By-catch reduction devices in the European Crangon fisheries.

Transcription

1 This paper not to be cited without prior reference to the authors INTERNATIONAL COUNCIL FOR THE EXPLORATION OF THE SEA ICES CM 2001/R:10 Theme Session R By-catch reduction devices in the European Crangon fisheries. by B. van Marlen 1, D. de Haan, A.S. Revill, K.E. Dahm, H. Wienbeck, M. Purps, J. Coenjaerts, H. Polet. 1 Netherlands Institute for Fisheries Research (RIVO) - Wageningen UR, P.O. Box 68, NL-1970 AB IJmuiden, The Netherlands, tel.: , fax.: , b.vanmarlen@rivo.wag-ur.nl - 1 -

2 Abstract EU-Study 98/012 Reduction of discards in Crangon trawls (DISCRAN) aimed at collecting catch and by-catch data, compare various designs and further develop effective and acceptable selective devices (veil nets and sorting grids) in the European Crangon fisheries. The project started with a literature review and model tests. Over the years 1999 and 2000 a total of 547 valid hauls were carried out in German, Dutch, Belgian and British fishing grounds on research vessels and/or commercial fishing boats. The project was guided through National Advisory Groups consisting of scientists, fishermen's representatives and practical fishermen. Both the sieve net and the grid are devices that can effectively sort out by-catch fish species, but they have to be rigged properly and blockage due to sea weed or jelly fish may impair their effectiveness. The loss in commercial shrimps in the German, Dutch and UK trials was between 5-20%. However, commercial trials in Belgian waters showed higher losses on average 37% probably caused by the unique Belgian catch composition and seasonal differences. A mesh size of 70mm and a grid bar distance of 20mm demonstrated proper sorting. Some general trends were observed. Grids and sieve nets work well on plaice, flounder, smelt, cod, and to a somewhat lesser extent on dab, bib and whiting. Fishermen expressed a preference for the non-rigid sieve or veil net. 1. Introduction 1.1. The problem Discarding and by-catch in C. crangon fisheries, essentially arises from the multi-species nature of the C. crangon fishing grounds and the non-selective nature of the small meshed nets, which are used to target and capture the C. crangon (Graham, 2001, Van Marlen et al., 1998, Revill et al., 1999). In 1998, Van Marlen et al. reported on a large pan-european C. crangon related discardsampling programme. The sampling programme was undertaken simultaneously during 1996/97 in the UK, Belgium, Germany and Denmark with 96 voyages and 497 commercial hauls analysed for by-catch. Summary results are detailed in Table 1 and Table 2. Survival of many of these discards is thought to be poor (Table 3). The major Crangon fishing ground s are depicted in Figure The estimated consequences resulting from this discarding The estimated biological and economic consequences arising from this discarding have been described. Estimates were made of the likely biological and economic consequences that were arising from this source of discarding (Revill et al., 1999). Specific predictive biological and economic models were developed in order to determine these consequences, the results of which are detailed in Table 4 to Table

3 1.3. The DISCRAN-project The level of fish discarding in the North Sea C. crangon fisheries has therefore been estimated to be significantly impacting in both biological and economic terms upon the North Sea plaice fishery. The unilateral introduction of the corrective technical measures (as detailed in EU Fisheries Council Technical Regulation No 850/98) into these fisheries is thereby supported by these findings. The DISCRAN project has sought to develop corrective technical measures (sieve nets (Figures 2 and 3) or sorting grids (Figures 4 to 7) for these fisheries in line with EU Regulation 850/ Materials and Methods 2.1. Overview of trials Table 7 details for each nation the experimental conditions such as experiment number, vessel used, period, beam width, configuration tested, fishing ground and number of valid hauls. The usual fishing grounds of the German cutter MFV Ramona lie as for a number of other cutters based in Cuxhaven rather close to the port in a shallow area called Medem Sand. RV Isis fished in grounds in the Wadden Sea and along the Dutch coast in the 1999 trials and in grounds in the Eastern Scheldt estuary and also along the Dutch coast in the 2000 trials. These grounds were selected in cooperation with local fishermen. RV Isis has a length overall of 28.0 m, a beam of 7.4m and an engine power of 590kW. The Belgian sea trials were undertaken onboard RV Belgica, as well as on the commercial charters MFV Bisiti (O-700) and MFV Benny (O-101) on the Flemish Banks off the Belgian coast in the period between August 1999 and Jan The RV Belgica has an overall length of 50.9m, a GRT of 765t and an engine power of 1154kW. Both commercial vessels have the same dimensions, i.e. a length of 16.8 m and a beam of 5.06 m, and are powered by a 177 kw engine. The vessels are specifically fitted for catching shrimps (rotating shrimp riddle, cooker etc.) and have a three-man crew. For the commercial sea trials, the choice of fishing grounds and fishing practices was left to the skipper in order to assure commercial conditions. The UK sea trials were undertaken during the shrimp fishing season in 1999 and 2000 onboard the commercially registered shrimp fishing vessels MFV Prospects (BM-61) and MFV My Quest (H-44) in the Wash on the English east coast

4 2.2. Gears used General The gears used vary in beam width. In the UK a width of 6m was taken, in the other countries the most commonly used width of 9m was used. Mesh sizes were around 20mm as usual. Commercial gears were used. Details are given below and an overview in Table Germany The fishing gear used on both sides of the beam trawler was a beamtrawl of ca. 12m length in the tapered part and 8.85m frame line length (beam width 9m). It consisted of PA 210/24 with 24mm mesh length in tapered part. The small wedge in the side was 77 meshes wide and 130 meshes long with the forward edge being cut in in 3T2B ratio. The two seam codend was 4.4 m long if a veil net and 7.04 m long if a sorting grid was tested. It consisted entirely of PA 210/24 with a mesh length of 22mm. The circumference of the codend was 250 meshes. The escaping fishes and shrimps were collected in a collector bag (sock) with a length of 7.04 m and a circumference of 180 meshes. Strengthening bags of 3.76 m length, made from PA 210/78 double with a mesh length of 80 mm protected the codend and the end of the collector bag The Netherlands Details and dimensions are given in Table 8. The Crangon nets were hired from the company C.I.V. of Den Oever, The Netherlands to ensure commercial representation. The nets are typical Dutch shrimp trawls for a beam width of 9m. The mesh size was 20mm throughout most of the nets, with 25mm in the upper square. A grid of 900x600mm was installed in the experimental net. Bar spacings of 15mm and 20mm were chosen. After first experience floats were attached to the grid to keep it in the right position. The sieve nets used were of 60mm (material PA) and 70mm (material PE) measured mesh size with a cutting rate of 1N11B Belgium The two fishing gears used (standard and experimental) were typical 8m shrimp beam trawls as they are used in the commercial shrimp fishery. The standard net and the net used to fit a sieve net were identical. The only difference between the standard net and the net fitted with the grid was a tunnel, 75 meshes deep and 30 meshes from the cod-end, in which the grid was inserted. The overall characteristics were the following: The net was attached to an 8m beam. The overall cutting rate was 1N2B, except for the cod-end. The cod-end was 200 meshes deep and 280 meshes round. The nominal mesh size of the cod-end was 22mm. The mesh sizes of the other parts of the net varied between 24mm, just above the codend, 26mm in the middle and 28mm at the top of the net. The trials aboard commercial vessels were always carried out with an outlet cover. This cover had a nominal mesh size of 80mm, which is the legal minimum mesh size for the sole fishery

5 The sorting grid used was also 900x600mm with bars parallel to the 900mm side and parallel to the longitudinal axis of the net. A stainless steel grid was used because of its firm properties. The grid existed of a hollow tube frame (18 mm diameter) and filled with 21 solid bars of 6 mm diameter fitted with 20 mm bar spacings. Five floats (in a later stadium three) of 1 l were attached to the top bar of the grid to prevent scraping on the bottom. The grid was first installed in the net under an angle of approximately 51 and then reduced to 47. The angle is in that way oriented so that the top comes before the underside. The grid itself is braided to the net. The part of the catch that does not pass through the grid will be deviated to the bottom of the net where an escape opening is cut. This escape opening is covered with an 80mm outlet cover to retain commercially sized fish. Since the sorting grid was not commercially used at the time of the trials and no standard design existed, the design of the grid (setting angle, guiding panel, outlet) was changed throughout the sea trips in order to optimise it. The first design was based on successful tests aboard the Belgian research vessel. Several solutions have been sought to reduce the clogging and the loss of commercial shrimps. Sieve nets are rarely used in the Belgian shrimp fishery. For the design of the sieve net to be used in the Belgian trials, the net plans used by the Dutch (for a 9 m beam) and the UK (for a 6 m beam) partners were scaled to the size of an 8m beam. After scaling, both sieve nets were almost identical. Also the position in the net was copied from the other partners. The sieve was placed with the front edge at the join between the panels with 28mm and 26mm mesh, which is about one third from the beam. The global characteristics of the sieve net were the following: the nominal mesh size is 70 mm. 116 meshes wide at the front, 16 meshes wide at the rear end and 60 meshes deep. the overall cutting rate was 1N10B. the opening of the sieve net is placed in the belly of the net. the front part of the sieve net was attached at the 400-mesh join (28mm into 26mm mesh). The sieve net was then stretched, together with the net, and the aft part was put 22 meshes back where an opening of 48 meshes wide was cut in the shrimp net where the end of the sieve net was laced around. The escape opening was covered with an 80mm outlet cover, on a commercial vessel, to retain commercially sized fish. On the RV Belgica the escape opening was covered with an 11mm outlet cover in order to retain all the fractions of the catch that normally escaped. The cod-end was blinded with an 11mm blinder. Since the sieve net was already being used abroad in the commercial fishery at the time of the trials, an existing successful design has been used. This design was used throughout the trials in order to test its performance in different conditions

6 United Kingdom This design of Sieve net UK No 1 is currently produced and sold commercially by the net-makers of Lauwersoog (The Netherlands). The net used was constructed by a netmaker from the port of Lauwersoog. The sieve net was constructed from nylon (polyamide) (weight 210/42) with an overall full mesh size of 53mm. Sea trials were conducted with this design of sieve net during October Sieve net UK No 2 was based on designs that were have been used in the Danish fishery. The sieve net was constructed from twisted polyethylene (PE) (weight) with an overall full mesh size of 64mm. Sea trials were conducted with this design of sieve net during November Sieve net UK No 3 was a simple modification of sieve net design UK No 2. The amount of netting material used to construct the sieve net was reduced in bulk, in order to stretch the meshes more open and increase water flow. The aim of this design was reduce the loss of marketable shrimp that was observed with sieve net design UK No 2. Sea trials were conducted with this design of sieve net during February Sieve net UK No 4 was based on sieve net design UK No 1 (Dutch) and was constructed at the University of Newcastle (UK). A larger mesh size was used (68mm) and all cutting rates, lengths, etc were adjusted to keep the overall dimensions the same as sieve net UK No 1. The veil net design UK No 4 therefore was designed to be as similar as veil net UK No 1 in design as far as possible, apart from the mesh sizes. The netting used to construct the sieve net was obtained from the fishing co-operative in Oostende in Belgium, where it is used to construct beach sole nets. Sieve net UK No 4 was constructed from nylon (PA) (weight 210/30) with an overall full mesh size of 68mm. Sea trials were conducted with this design of sieve net during February Data collection and analysis Data recorded The following data were recorded for each haul: Haul number, date-time of shooting, date-time of heaving in, description of standard trawl, description of experimental trawl configuration, shooting position, heaving position, towing speed, warplength, towing direction (B only), fishing ground, windspeed and direction, direction of the current (B), waterdepth (shooting and hauling), water surface temperature (D only), sea state (B), availability of precipitation (B), light intensity with a difference between bright/dusk/dark (B), and local name of fishing ground Sampling procedure Two different approaches were taken, either sampling from the sorted catch, or from the unsorted catch. In Germany at first all riddle sieve fractions and the content of the sock codend were weighted. The catch from the sock codend was only subsampled if its total amount - 6 -

7 exceeded two buckets (ca. 15 ltr), if not, it was stored in total. A subsample of usually 20 kg was taken from the commercial shrimp fraction from the riddle sieve. This subsample was immediately examined for fish, and all fish and a sample of 1 ltr of shrimps was stored for later evaluation. A sample of one bucket was taken from the discard shrimp fraction. A subsample of two buckets was taken from the fish fraction of the standard trawl net and of one bucket from the fish fraction of the experimental trawl. This mirrored the apparent diminution of the fish by-catch by selective devices. All samples were deep frozen immediately after the daily return of the ship to the harbour, stored in Cuxhaven and transported to Hamburg for further analysis. After thawing they were at first sorted for species if this had not been done yet onboard. The species comprised in the sample were then weighed. All fish were measured on electronic measuring boards and the data stored to a file specially marked for later identification. Shrimps were again subsampled after weighing. The usual size of such subsample was half a litre or ca 300 animals. Length measurement of shrimps was done with the help of a newly developed video image processing system which produced an ASCII-file containing the length distribution of the respective sample. The weights of catch components, sample weights and subsample weights necessary for the calculation of the rising factors were stored to a separate file. In The Netherlands catches of both nets were sampled and measured according to RIVO's BTS-protocol. The catch of both nets is stored in the two fish bins of the sorting device on the working deck of the vessel. Shrimps and fish of both codends, and in the 1999 experiments from a collection bag mounted around the escape hole of a veil net or a grid, were then taken from the unsorted catch. For smaller fish a sub-sample is then usually taken to determine the total catch in weight and numbers of non-target and undersized target fish. The size of this sample was chosen with the estimated processing time taken into account, ensuring that the catch was measured before the next haul. Commercial species and larger specimen of other species were all measured. Fish and shrimp species were determined and numbers in each length class counted to obtain length frequency distributions. Shrimps were measured in the first three weeks in the 1999 trials, but later stored and frozen in plactic boxes for detailed analysis. Digital pictures of these samples were taken at BFAFi, Hamburg and the length frequencies calculated using image analysing software. In this case a larger sample was taken (e.g. some 250 individuals) and no extra time was lost between hauls for the time consuming measurement of these shrimps. Fish data was typed in directly during measurement using program Bessie Turf. An experiment is defined as a set of hauls for which no changes in gear or rigging were made, but may cover more than one fishing ground. In Belgium one of the two nets was used as experimental net and was rigged with a sorting grid or a sieve net. For the catch comparison aboard the commercial vessels, the experimental net was towed from the starboard side of the vessel while the standard net was towed from the port side. At regular time intervals the skippers were asked to check during the commercial fishery whether the two sides of the vessel had the same catching efficiency in order to avoid bias.the catching efficiency of both sides was quite similar. Commercial mesh sizes were used. On board of the RV Belgica, only one gear could be towed. Therefore, the cod-end of the experimental net was blinded with an 11 mm mesh blinder and the outlet was covered with an 11 mm outlet cover

8 At the end of the haul the contents of the standard and the experimental cod-ends were released into a repository divided in two. To keep the experimental and standard catch separated, baskets in two different colours were used. When the catch came into the repository the crew first sorted out all the marketable fish. Through a conveyer belt the rest of the catch was fed into a rotating shrimp riddle. Out of the riddle came three fractions: commercial shrimp fraction! mainly shrimp above 4.5 cm! occasional small flatfish and non-commercial fish discard shrimp fraction! shrimp less than 4.5 cm! small flatfish and non-commercial fish trash fraction! larger fish! benthos! debris The total volume of each fraction was recorded and samples of 1.5 ltr were taken from the commercial and discard shrimp fraction to determine the length frequency distribution of the shrimps and to determine the numbers and weight of the different invertebrates and non-commercial fish present in that fraction. From the trash fraction a 6 ltr sample was taken to determine the numbers and weight of the invertebrates and non-commercial fish. The commercial fish species were sorted from the three fractions and measured to the cm below until a sufficient length frequency was obtained (which was not always possible for all species). As a rule, the fish were measured on board. The only difference between the handling of the experimental and standard catch was caused by the presence of an 80mm outlet cover on the experimental net. The content of this cod-end (mostly commercial fish) was released in a basket and the fish was measured on board. The shrimp samples were stored in a 5% formaline solution while the trash samples were frozen once ashore. Occasional fish samples were measured by hand to the cm below. The trash samples were sorted out by species. For each species the numbers and aggregated weight were determined. Shrimp samples were rinsed and sorted out by species. All the different fractions were weighted and counted. The Crangon crangon fraction was weighted and subsampled (at least 250 individuals). The length measurement of the shrimps was carried out by the image processing system developed in Germany. Measurements were recorded with a 1mm level of accuracy. The sampling procedure on board of the RV Belgica was analogue to the sampling procedure on board of a commercial vessel except for the following difference: onboard of the research vessel fishing is only possible with one beam trawl instead of two. Therefore paired experiments between a standard and an experimental net are not possible. In order to enable the determination of the selective properties of the - 8 -

