The unaccounted mortality associated with commercial fishing has been assessed and reviewed by several authors (e.g. FAO, 1994; Chopin and Arimoto, 1995; Wileman et al., 1999; Davis, 2002; FAO, 2003) and study groups of the International Council for the Exploration of the Sea (ICES, 1994; 1995; 1997; 2000). Most of these reviews have limited their scope to one particular aspect of unaccounted mortality and, in most cases, only to towed gears. By far the most detailed and extensive description of unaccounted mortality is that of the ICES Topic Group on Unaccounted Mortality in Fisheries (ICES, 2000).
Studies on the capture-related mortality of fish in commercial fisheries can be classified into two broad categories according to capture and release process: (a) survival of fish that escape from fishing gears during the capture process; and (b) survival of fish caught by fishing gears and then discarded on deck (Figure 1). This chapter attempts to review and highlight the key principles and factors that evoke stress, injury and mortality in fish that escape from a fishing gear, in particular a trawl codend. The next chapter explores the issues specifically dealing with discard mortality from trawl fisheries, followed by a review of the mortality associated with other fishing gears. This presentation does not attempt to provide an exhaustive discussion of all the potential factors causing mortality in fisheries - instead it focuses on the major factors, particularly those that can be affected.
It is notable that fish escape from many parts of the gear, and may therefore be affected differently. In the case of towed fishing gears, however, fish usually escape from the codend (e.g. Wileman et al., 1996), and most survival studies have therefore focused on the mortality codend escapees. However, the zone of influence of a fishing gear is not limited to where the fish are retained and where they escape; it also includes those parts of the gear that herd and scare fish. Very little is known about the fate of fish escaping from these areas but apparently, the stress, injury and mortality induced is substantially smaller than in those fish escaping from codend.
Most assessments of escape mortality have been made for commercially important species escaping from towed fishing gears, mainly demersal trawls. These investigations have usually attempted to compare the relative benefits of using different mesh sizes or selective devices in codends. The most studied group of fish are the gadoids, particularly haddock, whiting and cod (e.g. Main and Sangster, 1990; 1991; Soldal, Isaksen and Engås, 1993; Sangster, Lehmann and Breen, 1996; Wileman et al., 1999; Suuronen et al., 1996a; Ingolfsson, Soldal and Huse, 2002). Some work has also been conducted with pelagic species such as herring, vendace, capelin and walleye pollock (e.g. Treschev et al., 1975; Efanov, 1981; Suuronen et al., 1995; 1996b; Thorsteinsson, 1995; Pikitch et al., 2002), with flatfishes (DeAlteris and Reifsteck, 1993; Robinson, Carr and Harris, 1993) and with species such as red mullet, sand whiting and yellow-fin bream (Broadhurst, Kennelly and Barker, 1997; Broadhurst, Barker and Kennelly, 1999; Metin et al., 2004).
It is clear from these studies that the robustness and ability of various species to withstand physical injury and fatigue associated with capture and escape vary markedly. Generally, relatively low escape mortality has been observed in many gadoids (e.g. cod, haddock, whiting, saithe). For instance, Soldal, Isaksen and Engås (1993), Suuronen et al. (1996a), Suuronen, Lehtonen and Jounela (2005) and Ingolfsson, Soldal and Huse (2002) observed negligible (< 3 percent) mortality among cod escaping from trawl codends under normal fishing conditions. The observed escape mortality of saithe was at the same level as that of cod (Jacobsen, Thomsen and Isaksen, 1992; Ingolfsson, Soldal and Huse, 2002). The mean mortality of haddock and whiting escaping from a 100-mm diamond mesh codend was less than 15 percent (Wileman et al., 1999).
Low mortality rates (mostly < 10 percent) have also been observed with flatfishes such as winter flounder, yellowtail flounder and American plaice (DeAlteris and Reifsteck, 1993; Robinson, Carr and Harris, 1993). Broadhurst, Kennelly and Barker (1997) and Broadhurst, Barker and Kennelly (1999) showed low mortality (< 3 percent) associated with sand whiting and yellow-fin bream that had past through the square mesh panels of a trawl codend. It is notable, however, that some of these species have also shown relatively high mortalities.
