ICES Journal of Marine Science: Journal du Conseil Advance Access originally published online on July 12, 2007
ICES Journal of Marine Science: Journal du Conseil 2007 64(8):1543-1550; doi:10.1093/icesjms/fsm099
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Separating species using a horizontal panel in the Scottish North Sea whitefish trawl fishery
Fisheries Research Services Marine Laboratory, PO Box 101, 375 Victoria Road, Torry, Aberdeen, AB11 9DB, UK
Correspondence to R. S. T. Ferro: tel: +44 1224 876544; fax: +44 1224 295511; e-mail: ferro{at}marlab.ac.uk
Ferro, R. S. T., Jones, E. G., Kynoch, R. J., Fryer, R. J., and Buckett, B-E. 2007. Separating species using a horizontal panel in the Scottish North Sea whitefish trawl fishery. – ICES Journal of Marine Science, 64: 1543–1550.In the North Sea, Scottish vessels target haddock, cod, whiting, monkfish, saithe, and flatfish in a mixed whitefish trawl fishery. These species mature at different sizes and hence have a range of minimum landing sizes. Their different shapes and swimming capabilities imply different selection characteristics when escaping from trawl gear. However, they are often caught at the same time on the same grounds. Optimal exploitation can only be achieved by ensuring that the selection of each species varies appropriately with length during the fishing process. This paper describes one part of a large European project to develop species-selective trawl gear to improve the exploitation pattern of North Sea cod, while maintaining the catch of other important commercial species. A gear suitable to the Scottish mixed whitefish fishery was fitted with a horizontal panel in the tapered part of the net to separate species into an upper and lower compartment. Trials were conducted on research vessels to measure separation performance for nine species in different light conditions, at different towing speeds, and with different lengths of panel. Most haddock, whiting, and saithe pass above the panel, whereas most cod, flatfish, and monkfish pass below it. Towing speed and panel length had no significant effect on separation. At lower light levels during the night (April at latitude 58° to 61°N), fewer dab, sole, plaice, and cod pass below the panel. Observations and measurements of fish behaviour using acoustic methods are described. They suggest that the height at which fish enter the net mouth may be influenced by light level and water clarity.
Keywords: cod, demersal otter trawl, light level, separating panel, species separation, towing speed
Received 1 September 2006; accepted 15 April 2007; advance access publication 12 July 2007.
 |
Introduction |
|---|
 Top Introduction MethodsResultsDiscussionReferences |
|---|
Cod (Gadus morhua) stocks in some European waters have been declining for several years. In principle, large increases in effective mesh sizes of gears that catch cod could reduce the fishing mortality rate of this species, particularly juveniles. However, to have a meaningful impact, the required increases for the main fleets exploiting cod would severely affect the catch of several other target species of roundfish, flatfish, and monkfish (Lophius piscatorius), thereby risking the viability of mixed fisheries in the Northeast Atlantic (Anon., 2001). Furthermore, the potential for discarding cod would be high, if vessels continued to target other species with available quota once the cod quota was taken. So to protect cod, it might be necessary to reduce the quotas of other species, close areas, or cease all fishing once the cod quota had been taken. These measures, although potentially effective, might result in an underutilization of resources, threaten the viability of fisheries, and, in the case of area closures, might displace fishing effort elsewhere.
An alternative option is to modify the fishing gear to reduce the catch of cod while maintaining the catch of other target species. The so-called "species selective" gears make use of fish behaviour to separate the different species during the catching process so that each is subject to a selection process more appropriate to its size and morphology. The result should be that exploitation patterns are closer to optimum for each species.
This paper reports the results of initial trials to develop a species-separating trawl suitable to the Scottish mixed whitefish fishery in the northern North Sea. The fishery targets haddock (Melanogrammus aeglefinus), whiting (Merlangius merlangus), cod, monkfish, and flatfish. This work formed part of an EU-funded project "Recovery", started in 2002 to develop such gears in this whitefish fishery. An important aspect of the project was the close cooperation with industry through an industry liaison group.