9 experimental net, a small mesh blinder (11mm) was rigged into the cod-end and the outlet was covered with an equally small meshed cover. Sampling procedures, however, remained equal. In the UK in all experiments, the total catch was sorted onboard using the traditional sorting method (i.e. on-board deck riddle). All catch components were subsequently weighed (to 1kg below). All commercial species of fish were removed from the catch and measured (to 1cm below). There was no sub-sampling of the catch and all fish were measured individually. The total weights of the shrimp fractions (i.e. discarded and retained) were recorded. No codend covers were used in the experiments as it was assumed that the small meshes used in the body and cod ends (20mm) of the trawls are essentially non-selective in respect of the species of interest. This technique therefore removed the influence of codend covers upon water flow, and allowed for simultaneous catch comparison and selectivity analyses of the data Statistical analysis In Germany all files were stored in a purpose-designed database in ACCESS, that was used for analysis. The SAS (SAS Institute, Cary, NC, USA, 1992) statistical package was used to analyse the Dutch data in more detail. Files stored onboard were converted into a large SAS-datafile. A special code was written to extract the numbers and weights in kg per unit of fished area (10000m2) for the categories: shrimps (landed), shrimps (discarded), fish (landed), fish (discarded), by-catch species as well as commercial species sole, plaice, dab, flounder, brill, turbot, whiting, cod, bib and gurnards taken separately and with a split in sublegal and legal sizes. The routine calculated the numbers and weights per unit of area (10000m2) for both port and starboard gears, as well as the mean difference and the p-value. Length frequency distributions for both gears can also be extracted for each species of interest. Length-weight keys were used to calculate the weight of fish and shrimps. The following categories were used in the analysis: Non-marketable or sublegal commercial target species: sole (Solea solea L.), plaice (Pleuronectes platessa L.), dab (Limanda limanda L.), flounder (Platichthys flesus L.), brill (Scophthalmus rhombus L.), turbot (Scophthalmus maximus L.), whiting (Merlangius merlangus L.), cod (Gadus morhua L.), smaller than the minimum landing size. Other species without a minimum landing size, such as tub gurnard (Trigla lucerna L.), grey gurnard (Eutrigla gurnardus L.), red gurnard (Aspitrigla cuculus L.), streaked gurnard (Trigloporus lastoviza L.) and bib (Trisopterus luscus L.) were counted to this fraction according to national practice in the fish markets (See Table 10). Marketable or legal commercial target species. Same species list as above, bigger than the minimum landing size or equivalent. By-catch: - 9 -

10 All non-commercial species in this fishery among which: hooknose (Agonus cataphractus L.), dragonet (Callionymus lyra L.), solenette (Buglossidium luteum L.), goby (Pomatoschistus L.), eelpout (Zoarces viviparus L.), shorthorn sculpin (Myoxocephalus scorpius L.), sea snail (Liparis liparis L.), four bearded rockling (Enchelyopus cimbrius L.), five bearded rockling (Ciliata mustela L.), sea lamprey (Petromyzon marinus L.), herring (Clupea harengus L.), sprat (Sprattus sprattus L.), poor cod (Trisopterus minutus capelanus L.), lesser weever (Echiichthys vipera L.), sandeel (Ammodytus spp. L.), scald fish (Arnoglossus laterna L.), spurdog (Squalus acanthias L.), etc. Discards: All sublegal commercial species plus all by-catch species. Landings: All legal commercial species. Benthos: Major species caught: swimming crab (Liocarcinus spp. L.), masked crab (Corystes cassivelaunus L.), common starfish (Asterias rubens L.), starfish species, (Astropecten irregularis L.), brittle star (Ophiura spp. L.), hermit crab (Pagurus bernhardus L.), but in weight expressed including all other species found. The minimum landing sizes used in these analyses are given in Table 10. In some cases the towing speed was not recorded for each haul and the area fished in 10000m2 was calculated with an estimated average speed with tow duration of each experiment, and gear width (NL: 1999 experiments, D). In other cases counts were recorded of nautical miles fished for each haul and these were used to calculate the area fished for each experiments (NL: 2000 experiments). In the Netherlands and Belgium shrimps are sorted out using carapace width with a minimum of 6.5mm. This was converted to an equivalent length by using the expression: carapace width (mm) * 100 = 15.6 * length (mm) 72.5 For 6.5mm width this will thus lead to an equivalent length of mm. As length classes were taken at 1 mm steps, all values above 45mm were taken as commercially sized Contacts with the industry Intensive contacts with the fishing industry took place during the project. National advisory groups consisting of practical fishermen were established in Germany, The Netherlands and Belgium. Two meetings were organised in Germany, The Netherlands, and Belgium. A total of 32 fishermen were consulted in the United Kingdom (East Coast). The purpose of these communications was to enable development of practical and acceptable solutions, and to stimulate co-operation between researchers and scientists

11 3. Results 3.1. General A standard table and figure format was designed for presenting the results per experiment. Catches of the overall categories shrimps (landed and discarded), commercial fish species (landed and discarded), and benthos (in weight), as well as detailed per species: sole, plaice, dab, flounder, brill, turbot, whiting, cod, smelt, bib, gurnards (sublegal and legal) of the standard and the experimental net are given in Table 12 to Table 53, expressed in number per 10000m2 and in kg per 10000m2. Also the ratio between the two and the p- value of a 95% t-test. Significant values are printed in boldface Germany General Sea trials during the reported period have been carried out as follows: 6 days at sea in the period from 18 July to 30 July days at sea in the period from 23 September to 1 October days at sea in the period from 22 November to 27 November days at sea in the period from 28 March to 6 April days at sea in the period from 13 June to 20 June days at sea in the period from 4 July to 7 July 2000 The results of these experiments are summarised in Table 12 - Table 20. Samples from 198 hauls were collected of which 120 are considered to be valid (Table 7). The following results are based on the weight analysis as taken aboard and the length frequency distributions originating from the sample analysis Experiment 1 (Table 12) This experiment carried out in July, still outside the main catching season, yielded rather moderate catches of fish and shrimp. Most of the shrimp are still small as the high total portion of discard shrimps shows. The 20 mm sorting grid applied reduced the fish bycatch by 50%, the total bycatch by 55%. But the shrimps of commercial size were also reduced by 30% in weight. Plaice, dab and flounder are the dominant flatfish parts of the fish bycatch, smelt being the dominant roundfish. Both the reduction in fish and shrimp is significant, even in landed shrimp. That applies as well to numbers as to weights Experiment 2 (Table 13) A further experiment carried out in July as well, but with a 30 mm sorting grid resulted understandably in lower losses of commercial-sized shrimp (19%) but also of undersized fish (36 %). The same applies as in experiment one; the share of undersized shrimp is remarkably high. Though slightly better than in the first test experiment the catches still

12 remain at a comparatively low level. Plaice and dab are the dominant flatfish bycatch, whiting and smelt the dominant roundfish components. It strikes that the sorting grid sorts out more efficiently roundfish than flatfish. The reduction in fish and shrimp is significant both in numbers and weights, even in landed shrimps Experiment 3 (Table 14) This experiment repeated the test with the 20 mm sorting grid at the start of the peak catching season in September. Catches have risen to the 1.5 fold of the July catches. Though with an even higher share of undersized shrimp the loss of marketable shrimp remains at moderate 19%. The loss of fish bycatch (35%) is less compared to the July results with the same gear. Plaice and flounder are the dominant flatfish components of the bycatch and smelt the dominant roundfish. The reduction in fish and shrimp is significant both in numbers and weights, even in landed shrimps Experiment 4 (Table 15) This experiment was a repetition of the July experiment with the 30 mm sorting grid again at the start of the peak catching season in September. Catches remain high compared to the July catches. The loss of commercial shrimp with this gear is mere 12% and that of fish bycatch only 16% in total. Obviously the sorting efficiency against roundfish is higher (33%) than against flatfish (plaice 16%, flounder 4 %). Plaice, flounder, smelt dominate the fish bycatch. The reduction in fish and shrimp is significant both in numbers and weights, even in landed shrimps Experiment 5 (Table 16) This experiment, performed in October, in the middle of the best catches season investigated the effect of a 70 mm sievenet. Sievenets had been mandatory for some time from 1 October on. Catches of commercial shrimp are exceptionally good, nearly threefold the amount of the July catches. Interestingly, the relation of commercial shrimp to discard shrimp has shifted in favour of the bigger shrimp. The same fish community plaice, flounder and smelt dominate the fish bycatch. The efficiency of the selecting device is comparable to a 20 mm grid (total bycatch reduction 45 %) with the interesting feature that both flatfish and roundfish are sorted out with comparable efficiency. The reduction in fish and shrimp both in numbers and weights is significant, except for the landed shrimp where it is insignificant for the numbers Experiment 6 (Table 17) This experiment showed the rapid decline of the catches occurring in late autumn when with colder water temperature the shrimp return to the open sea. A 20 mm sorting grid was used again. Losses of commercial shrimp of 28% were certainly outside the limits commercial fishermen would willingly accept, even if the bycatch reduction was exceptionally good. Plaice, flounder and smelt again dominated the bycatch. Now the bycatch reduction was even good for flatfish probably because of their bigger average

13 size. The reduction in fish and shrimp is significant both in numbers and weights, even in landed shrimps Experiment 7 (Table 18) In this experiment, also carried out in November, the selecting device was again a 70 mm sievenet. To test if a ship influence could be seen in this experiment opposite to all other experiments done so far and later the experimental trawl was on the port side and the standard trawl an the starboard. The level of catches and the species composition of the bycatch remained similar to the results of experiment 6. Though, the loss of commercial shrimp was only 18%. As in the preceding experiment also plaice, flounder and smelt dominated this time the bycatch. All three showed sufficient bycatch reduction, with flounder demonstrating the best results. The reduction in fish and shrimp is still significant both in numbers and weights, even in landed shrimps Experiment 8 (Table 19) This experiment at end of March/first days of April finished the collection of data from a 20 mm sorting grid on different seasons. It ought to be noted that this experiment was carried out outside the catch season of the commercial fleet. Catches were the lowest obtained so far. Interestingly, though not unusual, shrimps of commercial size were more abundant than undersized. Except of the never missing smelt all other fish bycatches had gone down remarkably. Less reliable estimates of the sorting efficiency of the grid may partly be caused by this, partly by the fact that small fish were far less abundant than e.g. in September. Less small fish clogging the bars may also have lead to rather low losses in commercial shrimp. Consequently, the reduction in shrimp is insignificant, both in numbers and weights. Fish, however, are significantly reduced Experiment 9 (Table 20) This experiment, carried out at the beginning of April 2000, was a repetition of experiment 8. The difference was the renunciation of the covering bag over the escape hole of the sorting grid. It ought to be noted that this experiment, too, was carried out outside the catch season of the commercial fleet. Catches were even lower than in experiment 8. Thus, the difference observed concerning the loss of commercial shrimp (21% instead of 9 %) is as well not very convincing as the increase in catch of smelt. Concerning the significance of the catch reduction the picture is somewhat ambiguous. There is a significant reduction of landed shrimp, both in numbers and in weight. Discarded shrimp on the other hand are only insignificantly reduced. The reduction of fish is significant in weight but not in numbers Experiment 10 (Table 21) This experiment was planned to show differences in the sorting efficiency of an alternative sievenet design as it used at the northern Frisian coast in Germany. Though 31 hauls were successfully made, computer problems destroyed a great amount of the data evaluated. Only 9 hauls could be recovered completely. According to these rather low

14 catches were made. Plaice, flounder and smelt dominated the fish bycatch again. The experiment was one of the two cases where whiting appeared in sufficient numbers to be evaluated. There is a significant fish bycatch reduction in weight but not in numbers. Only landed shrimp showed a significant reduction, not in numbers but in weight Experiment 11 (Table 22) The last experiment of the project tested a sievenet design as used by the Belgian participants and, thus, offered the opportunity to test the same design at different locations of the North Sea. The catch situation has changed somewhat to the June situation with a lot more small shrimp coming into contact with the gear. Plaice, flounder and smelt, dominates the fish bycatch again but whiting keeps its unexpected high presence. The catch reduction by the veil net is insignificant both for fish and shrimps except for the weight of fish Detailed species analysis Length analysis concentrates on the target species, on plaice as the most endangered species in the area and on the more frequent flounder and smelt, since most other fish were available at the chosen fishing place only occasionally at numbers exceeding 10 per m2. A few length distributions of dab and whiting from times when available in greater numbers are added Shrimp (Crangon crangon L.) There seems to be a one-peak length distribution of shrimp from 1.5 to 7 cm on the chosen fishing place throughout the year with the peak concentrating more or less on 4 cm length. Slight deviations from the bell-shaped distribution occur in early summer with the mass invasion of small shrimp onto the fishing ground. No real length-related sorting effect is apparent in all devices tested. Though, occasionally, heavier release rates seem to occur at the extremes of the distribution with the very small and the very big shrimp Plaice (Pleuronectes platessa L.) Plaice from 4 to 20 cm are apparently available at the fishing ground chosen in a onepeak distribution. The maximum lies around 7 to 8 cm length and remains rather unchanged throughout the year. All devices tested demonstrate a length- related sorting efficiency. It is hard to distinguish between the effects of the 20 mm or 30 mm sorting grid. It appears, however, that the 70 mm veil net release 50% of the plaice caught at a lower length than the sorting grids tested Flounder (Platichthys flesus L.) Flounder occur throughout the year at the chosen fishing place with variable length composition between 4cm and up to 30 cm. In early summer apparently a new year class enters the fishing ground, which replaces the bigger fish in the course of the summer. The sorting efficiency of all sorting devices tested seems to be length- related. With 20 mm grids a sharper selection is obtained as with 30 mm. The sorting efficiency of the 70 mm veil net resembles to that of the 20 mm grid

15 Dab (Limanda limanda L.) Dab from 4 to 25 cm length were found in the low-catch seasons in the fishing ground. Their one-peak length distribution is clearly dominated by a maximum at around 5 cm length. A length related efficiency of the sorting devices tested is to be assumed. Too few numbers were caught to prove this assumption Smelt (Osmerus eperlanus L.) Smelt ranging from 4 cm to 25 cm occurred in the most variable length composition of all fishes encountered on the fishing ground. A length- relationship of the sorting efficiency of all devices tested seems highly probable. However, due to its slim body shape the length where 50% of the caught fish are able to escape is hard to guess because it lies in a range where only few bigger fish are caught. It is easiest in off-season conditions and then lies between 15 and 20 cm. Due to this particular conditions clear differences between the three devices tested are scarcely visible Whiting (Merlangius merlangus L.) Small whiting, ranging between 5 and 25 cm body length with an maximum around cm are entering the fishing ground only in greater number in early summer. A lengthrelated efficiency of the sorting devices tested is also to be assumed in their case Netherlands Sea trips NL 1999 An extended series of trials was carried out on RV Isis in Weeks from from the ports of IJmuiden, Den Helder and Den Oever. The trials were accompanied by 10 commercial fishermen on several days in the first three weeks. A total of 83 hauls were made with the configurations listed below of which 12 were unvalid. Fishing was done mostly during the day in positions in the North Sea and the Wadden Sea. The results are given in Table 23 - Table 27 for the experiments 1-5 of the trials in Experiment 1 (Table 23) The 60mm sieve caught more shrimps and more discard fish in the Wadden Sea, but far less fish of commercial sizes (e.g. landings). The sieve did seem to work in plaice, flounder and whiting, releasing many (legal) fish, but not for smelt and juvenile dab Experiment 2 (Table 24) The 60mm sieve used in the Dutch coastal area caught less shrimps and fish in both the landed and discard fractions, statistically significant in legal dab, flounder, and legal and sublegal cod. Discards in numbers were about 25% compared to the standard net, against 61% of landed shrimps Experiment 3 (Table 25) The 15mm grid fished in the Wadden Sea reduced the catch of landed shrimps (ratio 68% in numbers and 72% in weight), and discard shrimps (ratio 50% in numbers and in weight) without any clear effect on the total number and weight of discards. Landed fish

