Reduction of harbour porpoise (Phocoena phocoena) bycatch by iron-oxide gillnets

Fisheries Research 85 (2007) 270–278

Reduction of harbour porpoise (Phocoena phocoena) bycatch by iron-oxide gillnets Finn Larsen a,1 , Ole Ritzau Eigaard a,∗,1 , Jakob Tougaard b,1 a

Danish Institute for Fisheries Research, Department of Marine Fisheries, Charlottenlund Castle, DK-2920 Charlottenlund, Denmark National Environmental Research Institute, University of Aarhus, Frederiksborgvej 399, P.O. Box 259, DK-4000 Roskilde, Denmark

b

Received 10 January 2006; received in revised form 7 February 2007; accepted 12 February 2007

Abstract Reduction of harbour porpoise bycatch by use of high-density iron-oxide (IO) gillnets was tested in sea trials in the Danish North Sea bottom set gillnet fishery in September–October 2000. The trials were conducted as a controlled experiment with conventional gillnets as the control group. Eight porpoises were caught in the control nets and none in the IO nets, a highly significant reduction (P < 0.01). Of the four fish species analysed only catch rates of cod (Gadus morhua) were significantly (P < 0.01) different between the two net types, with CPUE in the IO nets being ca. 70% of the CPUE in the control nets. Subsequent investigations in seawater tanks revealed that the difference in acoustic target strength of the two net types was not significant and that the nets behaved similarly under various water flow conditions. Based on laboratory tests of twine samples and analyses of catch composition we conclude that it is the mechanical properties of the IO nets, primarily the measured increase in stiffness, that are the main reasons for the differences in catch rates for cod and for porpoises between IO and conventional nets. © 2007 Elsevier B.V. All rights reserved. Keywords: Acoustic detectability; Bycatch reduction; Echo-location; High-density gillnets; Phocoena phocoena; Target strength

1. Introduction The documentation of high bycatches of small odontocete cetaceans in various gillnet fisheries in the last two decades has led to the development of different types of acoustic alarms (pingers) whose function is to deter animals from nets, thereby reducing bycatch. A number of trials in commercial fisheries have shown that pingers can indeed reduce bycatch considerably (Kraus et al., 1997; Larsen, 1999; Gearin et al., 2000; IWC, 2000; Barlow and Cameron, 2003). However, pingers are active electronic devices, and as such they have a number of disadvantages, including the need for a continuous source of energy and sensitivity to physical impacts. In addition, concern has been expressed about the effects of widespread pinger deployment on target as well as non-target species and potential habituation by cetaceans to the alarm signals (IWC, 2000; Cox et al., 2001). Despite this, large sums have gone into developing and testing pingers, and pingers are now routinely used in a number of fish-

∗ 1

Corresponding author. Tel.: +45 33963388; fax: +45 33963333. E-mail address: [email protected] (O.R. Eigaard). Authorship equal.

0165-7836/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2007.02.011

eries, e.g. in the California drift gillnet fishery and in bottom set gillnet fisheries along the US east coast and in the North Sea (Rossman, 2000; Larsen et al., 2002; Barlow and Cameron, 2003). Comparatively little effort has been invested in modifying the acoustic properties of conventional nets, to increase their detectability to echo-locating odontocetes. A number of trials were conducted in the 1980s, but they were largely unsuccessful. Either they failed to take the acoustic capabilities of the odontocetes in question into account or enhancement of acoustic detectability had severe side effects, namely reduced catches of the fish target species (Perrin et al., 1994). Increased acoustic detectability has a number of advantages relative to pingers, of which the most important are: (a) habituation is irrelevant; (b) no noise pollution; and (c) no need for an energy source. However, reducing bycatch by increasing the detectability of nets rests heavily on the unproven assumption that odontocetes are entangled because they fail to detect the nets. There can be several possible reasons for an animal failing to detect nets: (i) the animals do not use their sonar to scan for obstacles sufficiently often (or fail to pay attention, even though emitting sonar signals); (ii) animals orient themselves so the net is out of the sound beam (e.g. when bottom feeding vertically) and drift

