Reduction of Bycatch in West Coast Prawn Trawl Fisheries
NOAA Fisheries Logo


Contact:  Nicholas Lowry, University of Washington, School of Fisheries, P.O. Box 357980, Seattle WA 98195-7980.  E-mail:  Tel: (206) 860 6782, Fax: (206) 860 3394

Primary Collaborators: University of Washington, Northwest Fisheries Science Center (NWFSC), and commercial fishers

Active Participants: Nick Lowry, Prof. Don. Gunderson (UW), C. W. (Bill) West (NWFSC).



Bycatch of finfish in the West Coast fisheries for spot prawns (Pandalus platyceros) causes wastage of fish which cannot be landed and utilized and causes conflicts when the species of bycatch are desired by other fishery segments. Reduction of bycatch is especially critical for several species of rockfish (Sebastes sp.) whose stocks are in poor condition or where rebuilding efforts are underway.

Even where there is no substantial impact on commercially valuable species, fish bycatch imposes a burden on shrimp trawlers, increasing the need for sorting and reducing the quality and value of the catch.

Current efforts and innovations have shown some promise, but use of bycatch reduction devices is patchy. Acceptance has not been wide, partly due to reports of high catch loss and the lack of adaptation of the devices to local conditions.

The objective of this project is to evaluate existing shrimp/fish separator technology, to determine which methods are the most effective and practical for reducing bycatch in the fisheries on this coast without loss of a significant amount of the target catch.


Three different types of fish excluders, which have shown potential in shrimp/prawn fisheries elsewhere in the world, were tested. These were a Nordmøre grid, a separator panel and a "fish eye" (Figures 1-3 respectively).   [CLICK HERE TO SEE THE FIGURES AND TABLES FOR THIS FINAL REPORT.]

Nordmore Grid

This system was developed in Norway for separating shrimps from fish in pink shrimp (Pandalus borealis) trawls. It consists of a rigid grid, made of stainless steel or plastic, mounted at a 45-50 degree angle in the aft section of a shrimp trawl. The grid has a netting funnel in front of it to ensure that all catch initially comes to the bottom of the grid to increase the chance of small objects (i.e. shrimps) passing through the grid and into the trawl cod-end. Above the grid, the top panel of the net has an opening to allow large fish to escape. Normal bar spacing is maximum of 19 mm for shrimp fisheries in Norway. The performance of this device tends to be good in pink shrimp fisheries, average shrimp loss is 1-3% by weight and fish exclusion up to 95% for small fish, 100% for larger species. The Nordmøre grid is a device that has been extensively studied and has been shown to be effective. However, it is not popular in some fisheries due to its cost and perceived difficulties in handling.

The grid used in this project was manufactured from a ½" thick polyethylene sheet, into which the grid openings were cut using a milling machine. The openings were 3" in width. This was determined by estimation of the differences between the sizes of the pink shrimps targeted in fisheries using the grids and the size of spot prawns. The overall grid dimensions were 4' in height and 2' in width. It was mounted at a 45-degree angle in a pre made 160 mesh round extension section (2" mesh size), complete with an 85 mesh long guiding funnel. Four 6" floats were mounted on the top corners to ensure that it fished in the correct orientation, and the bottom edge was reinforced with nylon twine to avoid abrasion damage from contact with the sea floor.

Fish Eye

This device consists of one or a pair of elliptical framed openings, which are inserted in the top panel of the aft part of the trawl. The frames are made from stainless steel and have bars attached to ensure that they remain vertical and that the openings point forwards relative to the towing direction of the net. For a fish to escape it has to swim forwards and upwards from the back of the net and then exit. This device is the simplest of the three tested, and is already in legislation in some spot prawn fisheries. There are fewer variables that can be adjusted with the fish eye, the major question is where it should be positioned in the cod-end. Workers in the Gulf of Mexico have researched the position of this in the net for those fisheries, and the supplier suggested a similar position. Two fish eyes were mounted in the top of the net, 88 meshes from the end of the cod-end, with a single 12" float attached between them to ensure that they would remain on the top of the extension as planned.

