed cost, FC, together with variable cost VC (measured in cost per unit of effort) were considered constant throughout the years. As cost data was not available for the rest of the fleets, Gross revenue per unit of effort was use as a proxy of profit, i.e.:

a=A

P L a=0 a,y a,y,f

(5)

Eyf

The effort implemented by the fleets, each year, was calculated based on management advice, the TAC, and stock abundance. It was assumed that fleets fully complied with their TAC quota in terms of landings and that the discards represented a surplus of this management advice. Thus, to calculate the efforts the following system of nonlinear equations needed to be solved:

⎧  (1 − daf )qaf Ey+1,f ˇf TACy+1 = (1 − e−Za,y+1,f )Na,y+1 wa ⎪ ⎪ Za,y+1,f ⎪ ⎨

for all f,

a

 ⎪ ⎪ qaf Ey+1,f ⎪ ⎩ Za,y+1 = M +

(6)

a

(7)

where S stands for selectivity, r stands for all the factors affecting catchability that are not related to the fleet’s mesh size and ms subscript refers to the fleet’s mesh size. If catchability and selectivity-at-age were known for a certain mesh size, r could be calculated applying Eq. (7) and using it afterwards to estimate the hypothetical catchability for a mesh size for which selectivity-atage was known. Selectivity-at-length of trawlers with 100 mm and 130 mm mesh size and for Chilean Hake (Merluccius gayi gayi) were estimated by Gálvez and Rebolledo (2005). In principle, the external morphology of both Hake species does not show differences that could lead to significant variations in their selectivity. In the absence of more appropriate estimates these estimates were used to parameterize the model. These selectivity-at-length curves were transformed into age using the ALK. The selectivity-at-age curves for both mesh sizes are shown in Fig. 3. These curves together with historical catchability of trawlers with 100 mm mesh size were used to estimate a hypothetical historical catchability time-series for trawlers with 130 mm mesh size. In the projection, the average of the hypothetical catchabilities of the years 2004–2006 was used.

for all a.

f

where ˇ is the fleets’ quota share (in percentage), w is the mean weight of individuals and both parameters were considered constant over time. Landings and discards-at-age were then calculated using their respective fishing mortalities-at-age in the Baranov catch equation. 3.2.1.1. OM parameterization. Within the OM, stock numbers-atage, fleets’ efforts and prices were updated in each iteration and year, using the formulae described above. The rest of the OM’s parameters were equal to the average of their historical values,

7 Price flexibility and elasticity are used typically to define price dynamics, related to supply and demand. Price flexibility stands for: the percentage change in the price of a good, as demand increases by one percent (own price flexibility describes the percentage change in the price of a good, where the demand for only that good increases by one percent).

3.2.2. The management procedure The MPM was the part of the model that reflected the management process with the objective of providing management advice based on an ‘observed world’. In this particular case, it was divided into three sub-models: the observation model; the assessment model; and the HCR. Within the MPM, there were three different time-frames: year in which the management was undertaken, denoted as year y; year up to which observed data was available, y-1; and year for which management advice was provided, y + 1. 3.2.2.1. The observation model. The observation model linked the ‘real world’ and the ‘observed world’ where it simulated the necessary data to feed the assessment model. Within each year and iteration the model simulated a ‘catch-at-age’ matrix and a set of abundance indices and it was assumed that the biological parameters were known without error. The discards were not included in the observed data because the aim was to test the MP as it was defined in the Assessment Working Group and at that time discards

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

were not included in the assessment (ICES, 2007). Landings were assumed to be observed exactly as derived from Eq. (2), discardsat-age at fleet level were proportional to the catch-at-age, and thus the following matrix was used in the assessment:

(Lat )

⎛

⎞

=⎝

(1 − datf )Catf ⎠



a = 0, . . . , A t = 1, . . . , y − 1

f

(8) a = 0, . . . , A t = 1, . . . , y − 1

(Iat )

a = ai1 , . . . , ain t = yi1 , . . . , y − 1

= (qat Nat eεai )

a = ai1 , . . . , ain t = yi1 , . . . , y − 1

(9)

where each index I covered certain ages, ai1 to ain , and commenced in a particular year yi1 , as it occurs in reality where several indices are used in the tuning of the assessment model, each one having its own time-frame and covering some ages. N denotes real numbers at age. The error term εai followed a normal distribution, with mean equal to 0 and standard error  ai . The standard error  ai was equal to the standard error of the abundance indices’ log-residuals in the initial XSA fit. Seven abundance indices were simulated corresponding with those used in the initial XSA fit and with those used by the Assessment Working Group. 3.2.2.2. The assessment procedure. The assessment procedure consisted of an assessment model and a short-term forecast. The assessment model used was XSA (Shepherd, 1999), a tuned Virtual Population Analysis (VPA) model, which was that used by ICES to provide advice on the management of this particular stock until 2009. This procedure utilised the data obtained through the observation model, providing estimates of the number of individuals-at-age and fishing mortality-at-age for the years with observed data, 1,. . .,y − 1. In order to provide advice for year y + 1, it was necessary to project this observed population, up to the 1st January of year y + 1. Thus, the estimated population in y − 1 was projected until that date, using the same growth and catch equations used in the OM, but using an overall single fleet. In year y, it was assumed that the catch was equal to the TAC advised the previous year and that the selection pattern was equal to the average of the observed selection patterns in the previous 3 years: 1  Fˆat Sˆ a = 3 Fˆt y−1

ing mortality in year y was not 10% higher than FPA . If that happened the HCR forced a fishing mortality equal to FPA . Mathematically:

