When investigating the survival of fish escaping from a fishing gear under commercial fishing conditions, it is necessary to collect the escaping fish and then either hold them captive for a specific period of time, or tag and release them (preferably at the deph of capture) for later recapture. For the latter approach, very little adequate technology is available, so escapees are usually held and monitored in cages near the fishing ground where they were captured. It is noteworthy that most escape-survival experiments carried out so far have not fully simulated commercial fishing conditions in terms of tow duration, depth, catch size and season. Hence, these experiments may not reflect the full range of possible sources of injury and mortality encountered by trawl codend escapees under normal commercial fishing conditions.
Typically, the mortalities of codend escapees have been assessed in relatively shallow waters during summer months, and under relatively short trawl tows and low catch rates (e.g. Main and Sangster, 1990; 1991; Soldal, Isaksen and Engås, 1993; Lowry, Sangster and Breen, 1996; Sangster, Lehmann and Breen, 1996; Suuronen et al., 1996a; 1996b; Wileman et al., 1999). This chapter focuses on potential ways of improving techniques for reliably assessing the survival of fish that escape from trawl codends across a wider range of fisheries and environmental conditions. Laboratory methods for assessing various types of capture-induced injury and physiological stress are beyond the scope of this discussion.
Investigations conducted in the field have revealed that survival can be markedly affected by methods of collecting, transporting and monitoring the escapees. For instance, the sampling of escapees in the collection cover may cause them substantial damage and stress (Sangster, Lehmann and Breen, 1996; Suuronen et al., 1996b; Wileman et al., 1999; Breen et al., 2002), thereby resulting in overestimates of escape mortality. Suuronen et al. (1996b) observed that small Baltic herring that had escaped from a trawl codend were soon lying against the rear netting wall of the collection cover. The fish had been exposed to a continuous flow of water inside the cover, and were apparently unable to sustain the swimming speed necessary to maintain their position within it. Hence, they were forced against the rear wall of the cover (Figure 6), where they were highly vulnerable to skin abrasion, scale loss and suffocation. Similar observations were made by Breen et al. (2002) on young haddock captured in a cover. The authors suspected that, depending on the length of time spent in the cover (cover exposure time), such entrapment against the cover may cause serious injury and fatalities among fish.
Although many cases have shown that the use of codend covers may lead to overestimates of escape mortality (see above) - unless the necessary precautions are taken - it may also result in underestimates. Breen et al. (2002) demonstrated that the water flow around a codend with a typical cover was significantly lower (by circa 80 percent) than that observed around a normal uncovered codend. The reduced flow means that the codend experiences less drag, and thereby the netting forming the codend will be under less tension. As a result, the passage of a fish through the meshes of a codend may be easier and perhaps less injurious. The passage may be less hazardous also because of reduced water flow outside the codend. Clearly, the presence of the cover around the trawl codend may protect escaping fish from the injurious forces that are normally experienced during passage through a trawl codend. This mechanism means that the cover may contribute to the underestimation of true escape mortality.
Moreover, under normal commercial fishing operations, codend escapees are free to swim away into their normal environment to recover. Enclosure in a cage, often with other damaged fish and in non-favourable environmental conditions, is likely to cause additional stress that may further contribute to mortality, possibly also through crossinfection. Hence, it might be argued that the recovery of a damaged fish is hindered by captivity (for further discussion, see Main and Sangster, 1990; 1991). However, it could also be argued that fish within the cage are protected from predators and other potential hazards. Therefore, great caution is needed when interpreting past and present results. Clearly, in a survival experiment, a fish may die of causes other than capture and escape-related damage. The key issue in assessing the survival of escaping fish is to ensure that the collection and monitoring methodology does not induce stress and injury in the fish that are subject to investigation. Experimentally induced stress and injury should be eliminated to the extent possible. It is important to realize that controls cannot account for all the potential errors and flaws in experimental methodology and design. General standardized methods and protocols to generate reliable estimates of escape mortality rates in a wide range of fisheries and conditions are not yet available. The ICES Topic Group on Unaccounted Mortality in Fisheries (ICES, 2000) provided the following general principles, which should be adopted in survival studies:
The capture and maintenance of specimens should take place without any additional stress or injury to them.
