Gear-related conservation measures are based on the assumption that fish escaping from fishing gears survive and live on to promote the population. Relatively little work has been done to assess the effects of selective fishing gears on reducing the overall mortality of target and non-target populations. This chapter assesses the problems associated with estimating the magnitude and impacts of escape mortality, and demonstrates the processes by which mortality estimates may be included in fisheries management decision-making processes.
Improved selection means that a larger part of the fish population will escape from fishing gears. If escape mortality is low, the use of selective fishing gears might be assumed to bring long-term benefits. Likewise, if the mortality rates of escapees are high, there may be no advantages associated with changing selectivity. The assessment and prediction of the impacts of an increase in trawl mesh size require reliable fisheries, selectivity and biological data. These are complex tasks, particularly when fisheries involve several species and multiple gears (e.g. Macer, 1982; FAO, 1984; Kuikka et al., 1999; Breen and Cook, 2002; Halliday and Pinhorn, 2002; Tschernij, Suuronen and Jounela, 2004). Escape mortality data have seldom been applied to stock assessment models or included in fisheries management decision-making processes. Moreover, for many commercially-important fish species, there are currently no sufficiently accurate and reliable estimates of survival to allow assessments to be made. The following case studies demonstrate various approaches that have been taken in evaluating the impacts of escape mortality on overall mortality.
Kuikka et al. (1999) assessed the potential outcome of changes in exploitation level and mesh size in the Baltic cod demersal fleet (under different environmental conditions). The assessment consisted of three steps: (a) modelling of selectivity; (b) estimation of uncertainties; and (c) decision analysis by Bayesian influence diagrams. The authors used the trawl selectivity data presented by Tschernij et al. (1996) to model the retention of 120-mm and 140-mm diamond mesh codends. They assumed that all escapees survive. This assumption was relatively well supported by the results of escape mortality studies conducted with Baltic cod (Suuronen et al., 1996a). Their simulations suggested that the yearly loss of catch potential is significant, owing to overfishing and the non-optimal fishing pattern (i.e., poor selectivity). The average yield could be increased substantially (by 30 to 40 percent) by increasing the mesh size to 140 mm, decreasing the trawl fishing effort by 20 percent, and controlling gillnet fish mortality. The simulations also suggested that with a larger mesh size, preferred exploitation levels would become less sensitive to assumptions about future recruitment and growth levels. Increasing the mesh size to 140 mm would markedly reduce the frequency of dangerously low spawning stock biomasses and the need for very low catch quotas. A larger mesh size would be beneficial, irrespective of the assumed recruitment level.
Kuikka et al. (1999) emphasized that the increase in mesh size would not remove all uncertainties, but it would act as some insurance against their negative impacts. However, the authors did not fully assess several sources of uncertainty, for example: cannibalism could increase at low fish mortality rates more than they assumed, thereby reducing recruitment to the fishable stock; density-dependent effects may retard the growth of cod; and fish may migrate to other areas. Moreover, the mortality of escapees may not be zero, as was assumed. There are indications that under certain conditions the escape mortality of cod may be substantially higher, particularly if escape occurrs near the surface, where water temperature may be relatively higher (see Suuronen, Lehtonen and Jounela, 2005). Kuikka et al. (1999) pointed out that underestimation of uncertainties leads to overestimation of the system’s controllability. As a result, managers may develop an overly optimistic perception of their potential ability to reach the desired state of the system. Nevertheless, the overall effects of a mesh size increase in the Baltic cod trawl fishery appear positive, and in the absence of additional information, this would represent an appropriate and precautionary management strategy.
Kvamme and Frøysa (2004) used an age-length structured population model to assess the effects of changes in fleet selectivity on Northeast Arctic cod stock. They assumed that escape mortality is zero, which is a relevant assumption in the light of results of relevant studies done on Atlantic cod (e.g. Soldal, Isaksen and Engås, 1993; Ingolfsson, Soldal and Huse, 2002). Their simulations showed that there would be substantial gains, in terms of both stock size and catches, from increasing the mean retention length by 5 to 8 cm (from the present 47 cm). Catches of three- to four-year-old fish would decrease, while catches of fish of six years and older would increase within a few years. It is notable that Northeast Arctic cod reach maturity when they are six to 12 years of age and 65 to 105 cm long (Jørgensen, 1990). Hence, immature fish would be the most affected by the change in fleet selectivity.
