J. Cons. int. Explor. Mer, 44: 148-161. 1988
A comparison of size selection of haddock and whiting by square
and diamond mesh codends
J. H. B. Robertson and P. A. M. Stewart
Robertson, J. H. B., and Stewart, P. A. M. 1988. A comparison of size selection of
haddock and whiting by square and diamond mesh codends. - J. Cons. int. Explor.
Mer, 44: 148-161.
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
When loaded with fish the codend of a fishing net is pulled into a bulbous shape. Fish
escape mainly through the relatively small area of open meshes at the front of the
bulb, and forward of this most of the meshes are stretched and closed. If the codend
netting is hung so that the meshes are square they remain open during towing.
Comparative fishing experiments with trawls and seines to investigate the relative size
selectivity of square and conventional diamond mesh codends for haddock and
whiting are described. Several mesh sizes of each type were tested. It was found that
the 50 % fish retention lengths for square mesh codends were larger than those for
diamond mesh codends of the same nominal mesh size. For haddock, the selection
ranges of square mesh codends on trawls were narrower than those of diamond mesh
codends; on seines the ranges were similar. For whiting, the selection ranges of
square and diamond mesh codends on trawls were similar but wider for square mesh
codends on seines.
J. H. B. Robertson and P.A.M. Stewart: Marine Laboratory, P.O. Box 101, Victoria
Road, Aberdeen AB98DB, Scotland.
Introduction
Codends of many types of trawl have been observed
with remote-controlled underwater television and normally have the shape sketched in Figure la. The netting
is hung so that the meshes are diamond shaped and
elongate under tension. As the codend fills with fish the
end meshes are obstructed, water flow is diverted, and
the end becomes bulbous. In the central part of the
codend the meshes are stretched and closed, preventing
the escape of all but the smallest fish. In the forward
part of the codend the diameter broadens to join the net
and the meshes are more open. Most escapes are seen
to take place from the open meshes at the front of the
bulb. Only a small proportion of the meshes in a codend
are opened widely enough to allow fish to escape. This
was suspected during earlier work on codend mesh selection and was investigated by Beverton (1963), who
used divided small mesh covers on the various sections
of a codend to demonstrate that fish escaped mostly
from the front and rear parts and not from the central
section.
The degree of opening of the meshes of a codend
depends on various aspects of its construction. The authors have observed that both the number of meshes
round the circumference of a codend and the length of
the extension piece joining the codend to the net affect
148
(a)
(b)
>'*' Square mesh
Figure 1. The shapes of (a) a conventional "diamond mesh"
codend and (b) a novel "square mesh" codend, when attached
to a trawl and towed.
enclosing type recommended by Pope el al. (1975), i.e.,
1-5 times the size of the codend in length and width.
There is always some doubt about the use of covers,
since they are capable of obstructing the codend meshes
and preventing fish escape. Small mesh covers on
square and diamond mesh codends were studied by
Stewart and Robertson (1985), who found that the cover recommended by Pope et al. could operate non-obstructively. It has been observed that the diameter of a
conventional codend at its widest point is about the
same as that of a square mesh codend with the same
number of meshes round the circumference. As the
square mesh codend diameter is constant along its
length (see Figure lb), it is wider on average and thus
more likely to be masked by a small mesh cover.
The report of the ICES/ICNAF working groups on
selectivity analysis (Holden, 1971) presents the available data on codend mesh selection for the important
commercial species in the Northeast Atlantic. For each
mesh selection experiment, the available information is
listed on the codends used, the experimental methods,
the catch numbers and the measured values of 50%
retention length, selection range, and selection factor
(see Table 3 footnote for definitions). Examination of
these data confirms earlier work (Beverton and Holt,
1957) indicating that 50% retention length and selection range are proportional to mesh size. There is considerable variance in the data. This may arise in part
from the use of trawls of different design, codends of
different dimensions, different types and thicknesses of
twine in the codends, and different experimental techniques. Considerable effort was expended on investigating the effect of twine type on codend selectivity. Analysis of the accumulated data suggested that for a given
mesh size only polyamide codends gave consistently
higher selection factors, possibly owing to their greater
extensibility. To illustrate the variability in the complete
Table 1. Details of the experimental cruises.
Cruise
Date
Name
Vessel
LOA
HP
Gear
type
No. of
hauls
1
Dec 1982
17
280
Trawl
44
2
Feb 1983
17
410
Trawl
27
3
Sep 1983
17
250
Trawl
35
4
Mar 1984
22
550
Trawl
18
5
Feb 1985
22
550
Trawl
20
6
Dec 1983
24
600
Seine
25
7
Jun 1984
"Harvest Reaper"
FR 177
"Edelweiss"
FR 104
"Heather Sprig"
BCK 153
"Orkney Reiver"
K 49
"Orkney Reiver"
K49
"Boy Andrew"
WK 171
"Boy Andrew"
WK 171
24
600
Seine
46
149
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
selectivity. Increasing the number of meshes round the
codend and lengthening the extension piece both reduce
mesh opening. If selvedge ropes are attached to a codend, some of the tension in the netting is transferred to
the ropes, and under load the meshes are less stretched
and more open.