Fˆy+1 =

⎧ Fmax 0.9Fˆmax < Fˆy < 1.1Fmax . ⎪ ⎪ ⎪ ⎪ ⎨ 0.9Fˆy 1.1Fmax < Fˆy and 0.9Fˆy < FPA . ⎪ 1.1Fˆy 0.9Fmax > Fˆy . ⎪ ⎪ ⎪ ⎩ FPA

where C denotes catch and t subscript indicates year. Abundance indices were simulated assuming a linear relationship between abundance-at-age and the index, via a catchability parameter associated to the age and the index. Moreover, a multiplicative lognormal random error was added to each abundance index, which was associated to both observation and process uncertainties. Each abundance index can be expressed as a matrix:

A

(10)

t−y a=0

where Fˆat was the estimated mortality-at-age a in year t and Fˆt was the reference fishing mortality in year t (mean fishing mortality of ages 2–6). To project the recruitment geometric mean the last 20 observed recruitments were used while for the projection of the biological parameters (natural mortality, weight and maturity at age) an average of the previous 3 years was used. 3.2.2.3. The harvest control rule. The HCR was the same proposed by the EC in the consultation to the stakeholders. It was based on achieving and maintaining reference fishing mortality around the pre-specified fishing mortality, FMAX . Annual variation in fishing mortality was limited to 10%, provided that observed reference fish-

103

(11)

0.9Fˆy > FPA .

3.2.3. Uncertainties considered In the simulation several uncertainties were introduced into the system which are summarised below according to the categories defined in Francis and Shotton (1997): - Process uncertainty: This kind of uncertainty was present in the stock-recruitment relationship and the catchability. Stockrecruitment parameters varied from iteration to iteration and a random error was multiplied to the point estimates in each year and iteration. Catchability at age varied from iteration to iteration but was constant over time in the projection. This variability arose from the fact that it was estimated using the fishing mortalities obtained in the iterative fit of the XSA (Section 3.1). - Observation uncertainty: The abundance indices were simulated using a linear index and abundance relationship and a multiplicative lognormal error. This error could be associated to both observation and process uncertainty, but they are indistinguishable in the XSA fit. Landings and biological parameters were observed without error. Discards were not considered at all within the MP, so they could be considered as a source of implementation uncertainty. - Model uncertainty: It was not considered in this study but in the 2007 assessment several scenarios were simulated using different stock-recruitment relationships and it was concluded that the management procedure was robust to uncertainty in the stockrecruitment relationship (SEC, 2007b). - Estimation uncertainty: This source of uncertainty emerged naturally in the MP as a result of using XSA estimates and a short term forecast procedure to obtain the advice instead of using the real values from the OM. - Implementation uncertainty: Discards were not observed and not taken into account in the MP so they represented a surplus over TAC. - Institutional uncertainty: It was not considered. 4. Results Two scenarios were run, starting from an historical random population from 1978 to 2007, and projecting it until 2040 for 500 iterations. Both had the same settings except, for the catchability of FU14. In the first scenario, the Base Case (BC), the model’s parameterization was based on the historically observed catchability. In the second, the FU14130 scenario, a hypothetical catchability for FU14 was used which corresponded to a mesh size of 130 mm. In order to present the results, firstly, the BC is compared with the equivalent scenario produced in the biological impact assessment (SEC, 2007b), here denoted as SC07 scenario. Secondly, the BC and FU14130 scenarios are analysed and compared. The main differences in the dynamics of the BC and SC07 scenario were the inclusion of discards in the simulation of the population, together with the disaggregation of the overall fishery into 5 fleets. Table 2 lists median and CV of several stock and fleet indicators in 2007, for the BC and SC07 scenarios. The populations in SC07 and BC scenarios were generated under different assumptions; as such, the absolute values obtained in both scenarios in 2007 were different. Landings were assumed equal in both scenarios and all

104

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

Table 2 Median and coefficient of variation (CV) of the Recruitment (numbers × 1000), Spawning Stock Biomass (SSB) (tonnes), Biomass (tonnes), Landings (tonnes), Discards (tonnes), ratio between discards and landings (%), mean price (D /tonne) and profit in 2007 in SC07 and BC scenarios. Median