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

sideways into the net; (iii) echoes from the nets are masked by echoes from free swimming or entangled prey in and around the net; or (iv) the net itself is not detectable by the odontocetes at a sufficiently large distance to avoid entanglement. In the two former cases enhancing the detectability alone will not reduce bycatch whereas it could have a beneficial effect in the two latter situations. It is also possible that porpoises are well aware of the nets but do not perceive them as a hazard, in which case increasing detectability of the nets may not reduce bycatch. Studies of detection distances for porpoises and delphinids suggest that they are capable of detecting regular gillnetting (Au, 1994; Kastelein et al., 2000; Villadsgaard et al., 2007), although the detection distance can be quite short, particularly for porpoises, depending on factors such as ambient noise level and angle of incidence, as well as the net itself and attached material (floats and lead-lines). If odontocetes are entangled because they don’t perceive the nets as a hazard it could be because the echo from the nets is not sufficiently strong, in which case enhancing the detectability again could reduce bycatch. It seems from the above that there are good reasons to develop and test nets with increased acoustic reflectivity, and that controlled experiments with such nets could help in the choice between the competing theories on why odontocetes become entangled in gillnets. In the late 1990s a private manufacturer developed a highdensity monofilament, where a metal compound is added as filler in the polymer to increase the acoustic reflectivity and thus the detectability for echo-locating odontocetes. Nets made from such monofilaments were tested in the Bay of Fundy, Canada, in 1998 and 2000 showing reduction in bycatches of harbour porpoise (Phocoena phocoena) when compared to control nets. The sea trials also demonstrated unaltered catches of four primary target species. These results were ascribed to different target strengths between the acoustically modified monofilaments and the control nets of standard monofilament (Trippel et al., 2003). However, Cox and Read (2004) reported that the effect was more likely caused by a difference in stiffness of the nets, therefore the mechanism of bycatch reduction in this experiment is not clear. In the North Sea, the documentation of high bycatches of harbour porpoises in the Danish bottom-set gillnet fisheries (Vinther, 1999; Vinther and Larsen, 2004) led to the formation of the Danish action plan to reduce bycatches of porpoises in the North Sea (Ministry of Environment and Energy, 1998). The action plan recommends pingers as one of the principal mitigations measures, but also recommends that alternative measures be investigated. The high-density gillnets described above appeared to be an interesting alternative to pingers, and a trial was conducted in the commercial fishery for cod during the autumn of 2000. The objective of the trial was to determine if the high-density nets had a lower bycatch of porpoises than the conventional nets used in this fishery. Following this trial the mechanism of reduction of bycatch was further explored during the autumn of 2001 by observing the nets under controlled conditions in a flume tank, and by measuring the acoustic target strength of the nets. The results of the sea trial as well as the subsequent studies are presented here.

271

2. Materials and methods 2.1. Sea trials The sea trial was designed as a controlled experiment to compare conventional gillnets with the IO gillnets. To simplify interpretation of results, the IO nets were to be manufactured to the same specifications as the control nets regarding twine size, stiffness, colour, and mesh size. The only expected differences between the nets would be in acoustic reflectivity and in specific gravity. The higher specific gravity was attained by the addition of 20% IO to the polymer from which the net twines were made. The difference in specific gravity was to be compensated by an equivalent increase in the number of floats attached to the IO nets. However, when the IO nets arrived from the manufacturer they differed also from the control nets in colour and in stiffness/flexibility. It was decided to continue with the trials despite knowing that the interpretation of the results would be compromised. The specifications of the two types of nets used in the trials are given in Table 1. As a measure of twine stiffness we used the E-module, an international standard for stiffness. In this case we used the E-module for longitudinal stiffness (E-alpha), calculated from the relationship between elongation and applied force. This parameter proved to be substantially different between the two net types as seen in Table 1. The fishing gear used in the trials consisted of strings (each containing ca. 50 individual gillnets of 60 m length tied together) of either control or IO nets. The design required that comparative hauls included approximately equal numbers of control and IO nets fished within a restricted area in time and space in order to minimize as far as practically possible the natural variation in species availability between hauls. A commercial fishing vessel typical of the Danish North Sea gillnet fleet was chartered to conduct the experimental fishing. The RI324 (“Ingrid Frich” of Hvide Sande), a 45.39 GRT vessel, was used for all sea trials to eliminate between-vessel variation in the experiment. An independent observer was on board the vessel for the duration of the experiment. The principal tasks of the observer were collection of information on gear type, fishing effort and bycatch of cetaceans. In addition the observer measured total weight and size distributions by species for each net string. This was done for all catches including discards.

Table 1 Specifications for the two types of nets used in the trials

Twine size Float distance Hanging ratio Twine colour E-module Mesh size Acoustic reflectivitya Specific gravitya a

Control nets

IO nets

0.59 mm 2.46 m 0.4 Silvery green 784 MPa 156 mm – –

0.58 mm 2.16 m 0.4 Reddish brown 2617 MPa 156 mm +13 dB relative to control nets +11% relative to control nets

According to manufacturer.