During the transfer from the site of capture to the site of captivity, specimens should experience a minimum level of environmental change.
Conditions in captivity should be stable and mimic as closely as possible the ambient conditions in the wild.
The effects of captivity should be closely monitored; ideally this should involve a suitable control group of specimens.
A full description of any mortality occurring within the experiment must be made in terms of all possible explanatory variables, both experimental and environmental.
These principles are still highly relevant and should be followed to the extent possible. The following sections give suggestions and practical recommendations for putting these principles into practice.
The first critical task in a survival study is to find (or develop) a method for collecting the fish escaping through trawl codend meshes or other selective devices. Collection should be done without causing any extra stress and injury to the fish. There are many technical variations for capturing escaping fish. Until the early 1990s, the most common method was a traditional codend cover, in which the rear part was designed as a holding cage (Figure 6). The cover was closed at the beginning of the tow and released at the end, when the codend was at the surface. Therefore, escaping fish were sampled throughout the entire haul.
To reduce the effects of cover exposure on escapees, the haul duration was often limited to periods that were much shorter than those used in normal commercial practice (e.g. Suuronen et al., 1996b). However, sampling durations were in most cases too long from the escapees point of view (owing to cover exposure effects). For example, the smallest escapees tended to become trapped against the rear netting wall of the cover, sometimes only a few minutes after escape from the codend (e.g. Soldal, Isaksen and Engås, 1993; Suuronen et al., 1996b).
Breen, Sangster and Soldal (1998) and Breen et al. (2002) demonstrated that the period during which escaping fish are sampled and forced to swim in the cover (or in the cage attached to the cover) may have a significant effect on their subsequent survival. Cover-induced mortality is linked to many stress-inducing factors, among which the swimming ability of the fish is one of the most important. Hence the high mortality seen in some investigations, particularly among the smallest escapees (e.g. Suuronen, Erickson and Orrensalo, 1996), could be at least partly the result of cover-induced mortality. Small fish may become too weak and exhausted by the capture process to have enough energy left to swim within the cover; the water flow then forces them against the cover wall.
To reduce cover-induced mortality on young Baltic herring escapees, Suuronen et al. (1996b) reduced the sampling (and trawl towing) time to only a few minutes. The relevance of such an approach, however, should be questioned because it does not reflect commercial fishing conditions. Moreover, such a reduction of the sampling (and towing) time means that insufficient numbers of the target species are often caught and collected in the sample, particularly when working with groundfish.
Clearly, such approaches and techniques have not been adequate for assessing the survival of escaping fish under commercial fishing conditions. A sampling method is needed in which the duration of the tow is not restricted by the method used to collect escapees.
To avoid decompression injury and temperature shocks to the escapees, Soldal, Isaksen and Engås (1993) used acoustic, and Suuronen et al. (1996a; 1996b) mechanical, releasers to close and release the cages from the cover remotely (at the depth of capture). Nevertheless, these systems still had the problem that sampling of the escapees was possible only at the beginning of the tow; collection was limited to the very first part of the tow. To minimize injury from cover exposure and simultaneously permit realistic tow durations, later designs allow the sampling of escaping fish at any moment of a haul (e.g. Lehtonen, Tschernij and Suuronen, 1998; Erickson et al., 1999; Wileman et al., 1999; Ingolfsson, Soldal and Huse, 2002).
The system developed by Lehtonen, Tschernij and Suuronen (1998) has the following principle (Figure 7): the collection cage attached to the cover has two gates that can be closed separately during the tow to conduct the sampling. The collection cage can then be remotely detached from the cover with the captive fish contained within. With this technique, sampling can be conducted at any moment of a haul, and the sampling period can be controlled precisely and kept substantially shorter than in previous experiments. Hence, the sampling duration is not dependent on the tow duration. Likewise, the haul duration is not affected by the sampling process.
Survival can be assessed for short and long tows, and for small and large catch quantities. Cover exposure time (i.e. sampling time) can be kept short enough to avoid cover-induced injury, but long enough to provide adequate numbers of escapees. Suuronen, Lehtonen and Jounela (2005) successfully applied the system developed by Lehtonen, Tschernij and Suuronen (1998) to assess the survival of Baltic cod escaping through various trawl codend meshes (Figure 8).