Recruitment SSB Biomass F Landings Discards Disc./land. Price GRPUE

CV

SC07

BC

SC07

BC

212,874 140,796 245,951 0.27 53,800 – – – –

282,197 173,794 289,673 0.24 53,800 5565 10% 3993 2263

17% 5% 4% 6% 0% – – – –

22% 17% 13% 16% 0% 15% 15% 1% 9%

the iterations. Variability was higher in the BC scenario, especially in total biomass and SSB, for which variability was 3 times higher in BC. Fig. 4 shows the projected time-series of recruitment, SSB, F and total landings, relative to 2007s values which resulted in SC07 and BC scenarios. Indicators’ confidence intervals were broader in the BC scenario for the whole time-series but, in proportion, variability increased more in SC07 scenario in SSB and fishing mortality (F). SSB increased and stabilised more rapidly in SC07 scenario, but the increment in the long-term was the same. In the SC07 scenario, the recruitment remained almost stable in the whole projection, but in the BC scenario it increased almost 20% in the first 7 years. F

decreased more in SC07 scenario and it stabilised later than in BC scenario. The general trend of total landings was similar in both scenarios, but in the BC scenario the long term yield was 56% higher. In the BC there were three iterations which collapsed (distinguished by a dashed line in the graph). This happened due to a gradual increase in the fishing mortality in those iterations, as a consequence of an observation error in F in the MP. When the high F was detected in the MP, the SSB and recruitment were already low and, as such, the MP was not able to recover the stock. In reality, this situation would not happen, as stricter actions would be taken and the fleet would even stop fishing as it would not be profitable to continue. Thus, the depletion state obtained in these simulations in the long-term is not reliable. However, it is certain that the sustainability objective of the MP failed and that the probability of reaching unsafe bioeconomic limits was greater than zero. Table 3 shows number of years and number of iterations that fall below BPA and BLIM in the 3 scenarios. In SC07 none of the iterations fall below BLIM . In the first year of the simulation, a significant number of iterations (45%) were below BPA but then the SSB increased and was maintained above the precautionary point from 2011 onwards. In BC and FU14130 scenarios 9% of the iterations were below BPA the first year of the simulation and none of them were below BLIM. In the projection the trajectories of the probability of being below BLIM or BPA was similar in both scenarios. Afterwards SSB increased and in the medium term the proportion of iterations that were below BPA decreased to less than 1%. However this proportion increased in the long term and in the last year of the simulation 1.6% of the iterations were below BPA and 1.2%

F

0

−40

50

−20

%

%

100

150

0

SSB

2010

2015

2020

2025

2030

2035

2040

2010

2015

2030

2035

2040

2030

2035

2040

60 40 −20

0

%

20

40 20 −20

0

%

2025

Landings

60

Recruitment

2020

2010

2015

2020

2025

2030

2035

2040

2010

2015

2020

2025

Fig. 4. Time-series of variation in %, with respect to the 2007 level, from 2007 to 2040, for SC07 (black) and BC (red) scenarios, of Spawning Stock Biomass (SSB), recruitment, reference fishing mortality (F) and landings. Points and triangles represent the median in SC07 and BC scenarios, respectively, and the solid lines the 5% and 95% quantiles. The dotted lines correspond to 3 iterations that collapse in BC scenario.

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

105

Table 3 Number of years in which SSB is below BLIM and BPA . And number of iterations that fall below BLIM and BPA in the years indicated. Scenario

Numb. of years

2025–2033

2034–2040

SSB < BLIM

SC07 BC FU14130

0 26 26

2007–2015 0 1 1

2016–2024 0 9 9

0 15 16

0 26 30

SSB < BPA

SC07 BC FU14130

8 34 34

141 191 186

0 21 22

2 31 37

0 45 49

Table 4 Median of the variation in % of 2007 level and the mean level in the period from 2020 to 2040 of recruitment (Rcr), Spawning Stock Biomass (SSB), biomass (Bms), reference fishing mortality (F), landings (Lns), discards (Dsc), ratio between discards and landings (D/L), mean price (Prc), and Gross per Unit of Effort (GPUE) in the BC and FU14130 scenarios.

BC FU14130

Rcr

SSB

Bms

F

Lns

16% 16%

104% 103%

75% 74%

−26% −26%

36% 40%

below BLIM . Three of the iterations that ended the simulation below BLIM correspond with those that collapsed. The trend in these iterations was similar in the BC and FU14130 scenarios but the decrease started earlier in the FU14130 scenario. The BC and FU14130 scenarios started in 2007 from the same initial conditions. In the projection, both showed similar trends at the stock and fleet aggregated level; nonetheless, there were some differences in the absolute values of some indicators (Table 4). In both scenarios, the median of fishing mortality decreased by 26% and stabilised around the target and discards decreased by 5%. SSB and total biomass were 1% higher in the BC scenario while landings were 4% higher in FU14130 scenario. Overall effort, and thus costs, decreased in both scenarios but the decrease was 3% higher in BC scenario. The overall price decreased equally in both scenarios, an 11%. The lower effort in BC compensated the loss in landings and resulting overall profit was 6% higher in the BC. In Figs. 5–7, the interannual variability in landings, effort and profits, respectively for BC and FU14130 scenarios are shown. There was equal variability in landings for all the fleets due to the con-

0 −10

%

1

0

Landings (Interannual Variability)

2010

2015

2020

2025

2030

2035

2040

Fig. 5. Interannual variability in landings from 2008 to 2040. Points and triangles represent the median in BC (black) and FU14130 (red) scenarios, respectively, and the solid lines the 5%, 25%, 75% and 95% quantiles. The horizontal dashed lines indicated 15%, 0% and −15% variability limits.