Codends are imperfect selectors offish. Investigation
of the discard rates of haddock and whiting on Scottish
commercial vessels has shown that this is particularly
true for fish smaller than the permitted minimum landing size. Since most discarded roundfish die, any means
of sharpening the selectivity of the net and codend
would be beneficial. It was suggested by Robertson
(1986) that the selectivity of a codend might be sharpened by hanging the netting so that a larger fraction of
the meshes remains open during towing. A codend constructed with the netting rigged to hang with the bars
parallel and perpendicular to the flow has squareshaped meshes (see Fig. lb). Such a "square mesh"
codend retains its shape under load and the meshes are
not stretched and constricted. The use of netting hung
like this in a codend was first described by Holt (1895),
who reported that "it caught considerably less small fish
than an ordinary codend with mesh of the same size". In
a paper to ICES in 1929, J. Gelder of The Netherlands
described the insertion of a small square mesh section at
the end of the top panel of a four-panel diamond mesh
codend. This codend was held open by a wooden frame
and it was claimed that fewer small fish were retained.
This paper describes a series of comparative fishing
experiments in which the size selectivity of "square
mesh" codends was compared with that of conventional
"diamond mesh" codends. Codends of a range of mesh
sizes were tested on both trawls and seines. Fishing
trials took place on chartered commercial vessels. Small
mesh codend covers were used to retain the fish escaping from the test codends. The covers were of the totally
set of data listed by Holden (1971), 95% confidence
limits can be calculated for the 50% retention lengths
and selection ranges for a given mesh size. For example,
for a mesh size of 80 mm, the 50% retention lengths
were 26-6±6-6 cm for haddock and 30-8±5-7 cm for
whiting. The corresponding selection ranges at the same
mesh size were 5-2±3-8 cm for haddock and 5-8±3-5 cm
for whiting.
To be more selective than a conventional codend, for
a given mesh size, a novel codend would require a
narrower selection range at the same 50% retention
length. The objective of the exercise described here was
to measure and compare the 50% retention lengths and
selection ranges of the two types of codend over a range
12 -
of mesh sizes. The catches consisted mainly of haddock
(Melanogrammus aeglefinus) and whiting (Merlangius
merlangus), and only data on these species are presented.
Materials and methods
Cruises
The work took place between 1982 and 1985. Table 1
gives for each cruise the date, the name, type and size of
the vessel, the type of fishing gear used, and the total
number of hauls.
Haddock
12 r
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
a) trawl
Whiting
a) trawl
c
CD
O
OJ
X)
8
0
8
e
L-, 0-0-0-
16
24
32
40
48
8
o 32
QJ
cn
a
a>
o
b) seme
o
u
12
o
b) seine
QJ
cn
24
_g
"c
CD
Q_
Q)
f i
16
QJ
CL
0
16
40
24
32
Fish length (cm)
Figure 2. The length-frequency distributions of the total
catches of haddock retained in diamond mesh (O) a n d square
mesh (D) codends of 80-mm (nominal) mesh size; (a) in a trawl
on cruise 4, (b) in a seine net on cruise 7.
150
16
24
32
40
Fish length (cm)
48
Figure 3. The length-frequency distributions of the total
catches of whiting retained in diamond mesh (O) and square
mesh (D) codends of 80-mm (nominal) mesh size; (a) in a trawl
on cruise 4, (b) in a seine net on cruise 7.
Codends and covers
The codends and covers for the experiments are listed in
Table 2, which gives for each the measured mesh size,
length in meshes, width in meshes, and information on
the twine (Rtex (g/km), single or double, braided or
twisted). Some of the codends were used on more than
one cruise. All were made from polyethylene, mostly
coloured green with a few coloured orange. The mesh
sizes (stretched inside mesh lengths) were measured,
with the netting wet, using a 4-kg ICES gauge. In both
types of codend mesh sizes were measured in the same
way, in the "N" direction of the netting (ISO, 1974),
between opposite corners of a mesh. Direct comparison
of mesh sizes between the types can be misleading, as
the square mesh netting is hung differently from the
diamond. The diamond mesh codends were of the type
typically used in the Scottish fishing industry, i.e., twopanel construction and 120 meshes round at a mesh size
of 80 mm. Practical aspects of the design, construction,
rigging, and use, including liability to damage, of square
mesh codends are described in detail by Robertson
(1986). The covers were attached to the extension piece
three rows ahead of the front of the codend. The square
and diamond mesh codends of each mesh size contained
the same number of twine bars, were the same stretched
length, and had the same number of meshes around the
circumference.
Table 2. Construction of codends and covers.
1
Type
Diamond
Diamond
Square
Square
Cover
2
Diamond
Diamond
Square
Square
Cover
3
Diamond
Diamond
Diamond
Square
Square
Square
5
6
7
Meshes
long
123-5
86-5
89-4
67-9
49
70
136
179
82
120
120
155
27
681
122-3
87-1
87-4
74-9
49
70
139
162
27
681
85-8
99-5
121-8
67-9
74-4
86-0
71
55
50
179
162
141
Meshes
round
Twine
Rtex
Single/
double
Construction
4310
4310
4310
3050
Double
Double
Double
Single
Braided
Braided
Braided
Twisted
464
1720
Single
Twisted
82
120
120
150
4310
4310
4310
3570
Double
Double
Double
Single
Braided
Braided
Braided
Braided
464
1720
Single
Twisted
120
98
88
120
120
120
4310
4310
4310
4310
4310
4310
Double
Double
Single
Single
Double
Double
Braided
Braided
Braided
Braided
Braided
Braided
28-2
681
464
1720
Single
Twisted
Diamond
Square
810
78-4
71
162
120
120
4310
4310
Double
Double
Braided
Braided
Cover
260
656
460
1720
Single
Twisted
Diamond
Square
80-8
80-5
71
71
120
120
4310
4310
Double
Double
Braided
Braided
Cover
28-0
656
460
1720
Single
Twisted
Diamond
Diamond
Square
Square
Square
86-1
79-1
86-1
79-0
60-7
70
77
141
154
200
120
120
120
120
146
4310
4310
4310
4310
4310
Double
Double
Double
Double
Double
Braided
Braided
Braided
Braided
Braided
Cover
26.3
656
460
1720
Single
Twisted
Diamond
Diamond
Square
Square
Square
870
791
861
790
60-7
70
77
70
77
100
120
120
120
120
146
232
232
232
232
232
Double
Double
Double
Double
Double
Braided
Braided
Braided
Braided
Braided
Cover
27-0
656
460
580
Single
Twisted
Cover
4
Measured
mesh size
(mm)
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
Cruise
Note: The codends were constructed from polyethylene, the covers for cruises 1, 2, and 3 from nylon, and the covers for cruises 4,
5, 6, and 7 from polyethylene.