Substantially larger escape mortalities (of as much as 70 to 100 percent) have been observed in small-sized pelagic species such as Baltic herring (Suuronen et al., 1996b), although almost opposite results have also been reported with the same species (Treschev et al., 1975; Efanov, 1981). Medium levels of mortalities have been recorded with some pelagic species such as walleye pollock and vendace (Suuronen et al., 1995; Pikitch et al., 2002). In most studies, survival has been shown to depend on size.
Although clear species-specific differences in survival have been observed, one of the main findings of these investigations is the high variability in survival, even with the same species in the same experiments (e.g. Suuronen et al., 1996b; Suuronen, Lehtonen and Jounela, 2005; Wileman et al., 1999; Ingolfsson, Soldal and Huse, 2002). This variation has not yet been explained adequately. Moreover, mortality estimates have not usually been estimated with confidence intervals; typically only a range of mortality observations is given.
It is generally considered that increasing codend mesh size will automatically reduce the injury and mortality of escaping fish. It is assumed that the larger the mesh opening, the easier it is for fish to pass through, and consequently the less damage that occurs. Indeed, an inverse correlation between mortality rates and increasing mesh size has been reported in several investigations (e.g. Main and Sangster, 1991; Sangster, Lehmann and Breen, 1996; Lowry, Sangster and Breen, 1996; Wileman et al., 1999). Lowry, Sangster and Breen (1996) demonstrated that fish escaping through a larger mesh may sustain less skin and scale damage than fish escaping through a smaller mesh. However, some studies have suggested that codend mesh size has less influence on the survival of escaping fish (e.g. Suuronen et al., 1996b; Wileman et al., 1999). Clearly, the positive effect of an increase in mesh size on the survival of codend escapees has not yet been demonstrated conclusively. This does not mean, however, that mesh size does not play any role in survival; it is potentially important, but many other important factors affect survival simultaneously.
There are indications that mesh shape may play a more important role than mesh size in reducing the mortality of escapees. For example, haddock and whiting escaping through a square mesh codend were reported to have lower mortality than those escaping from a traditional diamond mesh codend of the same mesh size (Main and Sangster, 1990; 1991). Apparently, many fish may escape through open square meshes with less injury than through conventional diamond meshes because the latter may become almost closed owing to net tension during tow (Figure 2). In some investigations, escape through a special sorting device, such as a grid (Figure 3), has shown to result in lower mortality than mesh sorting (e.g. Suuronen et al., 1996b; Ingolfsson, Soldal and Huse, 2002), but this has not yet been demonstrated adequately. Intuitively, it would appear that survival should be greater when escape is easier. This may be true to some extent, but the passage of a fish through an open mesh or other selective device is not the only factor that can inflict fatal injuries. A potential advantage of square mesh panels and special sorting devices is that they can be placed in front of the actual codend so that escapees do not have to enter the rear part of the codend, which is probably the zone with the highest mortality for escapees. Clearly, effects of a selective device, whether it is a netting mesh or a sorting grid, on injury and mortality of escapees are complex and relatively poorly understood; many factors have to be considered when developing selective gear modifications and bycatch reduction devices that also guarantee a high survival rate for escapees.
Studies have been somewhat inconclusive regarding the relationship between escape mortality and fish size (length), but the general observation has been that escape mortality decreases with increasing fish length. For instance, Wileman et al. (1999) observed that the mean length of haddock that died after escape was significantly lower than that of surviving haddock, i.e. survival increased with increasing mean length. Similar length-related mortality has been described in many other studies conducted on gadoids (e.g. Lowry, Sangster and Breen, 1996; Sangster, Lehmann and Breen, 1996; Pikitch et al., 2002; Ingolfsson, Soldal and Huse, 2002) and species such as Baltic herring (Suuronen, Erickson and Orrensalo, 1996).
On the other hand, a significant increase in injury and death rates would seem likely in groups of fish whose passage through a mesh or grid is made difficult by their physical size (i.e. in the largest escapees). For example, fish that have a transverse morphology that is similar in size to the maximum mesh opening could be expected to sustain more injury than smaller individuals that can pass through meshes with less contact. However, very few investigations have documented a marked increase in injury and mortality among the largest escapees (Efanov, 1981; Ingolfsson, Soldal and Huse, 2002).
