Using a modified Nordmøre grid for by-catch reduction in the Portuguese crustacean-trawl fishery

Fisheries Research 71 (2005) 223–239

Using a modified Nordmøre grid for by-catch reduction in the Portuguese crustacean-trawl fishery Paulo Fonsecaa,∗ , Aida Camposa , Roger B. Larsenb , Teresa C. Borgesc , Karim Erzinic a

INIAP/IPIMAR—Portuguese Institute for Agriculture and Fisheries Research, Avenida de Bras´ılia, 1449-006 Lisboa, Portugal b The Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway c Centre of Marine Sciences (CCMAR), University of the Algarve, Campus de Gambelas, 8000-117 Faro, Portugal Received 3 October 2003; received in revised form 9 August 2004; accepted 9 August 2004

Abstract By-catch and discards are a common problem to all fisheries. Recent studies of the Portuguese crustacean-trawl fishery show that discards average 70% of the total catch. This situation poses a threat not only to the management of target and commercial fish by-catch species stocks, but also in terms of the ecosystem impact of the discarded species whose survival is generally low. Previous attempts to address this problem by using oblique separator mesh panels associated with square-mesh windows, or square-mesh windows alone, provided encouraging results, but these are unlikely to have commercial applicability at present. Therefore, we tested the efficiency of a modified Nordmøre grid in excluding some of the most captured non-commercial bycatch species, such as the blue whiting, Micromesistius poutassou, and the boarfish, Capros aper, evaluating simultaneously the losses of the crustacean target species (rose shrimp, Parapenaeus longirostris, Norway lobster, Nephrops norvegicus and red shrimp, Aristeus antennatus) and commercially valuable fish by-catch. A total of 41 valid hauls were carried out during 2001, and 15 in 2002 on board the R/V “Noruega”. Results from both years varied considerably, with higher by-catch exclusion rates, in weight, registered in the second survey (73–74 and 48–63% for blue whiting and boarfish, respectively). The corresponding percentage reduction for target species was 8–15% for the Norway lobster, 4–9% for rose shrimp and 7–10% for red shrimp. The level of exclusion of the non-commercial by-catch is encouraging, although short-term loss of Norway lobster, where the escapees comprised larger individuals, raises some concern with regards to fishers’ acceptance of mandatory use of grids. © 2004 Elsevier B.V. All rights reserved. Keywords: Trawl selectivity; Sorting grids; By-catch reduction; Nephrops norvegicus; Parapenaeus longirostris; Aristeus antennatus; Micromesistius poutassou; Capros aper; Portuguese waters

∗

Corresponding author. Tel.: +351 213 027 163; fax: +351 213 015 948. E-mail addresses: [email protected] (P. Fonseca), [email protected] (A. Campos), [email protected] (R.B. Larsen), [email protected] (T.C. Borges), [email protected] (K. Erzini). 0165-7836/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2004.08.018

224

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

1. Introduction Vessels operating in the crustacean-trawl fishery off the coast of Algarve (south Portugal), target Norway lobster, Nephrops norvegicus, rose shrimp, Parapenaeus longirostris, and red shrimp, Aristeus antennatus, using a minimum codend mesh size of 55 mm. From 1984 to 1993, the Norway lobster dominated the landings, while from 1994 onwards there was a shift to the rose shrimp and red shrimp as the stock of Norway lobster declined. Like in most crustaceantrawl fisheries throughout the world, there is also a substantial capture of by-catch in addition to the targeted species. Some of the by-catch is retained and sold, including high value species such as the European hake, Merluccius merluccius, and the monkfish, Lophius spp., that contribute to a significant proportion of the total revenues of this fishery. However, a large subset comprising a miscellaneous collection of fish, crustaceans and cephalopods is discarded at sea. An average of 70% of the total catch weight per haul was discarded in this fishery in 1995–1996 (Borges et al., 2001), while in subsequent years a lower figure of about 38% was reported by Monteiro et al. (2001). The blue whiting, Micromesistius poutassou, was found to be the most abundant discard in the latter study. Often, a certain proportion of discards corresponds to undersized individuals of commercially valuable species that form the basis of other fisheries. For these species the practice of discarding is viewed as a contribution to growth overfishing of stocks, causing conflicts among different m´etiers and between fishers and fisheries management authorities. In a considerable number of crustacean-trawl fisheries, the concern raised by the mortality of large quantities of by-catch and undersized target species has been addressed via the assessment of alterations to existing commercial trawls, designed to improve both size and species-selectivity. These alterations typically have involved changes in the size and shape of the codend meshes and/or the use of different types of devices, commonly referred to as by-catch reducing devices (BRDs) (review, Broadhurst, 2000). Although different types of BRDs have been successfully testedworldwide and commer-

cially adopted in some fisheries, e.g., in Argentina, Australia, Canada, Iceland, Mexico, Norway, USA (Anon., 1996, 1998), there is often considerable resistance from the industry to their enforced use. Some of the factors that have been shown to limit acceptance of modifications include: failure to operate efficiently over a range of commercial conditions (Christian and Harrington, 1987), undesirable effects on gear performance and handling (Oravetz and Grant, 1986), complex designs and high costs (Watson, 1989; Mounsey et al., 1995; Robins and McGilvray, 1999) and unacceptable losses of commercially valuable by-catch or of target species (Hanna and Jones, 2000). The improvement in codend selectivity of crustacean trawls has been previously addressed by INIAP/IPIMAR (Portuguese Institute for Agriculture and Fisheries Research). A considerable amount of information was gathered regarding how the most important species react both to an increase in codend mesh size and changes in mesh configuration (Campos et al., 2002, 2003). The results obtained in these studies support the idea that an increase in codend mesh size from the current 55 mm, as well as a change in mesh configuration from 55 mm diamond to 55 mm square mesh, would allow the exclusion of a high proportion of undersized individuals, together with noncommercial by-catch, but always at the expense of losing an important fraction of commercial-sized shrimp. Experiments using separator mesh panels (Campos and Fonseca, 2004) and square-mesh windows inserted in the upper belly or in the top of the codend (Fonseca et al., 1998) were also carried out on board research vessels, and therefore the validation of the results under commercial conditions is still needed. However, while the sorting panels had a complex design and rigging, losses of rose shrimp above the minimum landing size (MLS) were registered when using the square-mesh windows alone. Hence, both the latter devices are likely to raise objections from fishers. As a consequence, it was decided to pursue a different approach based on the use of a modified Nordmøre grid, as these types of BRDs have proven successful in a number of crustacean fisheries (without excessive loss of target species), and their use is now mandatory in several fisheries in the North Atlantic (Anon., 1996, 1998).