A relatively similar technology using a double set of acoustic releasers was developed by Wileman et al. (1999). However, instead of using a door to close the rear part of the cover, their experiments used a rope to close the rear part of the cover netting. This technique was further developed into a solid door system (Ingolfsson, Soldal and Huse, 2002). Erickson et al. (1999) further modified the technique of Lehtonen, Tschernij and Suuronen (1998) when investigating the survival of Alaskan pollock escaping through a square mesh upper panel. They used an underwater camera to monitor the cover and the collection cage during the tow.
The collection of escapees began only after an adequate catch had been accumulated in the codend. When there were adequate numbers of escapees in the cage, it was remotely closed and released. Hence, the sampling period was restricted to the time that was required to collect the sample. No additional stress and injury was caused to fish. When there is high variation in fish density, a compromise often has to be made between sampling time and sample size.
It is largely unknown how long individual fish swim in trawls and when they escape. Some may swim within the gear throughout the tow duration, while others may escape immediately after capture. It is possible that the survival of fish escaping at the beginning of the haul, when there are fewer abrasive objects within the codend and when fish have spent only a short time within the gear, is higher than that of those escaping at the end of tow, when the codend may be full of debris, fish and other substances, and when the flow patterns may be less favourable for controlled and oriented swimming. Moreover, fish escaping at the end of the tow may be more exhausted than those escaping earlier. Therefore, it is important to assess and understand when and where during the tow the escapees should be sampled.
Figure 9 demonstrates a typical four-hour commercial demersal trawl tow (upper figure, A). Fish are captured along the path of the tow. The amount of time spent inside the gear varies considerably among individual fish. Some fish escape almost immediately, while others may swim for a long time within the gear before escaping. The lower figure (B) demonstrates two sampling approaches that have generally been used in survival experiments. Tow b-1 presents a sampling approach where the tow ends over a shallow bank so that the collection cage can be released on to a sheltered site. Therefore, the experimental haul is conducted near the shore in an area where fish density is typically relatively low. Owing to methodological restrictions, fish are sampled during the entire tow, i.e. sampling time is the same as towing time. The advantage of this technique is that it does not require the transfer of cages and escapees after release. The disadvantage is that the catch rates may be small and the tow does not represent a typical commercial trawl tow regarding catch size, number of escapees and capture depth. This type of approach has commonly been applied in survival experiments (e.g. Suuronen et al., 1996a).
A) A 'typical' commercial 4 hour-tow
B) Sampling with short tows (sampling time = towing time)
Tow b-2 (Figure 9, B) presents the other typical approach used by experiments to collect escapees from trawls. In this approach, escapees are sampled in areas of high fish density and where a commercial fishery is typically operated. The sampling time is the same as the towing time (e.g. it is restricted to 30 minutes to minimize additional stress and injury to escapees from cover exposure). After the collection cage has been released, it may be transferred to a more sheltered cage site. The sample of escapees in this approach is likely to be more representative than that of Tow b-1, but the potential stress and injury induced by the cage transfer process may cause difficulties in the final interpretation of the results.
An exp. tow with optional sampling duration and timing
Figure 10 demonstrates a more advanced approach where technology is available to collect a sample of escapees from any point and for any length of time during a commercial tow (e.g. the technology developed by Lehtonen, Tschernij and Suuronen 1998). The sample can be taken at the beginning, the middle or the end of the tow. Clearly, this approach allows a more realistic assessment of mortality. However, when Suuronen, Lehtonen and Jounela (2005) applied it in their Baltic cod survival experiments, they sampled fish during only the last 20 minutes of the tow in order to release the cages at a sheltered cage site. This approach was chosen because these authors did not want to have to transfer the cages after release. If the cages could have been left at the site of release at any point along the trawl path, samples could have been taken at any moment of the tow (and without the transfer of escapees after cage release). However, this would have made the daily monitoring of cages difficult because of the long distances, and would have required an area in which no other trawl towing was conducted (because the cages could easily have been damaged or lost as a result of other fishing activity). Nevertheless, if technically and logistically possible, this last approach would be highly recommended. Using the methodological approach presented in Figure 10, it is possible to assess the potential difference in survival of escaping fish at any particular moment of the tow.
