Action

Fit a size-sorting escape grid (rigid or flexible) to a prawn/shrimp trawl net

How is the evidence assessed?
  • Effectiveness
    not assessed
  • Certainty
    not assessed
  • Harms
    not assessed

Source countries

Key messages

COMMUNITY RESPONSE (0 STUDIES)

POPULATION RESPONSE (0 STUDIES)

BEHAVIOUR (0 STUDIES)

OTHER (30 STUDIES)

  • Reduction of unwanted catch (30 studies): Seven of seven replicated studies (including one controlled) in the northeast Atlantic Ocean, North Sea, North Atlantic Ocean, Greenland Sea, Gulf of Thailand, Tyrrhenian Sea and the Skagerrak and Kattegat found that fitting rigid or flexible size-sorting escape grids, of various types and configurations, to prawn/shrimp trawl nets reduced unwanted fish catches (non-commercial species and discarded commercial species/sizes) by allowing the escape of unwanted sharks and the other fish species monitored. Two of two before-and-after studies in the Gulf of Maine and Pacific Ocean found that after the introduction of size-sorting escape grids to trawl nets in fisheries for shrimp, the capture of non-target and unwanted fish was reduced compared to before grids were used. Eleven of 20 replicated studies (including one controlled and 19 paired and controlled) in the Tasman Sea, Coral Sea, Gulf of Carpentaria, North Sea, Indian Ocean, Bay of Biscay, Skagerrak and Kattegat, Pacific Ocean, South Atlantic Ocean, Gulf of St Vincent and Persian Gulf found that prawn/shrimp trawls with size-sorting escape grids, of various types and configurations, had lower catches of all or all but one undersized or otherwise unwanted fish and shark/ray species monitored, and unwanted total catch (fish and invertebrates), compared to trawl nets without escape grids. Two found that escape grids reduced non-target catches of most sizes of whiting and plaice and larger sizes of total fish, but increased the retention of small cod and haddock. Three studies found a variable effect of fitting escape grids to shrimp/prawn trawl nets on unwanted fish catch compared to nets with no grids, and the effect varied with year, site and grid type. Three found that grids had no effect on the reduction of unwanted fish and catches were similar for all or most of the unwanted non-commercial and commercial fish species/groups and for the total unwanted catch (fish and invertebrates). The other study found that fewer unwanted fish of 10 of 11 species/groups were retained in a shrimp/prawn trawl net with an escape grid used in combination with a diamond mesh codend with the mesh orientation turned by 90°, compared to a conventional diamond mesh net with no grid. One replicated, randomized study in the North Atlantic Ocean found that the reduction in catch of unwanted sharks depended on the type of escape grid and shrimp/prawn trawl net used.

About key messages

Key messages provide a descriptive index to studies we have found that test this intervention.

Studies are not directly comparable or of equal value. When making decisions based on this evidence, you should consider factors such as study size, study design, reported metrics and relevance of the study to your situation, rather than simply counting the number of studies that support a particular interpretation.

Supporting evidence from individual studies

  1. A replicated study in 1989–1990 of shrimp fishing grounds in the Northeast Atlantic, Norway (Isaksen et al. 1992) reported that shrimp trawl nets fitted with rigid size-sorting escape grids allowed small unwanted fish and Greenland shark Somniosus microcephalus to escape capture and escape frequency of the small fish increased with fish length. Data were not statistically tested. Trawl nets fitted with an escape grid released more Atlantic cod Gadus morhua and haddock Melanogrammus aeglefinus than were retained in the codend and all individuals larger than 20 cm (data were presented as length frequency distributions). For redfish Sebastes spp. higher proportions of fish >13 cm escaped than were retained and all redfish >18 cm. For polar cod Boreogadus saida higher proportions >14 cm escaped and all were released from 22 cm. All flatfish (including Greenland halibut Reinhardtius hippoglossoides and long rough dab Hippoglossoides platessoides) >30–32 cm escaped capture, although this larger size compared to other species was due to the species’ swimming behaviour. It was also reported that with a grid of at least 1.0 × 1.5 m, Greenland shark up to 4 m escaped relatively easily. In addition, target shrimp Pandalus borealis escapes were 2–5%. In 1989–1990, experimental deployments (location and number not reported) were undertaken using trawl nets fitted with an aluminium (Nordmøre) grid mounted at a 48° angle and top escape outlet. Covers mounted over the outlet collected the escaping fish. Two grid designs were used: a coastal (1.35 × 0.7–0.8 m) and an offshore (1.5 × 1.0–1.3 m) grid for smaller and larger vessels respectively. A remote-controlled underwater vehicle monitored fish behaviour. Data from the two grids were combined. Details of study location were not reported.

    Study and other actions tested
  2. A replicated, paired, controlled study in 1991 of a seabed area fished commercially for prawns in the Tasman Sea, New South Wales, Australia (Andrew et al. 1993) found that prawn nets modified with a flexible (mesh) size-sorting escape grid (a Morrison soft turtle excluder device) reduced the capture of unwanted finfish compared to unmodified standard nets. Discards of undersized commercial finfish species were significantly reduced in modified nets, however, retained finfish catch was similar (data reported as difference in average log ratio of catch weights). Use of the flexible escape grid did not reduce catches of the target prawn species Penaeus plebejus (data reported as difference in average log ratio of catch weights). The weight of discarded finfish and invertebrates combined was an average of 32% (9kg/tow) lower in modified nets than unmodified (not statistically tested). In October 1991, fishing experiments were done on two prawn trawlers, each fitted with three trawl nets in a standard triple gear configuration. Four paired 90-minute deployments using the outer trawl nets only were carried out on each of six consecutive nights. One of the outer nets was modified with a large-mesh (197 mm) escape panel/grid, measuring 36.5 meshes across at the leading edge, installed on the inside of the net. An opening of 20 meshes was cut in the net at the end of the panel immediately in front of the codend to allow larger catch to escape. The other outer trawl net of the three was not modified. Codend catches were separated into retained (prawns and other important species of commercial size) and discarded (rest of catch including undersized individuals of commercial species) portions and weight and lengths recorded.