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

225

Fig. 1. Technical drawings of the trawl used in this study: (a) top panel; (b) side panel; (c) bottom panel.

2. Material and methods 2.1. The gear The technical drawings of the trawl used are in Fig. 1. The trawl was entirely made of braided polyethylene (PET), 47.2 m long from the wing tips to the codend joining row, with a circumference of 1112 meshes of 60 mm1 at the footrope level. The codend was 8 m long and was entirely made of twisted polyamide mesh of 20 mm size and 1.0 mm twine thickness. The gear was rigged with a 47.0 m long footrope 1 Throughout the text mesh size should be understood as stretched mesh.

made up of 16 mm combination rope covered with 16 mm PET, 6.9 m bridles, sweeps of 100 m and semioval Portuguese Sounete doors of 650 kg each (area: 4.2 m2 ). Trawl geometry and speed were recorded during the surveys using height, spread and speed sensors. 2.2. The grid system The grid was mounted in a 3.0 m long extension piece made of 50 mm mesh size PET netting installed between the last belly and the codend (Fig. 2). The selective system included a grid made of stainless steel, with overall dimensions of 1.50 m × 0.79 m in height and width, respectively. The grid bars were 10 mm in diameter with 25 mm spacing between them. In the lower

226

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

Fig. 2. Details of the selective system and its placement in the trawl: (a) trawl top panel; (b) grid details; (c) selective system.

part of the grid a 20 cm high section without bars was intended to allow Norway lobster to enter directly into the trawl codend. A triangular shaped escape hole (1.3 m side length) was located just above the grid at the top of the extension piece, to release fish that either react to the grid or are too large to cross the bars. A guiding funnel with a stretched length of 1.55 m (33 meshes of 50 mm) was located before the grid with the purpose of forcing all the fish to enter into contact with the lower half of the grid, irrespective of the height of their entrance in the trawl, thus preventing direct escape through the top hole. The guiding funnel consisted of a double PET 50 mm mesh size panel cut along a line of bars. The funnel was attached to the front end of the upper panel of the extension piece and fixed to its lower panel 0.5 m ahead of the grid, where an elastic rope restricted its opening. The grid was mounted into the extension piece in such a way to achieve a working angle of attack close to 50◦ . A grid sensor was used for measuring the grid angle with relation to the horizontal plane. This grid sensor also provided information on water speed in the grid zone. In order to minimize the possibility of a decrease in the grid angle, due to the weight of the catch in the codend, supporting ropes

were used mounted from the grid top corners to the lastridge ropes. A top cover, constructed according to the specifications in Wileman et al. (1996) and ending in a collecting bag, was placed above the escape hole to collect the individuals excluded by the grid. The cover was made of 45 mm twisted PET mesh size, while the collecting bag was made of the same material as the codend. During the 2002 trials, a collecting bag (inner codend) made of the same material as the codend was connected to the lower section of the grid in order to retain those individuals entering directly into the codend (i.e., without crossing the grid bars). 2.3. Experimental procedures and catch handling Two surveys were carried out from 5 to 19 May 2001 and from 26 April to 1 May 2002 (Fig. 3). A total of 41 and 15 one-hour hauls, respectively, were carried out at crustacean fishing grounds off the Portuguese south coast. The surveys were carried out on board the IPIMAR R/V “Noruega”, a 47.5 m overall length trawler with a 1500 HP engine. Fishing was completed during daytime, at depths varying approximately between 180 and 660 m, in 2001, and 200 and 730 m, in 2002.

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

227

dend (i.e., not size-selected) could be determined and thus excluded from analysis. For crustaceans, whenever the catch data structure permitted, the retention-at-length of the grid (rg ) for pooled data was modelled by four different twoparameter symmetric (logit and probit) and asymmetric (log–log and c-log–log) models, and a three-parameter (Richards) asymmetric model (Wileman et al., 1996):
rg (l) = Fig. 3. Location of the experimental hauls. Black circles: 2001; open circles: 2002.

Trawling was carried out at speeds between 2.5 and 3.0 knots. This represents the depth range normally exploited and the trawling speeds generally adopted by commercial trawlers. Catches taken from the codend and the cover (and the inner codend, in 2002) were handled separately, identified to species and weighed. For commercial crustaceans, carapace length was measured to the millimetre below, while for the most important commercial and by-catch fish species the total length was measured to the centimetre below. For most of the species caught, including the Norway lobster, the red shrimp and the hake, the whole catch was measured in all hauls, while for others such as the rose shrimp and the blue whiting, due to their abundance, sub-sampling was carried out for the majority of hauls. The length class frequencies were scaled up by the inverse of the sampling proportions. 2.4. Species- and size-selection Although the key objective of the present study was to evaluate the species-selective properties of the grid layout, this type of device also induces a size-selection both on target and on by-catch species. As a consequence, while the experiments carried out in May 2001 had the single purpose of evaluating the escape proportion of the by-catch along with the loss proportion of the target species, in the following year attention was also given to the investigation of the size-specific component of grid selection. For this purpose, an internal cover was adapted to the lower opening of the grid so that the proportion of fish entering directly into the co-

e(α+βl) 1 + e(α+βl)