272

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

2.2. Data treatment

2.4. Target strength measurements

Catch per unit effort (CPUE) for porpoises and the four fish species most frequently caught were analysed on a trip basis. Each trip constituted a paired sample of catches from iron-oxide nets and from control nets. To take into account potentially large between-trip differences in species availability the multiplicative model below was used for analysing the variation in the fish CPUEs:

Target strength was measured on one of each of the two different net types. Measurements took place in a seawater filled concrete tank of sufficient size to allow nets to be fully submerged and stretched as they would be if bottom-set for fishing. Transmitting and receiving hydrophones were placed at midwater depth on a line perpendicular to the net (0◦ incidence angle) and at an approximate distance of 0.5 m. The sound used was a previously recorded harbour porpoise signal (duration 200 ␮s, peak frequency 140 kHz) stored in a computer and digitized at 400 ksamples per second. The signal was transmitted through a Sonar Products HS150 spherical hydrophone. Echoes were measured by a Reson TC4032 spherical hydrophone, amplified and recorded by the computer (12 bit, 400 ksamples per second). The results were confirmed in a different set-up, located in a sheltered corner of a harbour (Kerteminde, Denmark). This setup offered better positioning of transducers due to an increased source level of the transmitted signal. Nets were hung vertically from a horizontal bar placed just above water level and were held down by a lead line. Measurements were done only on the net material itself, i.e. lead and float lines were kept well outside the beam of the transmitting hydrophone. Transmitting and receiving hydrophones were placed approx. 2 m from the net, 1 m below the surface. Sounds were 200 ␮s pulses of 140 kHz sinusoidal signals and projected by a Reson TC2130 transducer. The TC2130 transducer has a beam width between the 3 dBdown points of approximately 30◦ (±15◦ ) both in the vertical and horizontal planes. This directionality closely matches the directionality of the harbour porpoise sound producing apparatus (Au et al., 1999). Echoes were received by a Reson TC4032 hydrophone and monitored on oscilloscope. Target strength was calculated according to Urick (1983) from the ratio of echo sound pressure measured 1 m from the net to sound pressure of the ensonifying signal at the net:   SPLecho @1 m . TS = 20 log SPLsignal@net

CPUEg,t = Gear × Trip × εg,t where CPUE is given as number of fish per km-days, representing the length of nets fished multiplied by the time fished measured in days (soaking time/24 h). εg,t designates the statistical noise in the model. The two indices mark the effect from the gear (control net and IO net) and from the trip for each sample and are treated as class variables in the statistical analyses. The model was linearised by log transformation: ln(CPUEg,t ) = Gear + Trip + εg,t where εg,t ∼ N (0, σ 2 ), an uncorrelated random variable. This linearised model was applied to catch data for the 4 species most frequently caught and subsequently submitted to statistical analysis by the SAS software package. The model was then applied to the same catch data but with catch weight rather than catch numbers being used as a numerator of CPUE. The statistical analyses of porpoise catch data included testing the difference between the two net types by analyses of variance within and between the two means for each of the two types of net. In this case it was assumed that the porpoise availability was constant between the six trips. 2.3. Flume tank experiments To clarify whether differences in the behaviour of the two types of gillnets during fishing could induce differences in the catch of target species and porpoises, a series of measurements were made in a seawater flume tank at the North Sea Centre in Hirtshals, Denmark. The experimental design was to submit the two net types (reduced to a length of 34.5 m, a height of 2.5 m and mounted with buoyancy and weight as when fishing) in turn to increasing currents from zero to approx. 1 knot (0.514 m/s) and to observe and measure changes in behaviour of the nets. We were primarily interested in the degree of “collapse” of the nets, i.e. the reduction in net height with increasing water velocity. The net height was measured with a fixed camera at water velocity intervals of 0.25 knot, after the nets had stabilized upon changing the velocity. After the measurement at 1 knot the velocity was reduced to 0 knot for a measurement of the “recovery” of the net height. In addition to the observations of the sensitivities of the two net types to different water velocities, a detailed description and comparison of the rigging of the two types of nets was undertaken with respect to, e.g. buoyancy and hanging ratios (Table 1).

Measurements were averaged over six echoes from each of the nets. 3. Results 3.1. Sea trials The trials were conducted in ICES subdivision IVb in the North Sea during September–October 2000, and included six trips each of approximately three days of fishing (Fig. 1). The total length of net fished (approx. 100 km of each net type) was only half the effort originally considered necessary to determine whether the IO nets had a lower bycatch of porpoises than the control nets. This was due to a premature termination of the sea trials following indications that the stiffness of the IO nets caused a considerably lower catch of the target species (cod) compared to the controls. If stiffness was the reason for the lower catch, it could also be the reason for the observed

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

273

Fig. 1. Map of The North Sea with location and effort of each set on the six trips. Black lines are IO nets and grey lines are control nets. Trip 1 is enlarged to give an example of how effort was distributed between the two net types.