A statistically adequate sample of fish from various sizes groups should be collected in each cage - a goal that is not always easily accomplished. The number of fish collected depends on the escape rate and sampling duration. A compromise is usually needed between sample size and sampling duration (which should be as short as possible). The escape rate is not normally known and has been quantified in only a few cases. In addition, it is likely to vary substantially within the same tow and between tows. The escape rate depends on numerous factors such as fish species and size, fish density, catch size and composition, geographic location, time of year, water temperature, amount of available light, and selectivity of the experimental gear.
Pilot tows are often necessary in order to predict escape rates before experiments are initiated, and they may provide some idea of potential escape rates and appropriate sampling times. Generally, relatively high escape rates have been encountered in pelagic fisheries; it is not uncommon to have several hundred escapees in a period of only one minute (e.g. Suuronen et al., 1996b; Erickson et al., 1999; Pikitch et al., 2002). This may occur, for instance, when a large school enters a trawl; between schools there may be no escapes. On the other hand, groundfish fishery escape rates may be somewhat more constant, but may be of only very few individuals per minute (e.g. Sangster, Lehmann and Breen, 1996; Suuronen et al., 1996a). Occasionally, however, high escape rates have been encountered with groundfish (e.g. Soldal, Isaksen and Engås, 1993). Adjusting the sampling time according to the escape rate with the help of an on-line underwater video camera and remote-operated cage closure and release mechanisms is an alternative that deserves attention (see Erickson et al., 1999; Pikitch et al., 2002; Ingolfsson, Soldal and Huse, 2002). With such a technique, it is possible to collect a statistically adequate number of escapees in each cage. This would reduce the number of invalid tows to close to zero.
As described in previous sections, most survival assessment techniques use some form of codend cover to collect escapees. The cover (and the collection cage attached to it) should be designed so as not to induce skin injury and other damage to escapees. It should be made of soft knotless material to minimize abrasive injury, and have relatively small mesh size to retain small fish and prevent the meshing of fish. It is also necessary to restrict the water flow inside the cover. A supporting hoop or frame system (Figure 11) may be necessary to prevent the cover from collapsing against the codend (see e.g. Sangster, Lehmann and Breen, 1996; Wileman et al., 1996; 1999). Alternatively, special kites can be used to keep the cover open (Madsen, Hansen and Moth-Poulsen, 2001). There must be adequate space between the cover and the codend in the area where escapees are collected. The cover affects water flow in and around the codend (e.g. Wileman et al., 1996; Breen et al., 2002; Madsen and Holst, 2002). This may have a substantial effect on the behaviour and survival of escaping fish (see above). Therefore, the flow in and around the codend should approximate that of actual commercial situations, where there is no cover around the codend or selective device. In particular, the design should ensure that the water flow inside the cover, outside the codend, is as close as possible to a commercial tow, and that there is no flow in the rear part of the cover, where fish may recover from exhaustion. Water flow patterns can be affected by constructional details of the cover. The twine thickness and the mesh size of the cover, as well as the towing speed, should be carefully considered in each particular experiment. In addition, the visual contrast between the cover netting and the surrounding water may affect the escape behaviour of fish swimming in the codend.
Generally, the visual contrast of the front and middle parts of the cover should be minimized in order to obtain escape behaviour that is reflective of commercial conditions. Furthermore, the cover must be large enough in all dimensions. Erickson et al. (1996; 1999) showed by video observations that walleye pollock escaped along the entire length of the escape panel to 18 m in front of the catch bulge. Hence, for this species, a cover installed only a few metres ahead of the codend would not collect those fish that escape from the front part of the codend. Such fish may show different mortality from those escaping from the aft part of the codend. Therefore, if the experiment is supposed to study the overall mortality of escapees, the codend cover should be installed to cover the whole length of the codend and other potential selection devices (i.e. the whole selective area). It is also important that the collection process and the proper performance of the cover and collection unit are monitored with underwater camera; there is always a possibility that some critical gear component becomes twisted or blocked, preventing normal performance.