δ (1)

where δ is the asymmetry parameter; if δ > 1, the lower branch of the curve will be longer than the upper one, conversely for 0 < δ < 1. When δ = 1, it will be equivalent to the logistic model. The model with the lowest AIC—Akaike’s Information Criterion (Akaike, 1974) was chosen as the one that best fitted the data. Considering that fish species, unlike crustaceans, often present active avoidance behaviour when facing a gear element, a different approach was followed. In other words, it may be that not all the fish available to the grid contact it, escaping directly through the top opening. Tokai et al. (1996) and Zuur et al. (2001) modelled the probability of a fish encountering the selective device given that it was in its area of influence by using a parameter π. As a consequence, the available-selection curve (ag (l)) was estimated as ag (l) = π × rg (l) + (1 − π),

0≤π≤1

(2)

where ag (l) is the selectivity of the l length class fish, π the contact probability and rg (l) the contact-selection curve. Here rg (l) was assumed to be logistic:
rg (l) =

e(α+βl) 1 + e(α+βl)

 (3)

Estimation of parameters α, β, δ and π was carried out by maximum likelihood using the software CC2000 (ConStat, DK) for the crustaceans and an ExcelTM spreadsheet by Professor Tadashi Tokai (University of Tokyo) for the fish. Differences in mean size and length class distribution of each species among the different compartments (codend, inner codend and cover) were assessed by way of t-tests and Kolmogorov–Smirnov tests, respectively.

228

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

3. Results 3.1. Gear performance The mean values (and standard deviations) of the parameters characterizing the gear performance for both cruises are presented in Table 1. For a speed over the bottom of 3.0 knots (GPS data), mean speed through water (speed sensor) was about 2.8 and 3.0 knots (1.44 and 1.54 m s−1 ), in 2001 and 2002, respectively. Vertical opening and wingend spread data were rarely collected, probably due to some instability in the positioning of the sensors, which may be partially related with strong bottom currents occurring in the area. Transversal current, as measured by the speed sensor, was usually 0.5 knots (0.26 m s−1 ) or less. On some occasions this exceeded 1 knot, thus contributing to an eventual distortion in gear performance. Nevertheless, indicative results were obtained pointing to a vertical opening of 1.9–2.0 m and a wingend spread of about 28 m. These results are consistent with previous monitoring of the performance of this gear (Leite et al., 1990). The mean door spreads were nearly the same, 115 m in 2001 and 116 m in 2002, but in the former year the variability was considerably higher. Also, in 2001 a highly significant (P < 0.001) correlation with depth was found. Fig. 4 displays the variability in the mean working angle of the grid for both years, along with a confidence interval of approximately 95%, and maximum and minimum values. The mean angle was about 46◦ for both trials, 4◦ less than the theoretical value. In spite of the identical mean values, there was a much higher variability during the 2002 trials. This variability could not be attributed to any controllable factor, since the rigging of the grid was similar to the previous year, as were the gear operation, the fishing grounds and the season. In addition, the relationship between grid angle and haul depth, haul speed and codend catch was investigated

Fig. 4. Variability of the grid angle during the 2001 and 2002 surveys. Maximum, minimum and mean values are presented along with an approximate 95% confidence interval for the mean.

but no significant (P > 0.05) statistical correlation was found for either of the trials. Mean water speed through the grid was fairly constant, 1.0 knots (0.51 m s−1 ) in 2001 and 1.1 knots (0.57 m s−1 ) in 2002, varying from 0.8 to 1.2 knots (0.41–0.62 m s−1 ). 3.2. Catch data Fig. 5 shows the 2001 and 2002 catches, in weight, for the most important commercial and by-catch species. In spite of the inter-annual variability in the catch of the different species, the status of blue whiting as the most important species stands out (44.3% in 2001 and 31.2% in the following year). Another clear feature is the size of the ‘Other’ group (50% in 2002), the bulk of which is constituted by low-value or noncommercial species that are discarded along with blue whiting. The total weight of commercial crustaceans represented less than 15% of the total catch for both years.

Table 1 Scanmar data on gear and grid performance, mean haul depth and water temperature during the 2001 and 2002 surveys

2001 2002

Haul depth (m)

Headline height (m)

Wingend spread (m)

Door spread (m)

Grid angle (◦ )

Water speed at the grid (knots)

Water speed at the mouth (knots)

Temperature (◦ C)

435 460

2.0 (0.21) 1.9 (0.15)

24.1 (7.84) –

116 (17.10) 115 (7.10)

46 (0.97) 46 (3.25)

1.0 (0.13) 1.1 (0.10)

3.0 (0.26) 2.8 (0.25)

12.5 (0.51) 13.7 (0.51)

The standard deviations are in brackets.

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

229

ster (MLS: 20 mm) were always scarce. Rose shrimp were caught within the same length interval in both years (8–40 mm), with most of the catch between 20 and 30 mm. While in the first year there were two distinct modes, at 22 and 26 mm, in 2002 a single mode was found at 25 mm. Red shrimp was the most scarcely fished of all crustaceans, since only a few hauls were carried out at depths below 600 m, where the highest concentrations are found. Consequently, individuals were grouped in 5 mm interval length classes, from 16–21 to 61–66 with an additional 66–71 class in 2001. Larger individuals were captured in the first year, with a bimodal distribution characterised by a dominant mode (46–51 class) and a secondary mode (26–31 class), whereas in 2002 a single mode occurred at the 31–36 class. Hake, which were only caught in considerable numbers in 2001 at lengths ranging from 12 to 60 cm, were not included in the analysis. Blue whiting in 2001 ranged in length from 17 to 45 cm, with the bulk of fish concentrated between 19 and 30 cm. A main mode was found at 21 cm and a secondary mode at 27 cm. Although the catch ranges in both surveys were almost coincident, average length was smaller in 2002, with fish concentrated in the 18–21 cm classes and with a mode at 19 cm. 3.3. By-catch reduction Fig. 5. Catches in weight of the most captured species in 2001 (a) and 2002 (b).