151
Table 3. Selection parameters for haddock and whiting and total catch numbers for each trawl haul with diamond mesh.
Mesh
size
(mm)
4148
7 389
3 249
9681
5331
1014
7 449
3004
2 800
_
3 249
4 486
2181
4 424
1470
1208
1758
2 150
_
_
_
_
_
_
1371
1620
851
1049
1558
1926
2208
5048
_
_
_
_
128
351
239
307
221
157
46
30
87
398
251
781
344
_
_
_
-
1848
1 172
960
1 176
2 335
415
1368
427
361
_
1897
648
425
774
124
113
321
840
_
—
_
_
683
744
463
40
10
47
241
941
_
_
_
_
1 127
1460
1035
1305
1696
1013
793
1 134
891
444
153
3 243
3357
_
_
_
-
5712
7609
1482
12383
1270
_
3316
3726
904
601
2040
888
543
381
504
1872
3056
3297
755
2660
840
1350
1440
1458
1 106
771
—
_
_
_
—
_
568
457
970
864
_
—
_
99
40
27
51
—
_
_
_
_
127
401
773
260
9240
14716
1127
6412
2615
_
3294
3880
352
330
8610
1984
2806
473
3046
323
3701
2077
390
2105
182
92
414
360
2 833
1890
_
_
_
_
_
1781
417
1051
1497
_
_
_
2 156
_
1 118
353
2 579
_
_
_
_
1016
241
775
212
50%
Length
25-0
22-0
20-4
19-4
20-9
19-9
20-6
151
18 9
_
231
26 1
20-4
22-1
20-3
22-8
22-5
20-8
_
_
_
_
22-9
23-8
24-3
15-2
170
18-3
18-2
23-6
_
_
_
_
42-7
42-9
40-5
39-5
41-7
38-6
49-2
60-7
56-1
30-3
27-1
38-5
43-9
_
_
_
-
Haddock
Selection
Factor
Range
5-8
6-5
7-2
8-5
81
9-4
7-5
121
5-3
_
4-3
3-2
12-6
7-5
7-2
4-6
6-6
9-6
_
_
_
_
14-3
6-4
22-0
111
7-3
6-7
11-5
5-3
_
_
_
_
11-3
16-4
13-4
10-6
12-5
9-8
14-6
17-8
24-7
9-7
9-4
7-6
120
_
_
_
-
31
2-7
2-5
2-4
2-6
2-5
2-5
1-9
2-3
_
2-9
3-2
2-5
2-7
2-5
2-8
2-8
2-5
_
_
_
_
_
_
2-6
2-7
2-8
1-7
20
21
21
2-7
_
_
_
_
4-3
4-4
41
40
3-4
3-2
40
5-0
4-6
2-5
2-2
31
3-6
_
_
-
50%
Length
291
29-6
24-9
24-5
28-9
_
26-4
27-1
20-5
21-7
301
29-6
31-5
28-2
33-5
19-9
29-7
25-5
25-4
27-4
20-0
20-7
25-0
24 1
291
29-3
_
_
_
_
_
34-0
25-7
29-8
301
46-9
_
_
38-3
_
36-5
41-7
42-6
_
_
_
_
43-1
26-3
27-2
27-6
Whiting
Selection
Range
Factor
6-3
7-5
9-9
8-2
61
_
7-9
10-4
12-4
10-9
3-7
5-5
3-9
3-5
6-6
10-7
8-7
6-7
5-1
7-6
9-7
5-8
50
60
8-2
10-4
_
_
_
_
_
10-7
11-8
29-2
9-9
11-2
_
_
7-6
_
6-5
141
8-0
_
_
_
_
13-3
11-6
32-5
110
3-6
3-7
3-1
3-0
3-6
_
3-3
3-4
2-5
2-7
3-7
3-7
3-9
3-5
41
2-5
3-7
3-1
2-9
3-2
2-3
2-4
2-9
2-8
3-3
3-4
_
_
_
_
_
_
3-9
3-0
3-4
3-5
4-8
_
_
3-9
_
3-0
3-4
3-5
_
_
_
_
3-5
2-1
2-2
2-2
Definitions of selection parameters:
a) The 50 % retention length of a codend for a particular species is the fish length at which 50 % of the species entering the codend
is retained.
b) The selection range is the length difference between the 75% retention length and the 25% retention length.
c) The selection factor of a codend for a particular species is the ratio of the 50 % retention length and the mean measured mesh
size.
