ICES mar. Sei. Symp., 196: 12-16. 1993
Fish learning in response to trawl fishing
Alexander I. Pyanov
Pyanov, A. I. 1993. Fish learning in response to trawl fishing. - ICES mar. Sei. Symp., 196: 12-16.
Changes in defensive behaviour adaptive to trawl fishing were studied in some freshwater fish species under natural and experimental conditions. Under natural conditions, repeated hauls of a trawl were made across aggregations of bream (Abramis brama) and individuals were tracked with ultrasonic transmitters. In aquarium experiments, a small scale model trawl and two dip nets of different sizes were used for catching lemon and white tetra (Hemigrammus caudovittatus) and rosy barbs (Barbus conchonius). Conditioned avoidance reactions to the fishing gears were established in the bream under test after one to two hauls. The reaction was weak in young bream and strong in adults. Experienced individuals escaped from the path of the trawler at a distance of about 50 m before it reached them. The results of the aquarium experiments confirmed and augmented those obtained under natural conditions.
Alexander I. Pyanov: Institute of Evolutionary Animal Morphology & Ecology, Leninsky Prospect, 33, Moscow 117071, Russian Federation.
Introduction Investigations under natural conditions Fishing is the oldest anthropogenic factor influencing In this work data on trawl catches of bream (Abramis populations of many fish species. However, to date the brama L.) were used as well as echograms and tracking question of fish behavioural adaptation to this factor has experiment data on individuals tagged with ultrasonic remained practically uninvestigated, although it is well transmitters. These investigations were part of a study in known that defensive fish behaviour significantly influ 1979-1985 on spatial and temporal distribution of bream ences the success of fishing. in Votkinskoe reservoir, an area where commercial Innate unconditioned reactions and learned avoid trawl fishing has never been practised before. ance both influence fish defensive behaviour. It has been A trawler of 180 reg. tons and 150 h.p. engine was experimentally shown that fish as well as higher ver used, equipped with a bottom trawl having 10 mm mini tebrates can easily acquire conditioned reflexes to any mum mesh size and 18 m between otter boards. Re external stimulus for which these animals have corre peated tows through the same fish aggregation were sponding receptors (Frolov, 1925). The speed of acqui carried out at intervals from 1.5 h to 3 days, and tows sition of a simple conditioned reflex is approximately were also shifted to be parallel to each other. The equal in all vertebrates (Voronin, 1954). duration of tows ranged from 5 to 20 min. The distances Having observed fish behaviour during fishing, some between parallel tows varied from 15 to 50 m. Catches scientists raised the possibility that fish might have (fish/hectare) in 34 sets of two hauls and in 13 sets of four conditioned defensive reactions to active fishing gear hauls have been compared. (Golenchenko, 1955; Kuhorenko, 1977). Individual bream were tracked according to the stan It is well known that birds and mammals quickly learn dard underwater telemetry method. “Pinger” transmit to avoid hunters (Severtsov, 1922; Illichov, 1977). This ters 35 x 17 mm were used. Ten bream of 25-48 cm and kind of learning in fish has been demonstrated only as far 0.5-2.8 kg were caught with trawl and gillnets for tag as angling is concerned (Beukema, 1970; Fernö and ging. All 10 tagged fish were released at their place of Huse, 1983), so I report here an attempt to investigate capture. After that, attempts were made to catch them learned avoidance in fish to active fishing gears under in the trawl some time later (from 30 min to 9 days). Two natural and experimental conditions. or three trials were made with each individual. Tracking
12 3.0 -i parallel transects, one of them being shifted some dis tance horizontally from the other, was tested. Catches from two groups of hauls (10 and 19 pairs) were tested; the distance between transects was 8-10 m and 15-25 m s z o respectively (Table 1). Correlation coefficients for log n2 ffl o and log ni in these groups were r = +0.96 (p < 0.01) and •Dc r' = +0.80 (p < 0.01). The difference between these o c groups is not significant (r-test). So, the dependence of CD C/D 0.5 - the catches occurred even when the parallel shift of the second tow was 8-25 m from the first one, though the zone of trawl capture (distance between the trawl 1.0 1.5 2.0 2.5 boards) was 17 m. First catch (log n,) Bream aggregations consisted of fish from 3 to 17 Figure 1. Dependence of the second trawl catch upon the first years old. Analysis of the size and age structure of this one in consecutive trawlings in the same area. species revealed a trend towards a decrease in the proportion of adult individuals in the second trawl data on 20 other individuals tagged in the fishing area