lack of bycatch of porpoises in the IO nets (see Table 2). If so, there was no reason to continue the sea trials, since these would not enable us to determine if the reduced bycatch of porpoises was caused by the increased density of the IO nets or by their increased stiffness. During data verification we discovered that the vessel in a number of cases had set conventional nets only on wrecks, i.e. without setting a comparable number of IO nets on wrecks as well. Because catch rates for cod are typically much higher in wreck sets, these sets were isolated by their positions and limited effort (typically only 8 or 12 nets in a string) and excluded from further analyses. Each net string included in the data analyses had an average total length of 2.8 km ± 1.4 km (S.D.) consisting of 46 ± 24 (S.D.) repetitions of 60 m long net units of either IO or control net fixed between two anchors and two buoys. All net strings

had an average soak time of approx. 16 h ± 4 h (S.D.) and were set at average depths of 63 m ± 28 m (S.D.). In Fig. 2 the effort with each of the two net types is compared. Table 2 presents the total effort on a trip basis as km–days, representing the length of nets fished multiplied by the time fished measured in days. Also presented is the total catch of porpoise and cod. In the following, all CPUE values refer to a standardised effort of 1 km and 24 h soak time. The CPUE for porpoise and the four species most frequently caught is given on a trip basis in Table 3. A total of 8 harbour porpoises were by-caught during sea trials. They were all caught in the control nets and distributed fairly evenly among trips and sets, with six individual sets each catching 1 porpoise and only one set catching 2 porpoises. Porpoise CPUE for the control nets varied between 0.09 and 0.20 except for trip 4 where no porpoise were by-caught. Average porpoise

Table 2 Effort and total catch (in numbers) of porpoise and cod for each of the six trips (IO = IO nets) Trip

1 2 3 4 5 6

Year 2000

Net strings

Km-days

Fishing dates

IO-nets

Control

IO-nets

11–12 Sept. 17–19 Sept. 24–27 Sept. 03–04 Oct. 15–18 Oct. 23–25 Oct.

5 5 7 4 8 4

5 7 6 4 8 5

13 8 14 6 16 11

Porpoise bycatch

Cod catch

Control

IO-nets

Control

IO-nets

Control

12 9 8 5 15 12

0 0 0 0 0 0

1 1 1 0 3 2

78 96 204 87 159 61

91 181 112 149 217 95

274

F. Larsen et al. / Fisheries Research 85 (2007) 270–278 Table 4 The ratio between catches in numbers (the ratios for CPUE in weight are given in parenthesis) in control nets and IO nets as calculated by the GLM procedure. P-values are found by least square means

Cod Saithe Haddock Plaice

Fig. 2. Average values and standard deviations of soaking time, length of net string and depths fished for each set in all 6 trips.

CPUEs for IO and control nets were 0 and 0.12, respectively, and are significantly different (ANOVA, P = 0.002). The fish CPUEs (in both numbers and weight) for the two net types proved to be significantly different (P < 0.05) only for cod although haddock (Melanogrammus aeglefinus) and plaice (Pleuronectes platessa) showed tendencies for differences in weight CPUE. According to Table 4 the catch rate for numbers of cod was ca. 1.5 times higher in the control nets compared to the IO nets and the weight was ca. 1.8 times higher (LSMeans, P < 0.05). Although the statistical results are in agreement, the ratios in Table 4 reveal some discrepancies between the CPUEs given in numbers of fish and those given as total catch weight. It seems that the cod caught in the control nets are larger than those caught in the IO nets. This is more evident when comparing the length–frequency distributions (LFD, Fig. 3) of the catches, which show relatively more small cod, caught in the IO nets and relatively more large cod caught in the controls. The differences for the other three species are less obvious, perhaps due to the fact that rather small numbers of fish were caught. The average lengths in Fig. 4 correspond to the LFDs in Fig. 3. The average lengths of cod and saithe (Pollachius virens) are significantly different (P < 0.05) between control and IO nets, whereas those of plaice and haddock are not. It seems, however, that for all four species the IO-nets catch fish of a smaller average size. In the case of cod and haddock this resulted in the CPUEs being higher for the control nets. For saithe and plaice the IO-

Control nets/ IO-nets

Significant

1.46 (1.82) 0.76 (0.86) 1.27 (1.36) 0.48 (0.47)

P < 0.05 (P < 0.05)

Non-significant

P = 0.54 (P = 0.68) P = 0.26 (P = 0.05) P = 0.12 (P = 0.06)

nets caught considerably more small fish in numbers than the control nets. This resulted in the CPUEs for these two species being higher in the IO-nets. For saithe and plaice none of the catch ratios (numbers and weight) are, however, significantly different from 1. 3.2. Flume tank experiments The flume tank experiments revealed no obvious differences in the sensitivity (the rate of height reduction) of the two net types to water current as judged by visual inspection. The final impression was that the two net types behaved similarly when exposed to water currents, as supported by the data points plotted in Fig. 5. There were no differences in the hanging ratios between the two types of nets. As intended, the IO net had 14% higher buoyancy than the control net, which should compensate the 11% increased specific gravity of the IO net. 3.3. Target strength measurements The target strengths were −53.5 dB ± 2.4 dB (S.D.) for the control net and −57.0 dB ± 1.3 dB (S.D.) for the IO-net, both measured at 0◦ incidence. The difference between nets was not significant (Mann–Whitney, P > 0.05). 4. Discussion The results of this study show a significant reduction in harbour porpoise bycatch in the IO nets compared to conventional nets, but also a significant reduction in catches of cod, the main target species. Differences between the two net types in acoustic