152
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
80-8
80-8
80-8
80-8
80-8
80-8
80-8
80-8
80-8
80-8
810
810
81 0
81-0
810
810
810
82-9
86-5
86-5
86-5
86-5
86-5
86-5
87878787878787878787878798-4
98-4
98-4
98-4
122-3
122-3
122-3
122-3
122-3
122-3
122-3
122-3
122-3
123-5
123-5
123-5
123-5
Catch numbers
Haddock
Whiting
Codend
Codend
Cover
Cover
Table 4. Selection parameters for haddock and whiting and total catch numbers for each trawl haul with square mesh.
Mesh
size
(mm)
1468
1899
1766
488
1344
_
_
_
372
244
_
830
849
2777
2338
1125
535
858
495
4466
2835
619
3776
90
4358
16156
537
1045
570
183
2465
2780
1594
880
980
2210
748
72
2051
391
_
_
576
511
102
586
212
312
842
668
267
461
561
43
138
_
_
_
_
_
_
_
77
61
_
340
928
172
189
187
92
344
99
151
865
429
1036
233
1383
2104
2324
1500
1448
612
484
510
832
992
840
383
253
55
2341
546
_
1036
710
330
1095
377
528
1493
1235
339
1193
1032
494
340
1040
1274
569
134
646
816
3437
448
752
62
426
522
_
_
463
1304
1690
732
438
1806
1310
1568
140
375
331
1340
1302
1813
217
904
907
_
_
_
_
_
47
2714
3094
436
4761
_
701
3636
415
1760
5088
712
5092
2088
1004
2868
_
_
_
_
1215
143
161
139
21
218
133
51
154
160
_
_
7440
1894
1154
1585
253
2043
1436
669
1136
1550
_
74
528
573
78
2047
_
214
9016
133
722
2091
183
2583
903
286
1118
50%
Length
Experimental procedure
During each cruise, the diamond and square mesh codends were used on the same net and exchanged at
convenient intervals, usually daily. The length of each
trawl haul was the same as that normally practised by
the vessel when fishing commercially, i.e., between one
23-8
23-3
22-9
250
26-3
_
Haddock
Selection
Range
Factor
3-6
4-6
5-6
3-4
2-4
_
_
_
—
_
_
26-8
22-2
26-5
29-9
25-3
23-5
26-4
26-7
26-9
26-8
19-5
26-9
27-1
25-2
30-4
27-2
23-8
291
27-1
30-5
29-7
251
19-9
28-3
25-7
29-4
26-4
280
29-2
300
300
_
31 5
30-8
32-9
31-3
30-6
30-9
30-7
30-8
—
50
150
—
6-3
11-8
3-4
5-8
41
3-9
51
3-9
6-3
4-3
6-5
7-3
3-4
5-3
5-8
6-3
11-7
7-9
4-8
5-5
10-0
5-6
8-6
80
4-9
4-7
5-6
4-6
6-2
_
7-6
7-2
3-6
6-4
6-8
5-2
4-8
6-4
3-5
3-4
3-4
3-7
3-9
_
_
_
_
_
_
3-6
30
_
3-5
40
3-4
31
3-5
3-6
3-6
3-4
2-5
3-4
3-5
3-2
3-9
3-5
30
3-6
3-4
3-8
3-7
31
2-5
3-5
3-2
3-4
3-0
3-2
3-3
3-4
3-4
_
3-6
3-5
3-8
3-6
3-5
3-5
3-5
3-5
50%
Length
27-2
26-8
27-7
28-5
27-9
29-5
28-7
30-2
28-6
26-8
_
20-4
31-6
30-6
31-6
30-7
29-4
_
—
_
_
—
29-2
31-6
31-7
33-7
30-5
—
32-3
17-4
31-9
32-5
33-9
34-2
30-3
30-6
32-3
30-4
_
_
—
_
_
_
34-9
29-4
42-1
44-5
33-9
36-6
35-7
35-4
35-9
36-4
Whiting
Selection
Range
Factor
4-5
6-3
5-6
4-3
4-8
30
5-9
2-8
3-2
5-7
_
11-7
7-4
71
5-7
8-6
17-7
_
_
_
_
5-3
50
5-5
6-7
7-6
—
3-4
27-3
70
12-2
121
5-7
6-9
9-3
7-3
110
_
_
_
_
_
_
9-4
5-9
17-9
15-5
5-7
8-9
8-8
70
9-3
9-2
40
3-9
41
4-2
41
4-3
4-2
4-4
4-2
3-9
_
30
4-2
41
4-2
41
3-9
_
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
67-9
67-9
67-9
67-9
67-9
67-9
67-9
67-9
67-9
67-9
67-9
67-9
74-9
74-9
74-9
74-9
74-9
74-9
74-9
74-9
74-9
74-9
78-4
78-4
78-4
78-4
78-4
78-4
78-4
80-5
80-5
80-5
80-5
80-5
81-0
810
810
81 0
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
87-4
Catch numbers
Whiting
Haddock
Codend
Cover
Codend
Cover
_
_
_
3-7
40
40
4-3
3-9
_
41
2-2
40
40
4-2
4-2
3-7
3-8
4-0
3-8
_
_
_
_
_
_
40
3-4
4-8
51
3-9
4-2
41
4-1
4-1
4-2
and three hours. After each haul, the catches in codend
and cover were boxed and sorted separately. Catches of
less than six boxes (40 kg per box, approximately) were
sorted completely into species. Most catches were
larger, however, and a sample of at least six boxes, and
sometimes up to twelve boxes, was taken from both
codend and cover and sorted into species. Sub-samples
153
were then taken from each species group and fish
lengths measured. The size distributions of the measured sub-samples of each species were then raised to
the totals for the whole catches in codend and cover.
different grounds. By comparing the distributions with
the market landing data the year classes in each figure
can be identified as: 2a - years 2 and 3; 2b - years 1, 2,
and 3 plus; 3a - years 2 and 3; 3b - years 1, 2, and
3 plus.
