A variety of terms are used in the literature related to wastage in fisheries. Unwanted catches can be understood as incidental catch of organisms that cannot be marketed because they have little or no economic value, or because of legal requirements (Maeda et al. 2017). Discards refer to the organisms of both commercial and non-commercial value that are caught during commercial fishing operations and returned to the sea, often dead or dying (Feekings et al. 2012).

Unwanted catches have been regarded as a key issue in commercial fishing worldwide (Kelleher 2005). Most unwanted catches are discarded and returned to the sea with a low survival rate, especially in the case of fish (Revill 2012, Guillen et al. 2014). Discards are influenced by a variety of factors that differ for each species and portion of discards (Feekings et al. 2012). For most species there is greater variability in discard rates across regions than across fisheries, suggesting that a region-by-region approach to discard reduction would be more meaningful (Uhlmann et al. 2014).

In the European Union (EU), discard levels vary considerably from one location and gear to another. For instance, for the Galician fleet discard rates range from insignificant (coastal trolling fleet) to 43.5% (offshore trawling fleet) of total catches (Vázquez-Rowe et al. 2011). The Nephrops trawl fishery in the Bay of Biscay discards about half of its catches in numbers, of which only 30% survive (Macher et al 2008). Tsagarakis et al. (2014) found that discards account, on average, for 18.6% of total catches in the Mediterranean Sea and are concentrated mainly in the bottom and shrimp trawls, despite their relatively low contribution to catches in weight terms. In addition, other factors such as water depth (Sánchez et al. 2004) and fishing intensity (Sánchez et al. 2007) also affect discards in the Mediterranean Sea.

All these concerns about discards were taken into account in the latest reform of the EU Common Fisheries Policy (EU 2013, Article 5), which includes a discard ban called the landing obligation (LO), which requires all catches of stocks subject to a total allowable catch regulation to be kept on board, landed and counted against quotas. This new regulation also states that fish under the legal minimum landing size only can be sold for fishmeal or other products not destined for direct human consumption. Some exemptions are established for protected species, for species with a high survivability and for discards that cannot be easily reduced through selectivity and avoidance measures (de minimis exemptions).

Although the LO will not be fully in force in all EU waters until 2019, its effects have already been studied in light of a bioeconomic modelling framework. For instance, Batsleer et al. (2016) applied the dynamic state variable model proposed in Clark and Mangel (2000) and state that restrictive quotas do not necessarily lead to a reduction in discards when a discard ban is not properly enforced. Prellezo et al. (2016) and Prellezo et al. (2017) use the FLBEIA simulation bioeconomic model (García et al. 2013, 2016) to estimate the effects of the LO on the Basque trawling fleet operating in the Bay of Biscay. Their results are mixed. The LO will probably have a negative short-term effect on the economic performance of the fleet, but at the same time there are likely to be private incentives to improve selectivity to reduce discards. Villasante et al. (2016a) used the stochastic age-structured optimization model developed in Da Rocha et al. (2012a, 2013) to investigate the impact of the LO in the Galician multispecies small-scale gillnet fishery. They found that the LO would result in short- and long-term losses of fishing days and yields. The future yield (catches) under the LO would be only 50% of catches expected without the LO. In an experimental study, Mortensen et al. (2017) show that relaxing technical regulations combined with proper incentives may help cope with the LO, reducing unwanted catches to some extent without negative effects on economic viability.

Assessing policies that seek to mitigate unwanted catches may help us understand what factors can contribute to the success of the LO. Various policies aim to reduce unwanted catches. Some focus on promoting the adoption of fishing technologies that improve age selectivity and effectiveness in harvesting target species (Catchpole et al. 2006, Kraak et al. 2013). Others focus more on market control by banning the sale or trade of unwanted catches, especially for pelagic species that are taken incidentally as by-catch (Tolotti et al. 2015). In fact, the LO policy promoted by the EU can be seen as a set of new rules that combine measures of this type. On the one hand, the LO compels regulated fisheries to land unwanted catches that could be reduced by using more selective technologies. On the other hand, under the LO catches below the minimum landing size cannot be sold for direct human consumption but can be sold for other uses. This possibility of trading undersized fish for other uses than human food may indirectly stimulate the black market of undersized fish for human food because the price for fish as reduction material is much lower than the price for human consumption (Catchpole et al. 2017). Stakeholders have also expressed alarm, fearing that the implementation of the LO could give rise to a black market for juveniles, thus neutralizing all the efforts made so far by the administration to address this problem (de Vos et al. 2016).