16 although low in numbers were removed from the trawl by the grid, but apparently many smaller fish were not sorted out as effectively. The grid was effective in releasing (small) plaice, dab, flounder and whiting Experiment 4 (Table 26) In the Dutch coastal area smaller numbers in landed (60%) and discard shrimps (60%) were found with the 15mm grid inserted. Legal commercial species were caught less (ratio 3%), but here also smaller catches of discard fish (89% in numbers and 62% in weight) were found, with a significant reduction in plaice, dab, flounder, whiting, cod and bib (juveniles), and indicative in turbot Experiment 5 (Table 27) Grid 20mm fished in the Wadden Sea caused significantly lower catches of shrimps, both commercially sized (ratio 56%) and discard (ratio 58%). Legal commercial species were sorted out effectively (ratio 2%). More discard fish were caught in numbers, but not in weight. Clear effects in numbers and weight were proven statistically in legal plaice, legal flounder, cod and whiting, in weight also in dab Sea trips NL 2000 Again an extended series of trials was carried out on RV Isis in Weeks from from the ports of Colijnsplaat, Scheveningen and IJmuiden. The trials were accompanied by one commercial fishermen in the first week. A total of 139 hauls were made with the configurations listed below of which 10 were unvalid. Fishing was done mostly during the day in positions in the Eastern Scheldt and Southern North Sea. The results are given in Table 28- Table 33 for the experiments 6-11 of the trials in 2000 (as in Section 3.3.1) Experiment 6 (Table 28) The 70mm sieve fished in the Eastern Scheldt reduced the catches of legally sized commercial fish as can be expected (ratio 21% in numbers and 8% in weight). The number of discards seemed less (ratio 92%), but in weight there was clearly a smaller catch for the sieve (ratio 80%). When investigating species separately most of the reductions were not significant, except for small bib, and juvenile whiting (in weight) Experiment 7 (Table 29) Landed (ratio 9% in numbers and 24% in weight) and discarded fish (ratio 54% in weight, but equal in numbers) were reduced with the 60mm sieve in the Eastern Scheldt. There was a clear reduction in juvenile plaice (ratio 51% and 34%), marketable dab (ratio 4% and 14%), marketable flounder (ratio 3%), juvenile whiting (ratio 26% and 20%), cod (ratio 4% and 3%), and small bib (ratio 21% and 17%). No effect on by-catch was found Experiment 8 (Table 30) Sieve 60 fished in Dutch coastal waters resulted in smaller catches of marketable fish (ratio 4% and 2%), and fewer discards in weight (ratio 65%), but again not in numbers. Fewer small plaice (27% in numbers and 16% in weight), dab, large flounder, turbot,

17 brill, whiting, cod and small bib were caught by the sieve. By-catch levels were about the same Experiment 9 (Table 31) The comparison between the two standard nets (the sieve cut loose in the starboard one) in Dutch coastal waters revealed that landings in fish were similar. Discards were somewhat higher for the net with the sieve installed (ratio 120% in numbers and 113% in weight), possibly caused by small higher catches of plaice, small sole, large dab and small cod Experiment 10 (Table 32) The 70mm sieve fished along the Dutch coast sorted out commercial species effectively (ratio 14% and 8%), but now also clearly lower catches of discard fish were recorded (71% and 44%). Most of the differences were significant. The by-catch was 76% in numbers and 64% in weight Experiment 11 (Table 33) Again these differences disappear when cutting loose the sieve and comparing both nets as standards. Now more small whiting and cod were caught by the starboard net (standard with sieve 70 cut loose), with no noticeable difference in by-catch Catch comparison Crangon crangon Experiment 1 in the Wadden Sea showed a substantially higher shrimp catch for the experimental gear, which may have been caused by patchiness of abundance. The Wadden Sea data showed higher numbers per area than the data from the coastal grounds. Both for the sieve and the grids shrimp losses in the 1999 trials are rather high, with maxima as much as 40%. In the year 2000 experiments on sieves, with many more hauls, this trend was not repeated, here the losses were 20% maximum, sometimes even nonpresent. The comparison between the two standard nets revealed no significant differences in performance Catch comparison marketable fish Very few marketable fish were caught in the Wadden Sea, but clearly the sieve and grids filtered out the majority. In the coastal areas larger fish were caught Catch comparison undersized fish Although less effective than for the larger specimen, juvenile commercial fish species were sorted out to a reasonable extent (ratio 20-40%) by the sieve and grids. Juvenile whiting seemed to be released better by the sieve net than by Grid 15, but for the larger bar spacing (Grid 20) similar results were found. Small dab are not very effectively released by the sieve nor the grid. Juvenile sole were caught in greater numbers by the sieve. Possibly the sieve hampers their escape. Although by-catch of commercial juvenile species is often reduced, when taking other species into account the overall number of discard fish is seldom clearly reduced. The only clear evidence was found in Experiment 10, where a significant reduction of 30% resulted, based on a total of 33 hauls

18 3.4. Belgium Sorting grid experiments An overview of all hauls carried out during the project is given in Table 7. Between July 1999 and January 2000, thirteen sampling trips were undertaken on commercial vessels and 1 sampling trip was undertaken on board of the research vessel Belgica. A total of 49 hauls were carried out with the sorting grid. The purpose of these trips was to test and optimise the grid. The alterations to the grid consisted of changing the setting angle (51, 47 ), changing the guiding panel, and the outlet as well the the amount of floatation Experiment 1 (Trip 11/08/99) A lot of blockage of the grid occurred. There was a relative low loss of commercial shrimp Experiment 2 (Trips 03/09/99 and 04/09/99) The design of the net was relatively simple and the results were more liable due to the fact that the nets were adapted to meet the standards used by the fishermen Experiment 3 (Trip 08/09/99) The angle of the grid is slightly diminished and a guiding panel is put in front of the grid, nevertheless the results don t improve Experiment 4 (Trip 09/09/99) On demand of the skipper the guiding panel was removed. The results did not improve and there is a greater loss of all the three fractions, indicating that the guiding panel had a positive effect on the loss of commercial shrimps Experiment 5 (Trip 15/09/99) The guiding panel was put back in front of the grid and there was now a clear sign of improvement of the results. Losses were, however, still high. This was probably due to severe clogging of the grid. Especially jellyfish could not escape through the opening and accumulated in front of the grid thereby preventing the catch to pass through the grid into the cod-end Experiment 6 (Trip 13/10/99) The escape opening was enlarged to enable jellyfish and large pieces of debris to pass through the escape opening and prevent clogging of the grid. In order to avoid too much loss of the catch a rectangular piece of netting with five 0.1 ltr floats were put beneath the escape opening. The aim was to keep the escape opening more or less closed, still allowing it to open when some pressure was exerted against it. There was no improvement in the results and losses where even higher. It was believed that the rectangular piece of netting did not sufficiently shield the escape opening. On the other hand there was no more blocking of the grid and jellyfish and debris could escape through the escape opening

19 Experiment 7 (Trip 28/10/99) To further reduce the loss of shrimp the following alterations were executed: two of the five floats at the top of the grid were removed in order to keep the grid closer to the bottom, thereby keeping the escape opening better shut. The guiding panel in front of the grid was further lengthened with 6 meshes. The rectangular piece of netting in the escape opening was tightened so that it better shielded the opening. The losses of the catch were still not significantly reduced but there was no blockage of the grid and the jellyfish were all accumulated in the 80mm cod-end Experiment 8 (Trip 03/11/99) The rectangular piece of netting at the escape opening was further tightened to better shield the escape opening. There were fewer losses of commercial shrimp and jellyfish could easily pass through the opening into the 80mm cod-end Experiment 9 (Trip 25/11/99 RV Belgica ) The rectangular piece of netting at the escape opening is further tightened, and this results that the escape opening is slightly sealed but when a slight pressure is put on to the rectangular piece of netting this gives an escape opening of about 15 to 20 cm. Bear in mind that in order to enable catch comparison an outlet cover with a mesh size of 11 mm was rigged over the escape opening (instead of the normal 80 mm cod-end). The average catch composition indicates a low loss of shrimp and a relatively high loss of trash. Especially larger fish escape through the escape hole into the escape cod-end, smaller fish pass through the grid into the cod-end. Although the loss of shrimp is low in general it can be noted that more small shrimps were lost compared to commercial shrimps Experiment 10 (Trips 10/01/2000 and 24/01/2000) As the net configuration worked fine on the research vessel it was used on a commercial vessel without making any alterations (except for the replacement of the small mesh outlet cover to the 80mm cod-end). There were no problems during fishing but loss of commercial shrimp were again rather high Overall catch comparison Table gives an overview of the comparison of catch volumes between the standard and the experimental net for each haul. This is given for the three fractions: discard shrimp, commercial shrimp and trash. Discard shrimp: the fraction of the catch containing shrimp usually smaller than 5cm, this fraction usually also contains most of the juvenile fish Commercial shrimp: the fraction of the catch containing shrimp usually larger then 5cm. Trash: the fraction of the catch containing the rest of the benthos. For the research vessel trials, the volumes in the cod-end and in the outlet cover are given. On the commercial vessels the experimental net catches less for each of the three fractions, i.e. on average over all hauls 39% (st. dev. = 24%) less trash, 36% (st. dev. = 20%) less discard shrimps and 32% (st. dev. = 14%) less commercial shrimp. On the research vessel the results are much better, with less than 5% loss (average 2%, st. dev. =

20 2%) of commercial shrimp and a reduction of the trash fraction with 44% (st. dev. = 14%) Catch comparison by species The numbers caught in the standard and the experimental net per m2 for each grid configuration are given for the agreed list of species in Table 34 - Table 43. The length frequency distributions for these species in the catches of the standard and the experimental net and the percentages released for each length class are given in Fig. s Crangon crangon The loss of commercially sized shrimps is for all commercial trials very high with an average loss around 30% and was in all cases but one statistically significant. The undersized shrimp loss is equal or lower compared to the large shrimps but for most experiments not significant. On the RV, however, the loss of commercial shrimp (-2%) and discard shrimp (-5%) was very low Marketable fish The grid is highly selective for marketable fish with about 75% of the sized fish that were guided to the outlet by the grid. The loss of marketable fish was, however, low at only 7%, thanks to the presence of the 80 mm outlet cover that retained most of the marketable fish selected out by the grid. The grid was most selective for marketable roundfish like cod and whiting with a reduction of 80%. For flatfish the grid was somewhat less selective. About 50 to 65% was guided through the outlet Undersized fish and benthos In general the grid was much less selective for undersized fish. Especially animals below 10 cm length could easily pass through the grid bars in quite large proportions. Selection started to improve between 10 and 20 cm and approached 100% above 20 cm. On average, only 23% of dab could escape through the outlet. Dab below 10 cm was caught in rather high numbers in all experiments. The RV trials (Fig ) clearly showed that the reduction of small dab (< 10cm) lied below 10%. This was confirmed in most of the commercial trials. Above 10 cm, selection started to improve and most fish above 20 cm could escape through the outlet. Due to the rather low numbers of plaice caught, the results for this species were less clear although they seem comparable to the results for dab. On average, 40% of the animals escaped. This percentage was higher as compared to dab but it should be born in mind that more large plaice were caught than for dab. On average 20% of the soles escaped. Most of the fish caught had lengths in a narrow range around 10 cm length. For these length classes, the selection looked slightly better than for dab and plaice with reductions around 50%, although the data points were rather scattered. The grid was, however, less selective for the larger fish (< 20 cm) compared to dab and plaice since many animals passed through the grid bars into the cod-end. The lengths of bib caught ranged from about 10 cm to 20 cm. The data points on catch reduction were very scattered and difficult to interpret. A length effect was not clear for the commercial trials. The RV trials, however, showed that bib of 10 cm could easily pass through the grid bars. Selection improved as fish got bigger to reach about 100% at 20 cm length. On average, 37% of the fish escaped. On average 26% of the undersized whiting

21 escaped through the outlet. The grid clearly was less selective for this species compared to bib. The catch reduction below 20 cm was very low (Fig ). Even above 20 cm many fish could pass through the grid bars into the cod-end. The reduction in benthos catch was 40% Sieve net experiments Between February 2000 and January 2001, fourteen sampling trips were undertaken on commercial vessels and two sampling trips on board of the research vessel Belgica (Table 7). A total of 101 hauls were carried out with the sieve net. The purpose of these trips was to evaluate an existing sieve net design in the Belgian shrimp fishery. At the end of February and the start of March 2000 the first trials with a sieve net were conducted on board of the RV Belgica. In April the first trials with a sieve net on board of a commercial vessel were conducted. The same design was used as on the RV Belgica except that the 11 mm outlet cover was replaced by the 80 mm one and that the fine mesh blinder was removed. Taking into account the problems with commercial catch loss observed with the sieve net design (which worked fine in the other shrimp fisheries) in the autumn and winter trials, the last trials were carried out with a slightly altered design Experiment 11 February March 2000 Belgica Trials with the standard sieve net Experiment 12 spring 2000 Trials with the standard sieve net Experiment 13 summer 2000 Trials with the standard sieve net Experiment 14 autumn & winter 2000 Trials with the standard sieve net Experiment 15 November 2000 (a) Belgica Trials with the standard sieve net Experiment 16 November 2000 (b) Belgica Trials with a sieve net with a different hanging ratio sieve net mouth contained 96 meshes instead of Overall catch composition Table gives an overview of the comparison of unsorted catch volumes between the standard and the experimental net for each haul. This is given for the three catch fractions (see section ). For the RV trials, the volumes in the cod-end and in the outlet cover are given. The RV trials gave very promising results, with only 7% (st. dev. = 2%) loss of the commercial shrimp fraction and a catch reduction for the discards shrimp and trash fraction of 17% (st. dev. = 6%) and 26% (st. dev. = 9%) respectively. The first five seatrips, carried out

22 between April and July more or less confirmed these results with again a rather low loss of the commercial shrimp fraction (12%, st. dev. = 7%) and a 27% (st. dev. = 15%) and 32% (st. dev. = 21%) reduction of the discard shrimp and trash fraction. The trials in the second half of the year, however, were rather problematic. Losses of commercial catch of over 30% (st. dev. = 13%) on average were observed. Also the catch reduction of the discard shrimp and trash fraction went up to 59% (st. dev. = 25%) and 49% (st. dev. = 18%) respectively Catch comparison by species The numbers caught in the standard and the experimental net per m2 for each grid configuration are given for the agreed list of species in Table 44 - Table 49. The length frequency distributions for these species in the catches of the standard and the experimental net and the percentages released for each length class are given in Fig. s Crangon crangon The average losses for the RV trials were very low, but each time statistically significant, with 5% or less for the commercially sized shrimps and 13% or less for the undersized shrimps. The results of the commercial vessels for the summer trials were quite comparable with the RV results with only 9% commercial shrimp loss and 27% undersized shrimp loss. During the spring trials, the losses were somewhat higher with 17% commercial and 40% undersized shrimp catch reduction. The results for the autumn and winter trials, comprising 38 hauls, were, however, consistently different from the other commercial trials and the RV experiments, with very high losses of shrimps. The commercial and undersized shrimp catch in the experimental net was on average reduced with 38% and 72% respectively. The retention points for the RV tests (Fig ) did not show a length effect for shrimp escapes. The summer trials on the commercial vessel (Fig ) show more scatter but again no length effect is noticeable. For the autumn-winter trials, however, a very clear length effect was observed (Fig ) with very high releases for the smallest shrimps and rather low releases for the biggest ones. These were also the trials with the highest shrimp catch losses. An intermediate length effect was observed (Fig ) for the spring trials when also an intermediate loss of shrimp catch was observed Marketable fish The sieve net was very selective for marketable fish. The marketable dab, plaice, sole, cod and whiting catch was almost entirely sorted out. Only whiting succeeded in penetrating the sieve net meshes to some extent with an average retention in the cod-end of about 25%. Again, the 80 mm outlet cover retained most of the marketable fish after they escaped through the outlet. The loss of marketable fish was rather low during the summer and autumn-winter trials with 9% and 19% loss respectively. The losses during the spring trials were higher, i.e. 35% Undersized fish and benthos The selective properties of the sieve net are most clear from the RV trials (Fig ). The experiments on the commercial vessels usually confirm these results but the data