Once the specimens have been collected in the cover or the collection cage, they usually have to be transferred into a sheltered and stable environment where they can be held without large risks of additional damage and be monitored efficiently. However, the transfer process can be risky (e.g. Lowry, Sangster and Breen, 1996; Erickson et al., 1999). Clearly, transfer should involve minimum environmental changes and stress for the escapees. Abrupt changes, for instance in water temperature, hydrostatic pressure, salinity, water flow or ambient light level, may affect the stress and survival of fish (e.g. Olla, Davis and Schreck, 1998; Wileman et al., 1999). Hence, all changes in environmental parameters should be avoided during transfer. Transporting should take place at a slow speed avoiding excessive water flow, and fish should be monitored throughout the transfer process. Fish should arrive at the cage site in good condition. To overcome some of the common problems of underwater fish transport, the collection cover/cage may be placed in a protective container for the towing process (Lowry, Sangster and Breen, 1996; Sangster, Lehmann and Breen, 1996). A rigid container can be towed far more easily and quickly to the cage site, while fish inside the cover are protected from excessive water flow. However, this method has several practical difficulties and can be very labour-intensive (see Lowry, Sangster and Breen, 1996). Controlled decompression protocols should be used if the transfer involves a marked change of fish depth. If the depth has to be changed, there is a risk of fishs gas bladders being overinflated to the point of rupturing as a result of depressurization. Erickson et al. (1999) ensured that cages remained continuously deeper than 10 m, and transferred them by using constant video surveillance and monitoring of the cage and environmental parameters (see also Pikitch et al., 2002).
The conditions where the fish are monitored should mimic as closely as possible those experienced by that species in its normal life (after escape). On the other hand, the monitoring site should be sheltered and practical for observing the cages and the fish inside. It is almost inevitable that some compromises have to be made in terms of normal natural (commercial fishing) conditions and practicability. To overcome, or at least mitigate, the stress induced by confinement in a cage, many field experiments keep the sampled fish in or close to the area where they were caught but in a specific cage site (e.g. Ingolfsson, Soldal and Huse, 2002; Suuronen, Lehtonen and Jounela, 2005). At cage site, fish are transferred from the collection cage to the separate larger monitoring cages (e.g. Main and Sangster, 1990; Suuronen et al., 1996b), or the collection cages are serving as monitoring cages, i.e. there is no transfer of fish from one cage to other (e.g. Suuronen et al., 1995; Suuronen, Lehtonen and Jounela, 2005). Fish can be held in cages that are suspended in the water column or placed on the sea bed (Figure 12). Suspended cages are best used with pelagic species (e.g. Suuronen et al., 1995; 1996b) but have also been used with demersal species (Jacobsen, Thomsen and Isaksen, 1992; Soldal, Isaksen and Engås, 1993; Suuronen et al., 1996a; Ingolfsson, Soldal and Huse, 2002). The major disadvantage with pelagic cages is that they are highly vulnerable to changes in environmental conditions.
Wave action and strong currents can force the cage to move considerably, causing additional stress to fish. Moreover, vessel traffic may cause severe problems by cutting off the mooring ropes of cages, sometimes dragging them for distances along the vessel. The use of pelagic caging with demersal species is questionable. Strictly, demersal species may need the sea bed for at least part of the day, and are likely to find much of their food there. Isolation from the sea bed could lead to additional captivity-induced stress. Sea bed cages are usually more stable in terms of environmental conditions. There are several critical choices to be made regarding cages and cage sites. Sea bed cages should be able to stay freely at the bottom; therefore, a rectangular shape is most suitable. A rigid frame is usually needed. Pelagic cages can be cylindrical and can be suspended; therefore, they can be constructed of hoops and netting, i.e. have a simpler construction. Cages should be large enough to allow a large sample of escapees to be held without any additional stress and injury. Some assessment should be made of the average number of fish to be held in cages and the amount of water required per fish. Netting for the cage should be as soft as possible, to avoid abrasions, and of relatively small mesh size, to avoid meshing. Environmental conditions, as well as fish behaviour, in cages should be constantly monitored. Abrupt changes in water temperature, salinity and quality should be avoided as much as possible by careful location of the cage site (see Suuronen, Lehtonen and Jounela, 2004). Fish held in cages should be offered suitable food regularly.