Kvamme and Frøysa (2004) pointed out that the change in selectivity would lead to more efficient exploitation of the stock’s growth potential, and more fish would have a chance of growing to mature size and spawning. This would increase the spawning biomass and result in greater and more stable catches within a few years. The total catch, however, would decrease during the first three years following implementation of a mesh size increase. Kvamme and Frøysa (2004) argued that age-length structured models (see Frøysa, Bogstad and Skagen, 2002) are highly suitable for simulating the stock effects of changing the selectivity of a fishery, because selectivity is linked to size. In age-length structured models, age at first capture will automatically be adjusted for changes in fleet selectivity by length. But for age-structured models, fleet selectivity by length has to be transformed into fleet selectivity by age, which may vary from year to year. Age-length structured population models offer versatile tools for assessing the effects of changes in the selectivity of fishing fleets. However, Kvamme and Frøysa (2004) pointed out that the estimation of natural mortality is one of the most uncertain points of their simulation.
Andreasson and Flaaten (1996) made a bio-economic analysis of the effects of using a size-selective sorting grid (55 mm bar spacing) in the harvesting of Northeast Arctic cod. Their analyses suggest that there is a great potential for economic gains by choosing an optimal selection pattern in the fishing fleet. They assumed that all escapees survive.
Rahikainen, Peltonen and Pönni (2004) applied length-specific selection and escape-mortality functions to estimate the total quantity of escapees that die and the actual removals from Baltic herring stock in the northern Baltic Sea. Based on the results of a herring survival study conducted by Suuronen, Erickson and Orrensalo (1996), they assumed that the smallest (< 12 cm) escapees have 100 percent escape mortality and that herring of 12 to 17 cm have escape mortality of 90 percent. Their analyses showed that more ages 0 to one year of herring die as a result of escape from trawl codends than are landed. They also demonstrated that because immature herring have a fast growth rate, escape mortality follows a marked seasonal pattern. Their analyses showed that the effect of fishing-induced escape mortality decreases as a function of age and size, so that the impact on estimated recruitment and fish mortality at age one is considerable, while it is almost irrelevant at age two and older (at present exploitation pattern). The actual fish mortality at age one was estimated to be more than twice as high as estimates of fish mortality based on the unadjusted data. But, the overall effect of escape mortality on the evaluation of stock status, the stock recruitment function and reference points was minor. Rahikainen, Peltonen and Pönni (2004) emphasized that correct catch and mortality data are necessary for age-structured assessment models. Such data may be biased, for instance owing to unaccounted mortality connected to escape from trawl gears.
Kuikka, Suuronen and Parmanne (1996) studied the impacts of increased codend mesh size on the northern Baltic herring pelagic trawl fishery with an age-structured population model. The long-term effects of an increase in trawl codend mesh size (from 20 to 36 mm) on catch weight, catch value per recruit and the stock biomass of herring in the northern Baltic were assessed. The high escape mortality (85 percent) of herring (from Suuronen et al., 1996b; Suuronen, Erickson and Orrensalo, 1996) was incorporated into the analysis. The length-based trawl codend selection curves (from Suuronen and Millar, 1992) were transformed into curves that described codend retention by age. It is known that the annual growth rate of herring varies, and consequently retention by age also varies from year to year. Therefore, yearly estimates of growth rates from commercial catch sample data for the years 1974 to 1992 were used in the analysis. By using age-length structured models, growth variation was handled directly. Natural mortality estimates were taken from the ICES multispecies population analysis. Kuikka, Suuronen and Parmanne (1996) also assessed the effect of mesh size increase on the economic value of the annual catches. Their results showed that under the conditions prevailing in 1974 to 1992, the increase in codend mesh size would have led to reduced catches and lower values of catch-per-recruit. The magnitude of the estimated reduction of catches varied greatly, according to the growth and natural mortality of the population. The calculation suggested that in order to make an increase in mesh size profitable for this fishery over the long term, the price of large herring processed for human consumption would have to be approximately six times greater than that of smaller “fodder” herring, or the survival of codend escapees would have to be increased to 80 percent from its current estimated level of about 15 percent (e.g. with the help of new fishing technology).