Many investigations have shown an inverse relationship between skin injury and the size of the escaping fish (e.g. Soldal et al., 1991; Soldal, Isaksen and Engås, 1993; Soldal and Isaksen, 1993; Suuronen, Erickson and Orrensalo, 1996; Breen and Sangster, 1997; Ingolfsson, Soldal and Huse, 2002), i.e. the smallest individuals show the highest injury rates. This supports observations that the smallest escapees often suffer the highest mortalities. Apparently, smaller fish with poorer swimming ability are less able to avoid injury when swimming within the gear and during escape. They may also have less physical strength to make active escape attempts, and may therefore stay longer inside the gear before escaping. The smallest fish are generally also more delicate than larger individuals, and are therefore more susceptible to all types of capture-induced injury. Their high vulnerability may be the result of a combination of exhaustion and injury. It is notable, however, that several studies have been inconclusive regarding the relation between skin injury and fish size (e.g. Lowry, Sangster and Breen, 1996; Sangster, Lehmann and Breen, 1996; Suuronen et al., 1996a; 1996b; Suuronen, Lehtonen and Jounela, 2005; Pikitch et al., 2002).
The mortality of escapees may also be at least partly a function of fish age. There is some evidence that in a particular year class the smallest haddock and whiting, i.e. those fish that have grown more slowly, are more susceptible (e.g. Lowry, Sangster and Breen, 1996; Sangster, Lehmann and Breen, 1996; Wileman et al., 1999). Hence, in a fish population, the fittest individuals from age group 1 may survive better than the less-fit individuals from age group 2. Therefore, the age and fitness of a particular size of fish, and not only its length, may play a vital role in its ability to survive codend escape.
In conclusion, there are complex relationships between the size of fish and their injury and mortality due to capture and escape. Owing to their sustained swimming ability, larger fish generally appear to suffer less injury and lower mortality than smaller fish. It is notable, however, that the methods used in the experiments may have seriously biased these estimates. In particular, the smallest fish may be highly sensitive to collection and handling (e.g. Suuronen et al., 1996b; Breen et al., 2002). Hence, results describing the relationship between the size of fish and their injury and mortality should be considered with great caution. Better techniques and more work are required in this area.
It is well demonstrated that the skin of fish performs a number of functions that are important to their survival and well-being. These include mechanical protection, osmoregulatory control, protection from pathogenic invasion, communication, sensory reception, and capture and predator avoidance. Depending on its severity, damage to the skin could result in the loss of one or all of these vital functions.
The abrasive qualities of netting materials suggest that fish may sustain severe injuries during the tow, especially in the codend where individuals are exhausted and crowded together. In a number of studies, gear-induced injuries have been observed in fish that escaped from a trawl codend (e.g. Borisov and Efanov, 1981; Main and Sangster, 1990; 1991; Suuronen et al., 1996a; 1996b, Wileman et al., 1999). During the escape, fish may sustain injury when they exhibit the common behavioural pattern of thrashing their tails in an attempt to swim while still confined by the mesh (e.g. Glass and Wardle, 1989; Main and Sangster, 1990). Fish can also sustain injury at other times; using video, Suuronen et al. (1996b) observed that Baltic herring scraped against the trawl netting along the trawl belly and codend extension prior to entering into the codend, which resulted in major scale loss and skin injury. Apparently, herring were not able to avoid contact with the trawl netting. Similar observations have been reported in vendace trawl fishery (Suuronen et al., 1995). Moreover, several studies have suggested that the proportion of other abrasive objects, such as spiny fish species, crustaceans and broken shells, in the codend may also injure escaping fish (Treschev et al., 1975; Main and Sangster, 1991; Wileman et al., 1999). Bublitz et al. (1999) demonstrated that many walleye pollack escaping from trawl meshes were bruised to some extent, and this bruising was correlated with subsequent mortality.