Dsc −5% −5%

D/L

Prc

−31% −33%

−11% −11%

Effort

GPUE

30% 28%

64% 58%

stant quota share over time There were slight differences between both scenarios. Landings, in median, decreased during 3 consecutive years at the start of the simulation, but the accumulated decrease was below 15%. Afterwards they started to increase year by year until after 2020 when they stabilised. The 5% quantile was below the −15% limit only in the first year of the simulation. In turn, the 95% quantile stayed above the 15% limit several years into the simulation. Both, 25% and 75% quantiles oscillated around −6% and 6%, respectively from 2025 onwards and the 5% and 95% quantiles around −12% and 14%, respectively. Effort variation depended on the fleets and the differences among scenarios were only significant for FU14. This fleet, as expected, was highly penalized with the increase in its mesh size. The reduction of 10% in F in the initial years of the projection did not correspond with a reduction of 10% in the effort of all the fleets, in any of the scenarios. All the fleets, except FU14 in FU14130 scenario, bore a decrease in median effort during the 5–6 years at the beginning of the simulation. The accumulated decrease in median effort oscillated between −22% in FU14 and −32% in FUXX. In some years the 5% and 95% quantiles surpassed the – 15% and 15% limits, especially at the start of the simulation and the FU03, FUXX and FU14 fleets. In the FU14130 scenario FU14 bore a sharp (12% in median) increase in effort in the first year of the simulation with only the 5% quantile below 0%. In the following years the variation was almost identical to that of the BC. In the short and long-term the 25% and 75% quantiles stabilised around the −6% and 6% quantiles, respectively. The profits were highly driven by the fleets’ efforts and their trend was opposite to the trends in efforts (Figs. 6 and 7). A high increase in profit was observed in FU03 in the fourth year of the simulation in both scenarios and a high decrease in FU14 in the first year of the simulation in the FU14130 scenario. In general, the profits were more stable than the efforts and the landings. Moreover, after 2020, the 5% and 95% quantiles were inside the −10% and 10% limits, respectively for all the fleets and the 25% and 75% quantiles fluctuated around the −4% and 4% limits. In Table 5, the variations with respect to 2007 of several indicators at fleet level are shown for the BC and FU14130 scenarios. Effort was lower in the BC for all the fleets, but the highest profit was not obtained in this scenario for all them. The gillnetters (FU03 and FU13) and FU14 obtained highest profits under BC. By contrast, FU14 and FUXX obtained similar profits in both scenarios. However the profits of all the fleets were higher than in 2007 in both scenarios. The difference in effort between both scenarios was lower or equal to 2% for all the fleets except for FU14 for which the difference was equal to 17%. The landings increased in both scenarios in

106

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

Effort (Interannual variability)

10 −30

−10

% −30

−10

%

10

30

FU04

30

FU03

2010

2020

2030

2040

2010

2020

2040

2030

2040

10 −30

−10

% −30

−10

%

10

30

FUXX

30

FU13

2030

2010

2020

2030

2040

2010

2020

−30

−10

%

10

30

FU14

2010

2015

2020

2025

2030

2035

2040

Fig. 6. Interannual variability in effort by fleet from 2008 to 2040. Points and triangles represent the median in BC (black) and FU14130 (red) scenarios, respectively, and the solid lines the 5%, 25%, 75% and 95% quantiles. The horizontal dashed lines indicated 15%, 0% and −15% variability limits.

the long term, but the increase was 4% higher in the FU14130 scenario. The increase in landings was equal in all the fleets because the quota share was constant for the whole projection. With the increase in its mesh size the discards for FU14 decreased by 10%. Nevertheless, this decrease was compensated for by the increase in discards of FU04 and FUXX. The mean price per kilogram in FU14 was 2% higher in FU14130 but was equal or lower (by 1%) for the remainder of the fleets.

5. Discussion Inclusion of discards, in the generation of the historical population, provided a different perception of the dynamics of the system and the performance of the MP. According to the median values for the population indicators, the MP candidate for long-term management of the stock led it far from over exploitation. Furthermore, discards decreased and profits increased for all the fleets, while

Table 5 Median of the variation in % of 2007 level and the mean level in the period from 2020 to 2040 of effort, landings, discards, price and profit for all the fleets in the BC and FU14130 scenarios. Profit corresponds with real profit in FU14 and with gross per unit of effort, used as profits’ proxy, in the rest of the fleets. Fleet