Length frequency distributions for the species analysed are shown in Fig. 6. All species were caught within a similar length class range in both years. Differences in mean size were generally small, with the exception of the red shrimp (45.7 mm in 2001 and 37.4 mm in 2002). Since both cruises were carried out in the same time of the year, the differences found are most probably due to differences in the fishing grounds surveyed. In fact, due to time constraints, the number of hauls, the areas and the depth range fished were more restricted in 2002, which may explain the considerable differences in the mean size for the red shrimp. For Norway lobster, the length ranges were very similar during both cruises (18–66 mm in 2001), although very few individuals below 18 mm were caught in the following year. However, the modes differed (37 and 40 mm), with the smallest individuals being caught in 2002. In any case, undersized Norway lob-

Fig. 7a–d and Tables 2 and 3 show the percentages of escape/retention, in both numbers and weight (total, above and below MLS), for the 2001 and 2002 trials. The blue whiting mean escape rate was the highest of all species, with about 73% of all fish entering the trawl being caught in the cover. Boarfish (Capros aper), the second major by-catch species, had an escape rate of about 47% in both weight and number, in the 2001 trials, while this percentage increased to 63% in 2002. For the crustacean species, differences in escapement were registered between trials, and within each trial varied according to whether they were based on numbers or weight. Fig. 8a and b displays the variability associated with the mean escape for Norway lobster, rose shrimp and blue whiting. Greater proportions of Norway lobster escaped compared to both shrimp species. While in 2001 Norway lobster losses were 8 and 4%, in 2002 they increased to nearly 15 and 11%, in weight and number, respectively. For rose shrimp, there was a similar increase from one year to the other,

230

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

Fig. 6. Length frequency distributions for the target species and blue whiting in 2001 (a–d) and 2002 (e–h). Thin line: total catch; thick line: inner codend; dashed line: codend; dashed line with circles: cover.

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

231

Fig. 7. Percentages of excluded (cover) and retained (codend and inner codend) for the target species and non-commercial by-catch. In number (a and b); in weight (c and d). Black rectangles: cover; grey rectangles: inner codend; white rectangles: codend. DSP: rose shrimp; NEP: Norway lobster; ARA: red shrimp; WHB: blue whiting; BOC: boarfish.

Table 2 Retained catch proportion for the species analysed in the different trawl compartments (codend, inner codend and cover) during the 2001 and 2002 surveys Year

Retained

2001

By number Codend Cover By weight Codend Cover

2002

By number Inner codend Codend Cover By weight Inner codend Codend Cover

Rose shrimp

Norway lobster

Red shrimp

Blue whiting

Boarfish

0.96 0.04

0.96 0.04

0.95 0.05

0.27 0.73

0.53 0.47

0.96 0.04

0.92 0.08

0.93 0.07

0.27 0.73

0.52 0.48

0.30 0.61 0.09

0.47 0.42 0.11

0.22 0.74 0.04

0.07 0.21 0.72

– – –

0.26 0.65 0.09

0.50 0.35 0.15

0.26 0.64 0.10

0.07 0.19 0.74

0.05 0.32 0.63

232

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

Table 3 Percentage excluded in number and in weight, below and above the minimum landing size (MLS), for the species analysed during the 2001 and 2002 surveys 2001 Rose shrimp (24 mm) Nb
Total

Codend

20383 8466 11917 223.4

19587 8107 11480 214.5

796 359 437 8.9

3.7a 4.2 3.7 3.9a

2.28 – – 2.85

1614 2 1612 79.9

1554 2 1552 72.6

60 0 60 7.3

4.2a 0.0 3.7 8.0a

27.70 – – 7.95

339 21 318 15.2

321 21 300 14.1

18 0 18 1.0

5.3 0.0 5.7 6.9

10342 1017.8

2775 257.0

7567 760.8

73.4a 73.3a

2203 62.9

1170 32.9

1033 30.0

46.9 47.7

Total

Rose shrimp (24 mm) Nb 2952
Cover

Excluded %

Cover

S.D.

– – – – 9.04 10.16 – –

Inner codend

Codend

Excluded %

957 516 441 8.0

1734 818 916 19.8

261 131 130 2.6

9.1a 8.9 8.7 8.9a

5.20 – – 2.98

675 0 675 27.7

548 5 543 19.3

77 0 77 6.7

10.7a 0.0 5.9 14.9a

13.30

54 14 40 2.1

182 42 140 5.0

9 2 7 0.8

3.7 3.4 3.7 9.6

180 12.9

981 49.5

2355 170.6

– 1.6

– 9.2

– 18.2

72.4a 74.0a – 62.8

S.D.

16.46 – – – 16.32 16.18 – –

Nb: catch in number; MLS: minimum landing size (in brackets after the species name; for fish—total length; for crustacean—carapace length); S.D.: standard deviation. a Mean value calculated from individual hauls excluded percentages.

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

233

Fig. 8. Variability in the excluded proportion of the Norway lobster, rose shrimp and blue whiting catch for 2001 and 2002, in number (a) and weight (b). NEP: Norway lobster; DSP: rose shrimp; WHB: blue whiting; 01 and 02 refer to 2001 and 2002.

with 4% in 2001 and 9% in the following year. Finally, for the red shrimp escapes in 2001 were in the order of 5 and 7%, in number and weight, respectively, while in 2002 the losses changed to almost 10% (in weight) and 4% (in numbers). Data from the 2002 trials in Table 2 provide further insight on the contrasting behaviour (avoidance) of each species towards the grid, by quantifying the percentage of individuals entering directly through the lower opening of the grid. By-catch species presented a lower percentage of fish caught after passing through the open section of the grid, with 5 and 7% (in weight) for boarfish and blue whiting, respectively. At the other extreme was the Norway lobster where 50% of the catch entered the codend directly via the open section of the grid, while for the shrimp species the value was 26%. 3.4. Selection at length Notwithstanding the fact that the major goal was to exclude by-catch fish species based on their differential behaviour without losing too much of the crustacean target species catch, the grid also induced size-selection among the population fished. Kolmogorov–Smirnov and t-tests (α = 0.01) were performed to compare the length distributions and mean size of the catches (except for boarfish) among the different compartments of the gear for both trials (Table 4). When comparing escapees with the retained catch, the results were almost coincidental for both years,