Most assessments on capture-induced skin injury have been conducted on gadoids and herring. Farmer, Brewer and Blaber (1998) assessed the scale loss of several tropical fish species escaping through trawl codend meshes. Scale loss was significant in all the species studied; the heaviest losses were observed for species with deciduous scales, such as perforated-scale sardine, pearly finned cardinal fish and sunrise goatfish. It is notable that escapees from a 45-mm square mesh codend were generally less damaged than those from a 38-mm square mesh codend.
To determine the potential injury to fish passing through netting meshes or other selective devices (e.g. sorting grids), some studies have attempted to simplify the observation process by conducting simulated laboratory (tank) experiments. Species studied include cod, haddock, saithe and sand whiting (e.g. Soldal, Engås and Isaksen, 1989; Soldal, Isaksen and Engås, 1993; Engås, Isaksen and Soldal, 1990; DeAlteris and Reifsteck, 1993; Jónsson, 1994; Broadhurst, Kennelly and Barker, 1997; Broadhurst, Barker and Kennelly, 1999). In most cases, mortality was indistinguishable from that of the control fish, which supports the view that the simple passage of a fish through a netting mesh or other selective device does not necessarily inflict fatal injury. However, results obtained in laboratory conditions cannot usually be directly extrapolated to the more complex nature of a commercial fishing process (see e.g. Soldal, Isaksen and Engås, 1993).
Clearly, while many studies have demonstrated that skin damage, particularly scale loss, among fish escaping from trawl codends is often the prevalent injury, it is not conclusive that this is the primary cause of mortality. The scientific literature provides few clues about the physiological mechanisms associated with skin injury and mortality (see the review of Smith, 1993). Skin injury may not account for all of the observed mortalities that typically occur within the first day after escape, but it may expose fish to secondary infections that significantly increase their longer-term mortality (e.g. Roberts, 1989). Mellergaard and Bagge (1998) observed a high occurrence of skin ulcerations in Baltic cod caught in the vicinity of the Danish island of Bornholm. The ulcerated fish were mainly 24 to 28 cm in length, and most of the ulcerations on the trunk occurred bilaterally. There were strong indications that the skin ulcers were induced by the trawl gears that are used in the area. Mellergaard and Bagge (1998) suggested that these fish had escaped from trawl codends and that many of them were likely to die of secondary bacterial septicaemia during the summer months of increasing water temperatures. Jones (1993) reported on gear-induced cuts, scrapes and scars on hoki caught from the New Zealand fishing grounds. He suggested that the presence of net-damaged fish provides evidence that hoki are injured by coming into contact with trawl gears. He also suggested that the absence of an epidermal layer in the gear-induced wounds of some hoki may indicate that re-epithelization by epidermal migration (see e.g. Bullock, Marks and Roberts, 1978) cannot cover the relatively large area of these wounds. Clearly, long-term observations are required to understand the role of skin injury in determining mortality.
Very little is known about the potential healing of gear-induced injuries. Although minor injuries may heal completely, and open skin lesions be replaced by scar tissue within a few weeks after injury (e.g. Engås, Isaksen and Soldal, 1990; Sakanari and Moser, 1986; Bell and Bagshaw, 1985; Jones, 1993), there may be secondary infections from potentially pathogenic bacteria and fungi (e.g. Bullock and Roberts, 1980; Copland and Willoughby, 1982; Roberts, 1989). Particularly at warmer temperatures, such infections may exacerbate the lesions and the fish will succumb to osmotic distress or septicaemia if the infection becomes generalized (e.g. Mellergaard and Bagge, 1998). On the other hand, Roberts (1989) showed that the healing of dermis may be slower in very cold water. Clearly, more work is required in this field.