In this article, we study the impacts of two measures aimed at reducing unwanted catches. First, we analyse the impact of age selectivity improvements on catches and discards. Second, we analyse the effects of imposing a ban on sale for different age ranges. To this end, we use a stochastic age-structured optimization model (SASOM). In particular, we compute reference points for a mixed fishery under several scenarios, representing the two measures as the steady-state solutions of a dynamic optimal management problem in which the present value of an economic indicator is maximized (Da Rocha et al. 2010, 2016). Following Da Rocha et al. (2012b), the optimization problem takes into account that species are caught simultaneously, in unselective fishing. The model is applied to the southern Iberian hake stock (SIHS) fishery and the analysis focuses on the effects of these measures on the yield, the stock and the discard rate.

A SASOM approach was used to assess the effects of mitigating the unwanted catches policy. The key aspects of the models that follow a SASOM approach are the following (Fig. 1):

– Optimality: solutions derived from SASOM always represent the optimal responses of agents given the economic and biological settings of the fishery.
– Age-Structured populations: the resource is structured in cohorts, i.e. in groups of fish that have the same age; in most cases, the length of the fish can be used as an indirect indicator of age (Cotter and Pilling 2007).
– Stochasticity: sources of uncertainty generate shocks affecting economic and/or biological aspects of the fishery.

The bioeconomic model that we used is based on the multispecies setting developed in Da Rocha et al. (2012b) which captures the age-structure and the optimization elements of the SASOM approach and can be extended to include the stochastic element by considering the main sources of uncertainty of the fishery. Figure 2 represents the logic of the model. It comprises two main model-boxes. One of them represents the biological aspects of the fishery, and has inputs such as recruitments and parameters representing selectivity, weights, maturity levels and natural mortality levels. The other box represents the economic model, which usually means the managers’ decision problem based on economic elements such as the cost and demand functions and the discount rate when the decision is a long-term one. Other social constrains such as the preservation of jobs in fleets can also be considered by the economic model. The managers’ decision problem is solved by taking into account the biological model, and as result optimal reference points (fishing mortality levels) are obtained. This optimal decision enables the stock to be evaluated in terms of economic indicators such as the present value of the yield.

The biological part of the model was represented as the standard multispecies age-structured model used for stock assessment. It was assumed that there were S species in the fishery. The stock of species s=1,…,S was broken down into A_s cohorts. i.e. in each period there were A_s–1 initial old cohorts for species s and a new cohort was born.

Let Z_s,a,t be the mortality rate that affects the population of species s at age a in period t. This mortality rate was decomposed into fishing mortality, F_s,a,t and natural mortality, M_s,a, so Z_s,a,t=F_s,a,t+M_s,a. While the fishing mortality rate varied from one period and one age to another, we assumed that natural mortality was constant over periods as the assessment groups tend to do (e.g. ICES 2015, 2016).

We also assumed that all S species were fished simultaneously in relatively unselective fishing operations, representing the catchability by q_s, and that the landing and discard selection patterns of each cohort of each species, $\overline{p}$ _s,a and $\overline{d}$ _s,a were constant. Therefore, for each unit of effort, E_t, the fishing mortality for each age and species was given as F_s,a,t=($\overline{p}$_s,a+$\overline{d}$_s,a)q_sE_t. This can be understood as a multiproduct technology in which different goods (fish of different ages and species) were caught with the same input (fishing effort, E_t) in different fixed proportions represented by ($\overline{p}$_s,a+ $\overline{d}$_s,a)q_s. This fixed proportion approach (Leontief 1944) enabled the fishing mortality multiplier to be defined with no loss of generality as effort F_t=E_t, and the original landing and discard selection pattern, p_s,a=$\overline{p}$_s,a q_s and d_s,a=$\overline{d}$_s,a q_s, could be rescaled to express the fishing mortality for each age and species as F_s,a,t=($\overline{p}$_s,a+$\overline{d}$_s,a)q_sE_t=(p_s,a+d_s,a)F_t.

The population of species was assumed to decrease at an exponential rate in accordance with the mortality rate, Z_s,a,t. Formally, N_s,a+1,t+1=e^–Zs,a,t N_s,a,t where N_s,a,t represented the abundance of species s for age a at the beginning of time t. Notice that by backward substitution N_s,a,t could be expressed as a function of recruitment (see Da Rocha et al. 2012b for more details).