Table 3 CPUE of porpoise and the four most frequent species by trip. CPUE is calculated as the total summed catch divided by the total effort (kilometres of net fished multiplied by soaking time measured in days) Trip

1 2 3 4 5 6

CPUE Porpoise

CPUE Cod

CPUE Saithe

IO-nets

Control

IO-nets

Control

IO-nets

0 0 0 0 0 0

0.09 0.12 0.12 0.00 0.20 0.17

6.1 12.0 14.7 14.3 9.9 5.5

7.9 21.3 13.5 31.7 14.6 7.9

0.4 1.5 0.0 0.5 7.3 0.5

CPUE Haddock

CPUE Plaice

Control

IO-nets

Control

IO-nets

Control

0.4 0.4 0.0 0.4 18.1 0.2

1.8 2.9 0.3 1.8 1.1 2.0

1.7 3.8 0.0 3.0 2.2 1.6

0.7 0.5 5.6 0.2 0.3 0.1

0.2 0.2 2.4 0.0 0.1 0.2

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

275

Fig. 5. Height of IO-net and control net in the flume tank when exposed to water velocities gradually rising from 0 to 1 knot and then rapidly dropping back to 0.

reflectivity, stiffness, colour and buoyancy could all contribute to the different catch rates, but we argue that the mechanical properties of the IO-nets, primarily the measured increase in stiffness, are the main reasons for the differences in catch rates for porpoises as well as for cod between IO and conventional nets. 4.1. Porpoise catches

Fig. 3. Relative length–frequency distribution of the four species most frequently caught. Note that the scales are different on the y-axes.

Fig. 4. Average lengths and 95% confidence intervals of the four most frequently caught species.

The a-priori assumption was that any observed differences in catch rates for harbour porpoise between net types could be explained by the claimed higher acoustic reflectivity of the IO nets. However, we found no significant difference in target strength between the two nets in our study. Our measurements of target strength are in good agreement with previous measurements made on a comparable net (0.57 mm diameter nylon monofilament, 140 mm mesh size, target strength −54.7 dB ± 3.3 dB), also measured using harbour porpoise signals (Kastelein et al., 2000). On the other hand, contrasting results are reported by Trippel et al. (2003). They measured target strengths of nets with barium-sulphate filler which were 4.2–5.2 dB higher than for nets without filler. This was measured with 200 kHz multi-beam sonar and their result is expressed as a relative difference only. Also Mooney et al. (2004) measured increased target strength of barium-sulphate gill nets compared to normal nylon monofilament nets, but only for angles of incidence between 10◦ and 30◦ . The lack of difference at 0◦ angle of incidence could point to other factors than the material itself as responsible for the increased target strength, such as differences in knot geometry. Our experimental nets contained iron oxide, not barium sulphate, and thus strictly we cannot know whether they also have an increased target strength at angles of incidence different from zero, as we only measured target strength perpendicular to the net, but it clearly remains a possibility. The target strength of an object is determined by a number of factors such as size, geometry, material and angle of incidence. The material of the high-density nets was manipulated by the manufacturer in order to increase target strength, relying on the fact that echo strength depends on the difference between the material and water densities, see Urick (1983) and Pence (1986)