    Study and other actions tested
  3. A replicated, paired, controlled study in 1991–1992 at one coastal seabed and one estuarine site in the Coral Sea, Australia (Robins-Troeger 1994) found that using a flexible (mesh) size-sorting escape grid (a Morrison soft turtle excluder device) inside a prawn trawl net resulted in variable reductions of unwanted non-commercial catch (fish and crustaceans) compared to nets without a grid. The reductions in weight (21–32%) of unwanted catch (fish and crustaceans) in nets with a flexible escape grid were significant at both sites in 1991, but not in 1992 (2–18%). In addition, larger unwanted non-commercial catch including stingrays Amphotistius kuhlii and shovelnose rays Rhinobatus batillum were not caught in nets with a flexible escape grid, but they were occasionally caught in nets without an escape grid (numbers not reported). Prawn catch (Metapenaeus bennettae, Penaeus plebejus, Penaeus esculentus) was significantly reduced (17–30%) in nets with a grid at one site in 1991 and both sites in 1992. Paired fishing deployments were done on a research vessel towing two prawn trawl nets at two sites in Moreton Bay off Queensland in May 1991 (low prawn catches) and January 1992 (high prawn catches). A flexible grid made of 150 mm monofilament mesh was installed inside one trawl net and the other net was unmodified. In each year, 17 to 23 deployments of 45–100 min were completed at each site. Codend catches were landed and processed separately.

    Study and other actions tested
  4. A replicated, paired, controlled study (year not given) of soft seabed at five coastal sites in the Coral Sea, Australia (Robins-Troeger et al. 1995), found that when a catch escape system, incorporating a flexible size-sorting escape grid (‘AusTED bycatch reduction device’) was fitted to a prawn trawl net, catch of unwanted fish was reduced at three of five sites, compared to a standard trawl net without a grid system. At three sites, fewer (33–59% weight) unwanted fish were caught in nets with grid systems compared to standard nets, whereas at two sites the reduction in weight (11–12%) in nets with grids was not significantly lower. The size (and quality) of the commercial catch of prawns Penaeidae was similar between nets (grids: 91–103% of standard net catch). Trials were done using twin prawn nets, one with a grid system and one without, towed simultaneously by a 15 m trawler. In one trawl net, a system incorporating a flexible inclined grid with an escape gap at the top, a large mesh panel and a guiding funnel (AusTED) was installed in front of the codend. Between 13–27 deployments of 60 min were conducted at each of five sites of southeast Queensland. The year(s) the study took place was not reported.

    Study and other actions tested
  5. A replicated, paired, controlled study in 1995–1996 of prawn fishing grounds in the Gulf of Carpentaria, Australia (Brewer et al. 1998) found that commercial prawn trawls fitted with size-sorting escape grids (rigid and flexible) typically caught fewer unwanted fish, sharks and stingrays compared to unmodified conventional trawls and the effect varied with grid type. Shark and stingray data were not tested statistically. In two of two experimental trials, two grid types (Nordmøre and AusTED) caught 27–35% less unwanted fish catch than unmodified trawls, but a third grid type (Super Shooter), while reducing unwanted fish catch by 17–21%, was not statistically different. In commercial trials, trawls with grids (Super Shooter and NAFTED) caught fewer sharks (with grids: 3–6, unmodified: 4–16), and stingrays (with grids: 0, unmodified: 0–15). Catches of the target prawn species were similar for two of the grids (Super Shooter and NAFTED) compared to unmodified trawls but were lower for the Nordmøre (50% less) and AusTED (22% less) grids. Fishing trials were carried out in the Gulf of Carpentaria, Australia, in February and October 1995 (experimental) and October 1996 (commercial). Standard prawn trawl nets (45 mm codend mesh) were fitted with one of four grid systems and replicate paired deployments with unmodified trawl nets were conducted by a dual-rigged trawler. Grids had either a top or bottom escape opening, and all had a guiding panel (full details of the grid system designs are provided in the original study).

    Study and other actions tested
  6. A replicated, paired, controlled study in 1994 of three areas of sand and mud seabed in the Coral Sea off Queensland, Australia (Robins & McGilvray 1999) found that prawn trawl nets fitted with a flexible size-sorting escape grid caught less unwanted fish catch and overall unwanted catch (fish and invertebrates combined) than conventional trawl nets with no grid. In one of one comparison in which fish catch was separated from invertebrate catch, nets with a grid caught 15% less fish (2.3 kg/tow) than conventional nets (2.7 kg/tow). In four of four comparisons, grid nets caught 15–49% less unwanted catch (14–50 kg/tow) than conventional nets (17–62 kg/tow). In addition, one shark (species not given) was caught in grid nets compared to three rays Rhinobatos spp. in conventional nets (results not tested for statistical significance). Target prawn Penaeidae catches were reduced (9–36% lower) in modified nets in two of four comparisons. In 1994, experimental trawl deployments were undertaken at three sites in the Queensland east coast fishery. One codend fitted with an AusTED II (a modification of the original AusTED design) escape grid and one conventional codend were towed simultaneously for 60–180 min. Codend catches were kept separate and the target and non-target portions were sorted and weighed for each net after every tow.