with significant differences for blue whiting and Norway lobster both in length distribution and mean size (bigger individuals in the cover). With the exception of mean size for red shrimp in the first trial, no significant differences were found for the shrimp species. For 2002 trials, the tests included the comparison among the three compartments. Data for blue whiting, Norway lobster and red shrimp confirmed the previous results. On the other hand, the situation for rose shrimp was distinct, with only the differences in length distribution and mean size between the cover and the codend (individuals crossing the grid bars) being nonsignificant. Both differences between the cover and the inner codend, and between the codend and the inner codend were statistically significant (bigger individuals were retained in the cover and codend when compared to the inner codend). Blue whiting and Norway lobster were the only species displaying clear length-selection patterns, with escapement through the exit hole increasing for larger individuals (Fig. 9a and b). For the Norway lobster, the length of 50% retention (L50 ) is 49.8 mm and the selection range (SR) is 9.1 mm. Comparing these parameters with those obtained from previous codend experiments carried in Portuguese waters, in both research vessels (Campos et al., 2002) and commercial vessels (Fonseca et al., unpublished data2 ), the much higher selectivity 2 Fonseca, P., Campos, A., Garcia, A., Cardador, F., Meixide, M., Padin, A., Theret, F., Meillat, M., Morandeau, F., 2000. Trawl selectivity studies in region 3. Final Report. Commission of the European Communities, DG Fisheries-C-1, Study Project No. 96/06, 178 pp.

– – – – 0.0992NS 0.8298NS (21.3; 21.4) 0.0005 <10−4 (19.8; 21.4) 0.2902NS 0.1921NS (43.6; 37.5) 0.5362NS 0.8399NS (37.1; 37.5) <10−4 <10−4 (43.9; 37.4) <10−4 <10−4 (35.9; 37.4)

<10−4 <10−4 (23.3; 22.3) <10−4 <10−4 (23.4; 22.3)

of the grid is striking. Considering the selection factors of 0.64 and 0.41 found in the former trials for square and diamond 55 mm mesh codends, respectively, the present results correspond to a 78 mm square mesh or a 121 mm diamond mesh codend. For blue whiting, the grid retention (=escape) pattern suggests that a high proportion escape through the upper opening without even entering into contact with the grid. In fact, even the smallest sized individuals were found to escape in a proportion greater than 0.6 indicating that this species exhibits strong avoidance behaviour regarding the grid. Therefore, instead of adjusting a contact-selection curve as for Norway lobster, an overall probability of contact with the grid was modelled and estimated (π = 0.34), and thus the curve displayed in Fig. 9b is actually an available-selection curve. The contact curve, modelling the selection of those fish effectively contacting the grid is also shown embedded. The selectivity parameters for the latter, L50 = 28 cm and SR = 5.0 cm are close to those obtained by Campos et al. (2003) using 55 mm square-mesh codends in the same fishery.

K–S t K–S t Codent vs. inner codend

Cover vs. codend

2002 Cover vs. inner codend

– – – – <10−4 <10−4 (21.3; 20.1) <10−4 <10−4 (21.3; 19.8) 0.1597NS 0.0972NS (43.6; 37.2) 0.1331NS 0.0859NS (43.6; 37.1) 0.1525NS 0.2872NS (23.2; 23.0) 0.1850NS 0.1758NS (23.2; 23.4) <10−4 <10−4 (43.9; 36.7) <10−4 <10−4 (43.9; 35.9) K–S t K–S t 2002 Cover vs. (codend + inner codend)

<10−4 <10−4 (30.9; 25.6) <10−4 <10−4 (23.8; 22.7) 0.0104NS 0.0044 (52.3; 45.4) 0.1252NS 0.0658NS (24.0; 24.3) <10−4 <10−4 (45.0; 38.2)

Blue whiting Red shrimp Rose shrimp Norway lobster Test

K–S t 2001 Cover vs. codend

Hake

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

Survey

Table 4 Results from the Kolmogorov-Smirnov (K–S) and Student’s t (t) tests comparing the length frequency distributions and mean size (in brackets) between the different compartments (codend, inner codend and cover) during the 2001 and 2002 surveys (NS—not significant at 0.01)

234

Fig. 9. Estimated size-selection curves: (a) contact-selection curve for Norway lobster (Richards’ model); (b) available-selection curve for blue whiting (probability of encountering the grid π = 0.34). Embedded: the contact-selection curve for blue whiting (corresponding to the selection of those fish entering into physical contact with the grid).

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

Fig. 10. Linear regression describing the proportion of the total number of Norway lobster entering the trawl retained by the inner codend (2002).

Simple regression analysis was carried out to evaluate the influence of the catch size, both in number and weight, on the escape and retention proportions. All regressions were non-significant with the exception of the proportion of Norway lobster retained in the inner codend (Fig. 10), which was found to be positively correlated with the number of individuals entering the trawl (r2 = 0.61).

4. Discussion The catch data clearly confirms the multi-species nature of the Portuguese crustacean-trawl fishery, where a large fraction composed of by-catch species creates both management (capture of commercial fish species) and ecological (high discard rates) problems. These questions may be partially addressed via the adoption of appropriate gear modifications, such as the use of sorting grids. The results obtained from the experimental fishing trials confirm that by making use of behavioural differences between fish and crustacean species, it is possible to efficiently separate one group from another. Fishes are much stronger swimmers than crustaceans, which are most often taken passively through the trawl. Therefore, while fish may display a more or less strong reaction to an obstacle, such as a grid or a panel, being guided without even entering into contact with it, the selection of crustaceans is essentially mechanical. Norway lobster is a benthic species and, although it displays an escape reaction when touched by a component of the gear rigging, after being caught it is carried more or less passively along the bottom panels of the trawl (Newland and Chapman, 1989). Data from