FU03 FU04 FU13 FU14 FUXX

Effort

Landings

BC

FU14130

−46% −25% −37% −13% −28%

−44% −23% −35% 4% −27%

BC

36%

Discards FU14130

40%

Price

BC

FU14130

– 1% – 13% −6%

– 3% – 3% −5%

BC −18% −8% −13% −3% −11%

Profit FU14130

BC

FU14130

−18% −8% −14% −1% −12%

115% 51% 74% 44% 57%

109% 51% 72% 22% 57%

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

107

Benefit (Interannual variability)

10 −30

−10

% −30

−10

%

10

30

FU04

30

FU03

2010

2020

2030

2040

2010

2020

2040

2030

2040

10 −30

−10

% −30

−10

%

10

30

FUXX

30

FU13

2030

2010

2020

2030

2040

2010

2020

−30

−10

%

10

30

FU14

2010

2015

2020

2025

2030

2035

2040

Fig. 7. Interannual variability in benefit by fleet from 2008 to 2040. Points and triangles represent the median in BC (black) and FU14130 (red) scenarios, respectively, and the solid lines the 5%, 25%, 75% and 95% quantiles. The horizontal dashed lines indicated 15%, 0% and −15% variability limits.

their effort decreased. However, the management strategy failed to maintain the stock above BLIM in 3 out of 500 iterations, contrary to what was observed in 2007. The increase in the mesh size of FU14 did not produce any change in fishing mortality and discards. However, total biomass and SSB were lower and landings higher; thus the stock was more productive in terms of landings, but less in terms of biomass. Under the present mesh size configuration, although the landings were lower, the overall GRPUE was higher. None of the fleets benefited from the increase in the mesh size of FU14. In fact, the gillnetters, FU03 and FU13, and the FU14 itself suffer a large decrease in benefits compared to the status quo situation. The increase in mesh size reduced significantly the discards of FU14, but the decrease was compensated with the increase of discards in FU04 and FUXX. It is not straightforward or even possible to isolate the effect of each individual parameter in relation to the increasing risk of collapse shown in this kind of nonlinear and complex system. In any case, a comparison between real and observed populations obtained year-by-year indicated that an underestimation of real fishing mortality within the MP was responsible for the collapse. In the MP, fishing mortality was underestimated sys-

tematically during several years, which provoked a high fishing mortality in the real population. This in turn caused a high fall in SSB and, thus, in recruitment. Finally, when within the MP a high level of fishing mortality and a low level of SSB was observed the HCR was not restrictive enough (note that the minimum F that could be advised when F was 10% higher than FPA was FPA itself) and the stock collapsed. Inclusion of discards in the real population, but not in assessment, and (to a minor degree) the evolution in time of the overall selection pattern, skew the perception of the stock and result in the failure of the management strategy. Fleets’ selection pattern in the projection was constant, but overall selection pattern changed because the relative contribution to the catch of each age did not change equally by fleet. The results at fleet level, obtained when the mesh size of FU14 was increased, were explained by the exploitation patterns of the fleets. The overlap of fleets’ exploitation patterns produced competition among fleets for the same fraction of the population in terms of age composition of the catch. This competition was reflected in the fleets’ economic performance. The standardised catchabilities of all the fleets in the BC and that of FU14 in the FU14130 scenario

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

5

108

FU04 FU13 FUXX

0

1

2

3

4

FU14 − BC FU14 − 130 FU03

0

2

4

6

8

Fig. 8. Standardised catchability at age by fleet. Black solid line corresponds with the catchability of FU14 in the BC and the black dotted line with its catchability in FU14130 scenarios. The rest of the fleets have the same catchability in both scenarios, solid circles correspond with FU03, empty circles with FU04, solid triangles with FU14 and empty triangles with FUXX.

are shown in Fig. 8. Catchability of FU14 moved towards the oldest ages when mesh size was increased. Thus, the competition for the same fraction of the stock between FU14 and FU04 and FUXX decreased and that between FU14 and the gillnetters, FU03 and FU13, increased. The effect of this competition was reflected, for instance, in fleets’ profits and discards. While the gillnetters lost, in economic terms, with the increase in the mesh size of FU14, the FU04 and FUXX did not experience any significant change. Further, the decrease in FU14’s discards was compensated for by discards increase in FU04 and FUXX. Paradoxically, the fleets that do not discard and have a ‘better’ selection pattern were those negatively affected in relation to their profits. The conclusions extracted from this study contrast to the classical theory of fisheries management, which recommend increases in the selection pattern in order to have a sustainable and productive stock (Armstrong et al., 1990; Froese et al., 2008; Myers and Mertz, 1998; Quinn and Deriso, 1999; Scott and Sampson, 2011). This theory focuses only upon stock performance and, as shown by other authors, its effect at fleet level could be negative with the effect on the stock marginal (Heikinheimo et al., 2006; Kuikka et al., 1996; Pascoe and Revill, 2004; Tschernij et al., 2004). This was the case in our study where, despite yield being higher when the selection pattern was increased, economic profitability was lower. Furthermore, biomass and SSB were also lower. The results obtained were dependent upon the parameterization used and assumptions made throughout the simulation. The conditioning of the model was based mainly upon the data available to, together with the assumptions made by, the Assessment Working Group of Hake which can be considered as the best source of knowledge about the dynamics of the stock. However, within the context of this study, four main limitations can be identified: the non-inclusion of other species in the analysis; the parameterization of growth; the parameterization of discards; and the fishery’s parameterization and segmentation. Many fleets targeting Hake catch other species simultaneously especially trawlers which catch a large number of different species. Within the context of EBFM, a complete assessment of the impact of technical changes in such a fleet should consider the effect on other target and non-target species exploited by it. However, inclusion of other stocks in the same detail as that undertaken for Hake complicates the simulation algorithm enormously. Even having the