    Study and other actions tested
  7. A replicated, paired, controlled study of a seabed area in the North Sea, between Scotland and Norway (Madsen & Hansen 2001) found that fitting a flexible size-sorting escape grid in a shrimp trawl net typically reduced the catches of undersized cod Gadus morhua, haddock Melanogrammus aeglefinus and whiting Merlangius merlangus and the overall numbers of Norway pout Trisopterus esmarki and saithe Pollachius virens but not monkfish Lophius piscatorius, compared to trawl nets without an escape grid. In two of two trials, total catch numbers of undersized fish were lower in nets with grids for cod (with: 365–382, without: 1,237–1,805) and haddock (with: 253–579, without: 4,617–5,315). Numbers of undersized whiting were lower in one (with: 15, without: 123) of two (with: 9, without: 11) trials. Overall catch numbers were reduced by 40–55% for Norway pout (two of two trials) and 73% for saithe (one of two trials). Monkfish catch numbers were similar between nets for two of two trials (with: 40–67, without: 43–66). Trials were conducted in the Fladen Ground shrimp fishery onboard a commercial twin-trawler in April–May (20 hauls) and August (30 hauls). A codend fitted with an escape grid was deployed on one side of the gear simultaneously with a standard codend on the other side. A collecting bag caught fish escaping through the grid. The flexible (polyamide) grid was 12 mm diameter bars of 19 mm spacing, fitted at 48° angle, with a fish escape hole at the top and a Norway lobster Nephrops norvegicus escape hole at the bottom (see paper for grid specifications). A panel of square mesh netting to retain marketable sizes of roundfish was fitted to the top of the net behind the grid. Fish from the codend and collecting bag were sorted separately into species and lengths recorded. The study year was not reported.

    Study and other actions tested
  8. A replicated, paired, controlled study in 2000 of an area of sand/mud seabed in the Indian Ocean, off western Australia (Broadhurst et al. 2002) found that a prawn trawl net fitted with a rigid size-sorting escape grid did not reduce the catches of seven of seven non-commercial and commercial fish species, or the amount of overall unwanted catch (fish and invertebrates), compared to an unmodified standard trawl net. Average catch numbers of six of six non-commercial and one of one commercial fish species (see paper for individual species data) were similar in nets with a grid to nets without a grid (with: 10–284 fish/tow, without: 5–266 fish/tow) and species weights were also similar (data reported as statistical model results). Average weight of all non-target catch was similar between nets (with: 55 kg/tow, without: 58 kg/tow). In addition, the average weight of all target prawn species Penaeidae was reduced (with: 11 kg/tow, without: 13 kg/tow). In August 2000, ten 40-min paired trawl deployments were done on established prawn fishing grounds in Shark Bay. A trawl codend end fitted with an aluminium Nordmøre grid (100 mm bar spacing, 45° angle, top escape opening) was towed simultaneously with a standard codend. All codend catches were sorted, counted and weighed and the species caught in sufficient numbers analysed.

    Study and other actions tested
  9. A replicated study in 1996–1997 of bottom fishing grounds in the North Sea off Belgium and the Netherlands (Polet 2002) found that shrimp trawl nets fitted with a rigid size-sorting escape grid (a Nordmøre grid) reduced the overall amount of fish catch and allowed the majority of undersized fish to escape capture. Across all trials, overall catch numbers of fish species were reduced by 72–75% in nets with a grid compared to without. For four of four species (plaice Pleuronectes platessa, sole Solea solea, whiting Merlangius merlangus and cod Gadus morhua) high percentages of the fish escaping through the grid opening were below the minimum landing sizes (data presented as length frequency distributions and selection curves). In addition, catches of undersized and marketable brown shrimp Crangon crangon were reduced by 17–45% and 15% respectively. Trawl deployments were made on a research vessel in November 1996 (24 tows) and during three trips on a commercial trawler in July and September 1997 (10 tows). An 8 m commercial shrimp beam trawl net was fitted with an 80 × 60 cm Nordmøre grid and top escape opening. The grid had 12 mm bar spacing in November 1996 and 14 mm spacing subsequently. Covers retained any catch passing through the codend and grid escape opening. All catch (codend and covers) was sorted, counted and lengths recorded. Full trawl details are given in the original study.

    Study and other actions tested
  10. A replicated, paired, controlled study in 1993–1995 of two sandy/mud fishing grounds in the North Sea, UK (Graham 2003) found that fitting rigid size-sorting escape grids inside shrimp trawl nets reduced the catch of unwanted whiting Merlangius merlangus and plaice Pleuronectes platessa, compared to nets without grids. For three of three designs of grid, total catch numbers of whiting (with: 263–839, without: 586–4,923) and plaice (with: 850–3,074, without 1,304–5,504), were lower in nets with a grid than without, representing 55–85% reductions in catch for whiting and 35–44% for plaice. In addition, for one of one grid designs, the selection length (the length at which fish have a 50% chance of escape) increased with increasing grid bar spacing for both whiting (10 mm: 7.7 cm, 12 mm: 10.5 cm, 14 mm: 11.8 cm) and plaice (12 mm: 9.2 cm, 14 mm: 10.4 cm). Total overall losses (8–10%) of target brown shrimp Crangon crangon in nets with grids were not statistically different to nets without grids. Three fishing trials, each testing a different design of grid, were done in the Humber Estuary and off the Lincolnshire coast in November–December and February–March from 1993–1995. Paired trawl deployments (12–22 tows for each grid) were done on commercial shrimp vessels towing a trawl net fitted with a grid and a standard trawl net simultaneously for 0.5–2 h. Grids were made from steel or plastics with 12 mm bar spacings and had top opening escape holes (see paper for full specifications of trawl nets and grids). Two further plastic grids with 10- and 14-mm bar spacings were also tested using a small mesh cover over the grid escape opening to collect the escaping fish.