The overall effect of capture-induced stress and exhaustion on the mortality of escapees is not clear. Stress response enables fish to avoid or overcome potentially threatening or harmful situations (e.g. Pickering, 1993; Chopin and Arimoto, 1995). How a fish reacts to a particular stressor will depend on the species, as well as the type of stressor and its severity. The type of stressors that fish are subjected to during capture by commercial fishing gears depends on the fishing method. In the case of towed fishing gears, stresses include confinement, overcrowding and severe exercise. Suuronen et al. (1996a) and Tschernij and Suuronen (2002) observed Baltic cod swimming in front of a demersal trawl until they became fatigued, after which they turned and were overtaken by the trawl. Xu, Arimoto and Inoue (1993) showed that small pollack became severely fatigued during the trawl-capture process, and Beamish (1966) suggested that muscle fatigue alone could cause mortality. Young vendace subjected to a trawl capture and escape process exhibited strong symptoms of exhaustion (Turunen et al., 1996); this may contribute markedly to the high mortality of escapees (Suuronen et al., 1995). Likewise, the dramatic reduction of liver glycogen of small (7 to 11 cm) Baltic herring escapees after forced swimming inside the trawl may have increased their susceptibility to stress, thereby contributing greatly to their high mortality (Suuronen, Erickson and Orrensalo, 1996).
Swimming exhaustion may be an important factor contributing towards mortality, at least in some specific cases, although very little direct scientific evidence exists. It is notable that haddock and cod survived severe swimming exhaustion and muscular fatigue in tank experiments (Soldal, Isaksen and Engås, 1993). It is likely that the towing speed has an effect on fishs swimming capacity within the trawl and during escape from the trawl, and thereby on mortality. However, there are no published data on the effect of towing speed on survival. The interactions between towing speed, fatigue, water temperature, the escape process, injury and survival are likely to be very complex. More work is needed to assess the importance of stress and swimming exhaustion on the survival of escaping fish.
It is generally assumed that increasing the catch size increases the likelihood of injurious mechanisms within the codend, owing to increasing the abrasive contacts with other fish, netting, debris and turbulent flow patterns. In survival experiments, however, no significant relationship between codend catch size and escape mortality has been demonstrated (e.g. Wileman et al., 1999; Pikitch et al., 2002; Suuronen, Lehtonen and Jounela, 2005). However, owing to the many methodological constraints, most survival studies have been conducted with small catches that typically do not reflect commercial situations. It is also notable that the effects of catch size and catch composition may be highly confounded by such variables as the towing time and environmental conditions.
It is notable that codend catch weight may reduce the selectivity of the codend owing to mesh blocking (e.g. Dahm, 1991; Dahm et al., 2002; Suuronen and Millar, 1992; Lowry, Sangster and Breen, 1996; Erickson et al., 1996). Therefore, the overall number of the fish escaping would decrease and the escape would become more difficult, leading to higher mortalities. However, it has also been demonstrated that as the codend fills with catch, the meshes in front of the catch bulge become more open and fish may therefore escape more easily. In fact, a codend may have poorer selectivity with smaller catch sizes, and selectivity may improve substantially as the catch starts to accumulate in the codend (e.g. Lowry, Sangster and Breen, 1996; ONeill and Kynoch, 1996; Dahm et al., 2002; Campos, Fonseca and Henriques, 2003; Herrmann, 2005). The risks of abrasive injury during the passage though a mesh may be reduced, at least in certain catch sizes. Clearly, the effects that codend catch size and composition have on mortality require further study.
Fishing vessels roll, pitch and heave in response to waves and winds. Sea state is generally known to affect trawl codend selectivity (e.g. Wileman et al., 1996; ONeill et al., 2003), so it can be assumed that increased sea state and vessel motion may also affect the survival of fish escaping from trawl codends. Trawl surging associated with increased sea state may alter the water flow within the gear and make it difficult for fish to orient towards selective panels or sorting devices. Fish may become stressed, meshed and injured but increased sea state may also cause increased escape during towing and haul-back (e.g. Engås et al., 1999). Wileman et al. (1999) demonstrated that the survival of whiting escaping from trawl codends decreased with increasing sea-state. However, no significant effects were observed in other experiments where greater changes of sea state occurred. More work is needed in this area.
Depending on the vessel, its operation and the design of the gear, a trawl codend may be under heavy tension during towing. Codend meshes may be almost totally closed during the capture process, and fish are not able to escape until the codend is at or near the surface during haul back (e.g. Tschernij and Suuronen, 2002). As a result of decompression problems, many fish may suffer injury to their swim bladders. Fish with ruptured swim bladders often have gas in their abdominal cavities and are trapped on the water surface. After escape, these floaters may be subjected to high predation by birds. Moreover, various mechanical forces on a codend floating at the surface can be extremely high, particularly in rough seas. Catch may become compressed in the aft part of the codend, and fish may become crowded, resulting in oxygen depletion, abrasion and injury. These fish may have been subjected to substantial differences in temperature, salinity and pressure. While there is little available information on the fate of fish escaping from codends near the surface, these factors suggest that the probability of survival would be quite low. Clearly, escape should take place during fishing and not during hauling.