Separator Panels

These devices are intended to separate shrimps from fish in shrimp trawls. They are adaptations of the Gulf of Mexico turtle excluder devices and are made entirely of netting or with only a few rigid parts. The concept is similar to the Nordmøre grid, shrimps pass through an angled mesh panel and fish are deflected upwards and out through an escape hole. Panel mesh sizes are normally 3-8 inches in USA fisheries. Performance of these devices tends to be less consistent than rigid grids. Shrimp losses have been reported between 2 and 30% and fish exclusion 30-80 % for small fish. They have advantages over rigid grids in being easier to handle on small vessels and cheaper to produce and repair. The separator panel has the most potential for performance problems caused by incorrect setup.

The separator panel used in this project was made from 5 1/2" polyethylene mesh and framed with a loop of 1/2" steel cable, which was wrapped with nylon cord to give protection from abrasion. A crosspiece of 1/2" cable was also installed to help the panel retain its shape. The panel was installed in a section of 2" mesh size netting, 160 meshes in circumference. Three 4" floats were installed inside the net at the back of the panel to help it retain its shape and ensure that it would fish the right way up.

Commercial net makers on the West Coast supplied all of the gear used. The separator panel was supplied by Astoria net shop, Astoria (George McMurrick), and the Nordmøre grid from Seattle Marine and fishing supply. As these are commercial products, the precise details of the designs are not presented here. The fish eyes were manufactured by Robert P Driscoll net services according to a standard design. The elliptical openings were 10" in height and 15" in width.

The experiment was designed to use a commercial fishing vessel during normal fishing operations. The vessels skipper was paid a small fee for participation in the experiment, but it was not a full charter. This limited the scope of the data collection, as it was necessary to fit in the experimental tows with a normal program of commercial tows. The weather conditions that prevailed during the period of the charter also affected the scope of the data collection.

The general approach was of first using trawl instrumentation characterize large-scale gear configurations and direct video observation to study small-scale separator geometry and the conditions that affect the operation of existing shrimp/fish separator designs. No new design work was attempted during this experiment. The catch composition was measured during the experimental tows to assess the degree of separation and catching performance for target species using these devices. Data collected from each net for each haul was the total catch weight, catch of each species (or species group where precise identification was difficult), and length frequency of major bycatch species.

The vessel that agreed to participate in the work was set up to use a double rig, i.e. it normally tows two identical trawls simultaneously from booms, one on each side of the vessel. This rig allows direct comparisons to be made between a net fished normally on one side and a net with a separator on the other. The instrumentation used on each net was identical, which enables direct comparison of fishing performance of each net.

Before starting the experimental part of the work, a calibration of the two nets used by the vessel was carried out to assess their fishing performance. Variables that were compared included wing end spread and door spread (indicators of swept width), headline height, and ground gear contact. Fish catches were compared by weight and species composition.

The vessel used was the F/V Nita H, based in Pt. Hueneme, California. This boat is a large shrimp trawler, approximately 105' in length. Two nets were fished, connected directly to a pair of wooden trawl doors. Rockhopper ground gear with approx. 10" discs was used.


General Conditions

During the trial period, the weather conditions were generally dreadful, with mostly strong winds from the north. This limited the amount of time which we could spend fishing on the more productive grounds. The 24-hour non-stop fishing schedule kept by the captain, also made it difficult to achieve the objectives as little time was available for altering the gear between hauls. The use of the trawls hung from extended booms on the sides of the vessel made it difficult to get access to the gear for making changes and for positioning cameras and trawl monitoring equipment.

Great difficulty was found in adjusting the nets to fish similarly. This was partly because the vessel's nets were not identical. Due to the nature of the fishery, nets are damaged frequently, resulting in them needing repair, and whole nets are lost occasionally. The vessel was using three nets interchangeably. Two had been manufactured identically but had subsequently been extensively repaired. The third was similar but not identical in design, including slightly different ground gear configuration. We attempted to adjust the nets so that their catches would be the same, but this aim was made difficult by the extreme variability in the catches from haul to haul. This inconsistency seemed to also apply to the variation between the catches from the two nets. Side to side variation was big enough to make it impossible to assess if the catching performance of the two nets were the same. After a reasonable time, we could still not be sure that the nets were fishing identically but we were confident that their performance was at least similar.