This study demonstrated that for species incurring a high mortality during capture and escape, there may be no biological or economic justification for a mesh size increase. Clearly, unless the level of escape mortality is known, the benefits of the change of selectivity could be largely overestimated. In the worst case, this type of unaccounted mortality can have a negative effect on fish stocks because overall fish mortality may be underestimated. The analyses of Kuikka, Suuronen and Parmanne (1996) answers the question of what is the adequate level of survival to justify the use of selective fishing gears; something few other analyses have attempted to address.
Harley, Millar and McArdle (2000) presented a modelling approach using estimates of selectivity to estimate discard and escape mortality, and applied their model to the Hauraki Gulf snapper fishery off the northeast coast of New Zealand. In their model, discard and escape mortality can take any value between 0 and 100 percent and can also vary with fish length. Their approach represents an extension of the work of Casey (1993), who provided a theoretical basis for estimating discard-at-age mortality. Casey, however, assumed that all discards die and all escapees survive. This does not allow examination of the trade-off between discard and escape mortality that occurs with towed fishing gears. By allowing escape mortality to be greater than 0 percent, the trade-off between escape and discard mortality relating to mesh size regulations can be examined. The method presented by Harley, Millar and McArdle (2000) requires certain specific information such as estimates of catch by year and fishing gear, selectivity at length, and the probabilities of mortality for fish that are discarded and those that escape from gears. The authors obtained estimates of escape mortality from the reviews of Muoneke and Childress (1994) and Chopin and Arimoto (1995). They included interannual growth variability in their analysis and assumed mean recruitment. Natural mortality was assumed to decrease with fish length among pre-recruits and to be constant for all mature fish. Fate diagrams provided a graphic representation of how the different components of fish mortality affect the population. Harley, Millar and McArdle (2000) found that escape mortality may not be a significant component of total fish mortality, although there are large numbers of escapees in fisheries conducted with towed fishing gears. They emphasized, however, that a large number of individuals just below the minimum landing size die as a result of discarding. The reasons for this type of heavy discarding practice should be explored, and practices should be changed accordingly. The use of more selective fishing gears seems likely to offer one solution.
Cook (1998) proposed a calculation method for translating the escape and discard mortality rates obtained in survival experiments into mortality rates of stock and at the fisheries level, and illustrated its use in North Sea haddock assessments. His approach was based on the method presented by Mesnil (1996), which includes length-based mortality data in age-based assessments. Data from Scottish survival experiments (Lowry, Sangster and Breen, 1996; Sangster, Lehmann and Breen, 1996; Wileman et al., 1999) were used to obtain escape mortality estimates by age. It was assumed that no discarded fish survive. According to Cook’s analysis, the mortality of young haddock due to escape peaks at two years of age. This represents about 40 percent of the mortality from fishing at that age. However, total mortality did not differ much when escape mortality was assumed to be zero. The difference at age two was only about 20 percent, suggesting that conventional estimates of fish mortality at age two may be biased downwards by about 20 percent. This relatively small difference is mainly because the estimated escape mortality of haddock is relatively low. Cook (1998) concluded that, for North Sea haddock, the inclusion of escape mortality in the assessment does not make a perceptibly major change to the state of stock; at least not with the fishing pattern that existed during the time of analysis. However, he emphasized that the analysis was very preliminary and was performed primarily for illustration purposes.
Breen and Cook (2002) updated Cook’s (1998) analysis of the impacts of selectivity on North Sea haddock assessments. Their model estimates discard mortality, escape mortality and retained catch separately using data from the ICES database. These were then transposed from length-based to age-based data using the age-length keys. Their simulation was run with discard mortality set at zero (no discards) and one (all discards die) and with varying escape mortalities (10, 25, 50 and 100 percent). The results were then partitioned into a single fish mortality value. Breen and Cook’s (2002) simulation showed that including discard mortality significantly increased the fish mortality estimates, particularly for ages one (94 percent), two (63 percent) and three (18 percent), and including escape mortality (assuming that 25 percent of escaping fish die) produced less significant but still substantial increases in fish mortality (38 percent at age one; 7 percent at age two; and 1 percent at age three). That is, their analyses showed that compared to escape during fishing, discarding has a far more profound effect on the fish mortality of haddock. Furthermore, they demonstrated that relative importance of escape mortality decreases as age increases. Clearly, the analyses of Breen and Cook (2002) provides a useful insight into the relative importance of the different components of fish mortality (landing, discard and escape mortality) with respect to the stock-assessment process. They emphasized that exlusion of escape mortality parameter estimates has the potential to introduce significant biases into the stock assessment process in particular if there are to be further increases in gear selectivity. These authors also assessed the benefits of increasing the minimum legal mesh size. Their analyses showed that this benefit is greatly reduced if, for instance, only 25 percent of escaping fish die. Significant benefits would be obtained only if most escapees survive.