23 contain more scatter (Fig ). As for the grid, the sieve shows a very poor selection for all commercial fish species with lengths below 10 cm (i.e. dab, plaice, sole, bib, whiting and cod). Only about 10% or less of these animals managed to escape through the outlet. Again, the larger the fish, the higher proportion that escapes. The sieve net had very similar selective properties for dab and plaice. Below 10 cm length the catch reduction was very low. The selection rose very steeply between 10 and 15 cm to reach 100% catch reduction around 15 cm. The retention points for sole were very scattered, although sufficient numbers of animals were caught in most experiments. The selection range was much wider compared to dab and plaice and 100% catch reduction was only reached at about 25 cm length. The smallest bib (about 10 cm) could easily pass through the sieve net meshes into the cod-end. Above that length selection improved to reach 100% around 20 cm length. For whiting a much shallower selection pattern was observed. As for bib, almost all fish of 10 cm length passed through the sieve net meshes but also many fish above 20 cm ended up in the cod-end. The 100% retention was only reached at about 30 cm. For cod, the data were scarce but the selection seemed comparable to bib. For fish selection, no seasonal effect was observed. The average benthos reduction on the RV trials was 16%, ranging from 21% in spring to 8% in winter. The lowest benthos reduction for the commercial trials was observed in summer (- 22%). In spring the reduction was somewhat higher (- 35%) and was highest in the autumn-winter experiments (- 68%). In experiment 16, a sieve net was used with a different hanging ratio. About 4% and 10% commercial and discard shrimps less were lost respectively. For fish, the selectivity was slightly less as well. For all species roughly 10% less fish were sorted out United Kingdom General haul data A total of 83 successful hauls in four different experiments were made during the UK sea trials during the period (Table 7). The catch comparison data on the UK sea trials are recorded in Table 50 to Table Reduction in the retention in the catch of unwanted species The catch comparison analysis of the catch fractions demonstrated substantial reductions in all unwanted portions of the catch (i.e. discarded shrimp, benthos and nonmarketable/sub-legal commercial discarded fish species)(see Table 50). The reduction in the retention of the unwanted portions of the catch is thought to directly result from the inclusion of the sieve net in the trawl. All four designs of sieve net tested were found to be effective in this manner

24 Loss of target species (C. crangon) It was observed that with sieve net designs UK No 1, UK No 3 and UK No 4 that the loss of target marketable size C. crangon was on average less than 10 percent (Table 54). Sieve net design UK No 2 resulted in large losses of target C. crangon of 21% (Table 54) Plaice (Pleuronectes platessa L.) by-catch reduction Catches of plaice during all of the UK sea trials were low. All four designs of sieve net tested (sieve nets designs UK No 1 No 4) were observed to reduce the level of plaice by-catch in the trawls as compared to that found in the standard trawl (Table 50 Table 53). The reduction in plaice by-catch was length dependent. All of the plaice caught in both the standard and experimental gears during the sea trials were below the current minimum landing size for this species (i.e. 27cm) Cod (Gadus morhua L.) by-catch reduction Catches of cod were very low during all of the UK sea trials. All four designs of sieve net tested were observed to reduce the level of cod by-catch in the trawls as compared to that found in the standard trawl (Table 50 Table 53). The reduction in cod by-catch appeared to be length dependent. All of the cod caught in both the standard and experimental gears during the sea trials were below the current minimum landing size for this species (i.e. 35 cm) Whiting (Merlangius merlangus L.) by-catch reduction All four design of sieve net tested were observed to reduce the level of whiting by-catch in the trawls as compared to that found in the standard trawl (Table 50 Table 53). The reduction in whiting by-catch was length dependent. Negligible numbers of whiting were caught during the UK sea trials, which were above the current minimum landing size for this species (i.e. 23 cm) Dab (Limanda limanda L) by-catch reduction Three designs of sieve net tested (sieve nets designs UK No 1, No 2 and No 4) were observed to reduce the level of dab by-catch in the trawls as compared to that found in the standard trawl (Table 50 Table 53). Design UK No 3 did not appear to reduce the levels of dab by-catch. The reduction in dab by-catch where present, was length dependent. Negligible numbers of dab were caught during the UK sea trials which were above the current minimum landing size for this species (i.e. 23 cm) Sole (Solea solea L) by-catch reduction Catches of sole were very low during all of the UK sea trials. Three designs of sieve net tested (sieve nets designs UK No 1, No 2 and No 4) were observed to reduce the level of sole by-catch in the trawls as compared to that found in the standard trawl (Table 50 Table 53). Design UK No 3 did not appear to reduce the levels of sole by-catch. The reduction in sole by-catch appeared to be length dependent. Negligible numbers of sole were caught during the UK sea trials, which were above the current minimum landing size for this species (i.e. 24 cm)

25 Flounder (Platichthys flesus L) by-catch reduction All four design of sieve net tested (sieve nets designs UK No 1 No 4) were observed to reduce the level of flounder by-catch in the trawls as compared to that found in the standard trawl (Table 50 Table 53). The reduction in flounder by-catch appeared to be length dependent. Flounder is of negligible commercial value in the UK and is only infrequently landed Benthos by-catch reduction All four designs of UK sieve net tested were observed to substantially reduced the level of benthos by-catch (Table 50 Table 53). 4. Discussion 4.1 Sorting grids The work presented here focused on sieve or veil nets and sorting grids. Sorting grids are used in many fisheries at present. In 1998 a global review of sorting grids usage throughout the fisheries of the world (Anon., 1998), around 1200 fishing vessels from ten countries involved in shrimp fisheries around the world were reported to be using sorting grid systems. A further 300 vessels were reported to be using sorting grids in eight other fisheries (fish and industrial species). The sorting grids were being used in some fisheries on a voluntary basis. Sorting grids are used in extensively in various pandalus fisheries (Brothers & Boulos, 1996, Isaksen et al., 1992). Their use has been investigated in the Faroese lemon sole fishery (Zachariassen and Jakupsstovu, 1997), gadoid fisheries (Isaksen et al., 1998), pelagic fisheries (Kvalsvik et al., 1998) as well as some hake, monk fish and panaeid shrimp fisheries (Anon., 1998) No sorting grids are used within the North Sea fisheries, with the exception of the fishery for Pandalus shrimp species in the extreme Northern Norwegian waters (Isaksen et al., 1992), although a grid for use in the North Sea Nephrops fishery is under development (Radcliffe, 1999 Unpublished data). No sorting grids were in use in the European C. crangon fisheries during 1998 (Anon., 1998). Graham (2001) describes sorting grid experiments developed for use in the UK (East coast) C. crangon fishery on two vessels, i.e. RV Osiris and MFV My Quest. Several grid designs were tested, steel elliptical grids, and flat grids constructed entirely from plastic materials. A rectangular flat grid of 0.8m x 0.6m set at 45 degrees angle with a bar spacing of 12mm was found to be the most effective design. Selectivity analyses performed using the SELECT-method on this grid resulted in an L50 of 7.5 cm and selection range (SR) of 5.3 cm for whiting. For plaice the L50 was 7.0 cm with a SR of 4.4 cm. The overall reduction in whiting for this grid was 85%, and in plaice 41%. Losses of C. crangon averaged at around 10%, but in some cases could be as high as 35%, due to blockage of the grid by sea weed

26 In Germany Wienbeck (1995) reported on experimental trials in the C. crangon fishery of the Elbe estuary using a stainless steel elliptical sorting grid with a 13mm bar spacing. The grid was housed in a cylindrical frame of 400mm diameter and was installed in the extension piece just in front of the cod end angled at 45 degrees, with a fish escape hole at the top. A series of 10 tows, of 15 minutes duration at a towing speed of 3 knots was performed. The catch of the main cod end was compared with the catch separated by the sorting grid. This achieved a reduction of 56% of plaice (Plueronectes platessa), 75% of flounder (Platichthys flesus), 99% of whiting (Merlangius merlangus), 94% of cod (Gadus morrhua) and 49% of smelt (Osmerus eperlanus). The loss of target species (C. crangon) was significant, with a mean loss of 43%. In subsequent catch comparison experiments, Wienbeck (1999a) reports on sea trials with similar larger stainless steel grids (diameter 65 cm) with bar spacing ranging from mm. The results are reported as % reductions in total catch weight, making it very difficult to draw any clear conclusions. The sorting grids reduced the total catch weight from between 18 to 38%. The commercial sized shrimp loss was reported to be around %. In the Belgian C. crangon fishery, preliminary experiments with a flat rectangular grid with a 14mm bar spacing produced some promising initial results. Flat and round fish reduction ranged from % and losses of C. crangon were limited to around 15% (Polet, 1998). One of the major problems identified with the use of sorting grids in the C. crangon fishery is that of blockages from weed and debris. Graham (1998, 2001) identified that clogging of the sorting grid with weed could sometimes present a problem in field trials. Roberts (1993) when testing a sorting grid in the Wash C. crangon fishery, also reports that debris, including brittle stars, crabs, weed, sand and mud often choked the grid allowing very little to enter the cod end and trials were subsequently abandoned. Polet (1998) also reported unacceptable losses of C. crangon when the sorting grid became clogged by the incidental catch of starfish. Wienbeck (1998a) reports that stainless steel grids have a low tendency to block with weed, algae or jellyfish in comparison to veil (sieve) nets. Large objects trapped in front of the grid could be removed by hand by fitting a zipper in the trawl and accessing the gear on hauling. A similar stainless steel grid was tested by Graham (1997) but was thought too costly to produce, and would meet with resistance from the fishermen. The selectivity of the cod-end for brown shrimp in a net with a sorting grid seems higher compared with a cod-end in a net without grid. Previous cod-end selectivity experiments (Polet, 2000) with a commercial brown shrimp beam trawl, without grid, gave an L50 of 39.4 mm ( mm) and a selection factor of 1.82 ( ). Based on the t- value of the difference between the two means, this is significantly lower (p < 0.001) compared with the present results (see Section 3.2). This is probably due to the fact that the cod-end catch size in a net with a grid is smaller and the catch composition more homogenous and contains less large benthic animals and debris that can hinder shrimp escapes

27 In a previous Belgian experiment (Polet, 2000) with a brown shrimp beam trawl, however, it was found that catch size and clogging of the meshes by seaweed and hydroids had a significant, negatively correlated effect on the cod-end selectivity of brown shrimp. Catches in a net with a grid have lower volumes and show less clogging of the cod-end meshes. Consequently, the higher selectivity parameters found in this study are in line with the previous results. The loss of marketable brown shrimp recorded during this study was comparable with the 12% loss found by Graham (2001) with a grid with 12 mm bar spacing. Wienbeck (1997) recorded a 15% loss with a grid with 20 mm bar spacing. Both authors also found a length dependence in the brown shrimp selectivity of the grid with higher losses for the larger animals. The high loss of the smallest brown shrimps as recorded in the Belgian experiments was, however, not observed. During experiments in the Norway shrimp fishery (Pandalus borealis) with a Nordmore grid with 19 mm bar spacing, shrimp losses up to 5% were recorded but no length dependence was observed (Isaksen et al., 1992). Graham (2001) found L50 values using the SELECT-method for the 12mm grid for whiting around 7.5 cm and plaice around 7.0 cm, which is, in comparison with the present study, quite higher for plaice and comparable for whiting. Starfish and debris were an important problem for the functioning of the grid. Especially starfish had a tendency to stick their arms between the bars and hold on. This reduces the selective area of the grid and decreases the opportunities for brown shrimps to pass through the bar spacing. Any object, plant or animal that accumulates on the grid will have such an effect. Seaweed, hydroids, big jellyfish, starfish, plastic bags, pieces of wood, etc., are very common throughout the year or in certain seasons in the southern North Sea catches and are likely to cause loss of commercial catch in many hauls. Therefore, a sorting grid as designed in this experiment, with a small selection area and a tendency to accumulate part of the catch, will probably be difficult to be accepted in the commercial fishery. During the cod-end selectivity trials in 97/08 almost no fish escaped through the cod-end meshes. In a previous study (Polet, 2000) the cod-end L75 for dab and sole was 4.8 and 8.7 cm respectively and no plaice with length above 4 cm escaped from the cod-end. The dab, sole and plaice caught during this cruise were not smaller than 5 cm, 7 cm and 5 cm respectively and, therefore, had very little opportunity to escape through the cod-end meshes. Consequently, the selectivity ogives were not possible to be calculated. 4.2 Veil nets. Veil nets (known as sieve nets on the continent) have been in mandatory use by all vessels engaged in the Danish fishery for some time. The mesh size of the Danish veil nets is set at a legal maximum of 60mm (Hansen, 1996). Other than in Denmark there is no requirement to use such devices in the European C. crangon fisheries

28 It is unclear from the literature as to where the designs for the veil nets, currently in use originate. One of the most likely forerunners of the veil nets in use today, was a modification of a French design described by Boddeke (1965). In this work, Boddeke describes a modified French otter shrimp trawl design, which was adapted to suit the Dutch beam trawler fleet. It featured a horizontal larger meshed separator panel and two cod-ends. The new Dutch design was based on a larger meshed sieve panel inside the trawl ending into a lower cod end to which juvenile fish were guided to escape, whilst shrimps were allowed to pass into the top cod-end. This design resulted in a substantial reduction of by-catch of juvenile fish, and higher catches of marketable shrimp. The results for 6 valid hauls were on average: +35.4% higher catches of marketable shrimps, 13.8% smaller catches of undersized shrimps, and 82.7% less small flatfish (all numbers refer to weights). In subsequent designs the lower cod-end was replaced by a discharge opening in the belly. In Germany, Mohr and Rauck (1979) describe experiments with an 8m brown shrimp trawl fitted with a selective funnel (or sieve net) ending in a discharge hole. Mesh sizes were 20mm in the cod end and 50mm in the funnel. The results over five valid hauls are given for several species. Practically all plaice larger than 10cm were released through the sieve, smaller individuals were filtered out by some 50%, about 20% of soles ranging from 4 to 9cm were released through the hole. Around 50% of cod smaller than 10cm were released, the discharge rate for crabs ranged from 60% to 90%, jelly-fish even higher i.e. 100%. In Germany (Wienbeck and Rauck, 1992) a veil net (44 mm mesh size) was tested. Results indicate that an 80% reduction in flatfish catches was observed. Shrimp loss was high at around 40% (by weight). It was suggested that veil nets of larger mesh sizes (60 or 70 mm) would reduce shrimp losses, but allow more flat fish into the cod end. In subsequent German experiments (Wienbeck, 1998b), used a 9m shrimp beam trawl testing veil nets of 50, 60 and 80 mm mesh sizes. The total catch weight was reduced from between 9-34 %, whilst the shrimp loss was between 6-15%. Detailed information of the catch compositions was not given, making it difficult to draw meaningful conclusions from the work. In France, Dupouy et al. (1998) report successful, if somewhat limited in number, experiments using unique French designs of trawls, aimed at reducing discard levels. The trawl designs are termed Asselin and Devismes trawls and the report indicates discard reductions of % (plaice), 60-87% (gadoids) and 45-75% (sole). Shrimp losses ranged from 0 17%. Generally, throughout the literature, weed does not appear to be as problematical in the functioning of veil nets as it does for sorting grids. In Germany, square mesh cod ends have been tested in the C. crangon fishery after reported observations of many small juvenile soles gilled in the meshes of the C. crangon

29 nets (Rauck and Wienbeck, 1990). The results indicated that the use of square meshed cod ends could reduce the retention of juvenile sole. 4.3 Other stimulation techniques. In a Flemish national project, Polet (1999) (in personal communications), describes trials currently underway in the Belgian C. crangon fishery to develop a trawl that uses electrical stimulation technology. The shrimp are stimulated to jump, using electrodes on the fishing gear and are subsequently caught in the trawl. There is essentially no conventional ground gear and expectations are that the by-catch of benthos and fish will be reduced as a result of the physical reduced ground contact. 4.4 Summary of selective gear use. Table 11 details the selective gears reviewed, and a brief summary of their estimated potential for use in the European C. crangon fisheries. Countries, results, problems and comments are given. It appears that at present the veil or sieve net will meet most acceptance from the fishing industry. The idea of using electrical stimulation in the Crangon fisheries, although tried extensively in the past without lasting commercial success, may hold future potential, particularly when a trawl design can be made with minimal bottom impact DISCRAN experiments General The experiments were carried out in four nations over a period of two years. In Germany and the United Kingdom only commercial vessels were used. In Belgium some experiments on a research vessel were done, and in The Netherlands all experiments were carried out on a research vessel, but with input from commercial fishermen. Differences in outcome may have resulted from this, e.g. in Belgium the results on the research vessel deviated from those found on the commercial charters. A possible explanation might be that commercial vessels operate in areas where high catches are experienced. Differences in performance based on detailed geographical knowledge and behavioural observations might occur. On the other hand in many cases deck space is limited onboard commercial vessels, and sampling may not inflict the catching operation too much. Those limitations are not felt on a research vessel. A vast array of species was sampled in the Dutch trials, larger than the list set for the project. This will enable more detailed ecological studies from the data. Another striking difference is the timing of the experiments. Growth patterns vary throughout the year and the various experiments will have met fish in different growth stadia. As the selection process is very much length dependant, the outcome may be quite different. In the original proposal for the study a cross-comparison was suggested between nations to be carried out in a third year. For budgettary reasons this was skipped from the