When a fish that has escaped from a trawl codend survives for a certain period in a monitoring cage, it has obviously recovered from its traumatic experience. If it dies during the monitoring period, there is often uncertainty about the cause of death. That is, fish held in monitoring cages may have died of causes other than capture and escape-related damage. The duration of the period for which escapees are monitored may have a large influence on the observed mortality (Wassenberg and Hill, 1993; Suuronen et al., 1996b; Pikitch et al., 2002).
Mortality has generally been shown to peak in the first two to three days after escape, the highest mortality usually occurring during the first day (e.g. Wassenberg and Hill, 1993; Chopin et al., 1996a; Sangster, Lehmann and Breen, 1996; Suuronen et al., 1995; 1996b; Wileman et al., 1999). The rate declines with time, usually reaching a minimum after one or two weeks. Mortality assessments over only a few hours may not be adequate for measuring short-term capture and escape-induced mortality. On the other hand, observation periods longer than one to two weeks may not be useful because of secondary infections and the stress connected with captivity (Figure 13). Interpretation of results may become more difficult the longer the experiment continues. Delayed deaths in monitoring cages are often correlated to the onset of various skin infections and other problems (e.g. deteriorated caudal fins, lesions and sores).
These types of secondary infections have been described for many species (e.g. Main and Sangster, 1990; Soldal, Isaksen and Engås, 1993; Suuronen, Erickson and Orrensalo, 1996; Erickson et al., 1999). In fact, for many fish speces, cage methodology may be suitable for studying only primary mortality. Clearly, assessing long-term mortality caused by secondary infections and predation may require other technologies (tagging, etc.).
The potential effect of captivity on escapees during the monitoring period can be assessed through the use of adequately captured and held controls. That is, a representative group of fish of the same species and size are held in captivity in similar cages and in the same area as the escapees, and their survival is assessed in the same way. If captivity has no lethal effect on the captive fish (over a specific period), there should be no observed mortality in the control group. When there is mortality among controls, it is important to know why they died. Death may be due to captivity stress, but it may also at least partly be due to natural mortality over that period, or to stress and injury from the capture of control fish. This must be known when interpreting the mortality for escapees held in cages. The equation in which control mortality is subtracted from trawling mortality to derive escape mortality may not always be quite true.
One of the major difficulties with controls is the question of what aspect of the experiment they are supposed to be a control for. The control fish should represent as closely as possible the population of fish that are in the test groups (treatments), except regarding the test variable(s), i.e. the capture and escape process. Usually, control fish have not experienced all of the aspects of the experiments, such as capture and confinement in the cover and collection cage and transfer to the cage site. In fact, in order to be able to measure the potential mortality caused by the collection and transport of escapees, there should be at least two different types of controls: those that determine the effects of captivity in the holding cage; and those that determine the effects of collection and transfer to the cage site (see Suuronen et al., 1996b; and Wileman et al., 1999 for more details).
In a survival experiment there is always a possibility of severe cumulative effects that are difficult to detect and measure, even when there is an effective arrangement of controls. For instance, a control sample may show no mortality, although the fish may have suffered substantial captivity-induced stress. On the other hand, for escapees that have experienced substantial initial stress from the capture, escape and transfer, the additional captivity-induced stress may be the final cause of mortality. That is, these escapees may have survived if they had not been held in monitoring cages. If the controls show no mortality, these escapees would be interpreted as dying as a result of the capture and escape process, and the escape mortality would therefore be overestimated (Figure 14).
This is a particularly high risk among the smallest size groups, because small fish are usually most sensitive to handling. Hence, the stress and injury caused by captivity should be minimized or, if possible, totally eliminated by improved methodological approaches (Figure 14).
Relatively little systematic work related to the capture of control fish has been carried out. In most survival studies, the capture of controls has been of secondary importance compared with the collection of treatments. However, the capture of adequate controls is one of the most important tasks in a survival study. Very often control samples are inadequate because they differ in body size from the treatment population. In particular, the smallest sizes are often missing from the control group. The lack of adequate controls has made the results of many survival studies practically useless.