235

the 2002 survey supports this knowledge since about 50% of the animals were retained by the inner codend. Consequently, it is crucial that the grid construction includes a lower bar-less zone in order that the maximum possible number of individuals may enter directly into the codend without being size-selected by the grid. The positive and significant relationship between the catch in numbers and the proportion of those passing through the lower opening of the grid is a further sign of the importance of this construction detail. Shrimps, on the other hand, are distributed in the water column, probably entering the trawl at different heights. Only approximately 26% of the rose shrimp and red shrimp were retained in the inner codend, with the majority contacting the grid and crossing it. In contrast, the retention proportion of fish by-catch species in the inner codend was very limited, not exceeding 7% for blue whiting. The consistence of the mean proportion of escape (over 70%) in both experiments for this species, suggests a very active behaviour which is in agreement with what was previously observed by Campos and Fonseca (2004) when testing the selective properties of oblique mesh panels and squaremesh windows. The contact-probably (with the grid) estimated from data (π = 0.34) provides an idea of the overall magnitude of the avoidance behaviour (1 − π) exhibited by this species, although it is most probable that it changes with size, being higher for larger fish. The behaviour of the different species inside the trawl and their reaction to the components of the selective system are crucial for its performance and simultaneously establishes the limits for further improvements. The escape proportion (in weight) of blue whiting was similar for both surveys, while in the 2002 survey it increased for all the other species. The reason for this increase in escape percentage for boarfish (47–63%), Norway lobster (8–15%), rose shrimp (4–9%) and red shrimp (7–10%) is not straightforward. Total catches were generally much lower during the second survey because of a lower number of hauls. Mean sizes were only slightly lower in 2002, except for red shrimp where the difference in carapace length was about 8 mm higher. Looking for a possible explanation for the observed differences between trials, the relationship between the escape proportions (both in weight and in number) and grid angle was investigated for rose shrimp, Norway lobster and blue whiting, for which haul by haul

236

Table 5 Experiments with standard Nordmøre grid sorting systems in different crustacean fisheries to exclude fish by-catch and/or undersized crustaceans Sorting grid and bar spacing

Target species loss

By-catch reduction

Reference

Norway Svalbad

NS, 17–21 mm NS, 19 mm

Pink shrimp: 2–5% ≥MLS Pink shrimp: 3–9%

Isaksen et al. (1992) Larsen (1996)

Canada Australia/New S. Wales Australia/New S. Wales Argentina Canada/Quebec Argentina/San Jorge Gulf North Sea, Fladen ground

NS, 28 mma NS, 20 mm NS, 20 mm NSD, 47/25 mmb NS, 19 mm NSD, 47/25 mmb NS, 47 mm NS (U + L), 19 mm

Pink shrimp: 2.9% School prawn: 10% School prawn: no effect Patagonia red shrimp: 0–14% Pink shrimp: 2% Patagonean red shrimp: 47% Patagonean red shrimp: 8% Pink shrimp: no effect, Norway lobster: 18% ≥MLS

North Sea

NS, 12–14 mm

Brown shrimp: 13% ≥MLS

U.K./East coast Sweden/Skagerrak-Kattegat

NS, 12 mm NS, 35 mm, plus square mesh codend NSL, 25 mm

Brown shrimp: 8.7% Norway lobster: 0–9%

Cod: 17–48% Redfish: 34–84%, Greenland halibut: 80–100%, Dab: 65–83%, Polar cod: 5–43% Fish bycatch: 72–83% By-catch: 76% By-catch: 77% Argentine hake: 80–98%, rays: 100% By-catch: 97% Argentinean hake: 77% Argentinean hake: 61% Pink shrimp: 54%
South Portugal

Rose shrimp: 3.9–9.1% Norway lobster: 8.0–14.9%, red shrimp: 6.9–9.6%

Hickey et al. (1993) Broadhurst and Kenelly (1996) Broadhurst et al. (1996) Ercoli et al. (1997) Brothers (1998) Pettovello (1999) Madsen and Hansen (2001)

Polet (2002) Graham (2003) Mats Ulmestrand (pers. commun.) Present paper

Pink shrimp, Pandalus borealis (+P. montagui, in Canada); brown shrimp, Crangon crangon; school prawn, Metapenaeus macleayi; Patagonian red shrimp, Pleoticus muelleri. NS—standard Nordmøre grid; NSU—modified grid with upper bar-less section; NSL—modified grid with lower bar-less section; NSD—double Nordmøre grid sorting system. a Twenty two and 25 mm bar-spacing also tested, with 24 and 44% losses registered. b Primary/secondary grids.

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

Country and/or area

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

information was available. Analysis was only carried out for 2002 data, where the use of an inner codend made the quantification of those individuals entering directly to the codend possible. This allowed the calculation of the escape proportion of those effectively contacting the grid. No significant statistical differences were found for blue whiting (P > 0.05). However, for crustaceans the relationship was statistically significant for rose shrimp (P = 0.0427) and Norway lobster (P = 0.0063), in weight, and for the latter (P = 0.0169), in number. The inverse relationship observed is in accordance to what has been known from the start of grid research, where the angle of attack has been identified as a key factor (Isaksen et al., 1992). These authors identified an ideal working angle of approximately 45◦ . If the angle is too low there will be an increased loss of crustaceans, if it is too steep the escape of by-catch species will be affected. Previous experiments with standard (or slightly changed) Nordmøre grids (Table 5) were mainly directed towards prawn (shrimp) fisheries, with data on Reptantia species such as Norway lobster being scarce or non-existent. For the latter species, Madsen and Hansen (2001) report a loss of 18% of individuals above the MLS in a fishery targeting the pink shrimp, Pandalus borealis. On the other hand, a maximum loss of 9% was reported during preliminary experiments carried out in the Swedish Norway lobster fishery (Mats Ulmestrand, IMR, Sweden, pers. commun.). Therefore, the results reported here for Portuguese waters are within the range of similar fisheries. Similarly, the percentage loss for shrimps varies considerably, with the majority of studies ranging from 0 to about 10%, which is the range of the present study. The exception are the experiments for the Patagonian red shrimp, where Ercoli et al. (1997) report losses of up to 14%, and Pettovello (1999) recorded a 47% decrease in catch. However, these values are associated with the use of a secondary grid for shrimp size-sorting. Any comparison with previous studies should be considered simply as indicative, since not only are the target species often not the same, but the management problems also differ. The Portuguese crustacean fishery has a further particular feature: it simultaneously targets Norway lobster and shrimp species, and fishers would like to retain commercial fish by-catch. This situation is similar only to the shrimp fishery in the Fladen Ground (Madsen and Hansen, 2001), where pink shrimp (P.