Dead fish were classified as dying from natural causes (M), discards (D), or valid catches for sale (C). So the dynamics of the fishery for any age and species could be expressed at any time as N_s,a,t–N_s,a+1,t+1=M_s,a,t +D_s,a,t +C_s,a,t, where

The above expression represents Baranov’s equation (Baranov 1918) for valid captures from the point of view of sale without considering discards. We refer to the yield of species s as the sum of the weights of all valid captures of all its ages,

where ω_s,a represents the individual fish weight for age and species s. Notice that discard patterns appear as an element of the mortality rate per age and species, Z_s,a,t, so they influence the fraction of the mortality that determines the yield for sale, p_s,a F_t /Z_s,a,t.

The size of a new cohort (recruitment) of species was determined by the stock-recruitment (S-R) relationship proposed by Shepherd (1982),

where

is the spawning-stock biomass, which depended on the individual fish weight for any age and species, ω_s,a, and the maturity fraction, µ_s,a; α_s, b_s and K_s are parameters with specific values for species s; α_s represents the slope at the origin of the S-R curve of species s, which is a measure of the maximum recruitment-per-unit-biomass attainable only at low stock sizes where density-dependent mortality (of pre-recruits) is presumably least; K_s measures a threshold of biomass of species s below which the population becomes increasingly vulnerable and the likelihood of collapse is increased; and b_s represents the degree of compensation that measures the power of the density-dependent effects to compensate for changes of stock size, particularly at large stock sizes (Shepherd 1982).

In this study, the economic model represented in Figure 1 was synthesized in a management tool that was used to find the reference points (fishing mortalities) that maximized the present value of the total yield of the fishery without considering discards. Formally, the present value of the total yield was defined as the discounted sum of the valid captures, in weight, of all species, that is

where 0≤β≤1 was the discount factor which represented the willingness of the manager (or society) to trade-off the value of fishing at the current moment against the benefits of increased yields in the future, measured by higher biomass and recruitment. The optimal reference point emerged as the steady-state solution of this dynamic problem. Da Rocha et al. (2012b) proved that this steady-state solution is just a generalization of F_msy.

The model presented in previous Section is applied to the SIHS in order to study the impacts of measures aimed at reducing unwanted catches in that fishery.

When the model presented in Section 2 had been parameterized for the SIHS, it was applied to study the impacts of two measures intended to reduce unwanted catches. Both measures were analysed first under the ceteris paribus criteria, and then the results of combining the two measures were studied.

The reform of the Common Fisheries Policy of 2013 was aimed at gradually eliminating the wasteful practice of discarding through the introduction of the LO. This radical change in fisheries management aims to improve fishing behaviour through improvements in selectivity and other measures. The LO requires all the catches of stocks subject to a total allowable catch regulation to be kept on board, landed and counted against quotas. Moreover, undersized fish cannot be marketed for direct human consumption purposes, while prohibited species (e.g. basking shark) cannot be retained on board and must be returned to the sea.

We implemented two measures for reducing unwanted catches in the SIHS fishery. First, we studied the impact on the fishery when age selectivity is improved. Second, we studied the effects of imposing a ban on the sale of juveniles. Both measures are in line with the European legislation that explicitly considers pilot projects on gears to increase selectivity and minimum conservation reference sizes as measures to be taken into account for the conservation and sustainable exploitation of marine resources (EU 2013, Article 7). More particularly, the LO considers that minimum conservation reference sizes may be established to protect juveniles of marine organisms (EU 2013, Article 15, point 10). Furthermore, the EU has allocated structural funds, such as the European Maritime and Fisheries Fund (EMFF, EU 2014), which can be used to support small-scale coastal fishermen to finance up to 80% of the cost of investment in new gear to improve selectivity and minimize unwanted catches (Directorate-General for Maritime Affairs and Fisheries (EC) 2017).

With respect to the first measure, our results show that reducing the selectivity parameters by 90% for the three early ages leads to the highest improvements in terms of yield of hake, total yield and hake discards compared with the status quo scenario. In particular, hake yield increased almost six-fold and discards of hake fell more than 20 percentage points in comparison with the status quo.