Most existing capture methods are designed to catch adult fish. Young fish that are needed as controls are often more difficult to catch. Moreover, young fish are often more fragile and sensitive than adults, and are easily injured and stressed during the capture and handling process. All major fishing gears involve some degree of stress and injury to fish. The severity of these injuries depends on the gear type, the fish species and size, and environmental conditions. Contacts between the fish and the gear should be minimal. In some circumstances, trap-net, seine net or hook and line fishing may offer the most practical capture method. Fish swim voluntarily into the trap. If properly constructed, a trap-net may catch fish with relatively little stress and injury. Wounding from hooking can be reduced by using barbless hooks, and potential skin abrasion in seine fishery can be reduced by using soft netting materials.
Clearly, the choice of capture method for controls is critical. Suuronen, Lehtonen and Jounela (2005) used hook and line, gillnets and eel traps to capture control cod, but had relatively little success. Demersal pots that were baited were much more successful in the capture of control fish. Pots were also used by Ingolfsson, Soldal and Huse (2002) to capture cod and haddock, and by Erickson and Pikitch (1999) to capture sablefish. Pikitch et al. (2002) captured walleye pollock controls with a purse seine. No significant differences in mortality were found between experimental fish and some of the controls, but differences were noted among seining procedures. This probably caused extreme variability of control mortality. Pollock held in the control cage that showed the highest mortality were crowded to such an extent during the seining process that they thrashed and boiled on seine webbing that was unintentionally pulled tight near the surface. Although the webbing was immediately released to provide space for swimming, this action probably caused extensive skin damage and stress (Pikitch et al., 2002). Seine-caught pollock held in another cage were not crowded during the capture process; mortality in this cage was only 2 percent, and almost none of the fish showed skin damage.
If technically possible, control fish should be caught through their own voluntary swimming into the control cage - various types of attraction device may be used to improve the catchability of a cage. Modified trap-nets and pots may be applicable in some fisheries (Figure 15). The problem may be poor capture efficiency, which can be substantially improved through the use of bait. Another issue is how to transfer the fish from the trap-net to the monitoring cage without causing any extra stress. The transfer distance should be as small as possible, and in the best case there should be no need for transport, i.e. the trap-net should be in the immediate vicinity of the cage. A modified trap-net may itself also act as a monitoring cage (see Suuronen, Lehtonen and Jounela, 2005).
The tag and recapture methodology has been used mainly to estimate the survival of discarded and released fish (e.g. Pacific halibut; Trumble, Kaimmer and Williams, 2000). This method may provide an accurate indication of long-term survival, but it does not explain much about those fish that have died (when, where and why). The methodology also suffers from being extremely labour-intensive. Tag and recapture methods may also be used when studying the survival of escapees. With current tagging technology, it is possible to mark fish with special tags that can be registered by instruments that are attached inside the trawl codend, for instance. With the help of such technology it is possible to register the passage of marked fish through the codend. Hence, by first tagging a certain number of fish within a fishing ground and then trawling in this ground it is possible to estimate how often a particular fish is captured in the trawl and how often it escapes successfully. It is likely that this type of new tagging technology will be used in the near future in many survival studies and other applications.
The wide variety of potential stressors in commercial fishing situations and the lack of controlled experimental conditions make it extremely difficult to conduct studies that systematically examine the interactions among stressors. To simplify the damaging mechanisms and repeat the treatments in a controlled environment, a variety of simulated laboratory (tank) experiments have been conducted to determine the causes of injury and death of escaping fish (e.g. Soldal, Isaksen and Engås, 1993; Broadhurst, Kennelly and Barker, 1997; Broadhurst, Barker and Kennelly, 1999; Davis, Olla and Schreck, 2001; Davis and Olla, 2001; 2002; Davis, 2002).
Laboratory studies provide a controlled means of investigating fish survival in regard to various stressors. It should be borne in mind however that the laboratory environment does not mimic the whole capture and escape process, nor does it exhibit the variability that is typically experienced in natural conditions. It is difficult, or perhaps impossible, to simulate all potential capture stressors in the laboratory. Fish that are held in laboratory conditions experience sensory-deprived environments, which can result in behaviour and stress responses that do not mimic normal responses in the field. Hence, the results of laboratory studies cannot be used to make direct conclusions about the survival of fish in commercial fishing conditions.