276

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

for details. The fact that the more dense experimental nets did not create significantly stronger echoes than did the controls in our measurements and in Mooney et al. (2004) at 0◦ of incidence, can be either because the increased density (11% relative to the control, according to manufacturers information) is not sufficient to produce a measurable effect, or because the effect of an increased density is counteracted by a parallel change in compressibility, which also affects target strength (Urick, 1983, p. 299; Pence, 1986). If porpoises and other odontocetes are in fact entangled because they fail to detect nets in time to evade, then increasing the target strength of the nets seems a viable strategy. Two problems stand out, however. First of all, it is not known whether failure to detect the nets is the fundamental problem for the animals, or at least a major part of the problem. Laboratory studies have indicated that detection may not be the problem per se, but rather the attention of the animal is the key. In the studies of Kastelein et al. (1995) porpoises kept well clear of a gillnet suspended in their pool for a long time, but eventually they were all entangled, perhaps due to lack of attention or distraction from the net. The second problem in manipulating target strengths of the nets, assuming it is indeed an answer to the problem, is that fairly large increases probably are needed in order to produce significant effects. According to the calculations of Kastelein et al. (2000), an increase in target strength of 10 dB (from −55 dB to −45 dB) is needed to increase 90% net-detection distance from 4 m to 7 m for porpoises. The reported increases in target strength of 4–6 dB (at 10–30◦ angle) for the barium-sulphate net result in an estimated increase of detection distance from about 3 m to about 4 m (Mooney et al., 2004). At present it is unclear what caused the substantial reduction in porpoise bycatch in the experimental IO nets. It does not, however, seem plausible that acoustic properties of the net alone can explain this difference, as the two types either did not differ (our measurements) or the detection distance of the iron-oxide nets was only 25% greater at some angles of incidence (inferred from the barium-sulphate nets). Neither do we believe that the four extra floats added to the float line of each IO net, resulting in the interval between floats changing from 2.46 m to 2.16 m, had any effect on the detectability of the nets. Also the general behaviour of the nets when set for fishing was irrelevant, as the two nets behaved in a similar way in the flume tank. There were other differences between the nets, however, most notably the colour and the mechanical properties of the twine, both of which could potentially influence bycatch. The possibility that the reddish-brown IO nets were more visible to the porpoises than the silvery green control nets is clearly present. Although vision has not been studied in detail in harbour porpoise, there is little reason to believe that they should not possess good visual capabilities, as does the well-studied bottlenose dolphin. However, as most fishing occurred at night, the colour is unlikely to be the main explanation for the elimination of bycatch in the IO nets. Although the exact circumstances of entanglement of porpoises in the wild are unknown, studies on animals in captivity indicate that entanglement occurs if the tail fluke, flipper or dorsal fin touches the net (Kastelein et al., 1995). Entanglement

is apparently facilitated by tubercles (small cartilaginous protuberances) often present on the leading edge of all fins. The tubercles effectively catch the net meshes at the slightest touch. It is conceivable that the increased stiffness of the IO net could make it more difficult for the net to cling to the fins of the animal and perhaps make it easier for the animal to escape if caught in the first place. This possibility could be investigated by deliberate entanglement of captive animals under closely controlled conditions, using different types of nets or twine. 4.2. Fish catches The reports received from the observer during the sea trial suggested that catches of target species in IO nets were as much as 30–50% lower than catches in control nets. Added to this were indications that fish caught in the IO nets were generally smaller than in the control nets, and that fewer fish in the IO nets were tangled than gilled compared to fish in the control nets. This information led us to suspect that the reduced catches in the IO nets should be explained by the stiffness of these nets. If stiffness was the reason for the lower catch, it could also be the reason for the observed lack of bycatch of porpoises in the IO nets. The analyses presented here show that the CPUE for cod, the target species, in the IO nets was only around 70% of the CPUE in the control nets. For the other fish species analysed, the differences in CPUE between the two net types were not significant but catches of haddock showed the same tendencies as cod whereas catches of saithe and plaice showed the opposite tendencies. We assume that the reduced cod catches in the IO nets were related to either physical properties of the nets or the behaviour of the nets in the water. Consistent differences in fish abundance in the areas where the two types of nets were used could also explain the observed catch patterns, but after excluding data from wreck fishing (as described in the Section 2), we are confident that the experiment was well balanced, i.e. that IO nets were not consistently used in areas with lower fish abundance (Fig. 1). The physical properties that conceivably could influence fish catches are twine size, colour, specific gravity, buoyancy, hanging-ratio and stiffness as well as mesh size. The data in Table 1 show that there was no difference in mesh size and only a minor difference in twine size. Hanging ratios were the same for the two types of nets. The difference in specific gravity was compensated by 14% higher buoyancy for the IO nets, attained by adding four extra floats to the float line. This should not influence catching efficiency of the net (stated by the manufacturer and supported by the tests in the flume tank). The reddish brown colour of the IO nets makes them very visible to the human eye compared to the silvery green of the control nets, but the experimental fishing was conducted primarily overnight and at depths where light levels in the North Sea are low. We thus exclude colour as a primary reason for the different catch rates of fish. This assumption is supported by Cui et al. (1991), who found that monofilament colour played no role in visibility to mackerel (Scomber scombrus L.) at the light levels found at even moderate depths at night in the North Sea. Furthermore, Wardle