Water temperature influences the physiological processes and behaviour of fish (e.g. He and Wardle, 1988; Özbilgin and Wardle, 2002), and so probably affects their escape and subsequent survival. Nevertheless, relatively little direct fieldwork has been done to quantify the effects of water temperature on the survival of escapees under commercial fishing conditions. Most observations on temperature effects are by-products of research in which other factors have been the main interest. Suuronen, Lehtonen and Jounela (2005) observed low mortality (circa 3 percent) in Baltic cod escapees at ambient (< 10 °C) water temperatures. At higher temperatures (> 15 °C), a substantially higher mortality (up to 75 percent) was observed, but there was high variation between hauls. Fish that are caught at greater depths or in cool waters by a towed fishing gear may be exposed to substantially warmer water temperatures for instance when gear is towed to shallower ground or during gear retrieval. If escape occurs at relatively higher temperatures, the stress, injury and mortality of escapees could be increased.
In laboratory studies, the mortality rates of sablefish, lingcod and Pacific halibut increased with increasing seawater temperature for fish that were first towed and then exposed to increased temperature, with 100 percent mortality at 16 °C for sablefish, 18 °C for Pacific halibut and 20 °C for lingcod (Davis, Olla and Schreck, 2001; Davis and Olla, 2001; 2002). Although there were substantial species-specific differences in mortality rates, these results demonstrate the marked effect of temperature. An abrupt temperature increase of several degrees induced high mortality in adult sablefish (Olla, Davis and Schreck, 1998). Exposure to warmer temperatures results in increased core body temperature, with smaller fish warming more rapidly. Davis, Olla and Schreck (2001) argued that the additional stress and mortality resulting from the interaction of capture, escape and exposure to increased temperature may be common in fisheries during warmer seasons of the year, or in areas where fish are caught in cooler deeper water.
Low water temperatures may also affect fishs behaviour and sensitivity to various capture stresses. Fishs swimming speed and endurance generally decreases at low temperatures (e.g. He and Wardle, 1988). Reduced swimming speed and endurance may influence the herding effect of various gear components and the ability of fish to escape from trawls. Davis (2002) argued that deficits in swimming performance and orientation in a trawl associated with low temperatures could cause fish to be injured more frequently. In conclusion, water temperature plays a critically important role in the mortality rates of escapees, and may magnify the effects of other stressors.
Vision plays an important role in the response of many species to trawls (e.g. Glass and Wardle, 1989; Olla, Davis and Schreck, 1997; Olla, Davis and Rose, 2000). Olla, Davis and Rose (2000) observed that there was a clear difference in the orientation and swimming behaviour of walleye pollock when the light level fell below that necessary for vision-mediated swimming. In darkness, captured walleye pollock swam less, passed along the trawl faster and did not orient to the long axis of the trawl. However, other field studies show that some species are able to orient in a trawl during dark hours, indicating that senses other than vision may play an important role (e.g. Engås and Ona, 1990).
Ambient light conditions may have a significant effect on fishs behaviour and ability to escape from codends. The available light may not always be adequate for visual mesh detection because fishing operations are commonly conducted in relatively deep waters, often at night and in turbid water. There is some evidence that loss of orientation and swimming ability under dark conditions may result in reduced ability to escape through trawl mesh and increased injury and mortality. Suuronen et al. (1995) observed that young vendace escaping a pelagic trawl codend at night sustained significantly higher mortalities than fish escaping during daylight hours. Apparently, fish were not able to orient towards the gear and the open meshes, and so incurred greater injuries. Laboratory experiments on walleye pollock towed in a net under light and dark conditions demonstrated that in light conditions fish could see the net and were able to swim up to three hours with no resulting mortality, whereas fish in dark conditions were not able to see and respond to the net and became pinned against net meshes, ultimately resulting in mortality (Olla, Davis and Schreck, 1997). Clearly, developing effective gear modifications to improve selectivity and the subsequent survival of escapees requires a good understanding of how fish react to gear under various light conditions; reactions both under light and dark conditions must be considered.