Separator trawls were first studied in Scotland in the 1970s and 1980s after the introduction of underwater television vehicles provided the means to observe fish reaction to fishing gear. Main and Sangster (1985) described the performance of a horizontal separator panel inserted the full length of the net from the groundrope centre, splitting the net into two compartments, and each ending in a separate codend (Figure 1). Other workers have studied separation, mainly in Northeast Atlantic fisheries (Hillis and Carroll, 1988; Thomsen, 1993; Wileman, 1995; Engås et al., 1998), but also worldwide (Hickey et al., 1995; Wardle, 1995). This paper describes the further development of the horizontal separator panel for commercial use and assesses separation performance at different speeds, under different light conditions, and with different panel lengths. A second phase of the project aims to develop a secondary size-selective device in the lower compartment to release cod.
|
|
Methods |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Sea trials
Trials were conducted on two research vessels on commercial fishing grounds in ICES Area IVa. From 5 to 16 April 2004, the Scottish stern trawler FRV "Scotia" (LOA 68.6 m) fished to the north and west of the Shetland Isles in depths of 102–117 m, and then in the Moray Firth in 70–75 m depths. From 16 to 26 November 2004, the Norwegian research vessel FRV "Johan Hjort" (LOA 64.4 m) fished in the eastern North Sea, to the west of Stavanger (123–190 m depth). The single trawl used on both vessels was a Jackson heavy rock-hopper net with a 470-mesh fishing circle and 450 mm diameter rock-hopper groundgear. The trawl was fished with 73 m single sweeps, 55 m double bridles, and spread by Morgère Type R otter boards, each weighing 2000 kg. The door spread was 69–85 m on FRV "Scotia" and 92–97 m on FRV "Johan Hjort".
On FRV "Scotia", observations were made using a remote-controlled towed vehicle (RCTV), equipped with a low-light silicon-intensified target video camera and multibeam sonar (Reson SeaBat, 455 kHz) orientated downwards to ensonify a 90° slice through the net (Figure 2). Outputs from the camera and sonar were viewed in real time, using a fibre-optic umbilical to the ship and recorded as digital video files.
|
On FRV "Scotia", the 80 mm (green twine) separator panel was initially rigged with the centre of its leading edge in line with the groundrope centre (Figure 3). Two ropes were rigged from the separator panel quarters to the quarters in the trawl's lower belly sheet to adjust panel height. This height was set at 0.75 m at the start of the trials. Sonar measurements estimated the panel to be
1.9 m above the bottom sheet at the selvedges and 1.1 m in the centre of the net. At the next join, 9 m behind the ground gear, the separation between the bottom sheet and the panel was 0.3–0.6 m; at 18 m, the separation was 0.2–0.3 m; and at 27 m, there was no visible gap between the panel and the bottom sheet.
|
The panel was then successively cut back to 11 and 19.5 m from the groundrope centre. In these cases, the panel was re-rigged with small wings on the leading edge, which were laced into the trawl belly down a mesh bar. By increasing or reducing the wing length, the height of the panel above the belly sheet could be adjusted (Figure 1). Sonar observations indicated that, with the panel cut back to 11 m, the leading edge was 1.2–1.5 m above the bottom sheet, and the gap was 0.2–0.5 m at the second join behind. When cut back to 19.5 m behind the groundgear centre, the leading edge was measured at 0.8 m above the bottom sheet. On FRV "Johan Hjort", data were gathered with the separator panel set at 13 m aft of the groundrope centre.
Haul duration was typically 2.5 h on FRV "Scotia" and 1.5 h on FRV "Johan Hjort". Towing speeds were classified as either normal (2.7–3.5 knots) or fast (over 3.5 knots). Hauls were undertaken during the day and the night. To quantify the separation between the upper and lower compartments, 40 mm codends were attached to the upper and lower extensions. Once hauling had taken place, the catch was sorted and total weights of each species noted. Length frequencies were taken for nine species, with subsampling when catches were large.
Behaviour data analysis
During selected hauls on FRV "Scotia", acoustic video was collected using the RCTV to examine the height of fish entering the net with different panel lengths and at different times of the day. The RCTV was maintained in one position above the net for 30 min or more. Self-recording, net-mounted, low-light video cameras and a digital still camera were also deployed in key positions on the net.