Results
Analysis
The catch data for each haul were treated separately.
The percentage of the catch at each length retained by
the codend was calculated and plotted against length.
Initially, selection curves were fitted by eye, giving
equal weight t o each point. It was found that this produced a bias towards narrower selection ranges in
square mesh codends, particularly for whiting. The
catch numbers at each length vary widely and there is a
relative scarcity of larger fish in the data. Thus the
higher percentage points in many selection curves are
relatively inaccurate compared with the lower points.
This may in part explain the bias found. It was also
noted that when a selection curve covered a length
range with large sample numbers throughout, the curve
was symmetric. A systematic technique seemed to be
required to avoid bias in estimating the selection parameters. The logistic function was chosen as a suitable
symmetric curve to fit the data (Holden, 1971). The
method of fitting used maximum likelihood and thus the
largest samples determined the positions of the curves.
As an illustration of the technique, Figure 4a shows
logistic curves fitted to the data from two hauls on cruise
5. These are for a square and a diamond mesh codend of
the same nominal mesh size. The measured size ranges
Table 5. Selection parameters for haddock and whiting and total catch numbers for each seine haul with diamond mesh.
Mesh
size
(mm)
79797979797979797979797987-0
87-0
87-0
87-0
87-0
87-0
87-0
87-0
87-0
87-0
154
Catch numbers
Whiting
Haddock
Codend
Cover
Cover
Codend
468
_
1311
1442
_
_
653
4 698
_
192
_
479
1561
184
_
_
_
258
1 197
584
2140
_
783
983
_
212
4732
_
1552
_
767
2272
984
_
_
_
924
6000
3030
3 873
1798
944
794
2 324
406
1968
_
3504
4 757
3853
3487
373
636
557
360
1008
826
2328
_
384
-
2 376
13662
3056
740
6538
2969
867
—
636
455
1375
3225
3 370
4 530
1709
32
1279
932
12690
_
3420
-
50%
Length
36-8
26-9
25-8
_
_
19-8
21-9
281
_
230
30-2
37-7
—
_
27-7
28-8
29-4
Haddock
Selection
Range
Factor
13-4
_
6-4
6-7
_
_
5-9
7-5
_
71
—
2-9
8-7
12-7
—
_
8-6
8-3
8-3
4-7
_
3-4
3-3
_
_
2-5
2-8
_
3-6
_
2-9
3-5
4-3
_
_
_
3-2
3-3
3-4
50%
Length
28-3
35-4
33-7
29-8
30-8
32-9
25-4
_
25-2
24-0
30-1
28-6
36-3
330
30-6
19-3
34-6
34-7
40-9
_
29-9
-
Whiting
Selection
Range
Factor
9-8
7-0
5-5
5-7
5-0
5-6
5-8
_
8-9
8-3
5-4
5-5
7-8
6-3
7-4
6-7
81
6-9
12-3
_
7-3
-
3-6
4-5
4-3
3-8
3-9
4-2
3-2
_
3-2
30
3-8
3-6
4-2
3-8
3-5
2-2
4-0
4-0
4-7
_
3-4
-
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
Catches were measured on all hauls in which the codend
was undamaged and at least half a box of haddock or
whiting was taken. It would not be practical to list the
complete set of length-frequency distributions in this
paper. Therefore, to illustrate the nature of the catches,
some representative length frequencies are presented in
Figures 2 and 3, and the total numbers of haddock and
whiting retained in both codend and cover in each haul
are given in Tables 3 to 6, for fish over 14 cm in length.
The size of the catches varied widely. It can be seen
from the tables, however, that average total catch levels
were high. Haddock and/or whiting constituted the major part of the catch in almost all hauls. Figures 2 and 3
present data for haddock and whiting respectively, from
cruise 4 (trawl, 1984, "Orkney Reiver") and from cruise
7 (seine, 1984, "Boy Andrew"). These figures give the
combined length-frequency distributions for all hauls
with diamond and square mesh codends of the same
nominal 80-mm mesh size. To aid comparison, the numbers at each length are expressed as percentages of the
total catch of all sizes retained in each type of codend.
On both cruises the two species sampled were found to
have more than one year class in the length range of
interest. The trawl and seine data in Figures 2 and 3
were collected in March and June 1984, respectively, on
Table 6. Selection parameters for haddock and whiting and total catch numbers for each seine haul with square mesh.
Mesh
size
(mm)
_
_
_
_
_
_
_
_
1359
295
555
1635
748
931
1050
_
380
590
663
_
190
265
182
_
_
_
_
516
876
500
546
1522
_
_
_
_
_
—
—
2 871
848
910
2904
1544
4239
2054
1679
3005
4285
_
3234
7100
4097
_
_
_
_
2744
5 668
322
490
681
4076
5 692
2666
5662
1432
318
2 640
1581
1332
8551
200
_
505
180
_
_
2071
1428
_
492
654
746
1092
_
_
_
293
674
177
2071
767
420
718
697
-
146
519
424
331
611
192
2188
3 749
8173
4 532
2 673
_
5177
680
_
_
3220
3876
_
1935
4316
5439
4187
_
_
_
2550
7209
1529
9824
2 632
6463
3500
3430
-
50%
Length
Haddock
Selection
Range
Factor
_
_
_
_
_
29-1
30-1
29-8
27-6
29-0
31-6
32-3
_
33-5
35-0
34-3
_
44-6
38-1
35-1
_
_
_
_
33-3
30-8
25-9
30-4
27-9
do not fully cover the selection curves, as was frequently
found. The transformed values and the corresponding
fitted straight lines are shown in Figure 4b.