appropriate algorithm, necessary data to parameterize it could not be available for many stocks. Age determination is a well known problem in European Hake ˜ (de Pontual et al., 2006; Pineiro et al., 2007). In this work we parameterized the model according to the slow growth assumption, a pattern used in the Assessment Working Group at the time of writing this work. Bertignac and de Pontual (2007) showed that trends in stock indicators and stock status with respect to precautionary reference points were almost the same. Thus, it could be expected that running the simulations with the faster growth parameterization would lead to similar results in relative terms. Discards play an important role in the model; however, the data used to parameterize them is not considered reliable by the Assessment Working Group (ICES, 2007). Available discards data relate to the trawlers which discard mainly undersize fish. However, it is known that gillnetters (FU03 and FU13) discard individuals impaired during the fishing operation, but this has never been quantified and therefore was not considered in this study. Introducing uncertainty in the estimates used could, in part, solve the problem. Due to a lack of data and knowledge on discards it was decided to avoid overuse of uncertainty, as postulating a probability distribution for them could be even worse than using a point estimate (Rochet and Rice, 2009). Regarding the fleets’ segmentation, the aggregated fleet FUXX combines two important fleets: the longliners, which catch large Hake individuals; and the Nephrops trawlers, which catch small Hake individuals and discard most of them. These two fleets have very different selection patterns, which could have skewed the results. Gross revenue per unit of effort was used as a proxy of profit for all the fleets except FU14. A rigorous economic impact assessment of the LTMP should include the true costs of all the fleets. Regarding selectivity, the curves used to parameterize the catchability of FU14 correspond with those of other Hake species. However, the external morphology of both Hake species does not show differences that could lead to significant variances in their selectivity. Finally, and related to selectivity, catchability in the projection has been considered constant while it is known that changes happen over time. In particular, an increase in catchability occurred in FU14 when its mesh size was increased to attenuate the increase in effort. However there was no basis to parameterize these possible changes so constant catchability was used. This study reflects the importance of including discards and fleets in the evaluation of management plans. The assumption of perfect knowledge in the MP skews the perception about the stock status and, as a consequence, the management strategy fails. We conclude that under current mesh size configuration, the LTMP will drive, with high probability, the stock to healthy and sustainable levels of biomass and at the same time will increase the profitability of all the fleets. It maintained the landings, the effort and, the profits quite stable in the medium and long term. The LTMP will produce, by itself, a reduction in discards. However, according to the results obtained, increasing the mesh size of FU14 would not lead to an additional reduction in discards, an increase in biomass or to an improvement in fleets’ profitability. The results obtained are dependent on the assumptions made and the parameterization used, but the comparison of both scenarios should, in relative terms, not be dramatically affected by them. The objective of increasing FU14’s mesh size was threefold, to increase the overall selection pattern of the fishery according to manager’s suggestion, to harmonize the mesh size of trawlers in areas VII and VIIIabd, to ease the fishing among areas and to decrease discards. Following the same argument the mesh size of FU13 could be increased and a simulation of the effect of the increase in the mesh size of both fleets could be carried out. With the increase in the mesh size of FU13 no variation in discards will be expected because it has been assumed that this fleet does not discard. But