Figure 2 shows a frame grab from an acoustic video file. The net volume and separator panel are distinguishable. Individual fish echoes are identified by their characteristic persistence and movement and are generally easily distinguishable, unless they occur in a very dense shoal. Such shoals are characteristic of pelagic species, whereas gadoids such as haddock and whiting tend to enter the net as less tightly packed groups. Apart from these characteristics, however, it is not possible to distinguish species on the sonar. Flatfish were assumed not to register. Any dense schools were excluded and echo counts were assumed to be gadoid species. The validity of this observation technique for detecting gadoids has been demonstrated by comparison of echo count and total gadoid catch (Reid et al., 2002). The video was analysed using proprietary coordinate grabbing software to give information on net dimensions and entry position of fish in relation to the panel and to the bottom sheet during both daylight and darkness.
Separation data analysis
The aim of the analysis was to assess whether there was a significant effect on separation caused by day/night, speed, or panel position. However, this was complicated by the lack of balance in the experimental design (Table 1). The two boats fished at different times of year and different gear combinations were tested on different fishing grounds and boats. The species and size compositions of the catches also varied greatly between fishing grounds, and there were few hauls with the same gear combination. The data were therefore analysed as if they came from three experiments, one on each ground. Differences in separation within fishing grounds were investigated first. For example, did speed affect separation around Shetland? Did moving the panel from 11 to 19.5 m affect separation in the Moray Firth? Only when there were no differences within grounds were differences between grounds considered.
|
The data were analysed using the standard generalized linear mixed model for size-selection data (Millar and Fryer, 1999). The proportions of fish entering the upper compartment in each haul were assumed to be binomially distributed and, following preliminary analysis, related linearly to fish length on the logistic scale. The parameters of the logistic curves were then assumed to vary randomly between hauls. The models were fitted using routines based on the generalized linear mixed modelling facilities in the GenStat statistical package.
The model can be written in terms of its random and fixed (or systematic) components as:
|
|
|
|
|
|
|
|
|
|
|
Results |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Gadoid behaviour
Water clarity on the Shetland fishing grounds was exceptionally good during daytime, with visibility of up to 20 m at depths over 100 m without artificial light. However, the numbers of gadoids seen in the four hauls examined were generally low, and species identification was not always possible. Limited video footage was obtained of the shortened panels because of the poorer visibility at this point in the net and the low height of the panel.
Acoustic data were collected from five hauls at Shetland and three in the Moray Firth (Table 2). In hauls 140 and 143, a full-length panel was present; in hauls 152 and 153, the meshes of the first section (9 m) were replaced with large white meshes, but the leading edge of the panel remained above the groundgear (these hauls were not used in separation data analysis); in hauls 159, 163, 164, and 169, the panel had been cut back and was not in view at the mouth of the net. The heights of the echo counts above the sea floor and, where relevant, above the panel were recorded.
|
On the Shetland grounds, with a full-length panel, gadoids entered the net at a greater height on average in daylight. In order to examine whether the presence of the panel itself influenced gadoid behaviour, height above the sea floor was compared between all eight hauls, including those where the panel had been cut back (Figure 4). Fish height was lower for daytime hauls on the Moray Firth grounds without a panel compared with those on the Shetland grounds with a panel present.
|
A linear model was fitted to investigate the effect of time of day and panel on fish height:
|
|
There was a highly significant interaction between time and panel (p < 0.0001) and additional significant variation between replicate hauls (daytime with no panel). It is therefore not possible to attribute the observed differences to the presence of the panel, the time of day, or an interaction between these two factors. The model was used to predict mean entry height for the eight hauls, and these are given in Table 2, along with mean length of the haddock and whiting catch, which made up the bulk of the gadoid catch. Mean length of haddock and whiting on the Shetland grounds was 34–38 cm. Separation for these species and saithe (Pollachius virens) was high (85–100%) for these hauls except for trawl 153 (51% of haddock, 78% of whiting by weight, entering the top compartment). Mean length of the haddock and whiting on the Moray Firth grounds was much lower, and separation into the top compartment was poorer; 71–81% by weight for haddock and 36–68% for whiting. This difference in species composition may contribute to the difference in entry height and subsequent separation, independent of light level. However, the water clarity in the Moray Firth was also significantly poorer than on the Shetland ground.
Fish separation results
Data from 19 hauls on FRV "Scotia" and eight hauls from FRV "Johan Hjort" were used to assess separation of nine species.