Nevertheless, laboratory experiments can be extremely useful and cost-effective in investigating stress responses and assessing injuries. They allow the systematic determination of the general behavioural, biological and physiological principles of stress response up to mortality in different species, and this is rarely possible in field conditions (see e.g. Davis, 2002). For instance, interactions between capture stressors (e.g. towing in a net and hooking) and temperature have been examined in laboratory conditions on walleye pollock, sablefish, Pacific halibut and lingcod using changes in behaviour, blood physiology and mortality as measures of stress (Davis, Olla and Schreck, 2001; Davis and Olla, 2001; 2002; Davis, 2002). In such experiments, researchers can focus on the mechanisms of interest and control all the others (see also Broadhurst, Kennelly and Barker, 1997; Broadhurst, Barker and Kennelly, 1999; Ryer, Ottmar and Sturm, 2004). Hence, cause and effect relationships can be established.
So far, most laboratory studies have been directed towards assessing the potential injury to fish passing through netting meshes or other selective devices, as well as assessing the exhaustion caused by forced swimming inside towed gear. The results of laboratory experiments could be used to calibrate measures of fish conditions (i.e. wounding, behavioural deficits) with mortality (see Davis, 2004).
Further observation of fish condition prior to caging in the field could be used to verify these calibrations. Estimates of the conditions of escapees and discards in the field could then be used to predict escapee and discard mortality without caging. Such an approach could expand significantly the range of fishing conditions that are studied, eliminate artefacts from caging studies and lead to more accurate estimates of unintended fish mortality. Increased mortality information could be used to develop new ways of minimizing unintended mortality in a wide range of fisheries.
The development of effective gear modifications to enhance the escape and survival of escapees requires a good understanding of how fish react to fishing gear and selective devices, including in conditions where vision is limited or inoperative. Species- and fisheries-specific variability in behaviour patterns is substantial and influences the types of gear modifications that are required. Few designs have the potential for application across different fisheries; significant fish- and fisheries-specific modifications are often required. Hence, observations and measurements of fish behaviour under various conditions in each particular fishery should be carried out in order to gain more specific understanding. Unfortunately, observations under limited levels of light require special technology that is not generally available.
Various types of light-sensitive video cameras have been used in many field studies, but in most cases they are not sensitive enough for the conditions that prevail during fishing. In many cases, it is necessary to use artificial light, which may however distort the behaviour of target species. Infrared technology has been used in laboratory experiments (e.g. Olla and Davis, 1990) but has rarely been applied in the field (Olla, Davis and Rose, 2000). Other possible techniques include acoustic and laser technologies, such as dual frequency identification sonar (DIDSON) technology. Underwater observation techniques have recently been reviewed by Graham, Jones and Reid (2004).
Most survival studies so far have resulted in qualitative mortality estimates of limited accuracy. Survival figures have been affected by inferior methods of collecting, transporting and monitoring escapees, and no investigation has been conducted without methodological compromises. Poor replication of commercial conditions is an additional concern. Stress, injury and mortality may vary, largely depending on fishing conditions and practices, so it is not sufficient to conduct survival studies under a single set of conditions and expect that the results will be applicable to the fishery in general. A survival study should cover the full range of fishing conditions for the gear type and fish species of concern, including variations in fishing practices. Clearly, there is substantial potential to improve the techniques and systematize the procedures used in survival experiments.
Control fish are used to measure the levels of mortality occurring among captive fish either as a direct result of the stresses of captivity or owing to natural mortality over that period. The use of control fish, and the size and design of cages and their layout on the sea bed should be carefully planned to take potential problems into account and to minimize uncertainties. Experiments should describe in detail what the controls are meant to demonstrate. Different factors and handling procedures may need separate controls for the final interpretation of results. Usually, the use of controls does not exclude all the potential error factors, but many uncertainties can be minimized.
Although laboratory studies can be criticized for lack of realism, they provide a controlled way of investigating fish injury and survival with regard to various stresses. Laboratory experiments can most effectively develop knowledge of key stressors and accumulative effects. To increase understanding of and predict overall escape mortality, mortality may - at least in some cases - best be addressed through a combination of field experiments under realistic fishing conditions and controlled laboratory investigations of various key stressors.