    Study and other actions tested
  11. A replicated study in 2001–2002 of a seabed area in the Atlantic Ocean, off southern Portugal (Fonseca et al. 2005) found that prawn/shrimp trawls fitted with a rigid size-sorting escape grid allowed most unwanted blue whiting Micromesistius poutassou and boarfish Capros aper to escape capture. Data were not tested for statistical significance. Overall, a total of 73–74% in number of blue whiting catch and 47–63% of the catch in number of boarfish, escaped through the grids. Losses of target shrimp species (rose shrimp Parapenaeus longirostris, Norway lobster Nephrops norvegicus and red shrimp Aristeus antennatus) were 4–15%. In May 2001 and April-May 2002 a total of 41 and 15 trawl deployments respectively were carried out on shrimp and Norway lobster fishing grounds off the Algarve. Trawl nets were fitted with a steel grid system (a modified Nordmøre grid) with 25 mm bar spacing, a guiding funnel, top escape opening and a 20 cm high section without bars at the bottom to allow target lobster catch to enter the codend directly (see paper for specifications). The escaping fish were collected in a mesh cover fitted over the escape opening.

    Study and other actions tested
  12. A replicated, controlled study in 2002 of two bottom fishing grounds in the Greenland Sea, off western Svalbard, Norway (Grimaldo & Larsen 2005) found that shrimp trawl nets fitted with size-sorting escape grids (two designs) allowed high quantities of Greenland halibut Reinhardtius hippoglossoides, cod Gadus morhua, haddock Melanogrammus aeglefinus and saithe Pollachius virens to escape capture, but the amounts and sizes of fish released did not differ between grid designs. Average overall reductions in catch weight for four of four species were similar between grid designs (new: 35–184 kg, old: 33–121 kg) representing releases of 81–99% and 69–99% of the fish captured for each grid respectively. The selection length (the length at which fish have a 50% chance of escape) did not differ between grid designs for all four species (new: 14.3–20.2 cm, old: 16.3–19.5 cm). In addition, average catch losses of target Nordic shrimp Pandalus borealis were low (4–5%) for both grids. Trials took place in the areas of Ice Fjord and Minke Bank in December 2002 in depths of 240–415 m. Deployments of trawl nets fitted with either a new design of grid (24 tows) made of lighter artificial materials (Cosmos) or an established grid (10 tows) of high-density material (HDPE) were made (see paper for grid specifications). Each grid had 19 mm bar spacings. A top cover collected escaping fish and shrimp. Cover and codend catches of fish and shrimp were weighed and fish lengths recorded.

    Study and other actions tested
  13. A replicated, paired, controlled study in 2001 of bottom fishing grounds in the Gulf of Carpentaria, Australia (Brewer et al. 2006) found that prawn trawl nets fitted with rigid or semi-rigid size-sorting escape grids reduced the amount of unwanted shark Selachii and ray Batoidea catch, but not sawfish Pristidae, compared to conventional trawl nets. Across all trawl nets fitted with escape grids, shark and ray catches were reduced by 13% and 31% respectively. Trawl nets fitted with upward-angled grids and top escape openings reduced unwanted shark catch by 20% and rays by 27%, while downward-angled grids and bottom escape openings reduced shark and ray catches by 9% and 35% respectively. No grid system reduced catches of sawfish. Total prawn Penaeidae sp. catch was reduced by 6% with grid-modified trawls, except trawls with an upward-angled grid, which caught similar prawn numbers to conventional trawls. Data were collected from up to 1,612 paired trawl comparisons (3,224 nets sampled over 442 nights of trawling) from 23 different vessels in August-November 2001, in which a wide range of catch reduction devices were tested. Nets with escape grids and nets without a grid system installed were towed simultaneously from one randomly assigned side of each vessel. Escape grid designs varied, with no two vessels having the same design. These included 14 downward-excluding grids and nine upward-excluding grids, made either of stainless steel or aluminium and with or without guiding panels/funnels (see paper for specifications). All codend catches were sorted and identified by species, weighed and counted.

    Study and other actions tested
  14. A replicated, paired, controlled study in 2005 of a seabed area in the North Sea, UK (Catchpole et al. 2006) found that prawn trawl nets fitted with a rigid size-sorting escape grid resulted in less non-target catch of most sizes of whiting Merlangius merlangus and plaice Pleuronectes platessa, but more small cod Gadus morhua and haddock Melanogrammus aeglefinus, compared to standard trawls without a grid. Average numbers of whiting (with: 25–126 fish/tow, without: 77–356 fish/tow) in the size ranges 20.6–35.5 cm (98%) and plaice (with: 0–8 fish/tow, without: 1–22 fish/tow) between 20.6–40.5 cm (91%) were lower with the grid than the standard trawl. Fewer cod of marketable size (35 cm) were caught in the trawl net with a grid (with: 0 fish/tow, without: 0–10 fish/tow), but average numbers of small cod (10.6–20.5 cm) were higher (with: 67–120 fish/tow, without: 25–64 fish/tow). More small haddock (10.6–15.5 cm) were caught with the grid (with: 20 fish/tow, without: 14 fish/tow) and catches above this size were typically similar (with: 0–12 fish/tow, without: 0–10 fish/tow). In addition, numbers of marketable sizes of the target species Norway lobster Nephrops norvegicus were greater in trawls with a grid (with: 3.8–4.4, without: 3.0–3.9 baskets/trawl) and total discards lower (with: 1.3, without: 1.7 baskets/trawl). In November 2005, a total of 12 paired deployments towing both a standard trawl net fitted with a metal grid (Swedish grid) and a standard trawl net were conducted in the Farn Deeps Nephrops fishing ground off the coast of England.

    Study and other actions tested
  15. A replicated, paired, controlled study in 2002 of a seabed area in the Bay of Biscay, France (Loaec et al. 2006) found that prawn trawls fitted with an experimental flexible size-sorting escape grid caught less discarded catch (fish and invertebrates) compared to a small-mesh trawl net without a grid. Data were not tested for statistical significance. The amounts of discarded catch (that included large quantities of horse mackerel Trachurus trachurus plus debris) were lower in nets with grids in five of six hauls (with: 12–32 kg, without: 14–51 kg) and higher in one of six (with: 59 kg, without: 32 kg). The total amount of discarded catch in all hauls was 161 kg in nets with grids and 184 kg in nets without grids. In addition, grid-fitted nets caught 88% fewer (with: 114 ind, without: 950 ind) undersized individuals of the target species Norway lobster Nephrops norvegicus, but also 61% less (with: 1033 individuals, without: 2632 individuals) of marketable size. A total of six, 2.5 h paired trawl deployments were done from a research vessel, simultaneously towing a trawl net fitted with a flexible grid (polyurethane) and a conventional trawl net with a fine-mesh inner lining. The grid had 20 mm bar spacing and was fitted at a 45° angle. Full details of trawl design are provided in the original study.