The mean weights per haul for each species were calculated for hauls under the same conditions (day/night, speed, and panel position; Table 3). The main commercial roundfish species haddock and whiting occurred in all areas. Plaice (Pleuronectes platessa) and lemon sole (Microstomus kitt) were present in reasonable numbers on both Shetland and Moray Firth grounds. Saithe, cod, and monkfish were more abundant at Shetland and in the eastern North Sea, whereas common (Limanda limanda) and long rough (Hippoglossoides platessoides) dabs were more plentiful in the Moray Firth.
|
No species displayed any differences in separation within grounds, i.e. between the 11 and 19.5 m panel position in the Moray Firth or between the normal and fast speeds in Shetland. However, significant relationships with ground, time of day, and fish length were identified. The results are illustrated in Figure 5.
|
For monkfish, there was no evidence that separation depended on fish length, ground, or time of day. An estimated 83% of monkfish entered the lower compartment.
For cod and the four flatfish species—long rough dab, common dab, lemon sole, and plaice—a greater proportion of fish entered the lower compartment during the day than at night (p = 0.04 for cod; p = 0.02 for lemon sole; p < 0.001 otherwise). During the day, more than 75% of the fish typically entered the lower compartment. However, at night, the proportion was sometimes as low as 50% (e.g. long rough dab and large plaice in the Moray Firth and cod in North Sea East). The only evidence of a systematic effect of fish length was for lemon sole and plaice (p 
0.001 for both species), with the proportion of fish entering the lower compartment increasing with length for lemon sole and decreasing with length for plaice. There were significant ground effects for cod (p = 0.04), long rough dab (p = 0.04), lemon sole (p = 0.01), and plaice (p < 0.001), but these effects were not consistent or interpretable (Figure 5).
Most saithe entered the upper compartment. There was weak evidence that the proportion entering the upper compartment decreased with length (p = 0.04), but the estimated proportion was > 70% for lengths up to 75 cm. There was no evidence that saithe separation depended on ground or time of day.
For whiting and haddock, linear logistic curves did not adequately describe the individual haul data, mainly in the Moray Firth, where many small whiting (22 cm) and haddock (23 cm) were caught. In particular, small whiting tended to enter the lower compartment more than predicted by the logistic model. The analysis, therefore, was restricted to lengths above 22 and 23 cm for whiting and haddock, respectively. For whiting, there was a highly significant interaction between length and ground (p < 0.001). An estimated 61 and 83% of 23 cm whiting entered the upper compartment in the Moray Firth and Shetland, respectively, with the proportion increasing only slightly with length. However, in North Sea East, the proportion entering the upper compartment increased from 80% at 23 cm to 99% at 40 cm. There was no evidence of a time of day effect. For haddock, there was strong evidence that the relationship between separation and length varied with time of day and ground (p < 0.001). There was no particular pattern to these interactions (Figure 5), but in all cases, the estimated proportion of haddock (above 23 cm) entering the upper compartment was more than 69%.
|
Discussion |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Panel design
The original design of the full-length separating panel was considered by the industry liaison group to be impractical and liable to suffer damage. During this project, a novel method of controlling the height of a shortened panel has been developed, using short wings running down a bar from the selvedges. The required width of the panel netting relative to the belly panel width depends on the design of the net, and further work is needed to develop general design rules.
Fish behaviour
The limited replication and differences in species composition and visibility between fishing grounds made it difficult to draw firm conclusions. In Shetland, where underwater visibility was good, the rising behaviour of gadoids appeared to be more pronounced in daylight. The entry height was lower during daylight hauls without a full-length panel present. However, these hauls were in an area where water clarity was poorer and the catch contained higher numbers of small gadoids. Separation for whiting was poorer in these hauls compared with the Shetland hauls, confirming that these fish remained low in the net. The cause may be a combination of behaviour of small fish and the poorer visibility reducing the visual stimulus of the panel.
Fish separation
No difference in separation caused by towing speed was detected for any species.
On the same ground (Moray Firth), no significant difference in separation was found between the panel starting at 11 m and at 19.5 m. At the other panel lengths of 0 m (Shetland) and 13 m (North Sea East), any panel effect is confounded with the change of fishing grounds.
A significantly greater proportion of cod and four species of flatfish entered the lower compartment during the day than at night. For other species, there was no evidence of a time of day effect except for haddock, for which the relationship between separation and length varied with time of day and ground.
Overall, regardless of panel length, fishing ground, time of day, or speed, fish of a particular species always tended to enter one compartment in preference to the other.