Fishing was carried out in three general areas, dependent on the prevailing weather conditions and catches. The first few hauls were made around San Nicolas Island, but catches were small, and so we moved to around Tanner Banks. The rest of the hauls were made further to the north, in the Santa Barbara channel, which was more sheltered from the wind (figure 4). Unfortunately, these changes in area almost coincided with the changes in the gear type tested, with the result that most of the Nordmøre grid tests were made at the Tanner banks and most of the fish eye tests in the Santa Barbara channel.    [CLICK HERE TO SEE THE FIGURES AND TABLES FOR THIS FINAL REPORT.]

The catches were generally small, and had a very variable species composition from area to area. On the Tanner banks, the catches contained a greater proportion of rockfish, fewer flatfish, more corals, sponges, and a higher proportion of prawns to total catch. The Santa Barbara channel area catches had a greater proportion of bycatch, and this consisted primarily of small rockfish (stripetail, shortbelly and splitnose rockfish), flatfish (dover, english, rex and slender sole), argentines and pacific hake. (Table 1). [CLICK HERE TO SEE THE FIGURES AND TABLES FOR THIS FINAL REPORT.]

For the comparison of catches, the amount of bycatch of individual species was too small to be meaningful. The catches are divided into four species groups by general type. Large rockfish are; Greenblotched, Blackgill, Bank, chillipepper, boccacio, cowcod and darkblotched. Small rockfish are; stripetail, greenstriped, splitnose, shortbelly, halfbanded, flag, pinkrose rockfish. Flatfish are primarily dover, english, rex, petrale and slender sole. Roundfish are; Pacific hake, ratfish, various cottids, grenadiers, eelpouts, and dogfish.

Separator Panel

The first tests were with the separator panel. The aim was to see both if the panel was functioning and to test the potential of the video system to work with natural light.

Results of these were somewhat unimpressive. The video system gave excellent pictures on the surface and down to substantial depth, but there was insufficient light when the net got to the fishing depth at >200 meters. The surface video pictures clearly show that the panel does not take up the anticipated shape but is actually bowed downwards like a spoon. It also appears to be at too shallow an angle for anything to easily penetrate the meshes. Some prawns, which had been caught up in the meshes further up the trawl, were seen to go out of the exit hole as the net was hauled back. The mesh size used may also have been too small to allow spot prawns to penetrate through to the cod-end in any case.

A second haul was attempted with this gear. There was a substantial quantity of hard invertebrates (corals, sponges, sea urchins and the like) and inorganic debris caught during this haul, which stuck on the meshes of the separator and blocked them. Any prawns that came down the bag would then be either caught up in this debris or deflected by water flow out of the exit hole.

Because this was a commercial fishing trip, the experiments with the separator panel were discontinued at this time. The initial performance was seen by the skipper as reducing the catch too much to be a viable fishing option.

Nordmore Grid

The second set of tests was with the Nordmøre grid. Video observations were made of a number of hauls. These showed that the angle of the grid during towing was approximately 45 degrees, which was as designed. The few prawns that were seen passing down the extension went straight through the grid without seeming to be significantly impeded. A number of fish were also observed to pass straight through the grid and into the cod-end. Some fish were seen holding station in front of the grid. It seemed to be necessary for these fish to make a definite upward escape attempt to get away. The majority of them did not do this, and were eventually exhausted and passed through he grid towards the cod-end. Some large fish were observed being deflected by the grid and out of the escape hole. A number of fish were observed escaping when the net pulsed, presumably due to being snagged temporarily on some bottom obstruction. This is further evidence that the water flow is what is holding the fish in the area in front of the grid and discouraging escape attempts. As it was necessary to make all of these observations with the aid of lights, the behavior of the fish cannot be considered completely typical of that under normal fishing conditions.

Some prawns were observed passing down the extension and through the grid. They all stayed right at the bottom of the net, where they were directed by the funnel, and passed with great ease through the grid. These observations, combined with the fact that only the largest fish were deflected by the grid, show that the bar spacing chosen was in fact rather too large. Better results in terms of fish exclusion would be expected with a smaller bar spacing.