It is worth noting that the whole aft section of a bottom trawl is often surrounded by a sand or mud cloud stirred by the otter doors and the ground gear (e.g. Pikitch et al., 1996; Tschernij and Suuronen, 2002). It is likely that this makes visual detection of open meshes or other selective devices difficult or impossible, even during the day (in light conditions). Very little is known about the potential effect of this on survival.
It could be assumed that towing time has a strong effect on stress, injury and mortality because a longer tow is likely to produce more catch, crowding, abrasion and swimming exhaustion, as well as more blocked meshes. However, Wileman et al. (1999) observed no clear effect of trawl towing time on the escape mortality of haddock and whiting. Treschev et al. (1975) reported that tow duration had an effect on the survival of Baltic herring, but the results they presented are not very convincing. The length of time that escaping fish spend inside the trawl is not generally known, and there may be substantial variation depending on the conditions, species and sizes. The general rule may be that fish that escape do so very soon after reaching the codend. For these fish, the length of trawl tow is not necessarily the most important factor affecting their subsequent survival. Those fish that do not manage to escape rapidly will soon become exhausted and be piled up in the catch, from where they are carried out with the trawl. Some of these fish may manage to escape later, particularly during the hauling when tension on the codend netting is reduced. They are likely to exhibit very different mortality rates compared with those of fish that escape immediately. The effect of tow duration on escape mortality should be explored in more detail. It may be influenced by the codend catch size and composition, the towing speed, the netting material of the gear, and hauling and handling processes.
The effect of towing duration on the survival of fish has been explored in greater depth in discard mortality studies (see page 21). However, it is important to realize that discarded fish have probably spent, on average, longer in the trawl gear than escaping fish have, and will have experienced additional stress owing to gear retrieval and on-deck handling processes. Hence, tow duration may be an important factor in determining discard mortality.
Few studies have investigated the effect of sex on the mortality of escaping fish. Wileman et al. (1999) showed that sex had no effect on the survival of sexually immature haddock and whiting. However, sex may be an important factor for adult fish, at least near and during spawning time, when females have a wider transverse morphology than males of the same length. Notwithstanding this, the issue may not always be relevant because selective fisheries usually attempt to sort out only young immature fish. It is notable that the survival rates of trawl-caught and discarded Norway lobster were found to be significantly lower among females than males (Wileman et al., 1999).
Marine predators are known to follow trawls during towing, and consume fish that escape from the codend meshes (e.g. Broadhurst, 1998). Under laboratory conditions, Ryer (2002) showed that walleye pollock subjected to capture and escape stress were more likely to encounter predators than a control group was. Ryer, Ottmar and Sturm (2004) emphasize that even a relatively small change in predator detection, avoidance, schooling and shelter seeking could have profound implications for survival. Clearly, increased vulnerability to predation may be an important, yet largely unobserved, source of mortality for fish escaping trawl gears. By not accounting for predation, many survival experiments may have underestimated mortality values. Moreover, fish may be transported for significant distances in front of and inside the trawl. When they escape, they may be in a very different environment (sub-optimal or inappropriate habitat), where they may be more vulnerable to predators. This may further reduce the effectiveness of innate behavioural responses to predators, make it more difficult to find shelter and food, and reduce shoaling and swimming. The influences of such effects are poorly understood and require substantially more investigation.
Very little is known about the effects of repeated capture and escape of organisms from trawls (but see Broadhurst et al., 2002). In grounds where fishing activity is high, young fish may repeatedly encounter trawl gear. If stress and injury are cumulative, fish may eventually die when they experience several escape processes within a short period, without sufficient time for adequate recovery between the events. This hypothesis clearly requires further investigation. Moreover, learning may play some role here. A fish that has been captured and passed through the meshes of a fishing gear may be capable of escaping more easily on subsequent encounters, and may thereby also have a higher probability of survival than a naive fish. Özbilgin and Glass (2004) provided experimental evidence that haddock can modify their behaviour based on prior experience; one mesh penetration tended to increase the probability of penetration in the next encounter. Özbilgin and Glass (2004) also emphasized, however, that the true nature and effect of the learning abilities of fish in codend mesh penetration behaviour during commercial fishing operations remain to be investigated.