    Study and other actions tested
  16. A before-and-after study in 1991–1996 of a large seabed area in the Gulf of Maine, North Atlantic Ocean, USA (Richards & Hendrickson 2006) found that the requirement to fit a size-sorting escape grid to shrimp trawl nets in a Northern shrimp Pandalus borealis fishery reduced the capture of non-target fish compared to the period before grids were introduced. During the four-year period after grids were introduced, averages fish catches (by type) were lower with grids for roundfish (after: 5 kg/hr, before: 11 kg/hr) and flatfish (after: 3 kg/hr, before: 7 kg/hr) compared to the previous two-years and indicated reductions of 59% and 61% for each group respectively. Reductions for individual species ranged from 9% to 62% (see paper for full list of species). In addition, average target Northern shrimp catch increased after grids were used (after: 11 kg/hr, before: 8 kg/hr). Fishery observer data were collected onboard Northern shrimp fishing vessels, fishing up to 182 m depths, during December–March of the 1991–1996 Northern shrimp fishing seasons. A total of 140 vessels were sampled after (643 tows with grids) and before (283 tows without grids) grids were made a requirement in April 1992. Regulatory specifications for the grid were that it must include a rigid or semi-rigid grid of parallel bars spaced no more than 2.54 cm apart, a fish escape opening or hole (top or bottom) in front of both the codend and grid, and a mesh funnel to direct objects to the bottom of the grid (optional in 1994–1996).

    Study and other actions tested
  17. A replicated study of an area of shallow water in the Gulf of Thailand, Vietnam (Eayrs et al. 2007) found that shrimp trawl nets fitted with a rigid size-sorting excluder grid resulted in the escape of a high proportion of immature unwanted fish and sub-legal sizes of three of three fish of value, in a Vietnamese shrimp Penaeidae fishery. Overall, the grid excluded by weight 73% of the immature fish and 16% of the valuable fish catches (data not reported). For three of three fish species of commercial value (Japanese threadfin bream Nemipterus japonicus, bartail flathead Platycephalus indicus and snakefish Trachinocephalus myops) numbers of fish below the 150 mm length minimum landing size were reduced by 70–78% by the grid compared to the total codend catch (grid: 4,885–8,593 fish, codend: 1,767–3,706 fish). Lengths at which 50% of fish escaped were 124–134 mm across the three species. In addition, 8% of the target shrimp species were excluded by the grid. Data were collected over five days from 15 × 3 h trawl deployments at 12–15 m depth near Phu Quoc Island. A grid of three rectangular hinged panels, two of steel construction with 20 mm bar spacing, and one of small mesh to stop fish re-entering the net, was fitted to a 15 mm diamond mesh codend shrimp trawl net. Fish escaping from the grid were collected in a small-mesh cover installed over the panels. Both cover and codend catches were sorted separately by species, weighed and fish lengths recorded. The year the study took place was not reported.

    Study and other actions tested
  18. A replicated, paired, controlled study in 2005 of a seabed area in the Indian Ocean, off Mozambique (Fennessy & Isaksen 2007) found that fitting a prawn trawl net with rigid size-sorting excluder grids reduced the overall amount of discarded catch (fish and invertebrates) and caught fewer sharks and rays, compared to a conventional trawl without a grid. For two of two grid designs, average catch rates of discards (90% fish, 10% invertebrates) were lower in nets with grids than without (with: 37–48 kg/h, without: 49–83 kg/h). Large sharks and rays were caught in fewer hauls with grids than without (with: 0–2 hauls, without: 4–9 hauls) but there was no statistical difference for smaller sharks and rays (with: 5 hauls, without: 5–9 hauls). In addition, average catches of the target prawn species Penaeidae were similar between nets (with: 12–24 kg/h: without: 13–23 kg/h). Trials took place in February 2005 by twin-rigged trawler. A total of 16 paired trawl deployments were done with a net fitted with one of two grid designs (both Nordmøre) and a conventional trawl without a grid. Grids were aluminium, 100 mm bar spacing, fitted with a guiding funnel either with or without a cover flap in front of the grid escape opening (see paper for specifications). Codend catches were sorted into commercial/non-commercial portions, counted and weighed, and lengths of selected species measured.

    Study and other actions tested
  19. A before-and-after study in 1981–2005 of an area of seabed in the Pacific Ocean off the coast of Oregon, USA (Hannah & Jones 2007) found that the requirement to use rigid or flexible (mesh) size-sorting escape grids in shrimp trawl nets led to an overall reduction in catches of unwanted fish in an ocean shrimp Pandalus jordani fishery, compared to historical pre-use levels. Data were not tested for statistical significance. For four different types of escape grid, the amount of unwanted fish catch in the period after grids were introduced (2002–2005) was 6.5–13.3% of the total catches, compared to 32–61% unwanted fish catch in the years before grids were introduced (1981–2000). In 2005, catches of unwanted fish were 77–88% lower than the years from 1981–2000. In addition, catch rate and percentage of unwanted catch was significantly related to grid type and bar spacing (see paper for grid types). The use of a rigid or soft-mesh escape grid device to reduce unwanted catch was fully mandated in the ocean shrimp fishery in 2003 but grids were in use prior to this. Fisheries catch data post-grid use were collected in 2002–2005 by observers deployed on vessels operating in the fishery off the coast of Oregon. Historical catch data from 1981–2000 for nets without a grid or panel, were obtained from published and unpublished research sampling and survey records. See paper for list of fish species caught.