30 programme. A fair amount of standardisation was aimed at, but gear widths and detailed net designs could not be matched entirely. The flat grids were kept practically similar, with a limited selection of bar spacings and overall dimensions being equal. The ellitical grid was only tested in Germany, but with a comparitive range of bar spacings. The sieves differed somewhat in design, but the range of mesh sizes was limited to 50, 60 and 70 mm. In some trials the number of hauls may be rather small. In later experiments it was emphasized for statistical reasons to be restrictive in the number of configuration to test. A number between 15 and 25 seems adequate. Particularly on commercial boats there is a tendency to alter gears if catches are below expectation. Another reason for breaking up a particular experiment was mal-functioning of the gear due to blockage by sea-weed. When this occured frequently more emphasis was put in finding a suitable remedy than to carry on experimenting knowing the results will remain to be biased by this event Germany During the first year the German participants concentrated on collecting data on the fish/ shrimp separator grids. Data on sieve nets were collected too but to a much lesser extent. The main reason for this choice was the consideration that little is known about the sorting properties of grids during different seasons. Nearly nothing was known also about how grids perform in the case of the seasonal mass occurrence of jellyfish and the occasional clogging of the sorting devices by drifting seaweed, one of the main problems in shrimp fishery in Germany. During the second year, the effort was concentrated on sieve nets and methodological questions arising in the course of the project with regard to a possible effect of the ship side where the experimental trawl was towed and also to effects of the collector bag. Three different gear dimensions were tested: 70 mm mesh opening in sieve nets and 20 and 30 mm bar distance in sortings grids. This resulted initially from temporal attempts of the EU-Commission to unify dimensions of selective devices in shrimp trawls. The Commission has given up these intentions during the start period of the project. The respective legislation was left to the discretion of the nations participating at the North Sea shrimp fishery. The respective authorities are, however, obliged according to Council regulation 1298/2000 to establish national regulations until end of June From previous experience with sieve nets, under the special conditions of the German shrimp fishery with regard to the meshsize of the sieve net, it was thought that the measures envisaged by the EU-Commission were likely to be close to the optimum between reduction of unwanted fish by-catches and loss of marketable shrimp. This mesh opening, though, has never been tested in reality before. Further, from previous experiments with grids it was also guessed that 20 and 30 mm bar distance would cover the range in a sorting grid producing a similar effect to the sieve net envisaged. The shift in focus of research during the second year was partly caused by the reluctant and critical attitude of practical fishermen towards grids. Only a few saw practical advantages in seasons with a high probability of clogging because the actual construction of the grid now allows for a quick change of the oval grid plate from the protection frame. All others claimed impediment of the fishing operation, even danger to the crew with

31 these rigid metal structures hanging over the deck when emptying the codend. Extra costs were also claimed which is undeniable but somewhat farfetched compared to the daily turnover in this fishery. A trained engineer is able to produce a grid in less than two days from about 12 kg stainless steel rods, diameter 8 mm. As mentioned above some concern arose during the course of the experiments with regard to possible hidden effects of the research methodology. One was the question if the ship had one better fishing side ( starboard or port). After an experiment with a 70 mm sieve net in October 1999 (Experiment 5) the same experiment was repeated in November 1999 (Experiment 7), now with the experimental trawl on port side instead of starboard as usual. In general, the biggest differences found were detected in discarded shrimp. However, this may have been caused by the different length composition of this catch component in the peak and the declining catch season. Single species demonstrated some variance concerning the sorting effect, too, but a systematic bias for one side could not be detected. Two experiments (8 and 9) where in one the 20 mm grid was fitted out with a collector bag over the escape hole and in the other the collector bag omitted also gave no reason to believe in a falsifying effect of this additional equipment. Except a possible increased loss of commercial shrimp without the collector bag. The analysis with regard to seasonal trends in losses is at the moment only possible with grids 20 mm where catch data from July, September, October and March-April are available. It demonstrates clearly a two-peak trend with minima occurring in September and April and maxima in October and presumably in early summer. These trends apply as well for general catch components as shrimps landed, shrimps discarded and fish in general as for single fish species as e.g. plaice, whiting, and smelt. The reason for this is unclear. It strikes that this coincides with a opposite trend in by-catch of invertebrates. Thus, one explanation would be that high by-catches of invertebrates impede the escape of as well shrimp as fish. It is, however, similarly probable that at seasons with low losses much of the by-catch has a size which prevents being sorted out Netherlands The Dutch trials suffered in the initial stage from inexperience with the shrimp trawl and time was lost to find a proper rigging of the sorting grids, e.g. the use of adequate floatation. The first series of trials in 1999 were conducted in grounds in the Wadden Sea and the coastal region in the northen part of the country in co-operation with fishermen from Den Oever. The duration of these trials was extended after the first three weeks, and this resulted in a better data series. The second series in 2000 were carried in southern grounds at the estuary of the Eastern Scheldt on request of fishermen from this region and along the Dutch coast south of IJmuiden. In this series particular emphasis was put on large data sets with the same configuration to enable a meaningful statistical analysis. In both sets commercial fishermen joined to locate commercial fishing grounds and give practical comments. The trials took place in the months of November and December, which are behind the top season for shrimping. Lack of availability of the research vessel hampered comparison in other seasons. (See also Table 7)

32 The between-gear comparison was carried out by cutting the sieve loose at the outlet but leaving it inside the net on the experimental side to save time. In retrospect this may have caused higher catches for some species, possibly by extra mesh blockage from the sieve net. It is particulary noticeable that sole catches in the standard gear with the sieve still in are higher (Experiment 9 Table 30, Experiment 11 Table 32) Belgium Pre-DISRAN trials by the Belgian partner with sorting grids with the outlet on top were problematic because of clogging with benthos and debris. Several designs were tried (different grid angles and bar spacings, guiding panels, secondary grid) to reduce clogging but none were successful. Therefore the grid with the outlet below was adopted for the DISCRAN project. Pre-DISCRAN RV trials gave good results with this design. On the commercial vessel, however, again clogging caused severe problems. Clogging by starfish and other small benthos, that caused most of the problems in the previous experiments, was reduced because of the positioning of the outlet below but probably was still responsible for some of the commercial shrimp loss. Large debris and jellyfish, however, caused many problems. Alterations to the outlet (large enough to allow debris to pass) and the introduction of a guiding panel resolved the problem of clogging but could only reduce the losses to a small extent. The larger size of the outlet was probably responsible for a higher outflow of water taking part of the catch with it before it could encounter the grid for selection. The trials on the RV, on the other hand, were quite successful, with very low losses of large shrimps. This shows that no errors were made to the rigging of the device. The main difference with the commercial trials was another fishing ground with a) cleaner catches with less clogging and b) a flat and hard sandy bottom. It is obvious that less clogging reduces the loss of shrimps. The effect of the topography of the fishing ground can, however, not be neglected. The commercial fishing grounds off the Belgian coast often have small and larger sand dunes. Underwater observations were not possible due to the very low visibility in the water but it seems logical to assume that the up- and downward movement of the net caused an unstable outlet and irregular outflow of water through the outlet. Probably part of the catch went through the outlet without encountering the grid and no selection could take place. Because of the large shrimp losses, almost each sea trip changes were made to the design of the grid. Consequently, many experiments were carried out with few hauls per experiment and the data were quite scattered. It was therefore difficult to draw firm conclusions on the selective properties of the grid. The losses of shrimp also pointed to the fact that the grid was not functioning properly and probably part of the catch did not even encounter the grid before it went through the outlet. This probably also contributed to the scattering of the data. The RV data do give more reliable selectivity data for shrimp and fish, though these were only based on 4 hauls. This information is, however, hypothetical for the Belgian situation since the sorting grid does not seem to function properly in a commercial situation. Based on the amounts of shrimp lost, the experiments with the sieve net can be subdivided into four groups: a) all RV-trials very low shrimp loss, b) summer trials low shrimp loss, c) spring trials intermediate shrimp loss and d) autumn-winter trials

33 high shrimp loss. The high shrimp losses for the latter came as a surprise since the previous commercial experiments and the RV-trials gave good results, which were quite similar to the results from the other project partners. Throughout the trials, the same net and sieve net were used, without alterations. Also, the commercial vessel and its skipper and crew were the same throughout the trials. The reason for this sudden change in commercial shrimp loss was not obvious. An obvious conclusion that can be made from the results is that high losses of commercial shrimp go together with a high reduction of each catch fraction, irrespective of the size of the animals (see Section ). This points at a none-size selective process going on in the net. Part of the catch seems to leave the net without encountering the meshes of the sieve net. All variables that could have led to this difference in performance were examined: a) all RV-trials: - 6% and 10% commercial and discard shrimp loss - 19% benthos catch reduction - very low catch of dab between 25 and 30 cm - intermediate catch of whiting between 25 and 30 cm - flat and hard sandy bottom - clean catches b) summer trials - 9% and 17% commercial and discard shrimp loss - 22% benthos catch reduction - almost no catch of dab between 25 and 30 cm - almost no catch of whiting between 25 and 30 cm - fishing ground with sand dunes - larger catches c) spring trials - 17% and 40% commercial and discard shrimp loss - 35% benthos catch reduction - intermediate catch of dab between 25 and 30 cm - intermediate catch of whiting between 25 and 30 cm - fishing ground with sand dunes - larger catches d) autumn-winter trials - 38% and 72% commercial and discard shrimp loss - 58% benthos catch reduction - highest catch of dab between 25 and 30 cm - highest catch of whiting between 25 and 30 cm - fishing ground with sand dunes - larger catches An extra factor that should be taken into account is that during the autumn-winter trials, whiting but especially dab were often observed to be gilled in the sieve net meshes around the outlet, distorting the shape of the outlet. The RV-trials were most successful with low shrimp and benthos losses and almost no large dab in the catch. So the sieve net does not seem to be very selective for benthos. Almost equally good results were observed during the summer trials, also with a low

34 benthos catch reduction and almost no large dab in the catch. During the spring trials, the shrimp losses more or less doubled. Large dab were present in intermediate numbers. Since large dab were gilled in the meshes around the sieve net outlet, these fish probably distorted the outlet. This distortion would let part of the catch escape through the outlet without it being subject to selection by the sieve net meshes. This is confirmed by the fact that also more benthos was released, which would not be logical if the sieve net functioned properly. During the autumn-winter trials, the shrimp catch reduction increased further, as was the case for benthos. In that period, however, high numbers of large dab were caught. This confirms again that the presence of large dab was probably the reason of these losses. The other partners did not observe such high losses of shrimp, most likely because large dab were not present in their catches at all. The effect of whiting is less clear. As for dab, large whiting were gilled around the outlet in the autumn-winter trials. These fish were also present in the RV catches, but were not gilled in that case. Possibly, gilled dab caused whiting to be gilled as well. In the absence of large dab, whiting could easily pass through the outlet. The loss of the commercial brown shrimp catch was estimated to be 13% on RV Belgica. This number should be interpreted with caution since it is an average of all numbers in a pooled range of length classes and depends on the catch composition. The losses of shrimps are length-dependent and the amount of shrimps larger than 60mm in the catches during the sea trials in this experiment was relatively low. Consequently, if the relative amount of large shrimps in the catches would be higher, the percentage commercial shrimps lost would also increase. With the conversion formula to calculate the carapace width of a brown shrimp from its total length, it was found that the largest shrimps in the catches in this experiment had carapace widths below 11.5 mm. Theoretically these shrimps should be able to pass through the bar gaps of a grid with a 12 mm bar spacing. The tests have shown, however, that this was not the case. Even with a 14 mm bar spacing, a significant amount of the larger shrimps escaped through the outlet. Probably for some of the animals encountering the grid with their bodies not parallel to the gaps, time is too short to obtain the right orientation to pass through the gaps or other material lying against the grid hinders their passage. The results on the length dependence of small shrimps lost through the outlet, however, seem illogical since they should pass through the gaps more easily than the bigger ones. A possible explanation could be that the smallest animals are taken with the water flow going through the outlet quite easily due to their low body weights. The numbers of shrimps caught in the lowest and highest length classes were rather low, however, and the confidence limits rather wide, which results in less firm conclusions on this matter. The length classes within the selection range of the sorting grid for fish lie far below the minimum landing size for all species observed. The grid gives the opportunity to the larger fish (age 1 and higher) to escape through the outlet and almost all marketable fish will be lost if no second large mesh cod-end is attached over the outlet. Age 0 fish, however, still pass quite easily through the grid bars (50% and more) and are caught in the cod-end. This has consequences for the applicability of the sorting grid. If the

35 unwanted by-catch of fish consists mainly of the larger fish, the grid can be useful. If mainly small fish are caught, like in nursery areas, the effectiveness of the grid is smaller. When comparing the results of RV Belgica and the commercial vessel, the variability in the data is higher on the latter, due to the rather low number of successful hauls. The grid selectivity results for brown shrimp and plaice were confirmed in commercial conditions. For dab, however, more small fish escaped in the case of the commercial vessel. Although the opposite seems to be true for whiting, such a comparison is difficult to be made since small fish were not present in the catches on the RV. The results for sole were unexpected, since the smallest fish (< 8 cm) showed a higher escape rate than the mid range of the length classes. It is possible that the smallest fish were more easily taken by the water flow through the outlet. It is, however, also possible that due to the lower catch volume in the cod-end of the experimental gear that the selectivity for sole improved when compared to the standard gear. The set-up of the comparative fishing experiment, however, does not allow to attribute the higher escapes of the small soles to one or the other possible cause. The selective properties for flatfish in grids and sieve nets are quite different. With the grid, more large flatfish could slide through the grid bars and end up in the cod-end. The increase in selection from smaller to larger fish is very shallow. In the sieve net, on the other hand, this increase is very steep and rises from about 10% catch reduction at 10 cm to 100% around 15 cm. As the length frequency distributions indicate (Fig ) many fish of marketable size are caught in the Belgian Crangon fishery. Contrary to the other Crangon fisheries, Belgian fishermen depend for part of their income on this catch, which they are allowed to land. The grid and the sieve net, however, sort out most of these fish. Therefore there is a need for the 80 mm outlet cover to retain this part of the catch, although some losses still occur compared to the standard net United Kingdom All sea trials in the UK were conducted during the shrimp fishing season in 1999 and 2000 onboard two commercially registered UK shrimp fishing vessels. The experimental design does not however, allow for assessing potential differences between the two vessels or differences between the two seasons. Any such effects are therefore incorporated into the estimates of the selectivity of the individual types of sieve nets. Likewise there was not made any attempts to relate the selectivity to other (external) variables, such as total catch weight, weather conditions etc.. Four different designs of sieve nets were evaluated, with an average of twenty successive hauls undertaken for each design. This appeared to be sufficient in number to provide data for analysis. However, the absolute fish numbers caught during the sea trials were relatively low, and as such necessitated the use of bootstrapping techniques to make determinations of the selectivity characteristics of the four different designs. Three designs of sieve net (Design No s UK No 1, 3 and 4) all gave rise to losses in target species of less than 10 percent. UK design No 2 resulted in commercially unacceptable