catches (Fig. 2). For bream less than 22 cm long (juven and nearby have been analysed. iles), the value of the ratio between second and first Additional data on fish behaviour and distribution catches log (n2/nj) fluctuated about one. during the hauls were obtained with an echo sounder, Thus, as a rule, the first catch from a bream aggre model “Skipper-406” (50 kHz). gation greatly exceeded subsequent catches, both in fish There was no commercial trawl fishing at Votkinskoe quantity and in mean fish length. reservoir; the main fishing gears used by local fishermen The control was as follows. At first, the vessel passed were gillnets and basket traps. Bream aggregations were above a fish aggregation without trawling, and then located in the channel zone of the reservoir at a depth towed the trawl in midwater. After that, consecutive from 8 to 14 m; this channel is the navigation route for hauls were carried out. The changes of catches in the vessels on the reservoir. control (six tests) and in the experiments appeared to be Comparing results of consecutive hauls through the the same. bream aggregations, it was found that the second catch The quantity and distribution of fish, before and after (log n2) was lower than, and depended significantly hauls, were registered in a series of tows with an echo upon, the first one (log ni) (r = +0.81, p < 0.01) (Table sounder fitted on the vessel. On the basis of these data a 1; Fig. 1). No correlation was found between the second catchability coefficient (CC) was calculated: CC = n/ and third catches (log n2 and log n3), nor between the N x 100%, where n is the catch and N is the fish quantity third and fourth catches (log n3 and log n4) (r = +0.27, in the trawl path. Some patterns of catchability changes p > 0.05; r' = +0.31, p > 0.05 respectively). are shown in Figure 3. In the daytime some bream The dependence of log n2 upon log ni regarding two located close to the bottom were not recorded, so the CC in the second example (day after first haul) turned out to be more than 100% (Fig. 3). However, for the present Table 1. Average fish catches in consecutive trawl tows purposes it is not the absolute values that are of import through bream aggregations. ance, but the trend of their changes.
No. of Average no. of fish Haul order tows caught per tow s.e. s.d.
Pair series : First catch 47 118.7 16.8 107.8 Second catch 47 43.4 7.8 50.2
Long series: c First catch 13 127.4 34.3 123.7 CN c Second catch 13 57.7 23.1 73.2 CD - 0.5 - Third catch 13 45.6 14.8 53.4 O _l Fourth catch 14 40.1 12.1 36.2 - 1 .0 - Laterally shifted hauls: 8-10 m First catch 10 178.2 52.8 158.5 - 1.5 Second catch 10 71.5 27.9 83.7 0 10 20 3040 50 15-25 m Fish length (cm) First catch 19 101.0 15.7 71.8 Second catch 19 50.8 10.3 47.1 Figure 2. Change in the value of the ratio of the second catch to the first one in different size groups of bream.
13 120-1 The experiments with white tetra were performed in a tank 500 x 70 x 35 cm. Two groups, 117 and 100 fish, 100 - c 25—15 mm long, from one aquarium, were used. The CD ö 80 - first group was transferred with water from the aquarium *=0 to the tank. Fish in the second group were captured with 8 60 - >< a lift net and with a dip net. Food was used to attract fish into the lift net. In the experiment, white tetra were I 40- « captured with a 1:30 scale model of the 18 m bottom -Co ra trawl. Two series of trials in each group, consisting of O three trials (hauls) at 1.5 h intervals, were carried out at 5 day intervals between series. Food was used to aggre 1 2 3 gate fish in the middle of the tank for trawl fishing. Haul Rosy barb, 589 fish, 25-40 mm length, were raised in Figure 3. Change of the catchability coefficient in consecutive an aquarium 273 x 97 x 45 cm from birth and had never hauls. * - during one day; + - in the next day after the first haul; been caught before the experiment. The fish were cap □ - during one night. Haul number 1 is first, and so on. tured with the same scale model trawl as used for white tetra. In all, three trials at intervals of 1.5 h were carried Defensive schooling behaviour is unknown in bream; out. probably they escape danger individually. Moreover, The velocity of gear movement was about 0.5 m s_1. fish can react to danger by detecting the signals of other Captured fish were not used in further trials. Fish behav escaping fish and their panic movements (Gandolfi et al., iour was recorded with a video camera. 