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

et al. (1990) demonstrated that there is not a simple correlation between colour differences in air compared to differences in water, even at high light intensities. There is a clear difference in stiffness between the two net types, and we believe that this could have a large influence on the composition of the catch. The entrapment of the individual fish in the meshes of gillnets is not a homogeneous process (Gulland and Harding, 1961; Regier and Robson, 1966). For multi-monofilament nets there are three major ways cod can be trapped: The fish can be gilled, enmeshed by their maxillae or trapped by their teeth (Hovgaard et al., 1999). Equivalent categories for monofilament gillnets have not been established to our knowledge and neither has the relationship between stiffness of the filament and the capture process. However, we find it reasonable to exclude the teeth category of entrapment, when dealing with large diameter monofilament twine, and furthermore conjecture that fish will not easily enmesh in nets with an increased stiffness. Since enmeshed fish tend to be larger than gilled fish (Hovgaard et al., 1999), catch rates will be reduced in the stiffer nets measured both in numbers and weight. This could explain the observed differences in catch number and catch weight for cod between the two types of nets (Table 4). It was also noted by the observer that the ratio of enmeshed to gilled fish was lower in the IO nets compared to the control nets. With respect to the behaviour of the nets in the water, the test conducted in the flume tank showed little difference between the two types of nets, although there was a slight tendency for the IO nets to be more affected by water currents. The reason for this is probably the additional floats creating more hydrodynamic drag on the float line of the IO nets. However, the final conclusion of the flume tank testing was that the two net types behaved in much the same way, when exposed to a simulated fishing condition, and we found no reason to conclude that the difference in float numbers is important for the catching efficiency of the nets. Hence, we conclude that the most likely explanation for the different catch rates for cod is the difference in stiffness between the two types of nets. 5. Conclusions The IO nets tested in this study were found to have a significantly lower bycatch of harbour porpoises than the conventional nets used as controls. Unfortunately, the IO nets caught less of the target species, to such an extent that they are not a viable mitigation measure. Differences between the two net types in acoustic reflectivity, stiffness, colour and buoyancy could all contribute to the different catch rates, but we conclude that the increased stiffness of the IO nets is the main reason for the different catch rates, and that the difference in target strength is insufficient to have an effect on harbour porpoise bycatch. It is important to note, that a large increase in target strength cannot be obtained without substantial changes in either material properties (density and compressibility) or dimensions of twine. This is likely to change other mechanical properties of the nets with consequent effects on the catch of target species and/or ease of handling for the fisherman. It seems more worthy to explore other strategies for increasing target strength than

277

manipulation of the net material itself. This could be done by using small reflectors, such as glass or metal beads in all or some of the knots of the net. During the 1980s a number of trials were conducted with reflectors of various kinds attached to gillnets (see review in Perrin et al., 1994), but these reflectors were not designed specifically for gillnet use and either failed to reduce the bycatch or had a severe negative effect on the catch of target species. We believe, however, that it may be possible to develop reflectors that could enhance the detectability of gillnets without sacrificing target species catches, but this would require a structured approach, involving insight into the principles of underwater acoustics, fishing gear technology and porpoise biosonar. Acknowledgements We are grateful to the skipper of RI324 “Ingrid Frich” of Hvide Sande for his cooperation with this project and to the technical staff from DIFRES for their work. The results presented here would also not be available without the funding from the Danish Directorate for Food, Fisheries and Agri Business. Jakob Tougaard was funded by the Danish National Research Foundation, Centre for Sound Communication. We are also grateful for the comments and suggestions for improvement received from three anonymous reviewers. References Au, W.W.L., 1994. Sonar detection of gillnets by dolphins: theoretical predictions. in: Perrin,W.F., Donovan, G.P., Barlow, J. (Eds.), Gillnets and cetaceans. Report of international Whaling Commission Special Issue 15. I–ix+629pp., pp. 565–571. Au, W.W.L., Kastelein, R., Rippe, T., Schooneman, N.M., 1999. Transmission beam pattern and echolocation signals of a harbor porpoise (Phocoena phocoena). J. Acoust. Soc. Am. 106, 3699–3705. Barlow, J., Cameron, G.A., 2003. Field experiments show that acoustic pinger reduce marine mammal bycatch in the California drift gill net fishery. Mar. Mammal Sci. 19 (2), 265–283. Cox, T.M., Read, A.J., Solow, A., Tregenza, N., 2001. Will harbour porpoises (Phocoena phocoena) habituate to pingers? J. Cetacean Res. Manage. 3, 81–86. Cox, T.M., Read, A.J., 2004. Echolocation behaviour of harbour porpoises (Phocoena phocoena) around chemically enhanced gillnets. Mar. Ecol. Prog. Ser. 279, 275–282. Cui, G., Wardle, C.S., Glass, C.W., Johnstone, A.D.F., Mojsiewicz, W.R., 1991. Light level thresholds for visual reaction of mackerel, Scomber scombrus L., to coloured monofilament nylon gillnet materials. Fisheries Res. 10, 255–263. Danish Ministry of Environment and Energy, 1998. Action Plan for Reducing Incidental Bycatches of Harbour Porpoises. The National Forest and Nature Agency. Nature and Wildlife Section (j.no. SN 1996-402-0035). Gearin, P.J., Gosho, M.E., Laake, J.L., Cooke, L., Delong, R.L., Hughes, K.M., 2000. Experimental testing of acoustic alarms (pingers) to reduce bycatch of harbour porpoise, Phocoena phocoena, in the state of Washington. J. Cetacean Res. Manage. 2, 1–9. Gulland, J.A., Harding, D., 1961. The selection of Clarias mossambicus (Peters) by nylon gillnets. J. Cons. Int. Explor. Mer. 26 (2), 215–222. Hovgaard, H., Lassen, H., Madsen, N., Poulsen, T.M., Wileman, D., 1999. Gillnet selectivity for North Sea Atlantic cod (Gadus Morhua): Model ambiguity and data quality are related. Can. J. Fisheries Aquatic Sci. 56, 1307–1316. International Whaling Commission, 2000. Report of the Sub-Committee on Small Cetaceans. J. Cetacean Res. Manage., 2 (Suppl.):235–263.