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

the profitability of the fleets could change and also the biomass levels. The algorithm presented permits integrated bioeconomic impact assessments to be performed of management plans for stocks and fleets with similar characteristics to those considered here. The structure of the approach is flexible and a wide range of scenarios can be simulated. For instance, simulations performed using different selectivity scenarios for different fleets were presented in 2008 to the stakeholders in the NSWWRACs, where additional mesh size configurations were proposed by those present and then simulated by the model. Thus, it created a mechanism for interaction between scientists and stakeholders. In 2010 the Hake Assessment Working Group changed the method used to assess the stock from XSA to Stock Synthesis SS3 (Methot, 1990, 2000; ICES, 2010). The perception of the stock is now different and the previous biological reference points are no longer valid. Thus, a new impact assessment of LTMPs for this stock is expected. At this stage, it is not possible to integrate SS3 within the MSE algorithm but it could be used to parameterize the OM and then introduce estimation error implicitly based, for example, on retrospective patterns. Nevertheless, a variant of the simulation model presented here could be used to evaluate the performance of the stock and the fleets in an integrated way and to test the implications of changing the mesh size of the fleets. In order to have a complete evaluation of the effect on the ecosystem of changing exploitation patterns, future studies should try to incorporate the effect on accompanying species in the analysis. Further, regarding fleet segmentation, the aggregated fleet should be decomposed into (at least) 3 fleets, the longliners, the Nephrops trawlers and the remainder. This disaggregation will permit an assessment to be undertaken of the effect on Hake of the selectivity devices for Nephrops trawlers tested by Macher et al. (2008). This could be of great interest as discards of Hake juveniles are extensively generated by this fleet. Acknowledgments This work has been funded through the Basque Country Government (Agriculture and Fisheries Department) and several FP6 EU funded projects (EFIMAS, Operational Evaluation Tools for Fisheries Management Options, contract no 502516, COMMIT, Creation of Multi-annual Management Plans for Commitment, contract no 502289 and CEVIS, Comparative Evaluations of Innovative Solutions in European fisheries management, contract no. 022686). We ˜ would also like to acknowledge the work done by Inaki Quincoces (Azti-Tecnalia) in preparing the computer grid for running the simulations, Michel Bertignac (IFREMER) and Nelida Perez (IEO) for providing data, Richard Curtin and Ane Iriondo (Azti-Tecnalia) for supervising the English and creating Fig. 2, respectively. Finally we want to thank the anonymous reviewers for useful comments that helped to improve the paper. The paper is contribution no. 535 from Azti-Tecnalia (Marine Research Unit). References Armstrong, D.W., Ferro, R.S.T., MacLennan, D.N., Reeves, S.A., 1990. Gear selectivity and the conservation of fish. J. Fish Biol. 37 (Suppl. A), 261–1261. Arreguín-Sánchez, F., 1996. Catchability: a key parameter for fish stock assessment. Rev. Fish Biol. Fish. 6, 221–242. Bertignac, M., de Pontual, H., 2007. Consequences of bias in age estimation on assessment of the northern stock of European hake (Merluccius merluccius) and on management advice. ICES J. Mar. Sci. 64 (5), 981–988. Butterworth, D.S., 2007. Why a management procedure approach? Some positives and negatives. ICES J. Mar. Sci. 64 (4), 613–617. Butterworth, D.S., Punt, A.E., 1999. Experiences in the evaluation and implementation of management procedures. ICES J. Mar. Sci. 56 (6), 985–998. Casey, J., Pereiro, F.J., 1995. European Hake (Merluccius merluccius) in the North-east Atlantic. In: Alheit, J., Pitcher, T.J. (Eds.), Hake: Biology, Fisheries and Markets. Chapman & Hall, London, pp. 125–147.