However, our study also shows that only when age selectivity changes significantly are the results positive in terms of increased yield and reduced discards. This finding is consistent with the results of Mytilineou et al. (1998) and Prellezo et al. (2017), which show that large increases in the minimum mesh size are necessary to achieve significant impacts on biological and/or economic variables. This result may lead to the discouraging conclusion that unless new fishing devices are really selective in targeting sizes of species caught, it is better not to adopt them. Therefore, testing the effectiveness of new fishing practices and gear devices is an essential step towards adopting new solutions to improve selectivity. This recommendation is in agreement with that of Kennelly and Broadhurst (2002), who propose a five-step framework for solving by-catch fishery problems that includes testing the alternative techniques in field experiments as one of the stages.

The scenarios analysed for the age selectivity improvements should be understood as hypothetical because it will not be possible to implement them with the available fishing technology. Even thinking in the future, it is difficult to believe that technological advances may lead to these “ideal” situations. Selectivity improves only for a particular range of ages but not for the remaining ages, and the improvement of age selectivity is the same, in percentage terms, for all the species caught in this mixed fishery (Table 1). Nevertheless, we think that these scenarios may be of interest because they demonstrate how fishery trends change as age selectivity improves.

In the real world, several fishing practices and gear solutions have been proposed and tested to enhance age selectivity. However, the effectiveness of these modifications are gear- and fishery-specific (Broadhurst et al. 2007). For instance, some studies show that square-mesh codend improves age selectivity compared with diamond-mesh codend of a similar size in a variety of trawl fisheries (Kayka et al. 2009, Gorelli et al. 2017). It has also been proved that the T90 netting (a standard codend mesh turned 90 degrees) significantly increases the selectivity for some species in Northern European trawl fisheries (Herrmann et al. 2007, Bayse et al. 2016) and in the Mediterranean Sea (Deval et al. 2006, Tokaç et al. 2014). Square-mesh panels and escape windows have also shown great potential for improving selectivity in some trawl fishery (Madsen et al. 2002), although the results are not so positive for others (Alzorriz et al. 2016).

The modelling approach used for the simulations in the present article is generic and can be applied to any fishery under any selectivity scenario. This means that once new fishing practices or devices have been developed, new age selectivity parameters will be available for simulating specific scenarios. Our study can therefore be seen as an additional contribution to the studies that have investigated the costs and benefits of age-selective fishing technologies [see Suuronen and Sardà (2007) for a compendium].

Another shortcoming of our analysis is that it does not take into account the cost of modifying and adapting the fishing technology for improving age selectivity. This is a common feature of the few studies that analyse the cost-benefit of a selectivity change (Heikinheimo et al. 2006, Macher et al. 2008). Certainly, selectivity can be improved without much cost by simply making new spatiotemporal decisions (Dunn et al. 2011). However, most experts consider that it is difficult to improve selectivity without incurring high costs when doing so requires modifications and adaptations of the current fishing technology (Villasante et al. 2016b). Nevertheless, the increase yield resulting in our analysis from improving age selectivity can be seen as an approximation to the upper bound for the cost of incorporating the new technology into standard harvesting operations.

With respect to the policy for a ban on the sale of juveniles, our results show that banning the sale of ages 0 and 1 leads to the highest improvements in hake yield (21% up) and total yield (11% up) compared with the status quo scenario. Rather than improving selectivity measures, this policy raises discards by seven percentage points. Since discards do not involve any cost, fishers capture larger amounts as the ban covers more ages, even though only a small percentage of them can be sold under the ban. These positive results for hake yield and total yield are reinforced when the sales ban is accompanied by improvements in selectivity. However, the improvements are not sufficient to revert the rise in the discard rate.

Comparing the two policies, we can conclude that measures that improve selectivity give better results than sales ban strategies in terms of increasing yields and stock and reducing discards.

The model used to carry out this study falls within the framework of the SASOM approach, which allows biological and economic aspects of the fishery to be incorporated simultaneously considering potential sources of uncertainty (Da Rocha and Gutiérrez 2011). The extended information about the SIHS provided by ICES enables to include details of the biological elements (ICES 2015) although with some limitations, such as the recruitment of the secondary species, which was assumed constant. Moreover, the available information is not sufficient to confidently include a cost function for the fishery or to consider social concerns such as the preservation of jobs in fleets. All these constraints limit the analysis, and the conclusions must therefore be taken with caution.

Another weakness of our analysis is that the model does not consider that species can have different quotas, and this could lead to the “choke species effect” preventing the fleet from taking advantage of its fishing opportunities under the LO. This is especially important if any of the species have a very small quota (García et al. 2016) or if any of them show an unfavourable length distribution of the captures, with most catches under legal minimum size (Prellezo et al. 2017). Further research on these issues would be very advisable to complete the analysis.