36 losses of target species and was thought to be due to an excess of the number of meshes around used in the construction of the sieve net. Too many meshes around, can result in the meshes being more closed and lead to the direct funnelling of the catch through the escape hole. It will not allow for adequate sorting of the catch and losses of target species can result. The correct rigging and design of sieve net used is therefore critically important in terms of the economic catching potential of a vessel using such a selectivity device. All of the designs of sieve net were effective at reducing the unwanted by-catch of both fish and benthos. This was demonstrated to be both size and length dependent, with larger specimens being more readily eliminated from the net and thus avoiding retention after capture. Catch data from a large experiment involving two vessels at two cruises, testing four different types of sieve nets, have been analysed for selectivity properties. Three species were measured in each case. The primary deficiency of the analysis is due to the sparseness of the catches. It was not possible to use the traditional analysis techniques based on well-known properties of maximum likelihood estimation. Curves could be fitted for a number of hauls, but they appeared rather dubious or were only very vaguely determined. There were only minor problems observed in fitting curves to the pooled data, but this neglects the between-haul variation, which appears to be generally large for all cruises and all species. Apart from the obvious under-estimation of the variance, pooling of hauls also seems to affect the estimation of the parameters and in particular that of SR. Instead a bootstrapping technique was applied, with a scheme that accommodate both within as well as between-haul variation. Comparisons of the four designs are summarized by species in terms of L50 and SR. No comments are made on the split value as it is of less interest and do not seem to vary very much between species or between designs. For plaice there appears to be no significant difference in location between the four designs. Design No 2 has both the smallest L50 and shows also the smallest variation, whereas design No 4 has the largest L50 and also the largest variation. There are only minimal differences between the medians and mean values for all designs. Design No 3 and No 4 show the largest variation and do also seem to have large SR values than design No 1 and No 2. The differences are however not significant. The median value is close to 0 for all designs, The size and position of the L50 intervals and statistics for dab are very similar for design No 1 and No 2. In particular the size of the intervals are much smaller than for design No 3 and No 4, with No 4 having the largest interval, median and mean values. All intervals are overlapping so no difference between the designs can be proven. All four designs show very steep selection with both mean and median values close to 0. Design No 3 and No 4 differ from No 1 and No 2 by a much larger variation. For whiting design No 3 and No 4 appear similar with No 4 having a slightly larger L50. The variation of No 1 and No 2 seems bigger and do also have smaller L50 values. The

37 difference between No 1 and No 4 are close to being significant at a 5% level. Again all four designs show very steep selection with SR values close to 0. Design No 1 appears to have a slightly larger variation than the others, with design No 3 having the smallest. With the above remarks in mind, it is clear that all results inferred from these data should be taken cautiously. The sieve nets use a mesh size that is not easily penetrated by fish above cm. This implies that length classes, which are important for the determination of the selectivity, are not observed, as these are let out of the trawl Comparison between countries. The results obtained in each country can not be compared completely, as different gear widths, and deviating selective devices were used. In addition the seasons fished and the stocks fished were different. It is noticeable, that the number of shrimps caught in the German experiments is much larger than in the Dutch or Belgian experiments. The Dutch Wadden Sea data is comparable in numbers with the German data. The formulae used in the calculation were checked and no errors were found. A likely explanation is the difference in abundance of shrimps on differing fishing grounds in differing seasons. The abundance and landings figures in Anon., 2000 confirm this view and indicate that the numbers are realistic. In spite of attempts to standardise procedures, deviations in practice occured. In the Dutch data set the number of discard fish comprises of the number of all sub-legal commercial species and the number of by-catch non-commercial fish species added. Some species do not have a minimum landing size (smelt, bib, gurnards), but are sorted out by size in the auction. The small ones were also added to discards, using a minimum size practiced. The UK-data is based on weights rather than counts. A cross-reference between countries was originally planned in the study, but abandoned for budgettary reasons. Swapping gears between nations and comparing the performance is recommended. 5. Conclusions 5.1. General Both the sieve net and the grid are devices that can effectively sort out by-catch fish species with lengths over 10 cm, but they have to be rigged properly and blockage due to sea weed or jelly fish may impair their effectiveness. Juvenile fish species smaller than approximately 10 cm are not effectively released by the sorting devices tested. Effective selection devices were developed and tested in Germany, The Netherlands and the UK. The loss in commercial shrimps in these cases was between 5-20%. However,

38 commercial trials in Belgian waters showed higher losses on average 37% probably caused by the unique Belgian catch composition and seasonal differences. Both the sieve mesh size of 70mm and a grid bar distance of 20mm demonstrated proper sorting effects. Fishermen expressed a preference for non-rigid structures to install inside a net. The majority of experiments in the second year were therefore dedicated to this selection device. Grids in a rigid frame can have the advantage of quicker instalment, and deletion from the net in case of quickly changing circumstances on the fishing ground, e.g. the occurrence of sea weed. Some general trends were observed. Grids and sieve nets work well on plaice, flounder, smelt, cod, and to a somewhat lesser extent on dab, bib and whiting Germany Juvenile flatfish: The three tested selection devices brought a remarkable reduction of the catch juvenile flatfish. With plaice the catch of the experimental trawl was, compared to the standard trawl " With a 70 mm sieve net: 69.3% ( average of two seasons) " With a 20 mm sorting grid: 70.2% (average of four seasons) " With a 30 mm sorting grid: 77.6% (average of three seasons) With dab the catch of the experimental trawl was, compared to the standard trawl " With a 70 mm sieve net: 89.3% (average of two seasons) " With a 20 mm sorting grid: 44.2% (average of four seasons) " With a 30 mm sorting grid: 77.2% (average of three seasons) Shrimps: The three selection devices caused losses of shrimp in the following order of magnitude. Shrimps of commercial size: " With a 70 mm sieve net: 18.3% (average of two seasons) " With a 20 mm sorting grid: 24.8% (average of four seasons) " With a 30 mm sorting grid: 20.1% (average of three seasons) Discard shrimp: " With a 70 mm sieve net: 44.4% (average of two seasons) " With a 20 mm sorting grid: 24.1% (average of four seasons) " With a 30 mm sorting grid: 26.6% (average of three seasons) An observed bias in an order of magnitude of 7% in total caused by the experimental procedure (collector bags) might reduce the target species losses even further. No fish of commercial size were released because these normally do not occur in the German shrimp fishing grounds used for the investigation

39 Seasonal effects: A season related variation of by-catch reduction and shrimp losses could be observed. Both showed peak values in seasons with low catches. Gear recommendations: Apparently the suggested 70 mm sieve net fulfils the intended requirements of a remarkable by-catch reduction, particularly in endangered flatfish species. This is, however, accompanied by a loss of shrimp. Nevertheless, in a situation as the present with a voluntary catch restriction this seems a less urging argument. The grid solutions tested find resistance with commercial fishermen due to handling problems. Some fishermen, however, see advantages in their use due to the possibility of a quick cleaning in situations where drifting seaweeds impedes the fishery and cloggs the meshes of a sieve net in short time. It is recommended to leave the choice what selection device to use to the fishermen. A bar distance of 25 mm seems to be a good compromise between by-catch reduction and loss of target species The Netherlands. The highest sorting effect in juvenile plaice was measured with Sieve 70 (ratio EXP/STD =14%). The lowest figure with Grid 15 (65%). The only significant effect in dab was found with Sieve 70 (71%). Higher losses were found in commercial shrimps for the grids (ratio 68% for Grid 15; 56% for Grid 20). Sieve ratios were 90% for the 70mm and 20 for the 60mm case. Discard shrimp ratios were for Grid 15 was 50% and for Grid 20 58%. For Sieve 60 and Sieve 70 no significant differences were found. Both sieves show significant reductions in marketable plaice and cod, but there were hardly no data for these categories for the grids. For legal whiting the grids were most effective (ratio 0%), for Sieve 60 the ratio was 6% and for Sieve 70 23%. Unexpectedly the Grid with15mm bar size sorted out a higher percentage of commercial shrimps. This might have been caused by unobserved blockage. The sieves cause lower shrimp losses and better release of juvenile plaice and are more acceptable to the fishermen. The differences between the 60mm and 70mm sieve were not very large in most species, but a tendency was that the 60mm sieve sorted out some more commercial shrimps and overall more commercial fish species of legal sizes, but the results for species separately were not very consistent. The sieve with 70mm seems acceptable, because of the limited shrimp loss, i.e. 10% Belgium If the catch composition in the brown shrimp fishery does not cause clogging problems the sorting grid meets the purpose it was designed for. The reduction of age 1 and older fish in the catch was very satisfactory. Age 0 fish were sorted out as well, but to a lesser

40 extent. Most benthic animals are selected out by the grid. The commercial brown shrimp catch is reduced but usually not more than 15% gets lost. The cod-end catch consists mainly of brown shrimps and requires less sorting and the cod-end selectivity for brown shrimp increases. In this case, the sorting grid seems a very favourable choice to improve the selectivity of the brown shrimp beam trawl. If, however, material occurs in the catch that can cause clogging, the commercial brown shrimp catch is soon below the level acceptable to commercial fishermen. Starfish and debris were the main causes of clogging during these experiments. Seaweeds, hydroids, jellyfish and other debris, however, often occur in the brown shrimp catches in the southern North Sea and can be expected to cause the same problems. In this case it can be anticipated that acceptance by the fishing industry will be difficult United Kingdom Four different designs of sieve net were tested in the UK trials (UK designs No 1 4). The by-catch of juvenile plaice (below M.L.S.) was reduced, with an experimental/standard gear catch ratio of between 62-80% with these designs. Similarly, the by-catches of dab by-catch were reduced (with an experimental/standard gear catch ratio as low as 65%. Sieve nets designs UK No 1, 3 and 4 resulted in losses of commercial sized shrimp (the target species) of less than 10% (i.e. experimental/standard gear catch ratio of <90%). Design UK No 2 resulted in losses of 21% (ratio of 79%) and as such, this design was therefore considered to be economically unacceptable to the UK fishermen. The undersized (normally discarded) shrimp catch was reduced, with a experimental/ standard gear catch ratio of between 71-84% by the sieve nets. Commercially sized fish are not normally encountered in the UK C. crangon fishery and indeed only negligible quantities were of such fish were encountered during the UK sea trials. The analysis of the selective properties of the four different UK designs of sieve nets was undertaken and it was concluded that no significant differences could be demonstrated between the designs, even though the mesh sizes ranged from 52 68mm. It is therefore recommended that a mesh size not exceeding 70mm be used in the sieve nets in UK waters. This recommendation will ensure that effective discard reduction will be implemented in this UK fishery and will not adversely impact upon those UK shrimp fishermen already using sieve nets on a voluntary basis. It is strongly advised that the UK fishery regulatory authority (MAFF) consult with the UK shrimp fishermen regarding the negotiation of a seasonal derogation from this recommendation, in respect of the problems of weed as highlighted during consultation with the fishermen. Based on the sea trials conducted in the Germany, the Netherlands and Belgium, a separator grid with a maximum bar spacing not exceeding 20mm is recommended for use in UK waters as an alternative to the 70mm sieve net

41 6. References Anon., 1998c. Report of the study group on grid, grate) sorting systems in trawls, beam trawls and seine nets. ICES CM 1998/B:2 Anon., Report of the Working Group on Crangon fisheries and life history. ICES CM 2000/G:11. Bastin, A., Regionale sedimentologie en morfologie van de zuidelijke Noordzee en van het Schelde estuarium. Phd thesis Faculty of Sciences, Catholic University of Leuven Boddeke, R., New Dutch beam trawl stops flatfish slaughter. World Fishing, pp Brothers, G., and Boulos, D., Size sorting shrimp with an in trawl grid system. Working paper presented at ICES FTFB WG meeting 1996, Woods Hole, USA. Campos, A., Fonseca, P. and Wileman, D.A., Experiments with sorting panels and square mesh windows in the Portuguese crustacean fishery. ICES CM 1996/B:15. Dupouy, H., Vacherot, J.P., and Meillet, M., Selective trials for the by-catch of brown shrimp, Crangon crangon) in the estuary of the Vilaine, (Southern Brittany, France). Technical annex to ICES CM 1998/G:8 Fryer, R.J., A model of between-haul variation in selectivity. ICES J. Mar. Sci, 48: Graham, N., By-Catch reduction in the brown shrimp, Crangon crangon, fisheries using a separation (Nordmøre) grid. Fisheries Research (in press). Hansen, K., Fleet Inventory. Danish shrimp fishing fleet. Supplement to RIVO Report CO54/N7, Isaksen, B., Gamst, K., Kvalsvik, K. and Axelsen, B., Comparison of selectivity and user properties between Sort-x and single grid for two panel bottom trawl for cod, Gadus morhua). Working paper presented at ICES FTFB WG meeting, April 1998, La Coruna, Spain. Isaksen, B., Valdemarsen, J.W., Larsen, R.B. and Karlsen, L., Reduction of fish bycatch in shrimp trawl using a rigid seperator grid in the aft belly. Fisheries Research, 13, 1992, Kvalsvik, K., Misund, O.A., Gamst, K., Skiede, R., Svellingen, I., and Vetrhus, H., Size selectivity experiments using sorting grid in pelagic mackerel, Scomber scombrus) trawl. Working paper presented at ICES FTFB WG meeting, April 1998, La Coruna, Spain. Van Lancker, V., Sediment and morphodynamics of a siliclastic near coastal area, in relation to hydrodynamical and meteorological conditions: Belgian continental shelf. PhD thesis Faculty of Sciences, University of Gent Marlen. B. van, and Bouman, K., Inventory of the Dutch fleet fishing for brown shrimp. Supplement to RIVO Report CO54/N7, Marlen, B. van, Redant, F., Polet, H., Radcliffe, C.C.S., Revill, A.S., Kristensen, P.S., Hansen, K.E., Kuhlmann, H.J., Riemann, S., Neudecker, Th. and Bradant, J.C., Research into Crangon Fisheries Unerring Effect. (RESCUE) - EU-Study 94/044. RIVO Report C054/

42 Mohr, H. and Rauck, G., First results of German experiments with a selective shrimp trawl. ICES C.M. 1979/ B:7. Polet, H., Experiments with sorting grids in the Belgian brown shrimp, Crangon crangon) fishery. Working paper presented at the ICES FTFB WG meeting, April 1997, La Coruna, Polet, H., Provisional results from the national Flemish project to develop electric stimulation as mean of reducing by-catch in the Belgian C. crangon fisheries. Unpublished data and personal communications, Polet, H., Cod-end and whole trawl selectivity of a shrimp beam trawl used in the North Sea. Fisheries Research 48 (2000) Polet, H. and Redant, F., Report on the technical fleet inventory of the Belgium brown shrimp fishing fleet. Supplement to RIVO Report CO54/N7, Prawitt, O., Rosenthal, H., Thiele, W., Relationship between fishing power and vessel characteristics of German beam trawlers fishing for brown shrimp in the North Sea. ICES CM 1996/B:9 Prawitt, O., Kuhlmann, H.J. and Neudecker, T., Report on the technical inventory of the German brown shrimp fishing fleet. Supplement to RIVO Report CO54/N7, Rauck, G and Wienbeck, H., Square meshes in shrimp fisheries a possible means to reduce meshing of young soles. Inf.- Fischwirtsch.1992 vol.39, No.1, pp REDUCE project, EC, DG-XIV funded project, No FAIR CT 97/38/09 currently underway unpublished results. Revill, A., The UK, East Coast brown shrimp fishing fleet. Inventory of vessels and gear & analysis of effort. Supplement to RIVO Report CO54/N7, Revill, A., The economic and biological impacts of discarding in the UK (East Coast) Crangon crangon fishery. Report for MAFF, Chief Scientist Group, London, Food Research Centre, University of Lincolnshire & Humberside, Revill, A., 1998a. The biological and economic impacts of discarding in the UK, East coast C. crangon fisheries. Conference proceedings of the Xth annual conference of the European Association of Fisheries Economists, April 1998, The Hague, The Netherlands. Revill, A., 1998b. Discarding in the UK (East Coast) C. crangon fisheries, a bioeconomic impact assessment. Working paper presented at ICES FTFB WG meeting, April 1998, La Coruna, Spain. Revill, A.S., Pascoe, S., Radcliffe, C.C.S., Riemann, S., Redant, F., Polet, H., Damm, U., Neudecker, Th., Kristensen, P.S. and Jensen, D., The economic consequences of discarding in the Crangon fisheries. Study report to the European Commission No 97 / SE /025. Roberts, J.C., Juvenile fish by-catch study: Modified Larsen grid trials. Annual Research Report of the Eastern Sea Fisheries Joint Committee, p Schick, D.F., and Brown, M., Reduction of finfish by-catch in the Gulf of Maine northern shrimp fishery. Research on gear and implementation. Proceedings of the East Coast conference, Rhode Island, USA 7-8 April Rhode Island Sea Grant, , pp Thorsteinsson, G., The use of square mesh cod ends in the Icelandic shrimp, (Pandalus borealis) fishery. Fisheries Research, 13, 1992, p