    Study and other actions tested
  20. A replicated, controlled study in 2005 of a seabed area in the Skagerrak and Kattegat, northern Europe (Frandsen et al. 2009) found that fitting a rigid size-sorting escape grid to prawn trawl nets reduced the catches of larger-sized fish of all species but increased the retention of small cod Gadus morhua and haddock Melanogrammus aeglefinus, compared to a standard diamond mesh codend without a grid. Overall, total catch numbers of legal sizes of eight of eight fish species (see paper for full list of species and minimum landing sizes) were lower in the net with a grid (with: 1–229 fish, without: 57–3,283 fish) (data not statistically tested). Catch numbers were significantly lower for cod, haddock, whiting Merlangius merlangus and plaice Pleuronectes platessa smaller than legal size but longer than 25, 20, 17 and 22 cm, respectively (data not reported). Numbers of fish were significantly higher in the net with a grid for small sizes of cod between 10–19 cm and haddock between 11–15 cm (data not reported). In addition, retention of the target prawn species Nephrops norvegicus longer than 41.8 mm was significantly lower with a grid than without (data presented as retention probability curves). In September and October 2005, trials were done by commercial fishing vessel using twin-trawl net gear. Trawl deployments were carried out with a small mesh (40 mm) codend paired with either: a standard 90 mm diamond-mesh codend fitted with a steel grid (22 hauls) or a standard 90 mm unmodified codend (18 hauls). The grid was 35 mm bar spacing in the lower quarters and 80 mm spacing in the upper quarter (see paper for specifications). Catches in each codend were sorted by species and weighed. Total length was measured for commercially important fish species and carapace length for Nephrops.

    Study and other actions tested
  21. A replicated, paired, controlled study in 2009 in two areas of seabed in the North Sea, Scotland (Drewery et al. 2010) found that prawn trawls fitted with a rigid size-sorting escape grid caught fewer unwanted fish species compared to an unmodified reference trawl. A trawl net fitted with a grid caught lower overall numbers of six of six unwanted fish species (hake Merluccius merluccius, cod Gadus morhua, haddock Merluccius aeglefinus, whiting Merlangius merlangus, plaice Pleuronectes platessa and witch Glyptocephalus cynoglossus) than the unmodified trawl (data reported as statistical model outputs). In addition, overall catch rates of the target species Norway lobster Nephrops norvegicus were similar between trawl nets, but fewer lobster of legal landing size (> 41 cm) were caught in nets with a grid (data reported as statistical model results). In March and July 2009, a total of 22 paired trawl deployments were carried out in the South Minch and Fladen Grounds near Scotland, respectively. One trawl net (80 mm codend) fitted with a 1.52 × 0.36 m aluminium grid (Swedish grid) was towed simultaneously with a conventional trawl net with a small-mesh (40 mm) codend. The grid had 35 mm bar spacing, top escape opening and was positioned at a 45° angle 13 m in front of the codend. Trawls were deployed for 3–3.5 hours in depths 108–139 m and all catch counted, and lengths measured.

    Study and other actions tested
  22. A replicated, randomized study in 1995–1998 at multiple coastal sites in the Atlantic Ocean, USA (Belcher & Jennings 2011) found that the amount of unwanted shark catch in shrimp trawls fitted with a rigid size-sorting escape grid depended on the type of trawl net and grid used. Commercial mongoose and flat net shrimp trawls fitted with two types of rigid grids with bottom escape openings (Super Shooter or Georgia Jumper respectively) caught fewer unwanted sharks Elasmobranchii (2/h) than triple wing shrimp trawl nets fitted with only Super Shooter grids (23/h). Shark catch rates in mongoose trawl nets fitted with Super Shooter grids were not statistically different to other gear combinations, both with and without an additional supported escape opening (Fish Eye) in the codend (with: 15/h, without: 17/h). In addition, the duration and towing speed of the trawl deployments did not affect shark catch rates (data reported as statistical model results). From April 1995–January 1998 (except February and March each year) shrimp trawl discard data were collected by fishery observers onboard vessels in the penaeid Penaeidae shrimp fishery off Georgia. Vessels randomly selected one of three commercial shrimp trawl net designs to use: flat net, mongoose or triple wing trawls. Each was fitted with a ‘Super Shooter’ escape grid, except some mongoose nets that used a ‘Georgia Jumper’ grid. Both grid types were metal and oval, but differed in the angle of the bars. Some mongoose net/Super Shooter combinations also included a ‘Fish Eye’ escape opening in the codend. Full details of trawl and grid designs are provided in the original study.

    Study and other actions tested
  23. A replicated, controlled study in 2008–2009 of a seabed area in the Pacific Ocean off Chile (Queirolo et al. 2011) found that shrimp trawl nets fitted with a rigid size-sorting escape grid allowed more unwanted fish to escape capture than conventional trawl nets with no grid. The total percentages in number of unwanted catch (fish and invertebrates combined) that escaped capture was higher in trawls with a grid fitted (25%) than in trawls without a grid (2–3%). The percentage number of catch that escaped capture was also higher in trawls with a grid fitted than trawls without a grid for Chilean hake Merluccius gayi gayi (48 vs 1%), Aconcagua grenadier Coelorinchus aconcagua (20 vs 0–3%), cardinalfish (Apogonidae) (80 vs 1–4%) and elasmobranchs (Elasmobranchii) (39 vs 0%). Escapees were similar in trawls with and without a grid for cusk-eel Ophidiidae (100 vs 94%). Thirty-nine trawl deployments were made using two trawl nets with 56 mm diamond mesh codends, either with or without a rigid grid. The grid was a metal Nordmøre grid (1.2 × 0.8 m) with 35 mm bar spacing and top escape opening, fitted at a 45° angle in front of the codend. A mesh guiding panel guided catch to the grid. Covers over the grid opening collected the escaping catch. All codend and cover catches were sorted and weighed separately.