It is well known that large numbers of haddock and whiting and substantial quantities of cod caught in the North Sea are discarded every year. The majority of discarded fish are smaller than the minimum landing size (MLS) (which for 23, 30 and, 35 cm TL for whiting, haddock and cod, respectively). Garthe, Campyhuysen and Furness (1996) estimated the annual quantity of fishery discards of round fish in the North Sea to be about 260 000 tonnes. Total discards equated to about 22 percent of the total North Sea catch. There are several reasons for this situation. A larger minimum mesh size alone would not provide a suitable tool for achieving maximum yield-per-recruit for each species in the North Sea mixed species fishery. Graham and Kynoch (2001) demonstrated that with a 100-mm minimum mesh size, cod, haddock and whiting of approximately 23 cm in length entering the codend would have a 50 percent chance of escaping. Almost all fish of 30 cm and larger entering the codend would be retained. Macer (1982) estimated that for the North Sea mixed trawl fishery the mesh sizes required to give optimum yields are approximately 250 mm for cod, 140 mm for haddock and 90 mm for whiting. A minimum mesh size of 100 or 120 mm would be too small for cod and haddock, but too large for whiting. Requiring one mesh size to catch these three species inevitably results in discarding and/or high catch-losses. Nevertheless, an increase in the average mesh size would increase the average age-at-first-capture, and would therefore improve the overall situation, increasing the long-term total yield from the fishery even if a precise optimum were not achieved. This case study demonstrates the common problem faced in almost any mixed species trawl fishery. To attain marked improvements, there should be a selectivity system in which the different species are first separated from each other, and then sorted by size.
Most assessments of mesh size increases focus on the long-term effects. However, there are also short-term effects that may require attention, and the following example from the Baltic cod fishery demonstrates the importance of understanding and addressing these. In 2002, a highly size-selective 120-mm square mesh window (a “Bacoma window”) was enforced in the Baltic demersal trawl fishery (Madsen, Holst and Foldager, 2002; Tschernij and Suuronen, 2002). The decision to do this was based on long-term projections that suggested there would be a substantial increase in spawning stock size and a marked reduction in discards if a larger window mesh size was enforced (Anonymous, 2000). The short-term effects of a new selectivity pattern were modelled with a stochastic size-selective simulation model (Tschernij, Suuronen and Jounela, 2004). Vessel type-dependent selectivity estimates and catch-per-unit-of-effort (CPUE) data from the Baltic cod demersal trawl fishery were utilized to estimate the catch losses. The simulations suggested that when the window mesh size is increased by 15 mm (from 105 to 120 mm), the overall catch loss of fish of marketable size during the first month would be around 40 to 50 percent (with the same fishing effort). With a 120-mm window and a 38-cm minimum landing size, the discarding of undersized cod would decrease by about 70 percent. If fishers decided to compensate their loss in marketable catch by increasing their fishing effort, they would have to increase it by 55 to 90 percent.
Tschernij, Suuronen and Jounela (2004) suggested that fishers were unlikely to be able to increase their efforts to such a large extent. Instead, they might try to circumvent the regulations by intentionally decreasing the selectivity of their gear, i.e. by gear manipulation. In fact, widespread gear manipulation - legal and illegal - was observed in 2002 and 2003 in the main fishing grounds (Suuronen and Tschernij, 2003). Fishers were not able to adapt to heavy catch losses, which apparently were even larger than predicted by the simulations. In September 2003, the minimum window mesh size of the Bacoma window was reduced from 120 to 110 mm. This example demonstrates that even in a simple case where fishing targets almost exclusively one species, increasing the mesh size may be very complex even though the biological preconditions appear favourable. This case also demonstrates that too large an increase in selectivity is not commercially acceptable. Gears will be manipulated and rules will be circumvented if the losses are too large (see also FAO, 1984; Ferro and Graham, 2000; Halliday and Pinhorn, 2002). Clearly, short-term effects should be addressed in the management plans; it is not enough to assess only the long-term effects of a mesh size increase.