237

borealis) is the target species along with Norway lobster as by-catch. Here, there is also interest in keeping legally sized commercial fish species. In such complex situations, the exclusion of non-commercial by-catch will be inevitably accompanied by significant losses of some of the commercial species. The water flow through the selective system (funnel, grid, top escaping hole) is of major importance, influencing its performance, i.e., the retention of target species and by-catch exclusion. In spite of the importance of this issue, systematic research on the conditioning variables has seldom been undertaken. Riedel and De Alteris (1995) investigated the influence of webbing solidity and exit to entrance area ratio for the funnel, the angle of attack and the bar diameter and shape of the grid, at different towing speeds. According to these authors, ‘the ideal configuration for the Nordmøre system is a funnel that functions as both a concentrator and accelerator’ [of fish] ‘and a grate that does not disturb the water moving through it’ (thus minimizing the rejection of water through the top escapement hole). Data from water-speed sensors, placed at the centre of the headline and incorporated in the grid sensor, indicates that in the current experiments the water speed behind the grid was about 1/3 of the speed in the mouth of trawl (0.34 in 2001 and 0.39 in the 2002). These are much lower values than the 0.70 reported by Isaksen et al. (1992) in their original experiments. The reasons for this difference cannot be categorically explained at this point. They may be partially gear-related (very different designs), but are likely to be associated with the use of small mesh (20 mm) codends in the current experiments instead of the commercially sized ones (55 mm). However, the good results obtained in excluding fish by-catch reduce the concern for a possible general malfunction of the system due to poor water flow through the grid. Consequently, a decrease in the loss of target species may be expected when commercial (larger mesh size) codends are used.

5. Conclusions The potential of the modified Nordmøre grid to exclude fish by-catch was demonstrated. Still, the Portuguese crustacean fishery suffers from the shortcoming of having three target species with partially

238

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239

overlapping habitats, and different morphologies and mean sizes. Consequently, improvement in the performance of the selective system by altering grid dimensions or design may be a problem, since the right configuration for a species may negatively affect the catch of another. While losses of shrimp are undesirable, they are within expected values. On the other hand, the case of Norway lobster (average 15% weight, in 2002) is a cause for some concern. However, these results were obtained for a limited number of hauls where the grid displayed a greater variability in the angle of attack. In 2001, when more intensive sampling took place, both in terms of catches and in the fishing grounds covered, the escape percentage of all crustacean species was considerably lower. Variability in mean size of the catch, species composition, catch volume or gear performance due either to seasonal or area effects is to be expected and needs to be further assessed. Although the technical feasibility of the use of grids in this fishery has been established, it is vital to carry out tests on board commercial vessels, since it has long been recognized that conditions on board research vessels do not exactly match those found in commercial fishing. No less important is to guarantee that fishers may perceive the advantages of the system and have the opportunity to put forward their questions and suggestions. In spite of the overall advantages that can be drawn from the used BRDs, such as reduction in ecosystem impact due to minimization of by-catch, improvement in stock management of commercially valuable by-catch species and eventually also of target species, emphasis must be put on more immediate aspects. Reduction of sorting time on deck and increase in the quality of target species are prime candidates. These may be used to facilitate the eventual acceptance of drawbacks, such as short-term losses and increased time of gear handling. The experience of successful implementation of such devices, of which the development and enforcement of the original Nordmøre grid is a model, demonstrated the importance of the involvement of all stakeholders (fishers, scientists and managers) in this process. Acknowledgements The enthusiastic participation of students from the ‘licenciatura’ degree in Marine Biology and Fisheries

of the University of the Algarve is warmly recognised. In this regard, the assistance of Prof. Manuel Afonso-Dias (Univ. of Algarve, PT) is gratefully acknowledged. The thorough comments provided by one referee and the editor were of extreme value for the revision of the manuscript.Trials carried out in 2001 were partially supported by EU funds within the scope of the study project no. 99/058, “Managing by-catch and discards: a multidisciplinary approach (Bydiscard)”, while 2002 trials were carried out under the national project 22-05-01-FDR-00014, ‘Tecnologias da pesca/Fishing Technologies’.

References Akaike, H., 1974. A new look at the statistical model identification. IEEE Trans. Autom. Contr. AC-19, 716–722. Anon., 1996. Report of the study group on grid (grate) sorting systems in trawls, beam trawls and seine nets. Woods Hole, Massachusetts, USA, April 13–14. Int. Coun. Explor. Sea CM 1996/B:1. Anon., 1998. Report of the study group on grid (grate) sorting systems in trawls, beam trawls and seine nets. La Coru˜na, Spain, April 18–19. Int. Coun. Explor. Sea.CM 1998/B:2. Borges, T.C., Erzini, K., Bentes, L., Costa, M.E., Gonc¸alves, J.M.S., Lino, P.G., Pais, C., Ribeiro, J., 2001. By-catch and discarding practices in five Algarve (southern Portugal) m´etiers. J. Appl. Ichthyol. 17, 104–114. Broadhurst, M.K., 2000. Modifications to reduce by-catch in prawn trawls. Rev. Fish. Biol. Fish. 10, 27–60. Broadhurst, M.K., Kenelly, S.J., 1996. Rigid and flexible separator panels in trawls that reduce the by-catch of small fish in the Clarence river prawn-trawl fishery. Aust. Mar. Freshw. Res. 47, 991–998. Broadhurst, M.K., Kenelly, S.J., Isaksen, B., 1996. Assessment of modified codends that reduce the by-catch of fish in two estuarine prawn fisheries in New South Wales. Fish. Res. 27, 89– 111. Brothers, G., 1998. Estimates of the impact of actual and potential grid usage on discarded levels of non-target species in various fisheries in Canada. Report of the Study Group on Grid (Grate) Sorting Systems in Trawls, Beam Trawls and Seine Nets. La Coru˜na, Spain, April 18–19. Int. Coun. Explor. Sea. CM 1998/B:2, pp. 23–36. Campos, A., Fonseca, P., 2004. The use of separator panels and square mesh windows for by-catch reduction in the crustacean trawl fishery off the Algarve (South Portugal). Fish. Res. 69, 147–156. Campos, A., Fonseca, P., Erzini, K., 2002. Size selectivity of diamond and square mesh cod ends for rose shrimp (Parapenaeus longirostris) and Norway lobster (Nephrops norvegicus) off the Portuguese south coast. Fish. Res. 58, 281–301. Campos, A., Fonseca, P., Erzini, K., 2003. Size selectivity of diamond and square mesh cod ends for four by-catch species in the