Using this method of analysis, the 50% retention
length, selection range, and selection factor for each
haul were estimated. These parameters and the measured mesh sizes are listed in Tables 3-6 for haddock
and whiting. Not all hauls produced enough of both
species to justify measurement. Tables 3 and 4 present
the data from trawl hauls for diamond and square mesh
codends respectively. Tables 5 and 6 give the corresponding data for seine hauls. At the larger mesh sizes,
the relative scarcity of fish at the sizes around the 50 %
retention length reduced the accuracy with which the
selection parameters could be estimated. In the trawl
data, five haddock hauls and six whiting hauls were
excluded from further analysis. Similarly, in the seine
net data, four haddock hauls and three whiting hauls
were excluded. These hauls are not listed in the tables.
In three of the excluded haddock trawl hauls the 50 %
retention length was below 15 cm with an 87-mm mesh
size, and in the remainder of the excluded hauls the
_
_
_
6-9
9-7
150
7-2
5-6
7-0
11-5
7-5
8-1
6-7
13-8
9-4
9-2
_
_
_
_
5-3
5-3
10-8
8-3
9-5
_
_
_
_
_
3-7
3-8
3-8
3-5
3-7
4-0
4-1
_
3-9
41
4-0
_
5-2
4-4
41
_
_
3-9
3-6
30
3-5
3-2
50%
Length
18-3
25-7
25-3
24-8
28-6
28-7
30-2
31-8
34-9
35-1
35-9
_
40-4
33-0
_
36-3
361
31 -3
35-6
44-1
48-6
44-4
_
_
_
45-3
49 1
49-3
36-9
34-5
47-8
38-4
37-0
-
Whiting
Selection
Range
Factor
8-1
3-2
4-1
3-4
3-5
30
8-6
6-4
6-6
7-1
11-7
_
121
11-9
_
_
17-3
12-9
9-3
7-7
14-2
19-2
19-3
_
_
_
112
14-5
15-6
9-2
7-2
19-6
11-3
10-2
-
30
4-2
4-2
4-1
4-7
4-7
3-8
40
4-4
4-4
4-6
51
4-2
_
_
4-6
4-6
4-0
41
5-1
5-7
5-2
_
_
_
5-3
5-7
5-7
4-3
4-0
5-6
4-5
4-3
-
selection parameters had extreme values, since the 50 %
retention length was outside the measured length range
of fish.
Trawl
In Figures 5 and 6 the 50 % retention lengths and selection ranges for haddock and whiting caught in each
trawl haul are plotted against mesh size. As the data are
scattered, regression lines have been fitted to indicate
the trend in the dependence of the parameters on mesh
size. The regression lines were compared by an analysisof-variance method which examines differences between the slopes and intercepts. For both haddock and
whiting, the 50% retention lengths are higher for
square than for diamond mesh of the same measured
mesh size. The differences are significant at the level of
P<0-001. In the case of haddock, the selection range
was significantly smaller (P<001) for square than for
diamond mesh. For whiting there did not appear to be a
difference in selection range between mesh types. Re155
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
60-7
60-7
60-7
60-7
60-7
60-7
79-0
79-0
79-0
79-0
79-0
79-0
79-0
79-0
79-0
79-0
79-0
79-0
790
868686868686868686868686868686-1
861
Catch numbers
Whiting
Haddock
Codend
Codend
Cover
Cover
Figure 4. Mesh selection data for two hauls on
cruise 5 using square (D) and diamond (O>) mesh
codends of 80-mm nominal mesh size:
(a) Proportion retained (P) plotted against fish
length with logistic curves fitted.
(b) ln[P/(l-P)| plotted against fish length with
lines fitted by maximum likelihood.
1-O
1
0-8
0 6
g
ex
e o-2
0
: : ' . ' • • • •
| - "
10
20
30
Length (cm)
50
0
10
20
30
Length (cm)
50
3
„
2
^
1
£ o
c
~" - 1
-2
-3
moving the outliers from the data and recalculating the
regression lines made no significant difference to their
positions.
Seine
Fewer mesh sizes were tested on seine nets than on
trawls. The 50 % retention lengths and selection ranges
for each haul are plotted against mesh size in Figures 7
and 8 respectively. Regression lines are fitted only as a
crude aid to interpretation. The 50% retention lengths
are again significantly higher (P< 005 for haddock and
P < 0-001 for whiting) for square than for diamond
mesh. In the case of whiting, the selection range appears to be significantly larger for square than for diamond mesh ( P < 0-001) but not significantly different
for haddock.
156
Comparison between square and diamond mesh
The 50 % retention lengths found in square mesh codends were higher than those in diamond mesh, in both
trawls and seines. To catch the same size range of fish,
smaller mesh sizes must be used in square than in diamond mesh codends. In Figures 5 to 8 the square mesh
results cover a lower and smaller range of values of
mesh size than the diamond mesh results. It was not
practical to test sizes of square mesh netting larger than
90 mm as the catch retained would have been negligible, particularly for whiting. To aid comparison between the two types of codend, the selection factors and
selection ranges are shown as histograms in Figures 9
and 10 respectively. In Figure 9, the square mesh selection factors in all cases cover a higher range than those
for diamond mesh. This supports the conclusions drawn
from the regressions on Figures 5 and 7.