1968). Accordingly, experiments were conducted on Originally the lemon and white tetra were widely scat individual bream tagged with ultrasonic transmitters. tered and barbs gathered into a “ball”. No reactions were It was found that all tagged fishes, once captured in observed in lemon tetra to the appearance of an operator, the trawl, avoided it at the second encounter. They but they gathered around the net. White tetra at first started to move away from the approaching trawler in reacted to the operator and to input of food by fleeing, but advance, at about 50 m from the vessel. Three types of soon most of them aggregated at the feeding place. Barbs fish manoeuvres were noted: movement to coastal reacted to the operator by approaching him. waters (four fish), movement from the shore to the open After the first haul and during the next hauls, both area (four fish), and escape in the direction of the vessel tetras displayed school defensive behaviour. Lemon movement (two fish). Bream that had been caught once tetra made loop-manoeuvres, avoiding the net. White avoided the trawler both 0.5 h and 9 days after their tetra and barb started to avoid the trawl model when the release. In five cases, tagged bream swam in a new fish distance from the centre of the fish aggregation (or aggregation. The tagged bream avoided the trawler, “ball”) to the trawl became very short. while many fish in the same aggregation were caught in Both tetras continued school behaviour, whereas the the first haul. A decrease in catch occurred in consecu “ball” of barb dissociated for some time after the end of tive hauls. catching. These species restored their original distri Some other data were obtained with fish tracking: 18 bution in 3-5 min after the end of the trial. out of another 20 tagged fish left the fishing area during In the third trial lemon tetra, and in the second trial the 24 h following release, and the other two stayed white tetra and barb, changed their defensive behav there for two days and then left. iour: they came to form a school, to escape when the operator and gear appeared (lemon tetra) and to in Aquarium experiments crease avoidance distance to the trawl (white tetra, barb). The quantity of white tetra gathering at the The two strains of tetra (Hemigrammus caudovittatus feeding place decreased from about 80% in the first trial L.), lemon and white, and also rosy barb (Barbus to 5-10% in the last trial. They began to scatter from the conchonius Hum. buch., 1922) were used. Conspecific feeding place just after the trawl began to move (prob individuals belonged to one generation. ably reacting to the noise). A group of 106 lemon tetras, 15-30 mm body length, The values of catches and catchability coefficient was placed for two months before the experiment in tended to decrease in successive hauls (Figs. 4,5,6). For an aquarium 96 cm long and 38 cm wide, with a water white tetra in series 1 the catch was 20 fish and in series 2 depth of 42 cm. Two dip nets, small white (11.5 x it was 5. In the dynamics of lemon tetra catches in all 11.5 cm) and big black (19.5 x 19.5 cm), were used for trials, two phases could be distinguished. The biggest catching lemon tetra. Four series of trials were con values of catches and of the catchability coefficient were ducted at 3 day intervals. Each series consisted of three at the start of fishing and when the new, larger net was trials at 3.5 h intervals. Each trial consisted of five tested. Then the values of these indexes decreased (Fig. catches at 10 s intervals. 7).
14 70 7
1 0 - 6 Small white net Big black net c
20 -| r 16 of the experiment. They did not aggregate at the feeding - 14 place. Eight fish were caught in series 1, and seven in c a> 15 - - 12 series 2. o it -10 O
1 2 3 and that this second catch reduction occurred in an area Haul three times greater than the trawl capture zone. There Figure 5. Changes in the trawl model catchability coefficient fore, a thinning of the bream aggregation by the big first ( □ ) and in the distance of fish escapement (+) in hauls on barb. catch could not be the only cause of the decrease in the second catch. The time intervals between the trawl hauls varied from 1.5 h to 3 days, and 3 to 8 days were necessary to restore the large first catch and observe once again the 10 - reduction in the second catch. It is well known that after c CD cessation of action of a sudden stimulus, fish under O it natural conditions will soon restore their original distri CD o o bution pattern. Under the experimental conditions, the >. initial distribution pattern was restored within 3-5 min JD CO after hauling. Hence, frightening fish during the first £LO haul could not be the only reason for the second catch CC o decrease in parallel hauls. Echo-location data showed that fish quantity was
0 1 2 3 hardly changed after the first haul. At the same time the Haul trawl catchability in consecutive hauls decreased. Simi Figure 6. Changes in the trawl model catchability coefficient lar data were reported by other researchers for a marine (%) in experiments on white tetra. □ - exp. No 1 ;+ - exp. No fishing area (Zaferman and Serebrov, 1989). Hence, one 2. See text for explanation. more cause of the reduction of the bream catches was a decrease of fish vulnerability to the bottom trawl. The tagged bream avoided the trawler from a long distance. They left the fishing area after their release, The lemon tetra in an aquarium situated near that in but at the same time avoided the fishing vessel. Some of which the experiments with lemon tetra were carried out them were found in other areas 3-25 days later and did did not react to the operator's actions. The white tetra of not leave those fishing areas, but still avoided the group 2 displayed an avoidance reaction to the input of trawler. Malinin and Linnick (1983) investigated the food and to trawl-model noise from the start to the end defensive behaviour of 16 tagged bream. Some of them