109

De la Mare, W.K., 1998. Tidier fisheries management requires a new MOP (management-oriented paradigm). Rev. Fish Biol. Fish. 8, 349–356. de Pontual, H., Groison, A.L., Pineiro, C., Bertignac, M., 2006. Evidence of underestimation of European hake growth in the Bay of Biscay, and its relationship with bias in the agreed method of age estimation. ICES J. Mar. Sci. 63 (9), 1674–1681. Efron, B., Tibshirani, R.J., 1993. An Introduction to the Bootstrap. Chapman & Hall/CRC, Boca Raton, FL. Francis, R.I.C.C., Shotton, R., 1997. “Risk” in fisheries management: a review. Can. J. Fish. Aquat. Sci. 54, 1699–1715. Froese, R., Stern-Pilot, A., Winker, H., Gascuel, D., 2008. Size matters: how singlespecies management can contribute to ecosystem-based fisheries management. Fish. Res. 92, 231–241. Gálvez, M., Rebolledo, H., 2005. Estimating codend size selectivity of bottom trawlnet in Chilean hake (Merluccius gayi gayi) fishery. Invest. Mar., Valparaíso 33 (2), 151–165. Hauge, K.H., Nielsen, K.N., Korsbrekke, K., 2007. Limits to transparency exploring conceptual and operational aspects of the ICES framework for providing precautionary fisheries management advice. ICES J. Mar. Sci. 64 (4), 738–743. Heikinheimo, O., Setälä, J., Saarni, K., Raitaniemi, J., 2006. Impacts of mesh-size regulation of gillnets on the pikeperch fisheries in the Archipelago Sea, Finland. Fish. Res. 77 (2), 192–199. ICES, 2007. Report of the Working Group on the assessment of southern shelf stocks of Hake, Monk and Megrim (WGHMM), Vigo, Spain. ICES, 2010. Report of the Working Group on the assessment of southern shelf stocks of Hake, Monk and Megrim (WGHMM), Bilbao, Spain. Kell, L.T., Mosqueira, I., Grosjean, P., Fromentin, J.-M., Garcia, D., Hillary, R., Jardim, E., Mardle, S., Pastoors, M.A., Poos, J.J., Scott, F., Scott, R.D., 2007. FLR: an open-source framework for the evaluation and development of management strategies. ICES J. Mar. Sci. 64 (4), 640–646. Kell, L.T., Pilling, G.M., Kirkwood, G.P., Pastoors, M., Mesnil, B., Korsbrekke, K., Abaunza, P., Aps, R., Biseau, A., Kunzlik, P., Needle, C., Roel, B.A., Ulrich-Rescan, C., 2005. An evaluation of the implicit management procedure used for some ICES roundfish stocks. ICES J. Mar. Sci. 62 (4), 750–759. Kraak, S.B.M., Buisman, F.C., Dickey-Collas, M., Poos J.J., Pastoors, M.A., Smit, J.G.P., and Daan, N. 2004. How can we manage mixed fisheries? A simulation study of the effect of management choices on the sustainability and economic performance of a mixed fishery. ICES CM F: 11. Kuikka, S., Suuronen, P., Parmanne, R., 1996. The impacts of increased codend mesh size on the northern Baltic herring fishery: ecosystem and market uncertainties. ICES J. Mar. Sci. 53 (4), 723–730. Macher, C., Guyader, O., Talidec, C., Bertignac, M., 2008. A cost-benefit analysis of improving trawl selectivity in the case of discards: the Nephrops Norvegicus fishery in the Bay of Biscay. Fish. Res. 92 (1), 76–89. Methot, R.D., 1990. Synthesis model: an adaptable framework for analysis of diverse stock assessment data. Int. North Pac. Fish. Comm. Bull. 50, 259–277. Methot, R.D., 2000. Technical description of the stock synthesis assessment program. U.S. Dept. Commer., NOAA Tech. Memo, NMFS-NWFSC-43, p. 46. Murua, H., 2010. The biology and fisheries of European Hake, Merluccius merluccius, in the North-East Atlantic. Adv. Mar. Biol., 97–154. Murua, H., Quincoces, I., Garcia, D., Korta, M., 2010. Is the northern European hake, Merluccius merluccius, management procedure robust to the exclusion of reproductive dynamics? Method development and evaluation of stock reproductive potential of marine fish. Fish. Res. 104 (1–3), 123–135. Myers, R.A., Mertz, G., 1998. The limits of exploitation: a precautionary approach. Ecol. Appl. 8 (1), 165. Pascoe, S., Revill, A., 2004. Costs and benefits of bycatch reduction devices in European brown shrimp trawl fisheries. Environ. Resour. Econ. 27 (1), 43–64. ˜ ˜ R., 2007. Tag and recapture of European Pineiro, C., Rey, J., de Pontual, H., Goni, hake (Merluccius merluccius L.) off the Northwest Iberian Peninsula: first results support fast growth hypothesis. Fish. Res. 88 (1–3), 150–154. Punt, A.E., Donovan, G.P., 2007. Developing management procedures that are robust to uncertainty: lessons from the International Whaling Commission. ICES J. Mar. Sci. 64 (4), 603–612. Quinn, T.J.I., Deriso, R.B., 1999. Quantitative Fish Dynamics. Oxford University Press. R Development Core Team, 2009. R: A Language and Environment for Statistical Computing. Vienna, Austria. http://www.r-project.org/. Rademeyer, R.A., Plaganyi, E.E., Butterworth, D.S., 2007. Tips and tricks in designing management procedures. ICES J. Mar. Sci. 64 (4), 618–625. Rochet, M.-J., Benoît, E., Collie, J.S., 2009. Is selective fishing more harmful to marine communities than even exploitation? Theoretical investigations. ICES CM M, 07. Rochet, M.-J., Rice, J.C., 2009. Simulation-based management strategy evaluation: ignorance disguised as mathematics? ICES J. Mar. Sci. 66 (4), 754–762. Scott, R.D., Sampson, D.B., 2011. The sensitivity of long-term yield targets to changes in fishery age-selectivity. Mar. Policy 35 (1), 79–84. SEC, 2007a. Impact assessment of long-term management plans for northern hake (SGBRE-05-07). SEC, 2007b. Northern hake long-term management plans (SGBRE-07-03). Shepherd, J.G., 1999. Extended survivors analysis: an improved method for the analysis of catch-at-age data and abundance indices. ICES J. Mar. Sci. 56, 584–591. Suuronen, P., Sarda, F., 2007. The role of technical measures in European fisheries management and how to make them work better. ICES J. Mar. Sci., fsm049. Thunberg, E.M., Helser, T.E., Mayo, R.K., 1998. Bioeconomic analysis of alternative selection patterns in the united states Atlantic silver hake fishery. Marine Resources Foundation. Mar. Resour. Econ. 13 (1).

110

D. Garcia et al. / Fisheries Research 110 (2011) 98–110

Tschernij, V., Suuronen, P., Jounela, P., 2004. A modelling approach for assessing short-term catch losses as a consequence of a mesh size increase. Fish. Res. 69 (3), 399–406. WSSD, 2002. Report of the World Summit on Sustainable Development, Johannesburg, South Africa.

Zhou, S., Smith, A.D.M., Punt, A.E., Richardson, A.J., Gibbs, M., Fulton, E.A., Pascoe, S., Bulman, C., Bayliss, P., Sainsbury, K., 2010. Ecosystem-based fisheries management requires a change to the selective fishing philosophy. Proc. Natl. Acad. Sci. 107 (21), 9485–9489.