278

F. Larsen et al. / Fisheries Research 85 (2007) 270–278

Kastelein, R.A., de Haan, D., Staal, C., Nieuwstraten, S.H., Verboom, W.C., 1995. Entanglement of harbour porpoises (Phocoena phocoena) in fishing nets. In: Nachtigall, P.E., Lien, J., Au, W.W.L. (Eds.), Harbour porpoises, laboratory studies to reduce bycatch. De Spil Publications, Woerden, Netherlands, pp. 91–156. Kastelein, R.A., Au, W.W.L., de Haan, D., 2000. Detection distances of bottomset gillnets by harbour porpoises (Phocoena phocoena) and bottlenose dolphins (Tursiops truncates). Mar. Environ. Res. 49, 359–375. Kraus, S.D., Read, S.J., Solow, A., Baldwin, K., Spradlin, T., Anderson, E., Williamson, J., 1997. Acoustic alarms reduce porpoise mortality. Nature 388, 525. Larsen, F., 1999. The effect of acoustic alarms on the by-catch of harbour porpoises in the Danish North Sea gillnet fishery. Paper SC/51/SM41 presented to the IWC Scientific Committee Meeting, May 1999 (unpublished), 8 pp. The paper can be accessed at http://www.iwcoffice.org. Larsen, F., Vinther, M., Krog, C., 2002. Use of pingers in the Danish North Sea wreck net fishery. Paper SC/54/SM32 presented to IWC Scientific Committee, April–May 2002 (unpublished) 8pp. The paper can be accessed at http://www.iwcoffice.org. Mooney, T.A., Nachtigall, P.E., Au, W.W.L., 2004. Target strength of a nylon monofilament and an acoustically enhanced gillnet: predictions of biosonar detection ranges. Aquatic Mammals 30, 220–226. Pence, E.A., 1986. Monofilament gill net acoustic study. Report APL-UW-2-86 to the National Marine Mammal Laboratory. Applied Physics Laboratory, University of Seattle.

Perrin, W.F., Donovan, G.P., Barlow, J., 1994. Gillnets and cetaceans. Report of the International Whaling Commission (Special Issue 15), i-ix+629pp. Regier, H.A., Robson, D.S., 1966. Selectivity of gill nets, especially to lake whitefish. J. Fish. Res. Board Can. 23, 423–454. Rossman, M.C., 2000. Effectiveness of time/area closures and acoustic deterrents as harbor porpoise bycatch reduction strategies off the northeast coast of the United States. Paper SC/52/SM23 presented to the IWC Scientific Committee Meeting, June 2000, (unpublished), 10pp. Trippel, E.A., Holy, N.L., Palka, D.L., Sheperd, T.D, Melvin, G.D., Terhune, J.M., 2003. Nylon Barium Sulphate Gillnet reduces Porpoise and seabird mortality. Mar. Mammal Sci. 19, 240–243. Urick, R.C., 1983. Principles of underwater sound, third ed. Peninsula Publications, Los Altos, California. Villadsgaard, A., Wahlberg, M., Tougaard, J., 2007. Echolocation signals of free-ranging harbour porpoises, Phocoena phocoena. J. Exp. Biol. 210, 56– 64. Vinther, M., 1999. Bycatches of harbour porpoises (Phocoena phocoena L.) in Danish set-net fisheries. J. Cetacean Res. Manage. 1, 123–135. Vinther, M., Larsen, F., 2004. Updated estimates of harbour porpoise by-catch in the Danish North Sea bottom set gillnet fishery. J. Cetacean Res. Manage. 6 (1), 19–24. Wardle, C.S., Cui, G., Mosjewicz, W.R., Glass, C.W., 1990. The effect of colour on the appearance of monofilament nylon under water. Fisheries res. 10, 243–253.