Chopin, Inoue and Arimoto (1996) and Chopin et al. (1997) defined fish mortality as the sum of all fishing-induced mortalities occurring directly as a result of catch (capture), or indirectly as a result of contact with or avoidance of the fishing gear. They also recognized the following sub-components of fish mortality: landed catch; illegal, misreported and unreported landings; discard mortality; escape mortality; drop-out mortality; ghost fish mortality; avoidance mortality; and habitat degradation mortality. Except for landed catch, these sub-components of fish mortality are unknown or poorly known for the vast majority of fish stocks. They are therefore referred to as “unaccounted mortalities”. A significant amount of research would be necessary to estimate all the sub-components of fish mortality for any given stock, and such information is available for only a few stocks. Most of the work conducted so far in this field has been limited to two particular aspects of mortality - discard and trawl escape - as these are probably the most important components of unaccounted fish mortality, at least for trawl fisheries. Furthermore, questions about the potential negative effects of long-term size-selective fishing on the genetic composition of fished population are periodically raised. This subject, however, has remained largely inconclusive (e.g. Beverton, 1998; Law, 2000), and is beyond the scope of this study.
There are many problems associated with incorporating estimates of escape mortality into the fisheries assessment and management processes. Clearly, it will never be possible to predict exactly the outcome of a management decision that involves a change in gear selection, even when the escape mortality is known. It is evident that the biological and economic benefits that would be generated will always vary among species, fisheries and conditions. In most fisheries, there is a lack of suitable data and information to conduct relevant evaluations. Escape-mortality studies have been done for only a few species and in only a few fisheries. Extending investigations to other fisheries and quantifying this issue for a larger number of species would help to improve the reliability of predictive modelling. It would be particularly useful to extend investigations to fisheries where stocks are overfished or where there are large mortalities of escaping fish.
Very few studies have been able to verify their predictions with follow-up studies (after the measure has been introduced). The available analyses, however, help to provide an understanding of the general pattern. It has become increasingly clear that in many important fisheries, regulating gear selectivity alone is not enough to provide a sustainable exploitation pattern; control of the amount of fishing effort is essential also. Furthermore, it is apparent that in many fisheries the authorized mesh sizes remain too small to protect juveniles and young adults effectively, and insufficient numbers of fish survive to replace the lost spawning biomass. Moreover, density-dependent effects such as cannibalism, decreasing growth and migration of fish out of the areas exploited by the fleet may lead to lower benefits than are assumed. When escape survival is low, as it is for some small-sized pelagic fish such as herring, there are unlikely to be any biological benefits from using mesh selectivity as a management tool unless other gear modifications are concomitantly developed. Clearly, unless the level of escape mortality is known, the benefits of change in gear selectivity could be overestimated. In such a situation, the key question is what level of survival justifies the use of more selective gear. In mixed species trawl fisheries, increased selectivity should separate species first and then sort by size to account for minimum mesh sizes which differ with species.
Traditional stock assessments generally assume that all fish passing through a towed fishing gear survive the process. Few attempts have been made to incorporate escape mortality estimates into stock assessment processes. The above case studies demonstrate that ignorance of escape mortality may lead to underestimation of fishing mortality, particularly among the youngest age classes. It is evident that as the trend for using more selective gears continues, the relative importance of escape mortality on stock assessment will increase.
Further work is needed to assess adequately the true biological and economic costs of bycatch, and the benefits and costs of potential gear solutions. Technical measures such as mesh size increase are by nature long-term management measures (i.e. long-term gains are assumed when they are enforced). Substantial short-term economic losses, however, are often associated with their adoption. If short-term effects are not addressed, the assumed long-term gains may not be realized. The ecological as well as the socio-economic impacts of new measures and modifications should be comprehensively addressed before changes are implemented. In most cases, selectivity must be increased incrementally, otherwise short-term loss in commercial value for the fishery will cause fishers to not accept changes.