43 Wienbeck, H., Sorting grids -- an alternative to sieve- or veilnets in the brown shrimp fishery. Inf.-Fischwirtsch vol. 42, no. 1, pp Wienbeck, H., 1998a. First trials on the selection of a sorting grid in commercial fishery for brown shrimps. Anim.-Res.-Dev vol. 47, pp Wienbeck, H., 1998b. Analysis of the by-catch reduction by means of sieve nets in the commercial brown shrimp fishery. Inf-Fischwirtsch vol. 45, no 1, pp Wienbeck, H., Analysis of the by-catch reduction by means of sorting grids in the commercial brown shrimp fishery. Inf.-Fischwirtsch vol. 46, no. 1, pp Zachariassen, K, and Jakupsstovu, S.H., Grid sorting in a trawl for lemon sole. Working paper presented at ICES FTFB WG meeting, April 1997, La Coruna, Spain

44 Table 1 Estimated total numbers of four fish species discarded by the European C. crangon fleets in a single complete fishing year (1996/1997) Species No of fish (in millions) discarded in a single year ( ) Plaice 928 Whiting 55 Cod 42 Sole 16 Source: Van Marlen et al (1998) NB: All fish were aged 0, 1 and 2 years old Table 2 Estimated numbers of plaice discarded in a fishing year (1996/1997) in the European C. crangon fisheries per country National C. crangon fishing fleet Estimated number of plaice (in millions) discarded in one year by national C. crangon fishing fleets (1996/1997) Germany 724 The Netherlands* 158 Denmark 34 UK 10 Belgium 2 *Based on combined extrapolations of the Danish and Belgian discard data sets Source: Van Marlen et al. (1998) Table 3 Estimates of discard survival from the C. crangon fisheries Species Estimated discard survival rate Plaice 20% Sole 50% Whiting 0% Cod 0% Source: Revill et al, (1999)

45 Table 4 The estimated lost landings arising from current levels of discarding in the N. Sea C. crangon fisheries Species Estimated annual lost landings due to discarding in the N. Sea C. crangon fisheries (tonnes) (range shows 95% confidence limits of estimate) Estimated lost annual landings as a % of the 2000 North Sea TAC (tonnes) Plaice 7,349 18, % Cod 997 3, % Whiting 871 2, % Sole 153 1, % Source: Revill et al (1999) Table 5 The estimated value of the lost landings attributable to current levels of discarding in the North Sea C. crangon fisheries. Species Estimated lost annual landings (tonnes) Average prices (Euro/tonne)* Fish market gross value of estimated lost landings (million Euro) Plaice 7,349 18,749 1, Cod 997 3,198 1, Whiting 871 2, Sole 153 1,355 10, All four species 9,370 25, Source: * European Commission DGXIV Statistical Bulletin (Sept 1999)(EU trend prices (1998) Table 6 The estimated annual losses to the North Sea SSB s arising from current levels of discarding in the North Sea C. crangon fisheries Plaice Whiting Sole Cod Potential SSB spawners lost due to discarding (tonnes/ year) 14,300 37,400 1,600 4, , ,500 Potential SSB spawners lost due to discarding (% of last 3 year average SSB) 6 16% % % % Source: Revill et al (1999), NB: SSB (Spawning stock bio-mass)

46 Table 7 Overview of trials Cy Exp Vessel Period B (m) Subject Fishing ground # hauls (valid) D 1 MFV Ramona July Grid el20 Medem/Elbe 15 2 MFV Ramona July Grid el30 Medem/Elbe 4 2 MFV Ramona July Grid el30 Medem/Elbe 8 4 MFV Ramona Sept Grid el30 Medem/Elbe 10 3 MFV Ramona Sept Grid el20 Medem/Elbe 11 5 MFV Ramona Oct Sieve 70 Medem/Elbe 6 5 MFV Ramona Oct Sieve 70 Medem/Elbe 10 6 MFV Ramona Nov Grid el20 Medem/Elbe 9 7 MFV Ramona Nov Sieve 70 Medem/Elbe 15 8 MFV Ramona Mar Grid el20 Medem/Elbe 10 8 MFV Ramona April Grid el20 Medem/Elbe 5 9 MFV Ramona April Grid el20 Medem/Elbe MFV Ramona June Sieve 70 Medem/Elbe 9 11 MFV Ramona July Sieve 70 Medem/Elbe 13 NL 1 RV Isis Nov-Dec Sieve 60 Wadden Sea 8 2 RV Isis Nov-Dec Sieve 60 Dutch Coast 4 3 RV Isis Nov-Dec Grid 15 Wadden Sea 12 4 RV Isis Nov-Dec Grid 15 Dutch Coast 10 5 RV Isis Nov-Dec Grid 20 Wadden Sea 14 6 RV Isis Nov-Dec Sieve 70 East. Scheldt 5 7 RV Isis Nov-Dec Sieve 60 East. Scheldt 19 8 RV Isis Nov-Dec Sieve 60 Dutch Coast 24 9 RV Isis Nov-Dec Std Dutch Coast RV Isis Nov-Dec Sieve 70 Dutch Coast RV Isis Nov-Dec Std Dutch Coast 27 B 1 MFV Benny 11 Aug Grid 20 Flemish banks 2 2 MFV Bisiti 3-4 Sep Grid 20 Flemish banks 6 3 MFV Bisiti 8 Sep Grid 20 Flemish banks 3 4 MFV Bisiti 9 Sep Grid 20 Flemish banks 4 5 MFV Benny 15 Sep Grid 20 Flemish banks 2 6 MFV Bisiti 13 Oct Grid 20 Flemish banks 3 7 MFV Benny 28 Oct Grid 20 Flemish banks 3 8 RV Belgica 3 Nov Grid 20 Flemish banks 7 9 RV Belgica 25 Nov Grid 20 Flemish banks 4 10 MFV Bisiti Jan 00 8 Grid 20 Flemish banks RV Belgica Feb-Mar Sieve 1, 70 Flemish banks 8 12 MFV Bisiti Apr Sieve 1, 70 Flemish banks MFV Bisiti Jul Sieve 1, 70 Flemish banks MFV Bisiti Sep-Dec Sieve 1, 70 Flemish banks RV Belgica Nov Sieve 1, 70 Flemish banks RV Belgica Nov Sieve 2, 70 Flemish banks 6 UK 1 MFV Prospects Oct Sieve 1, 50 The Wash 20 2 MFV My Quest Nov Sieve 2, 60 The Wash 21 3 MFV My Quest Feb Sieve 3, 60 The Wash 22 4 MFV Prospects Feb Sieve 4, 70 The Wash 20

47 Table 8 Gear dimensions on RV Isis Item Standard net Modified gears beam length 9 m 9 m headline length m m gear weight kg kg footrope length m m rollers mm mm codend depth in meshes meshes meshes codend circumference 2 x 130 meshes 2 x 130 meshes codend mesh size mm mm codend material PA, 210/24 PA, 210/24 veil mesh size mm veil cutting rate 1N11B stretched length 5.01m length set 4.41m veil material PA, 210/30 join upper end 3 to 1 join lower end 2 to 1 sock mesh size grid dimensions bars ø grid material outer tubing ø MOD1 MOD2 MOD3 80mm 900x600mm 6mm INOX tubing 17.2mm veil 60mm inserted grid 15mm spacing grid 20mm spacing

48 Table 9 Gear details of the UK experiments Item Standard net Standard net fitted with sieve net design UK No1 Standard net fitted with sieve net design UK No2 Standard net fitted with sieve net design UK No 3 Standard net fitted with sieve net design UK No 4 Beam length 6m 6m 6m 6m 6m Cod end circumference (No of meshes round) Cod end length (No of meshes) Cod end material PA 210/30 PA 210/30 PA 210/30 PA 210/30 PA 210/30 Cod end mesh size 22.7 mm 22.7 mm 22.7 mm 22.7 mm 22.7 mm (full mesh) Cod end mesh size 20mm 20 mm 20 mm 20 mm 20 mm (between knots) Sieve net present No Yes Yes Yes Yes Table 10 Minimum Landing Sizes (MLS) used in the participating nations. (Source: EU Regulation Nr. 894/97 Technical measures, Annex II) Species D MLS used (cm) NL MLS used (cm) B MLS used (cm) sole (Solea solea L.) plaice (Pleuronectes platessa L.) dab (Limanda limanda L.) 23 - brill (Scophthalmus rhombus L.) 30 turbot (Scophthalmus maximus L.) 30 whiting (Merlangius merlangus L.) cod (Gadus morhua L.) bib (Trisopterus luscus L.) 25* 25 gurnards (Eutrigla gurnardus L.; Trigla Lucerna L.; 15* 15 Aspitrigla cuculus L.; Trigloporus lastoviza L.) shrimps (Crangon crangon L.) 0.45** 0.45 UK MLS used (cm) * No official MLS, but used to sort out in the market. ** Converted from length-width key, with minimum carapace width of 6.5mm

49 Table 11 fisheries. Summary table of selective gears and their potential for use in the C. crangon Selectivity device Veil nets Country where tested Used to varying degrees throughout Europe, notably in Denmark Results Problems Comments Commercially acceptable albeit with some reluctance Some loss of shrimps (10 15%) Probably the most commercially acceptable selectivity measure available at present Selection grids UK, Belgium & Germany Great variation of the results between the national trials Blockage of the grid and subsequent loss of catch A promising solution, but has a blockage problem. May meet with some resistance from the fishermen. Electrical stimulation Belgium Promising results in aquarium systems Square mesh cod ends Germany Some observed reduction in sole by-catch Practicality problems in the field Limited trials Early days, but may be too impracticable Unlikely to be as effective for larger flatfish and other species

50 Table 12 Catch comparison Germany Experiment 1 Table Gear STD P Day/Night N Nation D Gear EXP STB Time July Experiment 1 Fish. ground Medem Tot. hauls 15 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

51 Table 13 Catch comparison Germany Experiment 2 Table Gear STD P Day/Night N Nation D Gear EXP STB Time July Experiment 2 Fish. ground Medem Tot. hauls 12 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS > MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

52 Table 14 Catch comparison Germany Experiment 3 Table Gear STD P Day/Night N Nation D Gear EXP STB Time September Experiment 3 Fish. ground Medem Tot. hauls 11 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

53 Table 15 Catch comparison Germany Experiment 4 Table Gear STD P Day/Night N Nation D Gear EXP STB Time September Experiment 4 Fish. ground Medem Tot. hauls 16 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

54 Table 16 Catch comparison Germany Experiment 5 Table Gear STD P Day/Night N Nation D Gear EXP STB Time October Experiment 5 Fish. ground Medem Tot. hauls 15 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

55 Table 17 Catch comparison Germany Experiment 6 Table Gear STD P Day/Night N Nation D Gear EXP STB Time November Experiment 6 Fish. ground Medem Tot. hauls 9 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

56 Table 18 Catch comparison Germany Experiment 7 Table Gear STD STD Day/Night N Nation D Gear EXP P Time November Experiment 7 Fish. ground Medem Tot. hauls 15 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

57 Table 19 Catch comparison Germany Experiment 8 Table Gear STD P Day/Night N Nation D Gear EXP STB Time March Experiment 8 Fish. ground Medem Tot. hauls 15 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt bib gurnard by-catch

58 Table 20 Catch comparison Germany Experiment 9 Table Gear STD P Day/Night N Nation D Gear EXP STD Time March Experiment 9 Fish. ground Medem Tot. hauls 9 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS sole < MLS dab < MLS flounder < MLS turbot < MLS brill < MLS whiting < MLS cod < MLS smelt ??? bib gurnard by-catch

59 Table 21 Catch comparison Germany Experiment 10 Table Gear STD P Day/Night N Nation D Gear EXP STB Time June Experiment 10 Fish. ground Medem Tot. hauls 9 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS > MLS sole < MLS > MLS dab < MLS > MLS flounder < MLS > MLS turbot < MLS > MLS brill < MLS > MLS whiting < MLS > MLS cod < MLS > MLS smelt bib gurnard by-catch

60 Table 22 Catch comparison Germany Experiment 11 Table Gear STD P Day/Night N Nation D Gear EXP STB Time July Experiment 11 Fish. ground Medem Tot. hauls 13 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD % p STD EXP EXP/STD % p shrimps (landed) shrimps (discarded) fish (landed) fish (discarded) other (e.g. benthos) Commercial species: plaice < MLS > MLS sole < MLS > MLS dab < MLS > MLS flounder < MLS > MLS turbot < MLS > MLS brill < MLS > MLS whiting < MLS > MLS cod < MLS > MLS smelt bib gurnard by-catch

61 Table 23 Catch comparison The Netherlands Experiment 1 Gear STD 9m Day/night Day Nation NL Gear EXP sieve 60 Time Nov 1999 Experiment 1 Fish. ground Wadden Sea Tot hauls 8 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % 0.75 fish (landed) % % fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS >MLS dab < MLS % % >MLS % % flounder < MLS % % >MLS % % turbot < MLS >MLS brill < MLS >MLS whiting < MLS % % >MLS % % cod < MLS % % >MLS % %. smelt % % bib % %. gurnard by-catch % % 0.256

62 Table 24 Catch comparison The Netherlands Experiment 2 Gear STD 9m Day/night Day Nation NL Gear EXP sieve 60 Time Nov 1999 Experiment 2 Fish. ground NL Coast Tot hauls 4 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS >MLS dab < MLS % % >MLS % % flounder < MLS % % >MLS % % turbot < MLS % % >MLS % %. brill < MLS >MLS whiting < MLS % % >MLS % % cod < MLS % % >MLS % % smelt bib % % gurnard by-catch % % 0.245

63 Table 25 Catch comparison The Netherlands Experiment 3 Gear STD 9m Day/night Day Nation NL Gear EXP grid 15 Time Nov 1999 Experiment 3 Fish. ground Wadden Sea Tot hauls 12 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % 0.08 shrimps (discarded) % % fish (landed) % % fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS >MLS dab < MLS % % >MLS % % flounder < MLS % % >MLS % % turbot < MLS >MLS brill < MLS >MLS whiting < MLS % % >MLS % % cod < MLS % >MLS % %. smelt % % bib gurnard by-catch % % 0.832

64 Table 26 Catch comparison The Netherlands Experiment 4 Gear STD 9m Day/night Day Nation NL Gear EXP grid 15 Time Nov 1999 Experiment 4 Fish. ground NL coast Tot hauls 10 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % 0 fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS >MLS dab < MLS % % >MLS % % flounder < MLS % % >MLS % % 0 turbot < MLS % % >MLS % %. brill < MLS % >MLS whiting < MLS % % >MLS % % 0.05 cod < MLS % % >MLS % % smelt % bib % % gurnard by-catch % % 0.318

65 Table 27 Catch comparison The Netherlands Experiment 5 Gear STD 9m Day/night Day Nation NL Gear EXP grid 20 Time Nov 1999 Experiment 5 Fish. ground Wadden Sea Tot hauls 14 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % 0 fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS >MLS dab < MLS % % >MLS % % flounder < MLS % % >MLS % % turbot < MLS >MLS brill < MLS >MLS whiting < MLS % % >MLS % % 0 cod < MLS % % >MLS % % smelt % % bib gurnard by-catch % % 0.984

66 Table 28 Catch comparison The Netherlands Experiment 6 Gear STD 9m (60) Day/night day Nation NL Gear EXP sieve 70 Time Nov 2000 Experiment 6 Fish. ground E. Scheldt Tot hauls 5 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS % % 0.47 >MLS dab < MLS % % >MLS % % flounder < MLS >MLS % % turbot < MLS >MLS brill < MLS >MLS whiting < MLS % % >MLS % % cod < MLS % % >MLS % % smelt bib % % gurnard by-catch % % 0.098

67 Table 29 Catch comparison The Netherlands Experiment 7 Gear STD 9m Day/night Day Nation NL Gear EXP sieve 60 Time Nov 2000 Experiment 7 Fish. ground E. Scheldt Tot hauls 19 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % 0 fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS % % sole < MLS % % >MLS % %. dab < MLS % % >MLS % % flounder < MLS % >MLS % % turbot < MLS >MLS brill < MLS % % >MLS whiting < MLS % % 0 >MLS % % cod < MLS % % 0.03 >MLS % % 0 smelt bib % % gurnard by-catch % % 0.431