A major problem with the Nordmøre grid as with the soft panel excluder, was that any large sponges or other pieces of debris were stopped by the grid and remained there, blocking passage of anything through into the cod-end. This may cause loss of prawn catch if the blocked grid creates an upward flowing current that sweeps the prawns out of the exit hole. It was also somewhat difficult to clear the grid when it was hauled. This was partly due to the way it was handled on board the vessel. The second main problem with the grid was associated with the funnel used to direct the catch to the bottom of the grid. This became blocked with sponges and debris, and occasionally large rockfish were caught in the meshes by their spines, presumably as they were falling back down the net when they were exhausted. Using a different taper on the funnel may reduce some of these problems, but will not avoid them completely. Removing the funnel altogether risks a loss in prawn catch.

Comparisons were made of the catches obtained with the Nordmøre grid in comparison with the other trawl with no excluder installed. A total of 10 hauls were available for this comparison. These were hauls which were made without video camera installed to avoid fish behaviour differences caused by the lights, and on which there were no gear problems such as damage to the nets which would bias the catches on one side or other. Half of these hauls were made with the grid installed in the port side net and half with it in the starboard net, which should compensate any bias caused by net differences. The results are presented in table 2.  [CLICK HERE TO SEE THE FIGURES AND TABLES FOR THIS FINAL REPORT.]

The comparisons showed that the grid did not appear to cause any loss in prawn catch. It appeared to be effective at excluding large rockfish. No large fish were caught in the cod end of the net with the excluder installed, although some were caught lodged in the funnel in front of the grid. Catches of small rockfish were variable. For instance, on one haul over 500 pounds of shortbelly rockfish were caught in the side with the excluder, as opposed to less than 300 in the side without. Overall, the Nordmøre grid did not seem to have much effect on catches of small rockfish, flatfish or roundfish.

Statistical tests on the catches by side are inconclusive, the result for large rockfish is of marginal significance, This is not altogether surprising, as large fish could not get into the cod-end at all (paired t-test; P (one tailed) = 0.0539).

Fish Eye

The third device tested was the fish eye. Video Observations of this showed that there were clear openings in the top panel of the trawl, which were held open by the fish eye frame. It was not possible to deduce the strength and direction of the water flow around the fish eyes from the video. Large numbers of fish (mostly flatfish) were seen escaping through the fish eyes. The catch during the hauls observed consisted primarily of flatfish, with only a few small rockfish, so conclusions on the escape of rockfish through the fish eyes cannot be drawn from these observations. Some events where large numbers of fish escaped were seen when the trawl pulsed due to snags, as with the grid. Again, this is evidence that the water flow down through the fish eye is holding the fish back in the cod-end and preventing escape. It again seems that the behavior of the fish may be influenced by the presence of the lights. Some fish were apparently seen to attack the light before swimming off after escaping from the net.

Catch comparisons could only be made for four hauls (table 2). These were rather inconclusive, as would be expected with such a small sample size. Even though many flatfish were seen escaping on the video tows, this effect was not pronounced in the comparison (paired t test: P (one tailed) =0.0761). The fish eye did not appear to have a great effect of reducing bycatch of any of the other species groups.  [CLICK HERE TO SEE THE FIGURES AND TABLES FOR THIS FINAL REPORT.]

The fish eyes used in the experiment were manufactured from 1/4" stainless steel rod. This appeared to be very strong, but nonetheless all the fish eyes became bent after a few hauls of use, mainly due to their being caught on the rail as they were brought on board. Use of a stronger material and greater care during hauling would seem to be advised in the light of this experience.


In the configuration in which the separator panel was installed, it did not function adequately. This may be an inherent design flaw in the panel supplied, but it seems more likely to be due to the aft part of the trawl collapsing. In either case, it is necessary to re-think the design and installation. It is likely that mounting the panel in a pre made extension section as was done here is not appropriate. For this sort of design to work it needs to be in a part of the net which is more open than the aft part of the extension, and the meshes need to be under some tension to retain the shape of the panel. The most potential for this sort of separator is if it is designed as an integral part of the net and mounted in the aft of the belly where these conditions are met. However, for the spot prawn fishery, where there is often a substantial catch of debris, there needs to be a way to avoid the panel becoming blocked. It may be possible to achieve this objective by using a flapper on the bottom panel so that any heavy objects will fall out that way before they reach the panel. Alternatively, the panel could be installed to slope down from the top of the net to the bottom, which would direct heavy objects out of the escape hole. It would probably be necessary to use a guiding funnel with this arrangement, and that would have some additional problems.