15 were captured for tagging with a trawl and others with Thus summing up the data discussed above, it is gill nets. Individuals taken from trawl catches sub concluded that fast learning (1-2 trials) is the basis of the sequently avoided the trawler from about 50 m. Tagged adaptive changes in fish behaviour resulting in defensive fish that were originally caught with gill nets demon reactions to trawl fishing. strated a herding reaction to the approaching trawler and were caught by the trawl. More conclusive data were obtained on the tagged References bream fished by the trawl in fish aggregations. Both the Beukema, J. i. 1970. Angling experiments with carp (Cyprinus avoidance reaction of tagged fish and a sharp decrease of carpio L.). 2. Decreasing catchability through one-trial the trawl catches could be observed. Similar data on five learning. Netherlands J. Zool., 20(1): 81-92. Fernö, A., and Huse, I. 1983. The effect of experience on the bream have been obtained by another researcher (V. G. behaviour of cod (Gadus morhua L.) towards a baited hook. Yezov, pers. comm.). Fish. Res. 2: 19-28. A change in defensive behaviour of fish after they Fortunatov, M. A. 1959. Colour and transparency in Rybinsky were caught by trawl for the first time is evident from reservoir water. Proc. Inst. Reservoirs Biol., 2(5). Lenin these data. This change resulted from individual experi grad: 13-28 (in Russian). Frolov, U. P. 1925. Conditioned motor reflexes in freshwater ence and was retained for not less than nine days. It and marine fishes. In Proc. State Inst. Appl. Agronom., vol. served to increase survival of the fish, i.e. it could be 3, nos. 2-4, pp. 114-132. Moscow. 159 pp. (in Russian). considered as adaptive. Gandolfi, G ., Mainardi, D ., and Rossi, A. C. 1968. La reazione Bream and other species react to the approaching di paura e lo svantaggio individuale dei pesei allarmisti (esperimenti con modelli). Rend. Inst. Lombardo. Accad. trawl from 1-2 m when the water transparency in reser Sei. e Lettere., B102, NI: 8-14. voirs is 1.0-1.8 m (Fortunatov, 1959). It was confirmed Golenchenko, A. P. 1955. Speech. In Proc. Conf. Fish Behav. in the aquarium experiments that fish avoided gear from and Searching, pp. 53-54. Ichtiol. Comission, 5, Acad. Sei. a very short range during the first haul, but in the next URSS, Moscow. 236 pp. (in Russian). hauls increased the avoidance distance greatly. The Illichov, V. D. 1977. Behavioural management and spatial orientation in birds. In Management in animal behaviour, behaviour of bream changed after the first (second) pp. 12-20. Science, Moscow, 247 pp. (in Russian). haul. Subsequent catches were very small and consisted Kuhorenko, K. G. 1977. Formation of defence conditioned of young fish during the next few days. The dynamics of reflex to fishing gears in Atlantic mackerel. In Fish behaviour catches and catchability coefficients in aquarium experi studying in connection with fishing gears perfecting. Proc. Inst. Anim. Evol. Morf. Ecol., pp. 91-97. Moscow. 177 pp. ments and under natural conditions were similar. (in Russian). Aquarium data, of course, cannot be equally trans Leschova, T. S., and Zhuikov, A. Y. 1989. Fish learning: ferred to a natural situation. But only those behavioural ecological and applied aspects. Science, Moscow. 109 pp. (in patterns which were frequently observed under natural Russian). conditions, especially in studies of predator-prey re Malinin, L. K., and Linnick, V. D. 1983. On the behaviour of some freshwater fishes near the gears. In Fish behaviour in lations, were used. In the experiments, fish displayed connection with fishing and in mariculture. Proc. Inst. Anim. reactions to fishing gear that are well known as reactions Evol. Morf. Ecol., pp. 167-177. Moscow. 329 pp. (in Rus to a predator: gathering into a defensive school, “loop”- sian). manoeuvre of the school, increasing avoidance distance, Severtsov, A. N. 1922. Evolution and psychology. Ed. by Sabashnikov Brothers, Moscow. 54 pp. (in Russian). and escaping a dangerous place. Fish learn to produce Voronin, L. G. 1954. Conditioned reflex - universal adaptive these reactions after 1-2 attacks by a predator (Leschova phenomenon in animals. Moscow, Znanie, Serie 3, No. 43. and Zhuikov, 1989). So, it appears that fish learn to react 39 pp. (in Russian). to active fishing gear in the same way as to a predator. Zaferman, M. L., and Serebrov, L. I. 1989. On fish injuring when escaping through the trawl mesh. ICES CM 1989/B: 18.
16