Management issues
As noted in the introduction, the principal aim of this gear with a separator panel design is to reduce the fishing mortality of cod in mixed whitefish fisheries. The trials have demonstrated effective separation of the majority of haddock, whiting, and saithe into a top compartment, so retaining an important proportion of the catch. Further development is required so that, after separation, those species entering the lower compartment (principally cod, monkfish, and flatfish) can be confronted with a size-selective device, such as a grid, with a bar-spacing chosen to release a suitable size of cod. The selectivity of such a device for flatfish and monk would have to be determined, although larger individuals would be expected to be retained. This approach allows some flexibility to control the catch composition depending on the state of individual stocks, especially cod.
The project was undertaken in close collaboration with the commercial fishing industry, and the overall gears design and detailed panel design owed much to the input from fishers and netmakers. Nevertheless, introduction of species-separating gears will incur significant cost in net modification and maintenance, may increase gear handling times on each haul, and may cause short-term losses in the catch of some species. Effective enforcement of the use of this complex gear may not be easy. Monitoring of catch composition when the codend is emptied on deck may be an appropriate control mechanism. In these circumstances, some incentive to fishers to adopt the gear voluntarily could be helpful.
Although further testing of this separator design and especially of the selection device to release cod will be necessary, the work so far has demonstrated that a separator panel design may be acceptable to commercial fishers who want to maintain mixed whitefish fisheries while reducing cod mortality by a prescribed amount.
|
Acknowledgements |
|---|
The authors are grateful to the members of the Scottish Industry Liaison Group, who gave invaluable advice; to Jackson Trawls Peterhead, who designed the nets and contributed greatly during the sea trials; and to the crews of the research vessels "Scotia" and "Johan Hjort" for their cooperation. This work was partly funded by DGFish of the European Commission. This paper does not necessarily reflect the views of the Commission and in no way anticipates any future opinion of the Commission.
|
References |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
-
Anon. North Sea cod recovery plan. (2001) 3. Brussels. Report of the scientific meeting on improvement of selectivity of fishing gears arranged by European Commission 5–9.
Engås A., Jørgensen T., West C. W. A species-selective trawl for demersal gadoid fisheries. ICES Journal of Marine Science (1998) 55:835–845.
Graham G. N., O'Neill F. G., Fryer R. J., Galbraith R. D., Myklebust A. Selectivity of a 120 mm diamond codend and the effect of inserting a rigid grid or a square mesh panel. Fisheries Research (2004) 67:151–161.[CrossRef][Web of Science]
Hickey W. M., Brothers G., Boulos D. L. A study of cod/flatfish separation in otter trawls with the use of rigid grates. (1995) St John's, Newfoundland. 35. Canadian Technical Report of Fisheries and Aquatic Sciences 2027. Marine Environment and Habitat Management Division, Fisheries and Oceans Canada.
Hillis J. P., Carroll J. Further experiments with separator trawls in the Irish Sea. (1988) 51. ICES Document CM 1988/B.
Main J., Sangster G. I. Trawling experiments with a two-level net to minimise the undersized gadoid by-catch in a Nephrops fishery. Fisheries Research (1985) 3:131–145.[CrossRef][Web of Science]
Millar R. B., Fryer R. J. Estimating the size-selection curves of towed gears, traps, nets, and hooks. Reviews in Fish Biology and Fisheries (1999) 9:89–116.[CrossRef][Web of Science]
Reid D. G., Jones E. G., Copland P. J. The application of multi-beam sonar to the quantification of survey gear performance. (2002) Report of the Working Group on Fisheries Acoustics Science and Technology. ICES Document CM 2002/B: 05.
Thomsen B. Selective flatfish trawling. ICES Marine Science Symposia (1993) 196:161–164.
Wardle C. S. A review of fish behaviour in relation to species separation and selectivity in mixed fisheries. (1995) ICES Document CM 1995/B + G + H + J + K: 3.
Wileman D. A. Review of the progress made in Denmark in the development of species selective gears. (1995) ICES Document CM 1995/B + G + H + J + K: 2.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. A. Krag, R. Holst, and N. Madsen The vertical separation of fish in the aft end of a demersal trawl ICES J. Mar. Sci., May 1, 2009; 66(4): 772 - 777. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||