The Nordmøre grid produced more encouraging results than the separator panel. However, it seems that the 3" bar spacing chosen (by an educated guess), was quite significantly too large to be optimum in this fishery. The choice of bar spacing has to be a compromise between the risks of losing part of the target catch and blocking the grid and the benefits of deflecting bycatch. Further work would need to use much smaller grid spacing. 2" or less would be an appropriate starting point.

The use of a grid with smaller spacing does, however, increase the chance of the grid becoming blocked by some of the larger invertebrates (corals, sponges, kelp mats). This was a relatively serious problem in some of the hauls made during this experiment. Various solutions exist to combat this. Workers in Australia have used grids that are inclined downwards, and let fish out of the bottom rather than the top (Brewer et al 1998). This is more likely to be effective at getting rid of debris that reaches the grid. Alternatives include using a buoyant flapper over an escape opening in the bottom of the net so that the weight of heavy debris causes it to drop out, but lighter prawns will not have this option. The problem of blocking of the funnel in front of the grid still exists. This can be mostly eliminated by attaching the funnel further up the net where it is more open, rather than in the collapsed extension section. It may be necessary on some vessels to slightly change shooting and hauling procedures to accommodate these changes in the rig.

As it was necessary to make all of the observations of fish behavior in relation to the grid and the fish eye with the aid of lights, the behavior observed cannot be considered completely typical of that under normal fishing conditions. Many of the fish observed holding station in front of the grid may have been entrained to the lights and not reacting to the grid at all. However, it seems to give an indication that the fish which are caught in prawn trawls are not inclined to make escape attempts in an upward direction in the manner of for instance haddock and whiting in the Atlantic (Wardle 1993). Some Pacific hake were observed from within the cod-end to attempt to escape through the lower panel of the net. This seems to give further impetus to the use of escape panels in the lower, rather than the upper panel of the trawl.

The fish eye was the simplest of the devices used, and the video observations seemed to suggest that it worked to a certain extent. It is clear, even from the small number of hauls tested, that it is not a particularly efficient means of getting bycatch species out of the trawl. Improvements to its performance may be possible by moving it further back in the cod-end, although this risks loss of prawns. Its simplicity in use and installation are the main factors in its favor.

Other solutions, such as square mesh panels, may be an equally simple way to exclude small fish from the catch, although they will not release the large rockfish which are one of the species groups which may be significantly impacted by the spot prawn fishery. If the problems associated with the use of grids and the associated bad press which has resulted from this can be overcome, then the optimum solution is probably some sort of combination device. This could use a grid to let out the large bycatch and a second device such as a square mesh panel or fish eye to let out the rest of the small bycatch. The size of the target prawns is sufficiently large that a panel of rather large mesh could be used without risking loss of prawns. The use of a grid/panel combination would reduce the volume of bycatch and divert some of the water flow away from the end of the cod-end, potentially reducing damage to the catch and increasing its value on the dock.


Brewer D, Rawlinson N, Eayrs S and Burridge C. 1998. An assessment of bycatch reduction devices in a tropical Australian prawn trawl fishery. Fisheries Research 36 (1998) 195-215

Rogers D R, Rogers B D, de Silva J A, Wright V L and Watson J W. 1997. Evaluation of shrimp trawls equipped with bycatch reduction devices in inshore waters of Louisiana. Fisheries Research 33 (1997) 55-72

Wardle C S. 1993. Fish behaviour and fishing gear. In Behaviour of teleost fishes 2nd edn. Edited by TJ Pitcher. Chapman and Hall. ISBN 0 412 42930 6.

Privacy Statement | Disclaimer | Employee Locator

NOAA Fisheries, 1315 East West Highway, SSMC3, Silver Spring, MD  20910