68 Table 30 Catch comparison The Netherlands Experiment 8 Gear STD 9m Day/night Day Nation NL Gear EXP sieve 60 Time Nov 2000 Experiment 8 Fish. ground NL coast Tot hauls 24 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % 0 fish (discarded) % % 0 other (e.g. benthos) Commercial species: plaice < MLS % % >MLS sole < MLS % % >MLS dab < MLS % % >MLS % % 0 flounder < MLS % % >MLS % % turbot < MLS % % >MLS % % brill < MLS % % >MLS whiting < MLS % % 0 >MLS % % 0 cod < MLS % % >MLS % % 0 smelt % bib % % 0 gurnard by-catch % % 0.578

69 Table 31 Catch comparison The Netherlands Experiment 9 Gear STD 9m Day/night Day Nation NL Gear EXP 9m stn 60 Time Nov 2000 Experiment 9 Fish. ground NL coast Tot hauls 21 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS % sole < MLS % % >MLS dab < MLS % % >MLS % % flounder < MLS % % 0.99 >MLS % % turbot < MLS % % >MLS % % brill < MLS % >MLS whiting < MLS % % >MLS % % cod < MLS % % >MLS % % smelt % bib % % gurnard by-catch % % 0.048

70 Table 32 Catch comparison The Netherlands Experiment 10 Gear STD 9m Day/night Day Nation NL Gear EXP sieve 70 Time Nov 2000 Experiment 10 Fish. ground NL coast Tot hauls 33 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % 0 fish (discarded) % % 0 other (e.g. benthos) Commercial species: plaice < MLS % % 0 >MLS % % sole < MLS % % >MLS dab < MLS % % 0 >MLS % % 0 flounder < MLS % % >MLS % % 0 turbot < MLS % % 0 >MLS % % 0 brill < MLS % % >MLS whiting < MLS % % 0 >MLS % % 0 cod < MLS % % 0 >MLS % % 0 smelt % bib % % 0 gurnard by-catch % % 0

71 Table 33 Catch comparison The Netherlands Experiment 11 Gear STD 9m Day/night Day Nation NL Gear EXP 9m (70) Time Nov 2000 Experiment 11 Fish. ground NL coast Tot hauls 27 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD p STD EXP EXP/STD p shrimps (landed) % % shrimps (discarded) % % fish (landed) % % fish (discarded) % % other (e.g. benthos) Commercial species: plaice < MLS % % >MLS % % sole < MLS % % >MLS % dab < MLS % % >MLS % % 0.31 flounder < MLS % % >MLS % % 0.19 turbot < MLS % % >MLS % % brill < MLS % % >MLS whiting < MLS % % >MLS % % cod < MLS % % >MLS % % smelt % bib % % gurnard % by-catch % % 0.306

72 Table 34 Catch comparison Belgium Experiment 1 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment 1-11/08/99 Fishing ground Flemish banks Total number of hauls 2 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % --- shrimps (discarded) % --- fish (landed) fish (discarded) % --- other (e.g. benthos) % --- Commercial species: plaice < MLS % --- >MLS sole < MLS % --- >MLS dab <MLS % --- >MLS flounder % --- turbot whiting < MLS % --- >MLS cod < MLS >MLS bib <MLS % --- >MLS

73 Table 35 Catch comparison Belgium Experiment 2 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment /09/99 Fishing ground Flemish banks Total number of hauls 6 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.02 shrimps (discarded) % 0.27 fish (landed) % 0.12 fish (discarded) % 0.49 other (e.g. benthos) % 0.02 Commercial species: plaice < MLS % 0.34 >MLS % 0.36 sole < MLS % 0.25 >MLS % --- dab <MLS % 0.18 >MLS % 0.09 flounder turbot % --- whiting < MLS % 0.02 >MLS cod < MLS >MLS bib <MLS % 0.09 >MLS

74 Table 36 Catch comparison Belgium Experiment 3 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment 3-08/09/99 Fishing ground Flemish banks Total number of hauls 3 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.01 shrimps (discarded) % 0.26 fish (landed) % 0.02 fish (discarded) % 0.39 other (e.g. benthos) % 0.00 Commercial species: plaice < MLS % 0.21 >MLS % 0.31 sole < MLS % 0.13 >MLS % 0.48 dab <MLS % 0.06 >MLS flounder turbot whiting < MLS % 0.02 >MLS cod < MLS >MLS % --- bib <MLS % 0.25 >MLS

75 Table 37 Catch comparison Belgium Experiment 4 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment 4-09/09/99 Fishing ground Flemish banks Total number of hauls 2 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.08 shrimps (discarded) % 0.20 fish (landed) % 0.10 fish (discarded) % 0.28 other (e.g. benthos) % 0.16 Commercial species: plaice < MLS % 0.08 >MLS % 0.00 sole < MLS % 0.10 >MLS % 0.00 dab <MLS % 0.12 >MLS % 0.19 flounder turbot whiting < MLS % 0.14 >MLS cod < MLS >MLS bib <MLS % 0.10 >MLS

76 Table 38 Catch comparison Belgium Experiment 5 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment 5-15/09/99 Fishing ground Flemish banks Total number of hauls 2 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % --- shrimps (discarded) % --- fish (landed) % --- fish (discarded) % --- other (e.g. benthos) % --- Commercial species: plaice < MLS % --- >MLS % --- sole < MLS % --- >MLS % --- dab <MLS % --- >MLS flounder turbot % --- whiting < MLS % --- >MLS cod < MLS >MLS bib <MLS % --- >MLS

77 Table 39 Catch comparison Belgium Experiment 6 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Autumn Experiment 6-13/10/99 Fishing ground Flemish banks Total number of hauls 3 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.15 fish (landed) % 0.07 fish (discarded) % 0.20 other (e.g. benthos) % 0.18 Commercial species: plaice < MLS % 0.05 >MLS % --- sole < MLS % 0.18 >MLS % --- dab <MLS % 0.13 >MLS % 0.18 flounder turbot whiting < MLS % 0.01 >MLS % 0.24 cod < MLS % --- >MLS % --- bib <MLS % 0.18 >MLS

78 Table 40 Catch comparison Belgium Experiment 7 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Autumn Experiment 7-28/10/99 Fishing ground Flemish banks Total number of hauls 3 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.04 shrimps (discarded) % 0.02 fish (landed) % 0.16 fish (discarded) % 0.06 other (e.g. benthos) % 0.15 Commercial species: plaice < MLS % 0.35 >MLS % --- sole < MLS % 0.35 >MLS dab <MLS % 0.47 >MLS % 0.37 flounder turbot whiting < MLS % 0.07 >MLS % 0.12 cod < MLS % --- >MLS % --- bib <MLS % 0.22 >MLS

79 Table 41 Catch comparison Belgium Experiment 8 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment 8-03/11/99 Fishing ground Flemish banks Total number of hauls 3 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.02 shrimps (discarded) % 0.00 fish (landed) % 0.06 fish (discarded) % 0.09 other (e.g. benthos) % 0.09 Commercial species: plaice < MLS % 0.05 >MLS % --- sole < MLS % 0.26 >MLS % 0.25 dab <MLS % 0.28 >MLS % 0.17 flounder turbot whiting < MLS % 0.04 >MLS % 0.09 cod < MLS % --- >MLS % --- bib <MLS % 0.09 >MLS

80 Table 42 Catch comparison Belgium Experiment 9 Table Gear STD / Day/night fishing Day Nation Belgium Gear EXP / Time of the year Winter Experiment 9-25/11/99 Belgica Fishing ground Vlakte vd Raan Total number of hauls 4 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.08 shrimps (discarded) % 0.03 fish (landed) % 0.01 fish (discarded) % 0.00 other (e.g. benthos) % 0.01 Commercial species: plaice < MLS % 0.05 >MLS sole < MLS % 0.20 >MLS dab <MLS % 0.13 >MLS % 0.03 flounder turbot whiting < MLS % 0.03 >MLS % 0.03 cod < MLS >MLS % bib <MLS % 0.07 >MLS

81 Table 43 Catch comparison Belgium Experiment 10 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Winter Experiment /01/00 Fishing ground Flemish banks Total number of hauls 7 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.01 fish (landed) % 0.01 fish (discarded) % 0.01 other (e.g. benthos) % 0.02 Commercial species: plaice < MLS % 0.09 >MLS % --- sole < MLS % 0.17 >MLS dab <MLS % 0.03 >MLS % 0.03 flounder % 0.00 turbot whiting < MLS % 0.02 >MLS % 0.09 cod < MLS % 0.08 >MLS % --- bib <MLS % 0.07 >MLS

82 Table 44 Catch comparison Belgium Experiment 11 Table Gear STD / Day/night fishing Day Nation Belgium Gear EXP / Time of the year Winter Experiment Belgica Feb-Mar 2000 Fishing ground Vlakte vd Raan Total number of hauls 8 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.00 fish (landed) % 0.00 fish (discarded) % 0.00 other (e.g. benthos) % 0.00 Commercial species: plaice < MLS % 0.01 >MLS % 0.00 sole < MLS % 0.00 >MLS % --- dab <MLS % 0.03 >MLS % 0.02 flounder turbot whiting < MLS % 0.02 >MLS % 0.05 cod < MLS % --- >MLS % --- bib <MLS % 0.01 >MLS

83 Table 45 Catch comparison Belgium Experiment 12 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Spring Experiment Spring 2000 Fishing ground Flemish banks Total number of hauls 9 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.06 fish (landed) % 0.00 fish (discarded) % 0.00 other (e.g. benthos) % 0.02 Commercial species: plaice < MLS % 0.00 >MLS % 0.00 sole < MLS % 0.01 >MLS % 0.04 dab <MLS % 0.03 >MLS % 0.01 flounder % 0.07 turbot whiting < MLS % 0.00 >MLS % 0.01 cod < MLS % 0.02 >MLS % 0.01 bib <MLS % 0.03 >MLS % ---

84 Table 46 Catch comparison Belgium Experiment 13 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Summer Experiment Summer 2000 Fishing ground Flemish banks Total number of hauls 8 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.12 shrimps (discarded) % 0.02 fish (landed) % 0.00 fish (discarded) % 0.06 other (e.g. benthos) % 0.07 Commercial species: plaice < MLS % 0.06 >MLS sole < MLS % 0.00 >MLS dab <MLS % 0.27 >MLS flounder % 0.05 turbot whiting < MLS % 0.02 >MLS cod < MLS >MLS bib <MLS % 0.00 >MLS

85 Table 47 Catch comparison Belgium Experiment 14 Table Gear STD Port side Day/night fishing Night Nation Belgium Gear EXP Starboard side Time of the year Aut & win Experiment Aut & win 2000 Fishing ground Flemish banks Total number of hauls 24 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.00 fish (landed) % 0.01 fish (discarded) % 0.00 other (e.g. benthos) % 0.00 Commercial species: plaice < MLS % 0.00 >MLS % 0.00 sole < MLS % 0.00 >MLS % 0.03 dab <MLS % 0.01 >MLS % 0.00 flounder % 0.00 turbot % --- whiting < MLS % 0.36 >MLS % 0.00 cod < MLS % 0.02 >MLS % 0.00 bib <MLS % 0.00 >MLS % ---

86 Table 48 Catch comparison Belgium Experiment 15 Table Gear STD / Day/night fishing Day Nation Belgium Gear EXP / Time of the year Autumn Experiment Belgica Nov 2000 (a) Fishing ground Vlakte vd Raan Total number of hauls 12 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.00 fish (landed) % 0.00 fish (discarded) % 0.00 other (e.g. benthos) % 0.00 Commercial species: plaice < MLS % 0.00 >MLS % --- sole < MLS % 0.01 >MLS dab <MLS % 0.00 >MLS % 0.00 flounder turbot whiting < MLS % 0.00 >MLS % 0.00 cod < MLS % --- >MLS % --- bib <MLS % 0.00 >MLS

87 Table 49 Catch comparison Belgium Experiment 16 Table Gear STD / Day/night fishing Day Nation Belgium Gear EXP / Time of the year Autumn Experiment Belgica Nov 2000 (b) Fishing ground Vlakte vd Raan Total number of hauls 6 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % 0.00 shrimps (discarded) % 0.01 fish (landed) % 0.02 fish (discarded) % 0.01 other (e.g. benthos) % 0.00 Commercial species: plaice < MLS % 0.00 >MLS % --- sole < MLS % 0.05 >MLS dab <MLS % 0.00 >MLS % 0.02 flounder turbot whiting < MLS % 0.03 >MLS % 0.07 cod < MLS % --- >MLS % --- bib <MLS % 0.04 >MLS % ---

88 Table 50 Catch comparison United Kingdom Experiment 1 Gear STD Standard Day/night fishing Day Nation UK Gear EXP Sieve UK 1 Time of the year Oct 1999 Experiment 1 Fishing ground The Wash Total number of hauls 20 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % shrimps (discarded) % fish (landed) fish (discarded) & benthos % Commercial species: plaice < MLS % >MLS sole < MLS % >MLS dab < MLS % >MLS whiting < MLS % >MLS cod < MLS % >MLS flounder Turbot/ brill / smelt/ bib / gurnard

89 Table 51 Catch comparison United Kingdom Experiment 2 Gear STD Standard Day/night fishing Day Nation UK Gear EXP Sieve UK 2 Time of the year Nov 1999 Experiment 2 Fishing ground The Wash Total number of hauls 21 Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % shrimps (discarded) % fish (landed) fish (discarded) & benthos % Commercial species: plaice < MLS % >MLS sole < MLS % >MLS % dab < MLS % >MLS whiting < MLS % >MLS % cod < MLS % >MLS flounder % Turbot/ brill / smelt/ bib / gurnard

90 Table 52 Catch comparison United Kingdom Experiment 3 Gear STD Standard Day/night fishing Day Nation UK Gear EXP Sieve UK 3 Time of the year Feb 2000 Experiment 3 Fishing ground The Wash Total number of hauls 22 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % shrimps (discarded) % fish (landed) fish (discarded) & benthos % Commercial species: plaice < MLS % >MLS sole < MLS % >MLS dab < MLS % >MLS % Insufficient data whiting < MLS % >MLS % cod < MLS % >MLS flounder % Turbot/ brill / smelt/ bib / gurnard

91 Table 53 Catch comparison United Kingdom Experiment 4 Gear STD Standard Day/night fishing Day Nation UK Gear EXP Sieve UK 4 Time of the year Feb 2000 Experiment 4 Fishing ground The Wash Total number of hauls 20 Species Number per m2 Weight in kg per m2 STD EXP EXP/STD in % p STD EXP EXP/STD in % p shrimps (landed) % shrimps (discarded) % fish (landed) fish (discarded) & benthos % Commercial species: plaice < MLS % >MLS sole < MLS % >MLS dab < MLS % >MLS % Insufficient data whiting < MLS % >MLS % cod < MLS % >MLS flounder % Turbot/ brill / smelt/ bib / gurnard

92 Table 54 Experimental gear trials Percentage reduction in the catch fractions in the UK sea trials Reduction in the retention of landed shrimp (i.e. the principal target species) Reduction in the retention of shrimp of discard size Reduction in the retention of benthos & non - marketable / sublegal commercial discarded fish species Reduction in the retention of marketable fish (NB: reductions as a percentage of that fraction caught in the standard gear) Sieve UK 1 8% 26% 89% None caught Sieve UK 2 21% 29% 81% None caught Sieve UK 3 7% 16% 58% None caught Sieve UK 4 4% 22% 56% None caught NB: * The standard gear used for comparison in these trials was a standard net fitted with the sieve net UK No3 design with a cod end mesh size of 20 mm. Greyed areas indicate unwanted portions of the catch (normally discarded). Fish of marketable size are not normally found in the UK fishery

93 Figure 1: European (N. Sea) C.crangon fishing grounds European (N.Sea) Crangon fishing grounds

94 Figure 2: The veil (sieve net), seen from the front Figure 3: Seive net, view from above

95 Figure 4: Rectangular grid, tried in B, NL and UK, side view Figure 5: Rectangular grid seen from above

96 Figure 6: Elliptical grid tried in Germany, side view Figure 7: Elliptical grid tried in Germany, view from above