P. Fonseca et al. / Fisheries Research 71 (2005) 223–239 crustacean fishery off the Portuguese south coast. Fish. Res. 60, 79–97. Christian, P., Harrington, D., 1987. Loggerhead turtle, finfish and shrimp retention studies on four excluder devices (TEDs). In: Proceedings of the Nongame and Endangered Wildlife Symposium, Georgia, September 8–10. DNR, Social Circle, GA, pp. 114–127. Ercoli, R., Garcia, J., Salvini, L., Izzo, A., Bartozetti, J., 1997. Manual del dispositivo de selectividad de langostino con doble grilla—DISELLA II. INIDEP Informe T´ecnico Interno. Fonseca, P., Campos, A., Feitoria, J., 1998. Square mesh windows experiments in Portuguese waters. Int. Coun. Explor. Sea C.M. 1998/BB:12. Graham, N., 2003. By-catch reduction in the brown shrimp, Crangon crangon, fisheries using a rigid separation Nordmøre grid (grate). Fish. Res. 59, 393–407. Hanna, R.-W., Jones, S.A., 2000. By-catch reduction in an ocean shrimp trawl from a simple modification to the trawl footrope. J. Northw. Atl. Fish. Sci. 27, 227–233. Hickey, W.M., Brothers, G., Boulos, D.L., 1993. By-catch reduction in the Northern shrimp fishery. Can. Tech. Rep. Fish. Aquat. Sci. 1964, vi+41. Isaksen, B., Valdemarsen, J.W., Larsen, R., Karlsen, L., 1992. Reduction of fish by-catch in shrimp trawls using a rigid separator grid in the aft belly. Fish. Res. 13, 335–352. Larsen, R., 1996. Experiments with a new, larger type of fish/shrimp separator grid with comparisons to the standard Nordmøre grid. Report of the Study Group on Grid (Grate) Sorting Systems in Trawls, Beam Trawls and Seine Nets. Woods Hole, Massachusetts, USA, April 13–14. Int. Coun. For the Explor. Of the Sea. CM 1996/B:1, pp. 68–80. Leite, A., Ferreira, C., Fonseca, P., Henriques, V., 1990. Teste de redes de arrasto pelo fundo. Campanha de pesca experimental N/E Noruega – 02080889, 14 a 16 de Agosto de 1989. Relat. T´ec. Cient. INIP, Lisboa (26) Julho 1990, 50 pp., il. Madsen, N., Hansen, K., 2001. Danish experiments with a grid system tested in the North Sea shrimp fishery. Fish. Res. 52, 203–216.

239

Monteiro, P., Ara´ujo, A., Erzini, K., Castro, M., 2001. Discards of the Algarve (southern Portugal) crustacean trawl fishery. Hydrobiologia 449, 267–277. Mounsey, R.P., Baulch, G.A., Buckworth, R.C., 1995. Development of a trawl efficiency device (TED) for Australian prawn fisheries. I. The AusTED design. Fish. Res. 2, 99–105. Newland, P.L., Chapman, C.J., 1989. The swimming and orientation behaviour of the Norway lobster, Nephrops norvegicus (L.), in relation to trawling. Fish. Res. 8, 63–80. Oravetz, C., Grant, C.J., 1986. Trawl efficiency device shows promise. Aust. Fish. 45, 37–40. Polet, H., 2002. Selectivity experiments with sorting grids in the North Sea brown shrimp (Crangon crangon) fishery. Fish. Res. 54, 217–233. Pettovello, A.D., 1999. By-catch in the Patagonian red shrimp (Pleoticus muelleri) fishery. Mar. Freshw. Res. 50, 123– 127. Riedel, R., De Alteris, J., 1995. Factors affecting hydrodynamic performance of the Nordmøre grate system: a by-catch reduction device used in the gulf of Main shrimp fishery. Fish. Res. 24, 181–198. Robins, J.B., McGilvray, J.G., 1999. The AusTED II, an improved trawl efficiency device 2. Commercial performance. Fish. Res. 40, 29–41. Tokai, T., Omoto, S., Sato, R., Matuda, K., 1996. A method of determining selectivity curve of separator grid. Fish. Res. 27, 51– 60. Watson, J.W., 1989. Fish behaviour and trawl design: potential for selective trawl development. In: Campbell, C.M. (Ed.), Proceedings of the World Symposium on Fishing Gear and Fishing Vessels. Marine Institute, St. Johns, Newfoundland, Canada, pp. 25– 29. Wileman, D., Ferro, R.S.T., Fonteyne, R., Millar, R.B., 1996. Manual of methods of measuring the selectivity of towed fishing gear. ICES Coop. Res. Rep. 215, 126 pp. Zuur, G., Fryer, R.J., Ferro, R.S.T., Tokai, T., 2001. Modelling size selectivities of a trawl codend and an associate square mesh panel. ICES J. Mar. Sci. 58, 657–671.