In Figure 10, there is considerable overlap between
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
0
26
75
a) Haddock
a) Haddock
trawl
trawl
60
20
45
14
e 30
£
u
lengtn
o
DD
cn
c
p
15
80
95
110
125
48
b) Whiting
trawl
o
D
c
o
~o
0)
65
0
110
125
80
110
95
Mesh size (mm)
125
95
40
b) Whiting
trawl
Q)
oo
80
CO
30
LD
20
32
D
D
10
24
16 L
65
80
95
110
Mesh size (mm)
125
Figure 5. The measured 50% retention lengths (cm) plotted
against measured mesh size (mm) for (a) haddock and
(b) whiting caught in each trawl haul with diamond (O) and
square (•) mesh codends. The lines were fitted by leastsquares regression.
the selection range values for square and diamond
mesh. The mean values and standard deviations of the
selection ranges are given in Table 7. Analysis of variance of the selection ranges shows that the means for
diamond and square mesh are significantly different at
the level of P<0-01 for two cases: trawl-caught haddock and seine-caught whiting. In the former case the
selection range for square mesh is lower and in the latter
it is higher than for diamond mesh. This supports the
conclusions drawn from the regressions on Figures 6
and 8.
65
Figure 6. The measured selection ranges (cm) plotted against
measured mesh size (mm) for (a) haddock and (b) whiting
caught in each trawl haul with diamond (O) and square ( • )
mesh codends. The lines were fitted by least-squares regression.
Effect of catch size
The dependence of 50% retention lengths and selection
ranges on catch size was examined. Catch in the codend
and in the cover, the total catch, and the codend/cover
catch ratio were considered, but no evidence was found
to suggest that the selection parameters were dependent
on catch size.
Discussion
The catch data were highly variable, which complicates
comparison of the square and diamond mesh selection
11 Journal du Conseil 44
157
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
65
o
«
was made to maintain constant conditions during and
between cruises, there was often significant variation in
the selection parameters between consecutive hauls
with the same codend. Changes in environmental conditions (e.g., light intensity), or in gear performance
(e.g., towing speed and net height), may affect gear
efficiency and selectivity (Wardle, 1983). It should be
noted, however, that only those hauls producing extreme values of 50% retention length and selection
range have been excluded from the analysis. It is possible that in some of the hauls producing very low 50 %
retention lengths, despite the efforts made to have a
non-obstructive cover, masking of the codend meshes
was taking place and hindering fish escape. For exam-
a) Haddock
seine
32
E
a
en
c
C
800
82-5
850
16
87-5
a) haddock
o
a>
50
40
seine
12
b) Whiting
seine
D
D
30
£
o
20
o
10
L
56
64
72
80
Mesh size (mm)
88
Figure 7. The measured 50% retention lengths (cm) plotted
against measured mesh size (mm) for (a) haddock and
(b) whiting caught in each seine haul with diamond (<^>) and
square (D) mesh codends. The lines were fitted by leastsquares regression.
77-5
° 20
o
0)
0)
00
800
82-5
850
87-5
B
b) Whiting
seine
15
10
parameters. The parameters found for diamond mesh
codends on trawls are broadly the same as those reported by Holden (1971), which are also very variable.
This would seem to be a characteristic of data from
mesh selection experiments. More hauls to increase the
numbers of estimated selection parameters would have
been useful, particularly with mesh sizes above 90 mm.
Catch rates of haddock above 35 cm and whiting above
30 cm in length were low, however, compared with
those of smaller fish. Consequently, it would have required a very large number of hauls to amass enough
data to estimate 50 % retention lengths in these ranges
with a high level of confidence. Although every effort
158
0L,
56
72
80
Mesh size (mm)
88
Figure 8. The measured selection ranges (cm) plotted against
measured mesh size (mm) for (a) haddock and (b) whiting
caught in each seine haul with diamond (O) and square (•)
mesh codends. The lines were fitted by least-squares regression.
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
o
16
77-5
diamond
Haddock
trawl
Square
30 r
Figure 9. Histograms of the values of selection factor obtained
for haddock and whiting in all hauls with trawls and seines.
Whiting
trawl
20
i
10
10
0
0
1 2 3 4 5 6
1 2 3 4 5 6
Selection factor
diamond
Haddock
square
20 r Whiting
15
„
trawl
E
0 4 8 12 16 20
10
n
0
"0
5
r"
0 1 :
1 "|
0 4 8 12 16 20
i
4 8 12 16 20
u
seine
5
Figure 10. Histograms of the values of selection
range obtained for haddock and whiting in all
hauls with trawls and seines.
trawl
10
0
seine
- r~ !
1 !
•Hi
0 4 8 12 16 20
Selection range (cm)
159
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
seine
Table 7. Mean values of the selection ranges.
Trawl
Haddock
Whiting
Seine
Haddock
Whiting
Mean
s.d.
Square
Diamond
Square
Diamond
603
1001
8-16
9-58
2-45
4-67
4-65
5-77
Square
Diamond
Square
Diamond
8-71
804
10-30
7-12
2-74
2-83
500
1-82
160
Acknowledgements
The authors would like to record their appreciation of
the co-operation and assistance given by the skippers
(James Hay, John Smith, John Goodbrand, Andrew
Bremner, Norman Bremner, and Tom Harcus) and
crews of the vessels chartered for this work and for the
willing assistance of members of the staff of the Marine
Laboratory: Robert Rush, Christopher Whitworth, Wytek Mojsiewicz, Douglas Beveridge, William Leiper,
Kevin Peach, and David Emslie.
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
pie, in Table 3, in some of the hauls with 871-mm mesh,
the catch in the cover is very small.