    Study and other actions tested
  24. A replicated, paired, controlled study in 2007–2009 of a seabed area in the southern Atlantic Ocean, Brazil (Silva et al. 2011) found that only one of four designs of rigid size-sorting escape grids (Nordmøre grid) fitted to shrimp/prawn nets reduced the amount of unwanted fish catch in an artisanal canoe-trawl fishery, compared to nets without a grid. Average weight of unwanted fish catch was reduced by 50% in nets fitted with a small grid compared to no grid (with: 1.2 kg/tow, without: 2.5 kg/tow) but was similar for three other larger grid designs (with: 1.0–1.2 kg/tow, without: 1.5 kg/tow). Catch of the target species Atlantic seabob Xiphopenaeus kroyeri was similar for four of four grid designs (with: 1.7–3.9 kg/tow, without: 0.9–4.5 kg/tow). Two trials were done between July 2007 and November 2009 off the coast of Paraná. A total of 18 (small grid, 60 min tows) and 12 (three large grids, 30 min tows) paired deployments of two trawl nets, one with a grid and one without, were done from a motorized canoe. Four configurations of aluminium grids (Nordmøre) were tested; all with 24 mm bar spacing, but differing in size/weight (one small, three larger), bar type, presence or absence of a guiding panel and mesh size of the extension piece of netting (see paper for specifications). Trawl nets were randomly deployed on each side of the canoe. Codend catches were separated by species and the numbers and weights of fish recorded.

    Study and other actions tested
  25. A replicated, paired, controlled study in 2010 of a sandy seabed area in the Atlantic Ocean, off Brazil (Silva et al. 2012) found that shrimp trawl nets fitted with rigid size-sorting escape grids (Nordmøre) did not catch fewer unwanted fish in a canoe-trawl fishery, compared to conventional trawls with no grids, regardless of grid bar spacing. For three of three grid bar spacings, the average catch weight of all unwanted fish (see paper for list of species) was similar between trawl nets (grid: 1.0–1.3 kg/30 min, no grid: 1.2 kg/30 min). Numbers of four of four fish species with sufficient data were similar between nets (grid: 8–55 fish/30 min, no grid: 6–35 fish/30 min), but trawls with grids retained smaller sizes of two of those species (grid: 8.0–8.4 cm, no grid: 8.9–9.0 cm). In addition, there was no significant difference in the weights of retained target seabob shrimp Xiphopenaeus kroyeri catches (grid: 5.0–5.4 kg/30 min, no grid: 4.8 kg/30 min). In April–June 2010, data was collected from 24 paired deployments of each of six net pairings, towed in <15 m depth by a motorized canoe rigged with two identical trawls. Three conventional trawl nets were fitted with aluminium Nordmøre grids with 17-, 20- or 24 mm bar spacings and tested against each other and against one conventional net without a grid (see paper for specifications). Codend catches were counted and weighed by species.

    Study and other actions tested
  26. A replicated, paired, controlled study in 2002 of a coastal seabed area in the Coral Sea, Australia (Courtney et al. 2014) found that prawn trawl nets fitted with a rigid size-sorting escape grid (turtle excluder device) did not reduce the amount of unwanted fish catch, compared to a standard diamond mesh trawl codend. For one of 26 unwanted fish species with data (see paper for list of species), the average catch rate was lower with a grid than without (with: 94.7 g/h, without: 134.3 g/h) but for the remaining, average catch rates were either similar (23 species) or higher (two species) between nets (with: 1.3–95.7 g/ha, without: 1.1–72.2 g/ha). In addition, catch rates of the target eastern king prawns Melicertus plebejus were similar between nets (with: 291.1–274.4 g/ha). In July 2002, data were collected from 65 paired trawl deployments done over ten nights off the coast of Queensland. Three different codends were tested against a standard diamond mesh codend: a standard diamond mesh with a metal, top-opening grid (Wick’s turtle excluder device), a grid in combination with a square mesh codend, and a square mesh codend alone (see paper for specifications). Each codend design was randomly assigned to the two trawl nets every 12 tows. Each tow was two nm long, at 2.2 knots.

    Study and other actions tested
  27. A replicated study in 2012 of a seabed area in the Tyrrhenian Sea, western Mediterranean (Brčić et al. 2015) found that a prawn trawl fitted with a rigid size-sorting escape grid allowed a small proportion of unwanted blackmouth catshark Galeus melastomus of larger size to escape capture. In number, 182 catshark escaped through the grid, 263 escaped from the codend and 540 were retained in the codend. Catshark larger than 45 cm total length were more likely to escape through the grid and the estimated length at which half would escape was 53 cm, whereas individuals smaller than 20 cm were more likely to escape from the codend (data reported as probability/selection curves). In addition, for two of two commercial species, the grid-fitted trawl codend retained 39% and 94% of greater forkbeard Phycis blennoides and Norway lobster Nephrops norvegicus respectively. A conventional (50 mm diamond mesh codend) commercial bottom trawl net used in Mediterranean was fitted with an aluminium “Super Shooter” grid with 90 mm bar spacing, located 3.5 m in front of the codend at a 45° angle. Data were collected from six trawl deployments in April and July 2012. Two covers with 20 mm mesh were attached over the grid escape opening and the codend to collect escaped individuals. Total catch was sorted and weighed and fish lengths recorded.