There has been an overall trend towards larger and heavier ground gears in the demersal trawl fisheries, although in some fisheries the current trend is towards lighter or modified ground gears to reduce sea bed impacts (see Valdemarsen, 2004). During herding and capture, fish may collide with the ground gear (e.g. Walsh and Hickey, 1993; Tschernij and Suuronen, 2002). Very little is known of the magnitude of such collisions, although there is likely to be substantial variation in the type and degree of injuries sustained by fish. In Norway, scientists measured a high level of under-gear escape among young cod, and a significant number of these fish exhibit external injuries caused by collisions (Aud Soldal, personal communication). However, it could be assumed that in some cases a fish would have a better chance of survival when escaping through the free space between the rollers of a ground gear than when entering into a trawl and escaping through a codend mesh. In fact, for some fish species, such as cod, escape under modified ground gear could become an alternative to codend selectivity (see page 46).
It is evident that the damage and mortality incurred by fish during their capture and escape from trawls are often caused by a multitude of factors (stressors), and can only rarely be ascribed to a single cause. Some stressors are connected to ambient factors such as water temperature, light conditions, currents, pressure, fishing ground and sea state. Others are capture stresses such as crushing and wounding against trawl netting, collisions with other fish, sustained swimming until exhaustion, and the final passage through a mesh or selective device. Lack of oxygen in a towed fishing gear packed with fish could lead to conditions of anoxia. Finally, there are biological attributes such as fish species, size, age, behaviour and predators. The effects of all these stressors may be cumulative and lead ultimately to mortalities. For instance, while moderate damage to the skin is unlikely to induce initial mortality, injury may provide a sufficiently large physiological stress to - in combination with exhaustive swimming - induce the death of fish (e.g. through metabolic acidosis or osmoregulatory failure). Figure 4 summarizes some of the main factors that may affect a fish captured by and subsequently escaping from a trawl gear.
The results of experiments conducted on post-trawl mortality suggest that for many species it is relatively low, although among some species groups, such as small-sized pelagic fish, mortality may be high. Few studies, however, have adequately and quantitatively explained the full range of mortalities that can occur when fish escape from fishing gears under commercial fishing conditions. A number of mechanisms may cause injury, stress and mortality in escaping fish; the passage through a mesh or a selective device is not the only potentially damaging factor. Fish escaping from fishing gears may suffer immediate as well as delayed mortalities owing to physical injury, exhaustion, disease and predation. Moreover, changes in water temperature, pressure and light conditions may strongly affect the fate of escaping fish. The robustness and ability of various species to withstand the physical disruptions and fatigue associated with the process of capture and escape vary substantially. The smallest escapees often appear the most vulnerable. Until the effects on mortality of various critical factors and their interactions are better understood, there will be a lack of confidence in generalizing escape mortality results to a wider range of fishing conditions, gear designs and operations and fish species. Further work is required to identify the damaging mechanisms that cause injuries.
Research on the mortality of fish escaping from fishing gears has tended to focus on the mortality of fish kept in a sheltered environment such as a sea bed cage, for a relatively short time. Factors such as predation on injured fish and the ability of a fish to recover fully from its injuries or stress are more difficult to monitor, and are therefore poorly understood. The fate of fish after multiple encounters with fishing gears is largely unknown. Moreover, the cumulative effects of all stressors are likely to have a strong influence on the probability of long-term survival. These areas clearly require more investigation.
Methods for assessing escape mortality rates across a wide range of fisheries and environmental conditions are not yet adequate. It is necessary to develop appropriate methodologies, collect more realistic data and obtain a better understanding of the main sources of injury, stress and mortality under various conditions. It should be borne in mind that, owing to natural variation in environmental parameters and in the general condition of exploited fish, there will always be some variability in mortality estimates between experiments, tows and years. Detailed records of relevant environmental parameters and condition indices of target stocks should be collected routinely during survival experiments.