These results indicate that square mesh codends constructed from the same sheet netting as diamond mesh
codends are likely to have higher 50 % retention lengths
for both haddock and whiting caught in trawls and
seines. For the 50 % retention lengths for each species
to be the same in square and diamond mesh, it appears
that the mesh size of the square mesh codend would
have to be reduced by approximately 10 to 15 mm in the
75- to 95-mm range. The selection ranges found for
haddock in trawls were narrower in square than in diamond mesh, whereas those for whiting were no different. In seines it was found that the selection ranges for
haddock in square and diamond mesh did not differ,
whereas those for whiting appeared to be wider in
square mesh. As fewer mesh sizes were used and fewer
hauls made with the seine net, the results are less certain than those found for the trawl. Thus only in the
case of haddock in trawls do the results suggest that a
square mesh codend will have a sharper selectivity and
retain fewer undersized fish than a conventional codend
with the same 50% retention length.
A fish can escape through a mesh if its body section
can be squeezed through the available mesh opening.
The variation in fish girth at each length (partly related
to stomach fullness) ensures that not all the fish of a
given length will escape. Consequently, the selection
curve will not be knife-edged but sloping, and the girth/
length relationship will impose an upper limit on the
slope of the curve (Beverton and Holt, 1957). The compressibility of a fish body and the tension in the netting
also affect the ease of escape. Jones (1963) considered
the factors determining the shapes of selection curves.
He found by calculation that the girth/length variation
was not in itself enough to account for the low observed
slopes. He suggested that the number of escape attempts, the greater vigour of larger fish in squeezing
through the netting, and the range of mesh shapes in a
codend would also affect the slope of the selection
curve. His calculations suggested that the range in mesh
opening was likely to be the most significant of these
factors. In a square mesh codend there is little variation
in opening. All the meshes are fully open. Thus one of
the factors limiting the selection range should be removed and steeper selection curves should result. The
increase in 50 % retention length found in square mesh
codends may arise because the average mesh opening in
the escape zone of a diamond mesh codend is less than
in a square mesh codend.
The different effects of square mesh on the selection
ranges for haddock and whiting are difficult to explain.
Both species encountered the same conditions in the
codends. If the difference is real, it may arise from
different behaviour patterns in the two species. The
greater area of open meshes in square mesh codends
will only result in improved selection if the fish try to
escape at many places and particularly if they make
repeated attempts. Possibly the species differ in their
level of escape activity, which might account for the
difference. Also the body shapes of haddock and whiting are not the same, whiting being smaller and rounder
in section. This is reflected in the higher 50% retention
lengths for whiting. The more open shape of the square
meshes may present less of an advantage to haddock
than whiting.
It is noted that the seine net selection parameters
differed from those obtained on trawls. In the 80- to
90-mm range of mesh size the 50 % retention lengths
found in the seine catches were higher than those in the
trawl catches by about 6 cm for haddock and 4 cm for
whiting. This may arise from the different capture processes involved. In the trawl, fish are herded continuously in front of the gear. As they become exhausted,
fish of all sizes turn into the net and pass to the codend
(Wardle, 1983). Thus in moderate fishing conditions
there is a steady flow of fish into the codend. In the
seine, however, fish do not enter the net and codend
until a late stage in the haul. Many small fish tire, are
overtaken by the ropes, and do not reach the front of
the net. Further, a seine net is usually towed more
slowly than a trawl, allowing the codend meshes to be
less stretched and more open. The likelihood is, therefore, that fewer small fish are caught and that there is
more opportunity for escape if they do reach the codend.
Rpfprpnppc
Beverton, R. J. H., and Holt, S. J. 1957. On the dynamics of
exploited fish populations. Fishery Invest., Lond., Ser. 2,
19: 533 pp.
Beverton, R. J. H. 1963. Escape of fish through different parts
of a codend. ICNAF Special Publication, 2(5).
Gelder, J. E. 1929. A pamphlet by the Savings Trawl Net Co.
of London presented to the International Council for the
Exploration of the Sea in April 1929.
Holden, M. (Ed.). 1971. Report of the ICES/ICNAF Working
Groups on Selectivity Analysis. Coop Res. Rep. Cons. int.
Explor. Mer, Ser. A, No. 25.
Holt, E. W. L. 1895. An examination of the present state of the
Grimsby trawl fishery, with special reference to the destruction of immature fish. Marine Biological Assoc, Plymouth. New Ser., 3(5).
ISO. 1974. International Organization for Standardization,
Standard Ref. No. ISO 1107-1974(E). Fishing nets-Netting
- Basic terms and definitions.
Jones, R. 1963. Some theoretical observations on the escape of
haddock from a codend. ICNAF Special Publication, No. 5:
116-127.
Pope, J. A., Margetts, A.R., Hamley, J.M., and Akyuz, E. F.
1975. Manual of methods for fish stock assessment. Part 3.
Selectivity of fishing gear. FAO Fish. tech. Pap., No.41,
Rev. 1.
Robertson, J.H.B. 1986. The design and construction of
square mesh codends. Scottish Fish. Inf. Pamph., No. 12:
10 pp.
Stewart, P. A. M., and Robertson, J. H. B. 1985. Small mesh
codend covers. Scottish Fish. Res. Rep., No. 32: 11 pp.
Wardle, C. S. 1983. Fish reactions to towed fishing gears. In
Experimental biology at sea, chapter 4. Ed. by I. G. Priede
and A. G. Macdonald. Academic Press, New York.
Downloaded from http://icesjms.oxfordjournals.org/ by guest on March 6, 2012
161