    Study and other actions tested
  28. A replicated, paired, controlled study in 2012 in the Gulf St Vincent, Australia (Gorman & Dixon 2015) found that prawn trawls modified with a rigid size-sorting excluder grid, and a diamond mesh codend with the mesh orientation turned by 90°, caught fewer unwanted fish compared to conventional diamond mesh nets with no grid. Modified trawls caught fewer unwanted individuals than conventional trawls of rays Batoidea (2.6 vs 0.7/h), sharks Selachii (2.0 vs 0.3/h), porcupinefish Diodontidae (0.8 vs 0.0/h), bream Sparidae (0.2 vs 0.0/h), armourheads Pentacerotidae (0.1 vs 0.0/h), croaker Sciaenidae (0.1 vs 0.0/h), filefish Monocanthidae (22.4 vs 10.8/h), jacks and pompanos Carangidae (2.8 vs 2.2/h), dragonets Callionymidae (2.3 vs 1.0/h) and gurnard Triglidae (2.2 vs 1.3/h). Numbers of flatheads Platycephalidae caught in each trawl design were similar (data reported as statistical model results). Target western King prawn Penaeus latisulcatus catches were 15% lower in modified trawls (results not tested for statistical significance), although this was largely due to losses of small, low-value individuals. Twenty-nine, 30-min, replicate paired trawl deployments were undertaken at randomly chosen sites in May 2012 using modified and conventional trawl nets. Modified nets were fitted with a U-shaped plastic grid with 50 mm bar spacing and top escape opening and had a codend of 58 mm diamond mesh rotated 90° in orientation. Conventional trawls used 58 mm diamond mesh codends. Full details of trawl design are given in the original study.

    Study and other actions tested
  29. A replicated, paired, controlled study in 2013 of shallow, coastal waters in the Persian Gulf, Iran (Paighambari & Eighani 2016) found that shrimp trawl nets fitted with a rigid size-sorting escape grid (a Nordmore grid) caught fewer undersized fish compared to a trawl net without a grid. The net with a grid caught lower proportions of undersized individuals of three of three fish species: narrow-barred Spanish mackerel Scomberomorous commerson (with: 11%, without: 36%), tigertooth croaker Otolithes ruber (with: 29%, without: 43%) and silver pomfret Pampus argenteus (with: 13%, without: 15%). For each species, the length at which half were likely to escape was smaller than the minimum landing sizes. In addition, compared to a 32% reduction in average exclusion rate of undersized fish with a Nordmore grid, a trawl net fitted with a different design of grid (Nafted) and an additional large supported escape opening (Fisheye), reduced the undersized catches by 47%. Data were collected from a total of 30 valid trawl deployments (15 tows for each grid design) conducted by a commercial vessel in 2013. Test trawl nets fitted with grids were towed for 1.5 h alongside nets without grids in depths of 13–33 m. Codends were all of 30 mm mesh size. Both the Nordmore and Nafted grids had 60 mm bar spacing at were fitted at a 45° angle. The Nafted grid was fitted in combination with a Fisheye escape opening, a steel frame sewn into the top of the codend to provide an elliptical opening of 400 mm for fish escape.

    Study and other actions tested
  30. A replicated study in 2010 of a seabed area in the Kattegat and Skagerrak, North Sea, bordering Norway, Denmark and Sweden (Madsen et al. 2017) found that size-sorting escape grids of three designs fitted to prawn trawl nets all reduced the capture of unwanted small fish in a Norway lobster Nephrops norvegicus fishery. Overall, grids enabled 55–88% (225–6,766 fish) of undersized individuals of three of three roundfish (cod Gadus morhua, haddock Melanogrammus aeglefinus and whiting Merlangius merlangus) and 35–86% (337–3,463 fish) of undersized fish of two of two flatfish (lemon sole Microstomus kitt and plaice Pleuronectes platessa) to escape. In addition, grids reduced the catches of undersized individuals of the target Norway lobster by 5–17%, but there were losses above minimum landing size of 13–33%. Data were collected from 10–14 trawl deployments of 2–4 h at 42–71 m depth, for each of three grid systems in March 2010 using a twin-rigged trawler. Trawl nets of 90 mm mesh codend were fitted with grid systems of either: horizontal bars, vertical bars, or vertical bars and a mesh guiding panel. All grids were black in colour, 45 mm bar spacing, set at 45° angles and with a hole at the bottom part to stop debris (see paper for specifications). Small mesh covers attached over the grid escape opening collected fish escaping through the grid. Cover and codend catches were weighed and length measurements taken for all commercially important species.

    Study and other actions tested
Please cite as:

Taylor, N., Clarke, L.J., Alliji, K., Barrett, C., McIntyre, R., Smith, R.K., and Sutherland, W.J. (2021) Marine Fish Conservation: Global Evidence for the Effects of Selected Interventions. Synopses of Conservation Evidence Series. University of Cambridge, Cambridge, UK.

Where has this evidence come from?

List of journals searched by synopsis

All the journals searched for all synopses

Marine Fish Conservation

This Action forms part of the Action Synopsis:

Marine Fish Conservation

What Works in Conservation

What Works in Conservation provides expert assessments of the effectiveness of actions, based on summarised evidence, in synopses. Subjects covered so far include amphibians, birds, terrestrial mammals, forests, peatland and control of freshwater invasive species. More are in progress.

More about What Works in Conservation

Download free PDF or purchase
The Conservation Evidence Journal

The Conservation Evidence Journal

An online, free to publish in, open-access journal publishing results from research and projects that test the effectiveness of conservation actions.

Read the latest volume: Volume 18

Go to the CE Journal

Discover more on our blog

Our blog contains the latest news and updates from the Conservation Evidence team, the Conservation Evidence Journal, and our global partners in evidence-based conservation.


Who uses Conservation Evidence?

Meet some of the evidence champions

Endangered Landscape Programme Red List Champion - Arc Kent Wildlife Trust The Rufford Foundation Save the Frogs - Ghana Bern wood Supporting Conservation Leaders National Biodiversity Network Sustainability Dashboard Frog Life The international journey of Conservation - Oryx British trust for ornithology Cool Farm Alliance UNEP AWFA Butterfly Conservation People trust for endangered species Vincet Wildlife Trust