278 Vertical movements of “escaped” farmed Atlantic salmon (Salmo salar L.)—a simulation study in a western Norwegian fjord Ove T. Skilbrei, Jens Christian Holst, Lars Asplin, and Marianne Holm Skilbrei, O. T., Holst, J. C., Asplin, L., and Holm, M. 2009. Vertical movements of “escaped” farmed Atlantic salmon (Salmo salar L.)— a simulation study in a western Norwegian fjord. – ICES Journal of Marine Science, 66: 278 – 288. To study the vertical distribution of fish that had been allowed to escape, farmed Atlantic salmon were tagged with acoustic tags equipped with depth sensors, and then released on five different dates in the course of a year from two fish farms in the Hardanger Fjord in western Norway. Release stimulated the fish to dive to deeper than 15 m during the first hours or days postrelease, often down to 50 –80 m. However, during the following 4 weeks, most of the escapees spent most of their time above the pycnocline at depths of 0– 4 m. The fish were more widely distributed in the water column after release during winter, but still spent most of the time in the cold surface layers. There was a wide range in the vertical distribution of individual fish, and the proportion of detections below 14-m depth ranged from 0 to 90%. There was a significant diurnal cycle in all seasons except midsummer, when the fish were less abundant in the upper layer during daylight, especially on brighter days. The results suggest that salmon diving activity following escape may complicate the recapture of escaped fish at the farm site but that the subsequent tendency of most fish to stay near the surface, virtually irrespective of the time of year, may facilitate recapture. Keywords: escaped farmed salmon, swimming depth, vertical distribution. Received 6 June 2008; accepted 18 November 2008; advance access publication 12 January 2009. O. T. Skilbrei, J. C. Holst, L. Asplin and M. Holm: Institute of Marine Research, PO Box 1870, Nordnes, NO-5817 Bergen, Norway. Correspondence to O. T. Skilbrei: tel: þ47 55 236894; fax: þ47 55 238531; e-mail: [email protected]. Introduction Farmed salmon that escape from cages are a serious problem to the fish-farming industry, and to the management of both farmed and wild stocks. With an annual production of .700 000 t compared with a total catch of wild salmon of ,1000 t (ICES, 2008), the Norwegian salmon-farming industry has become so large, compared with the size of the stocks of wild salmon, that escape rates of even a fraction of 1% present major challenges to the management and conservation of the resource. The escapees may spread diseases and parasites, and interfere with the genetic make-up of wild stocks, if they manage to interbreed (Naylor et al., 2005; Jonsson and Jonsson, 2006; Skaala et al., 2006; Ferguson et al., 2007). There is a correlation between the incidence of escaped farmed salmon in rivers and the intensity of salmon farming at the county level in Norway (Fiske et al., 2006). A global assessment carried out by Ford and Myers (2008) suggested that salmon farming has reduced the survival rates of wild salmon and trout populations in many countries. It is therefore important to have a good understanding of how these fish behave after escaping, to be able to recognize and mitigate the environmental risks associated with escapes. Fish farmers in Norway are required by regulation to recapture escaped salmon (Regulation 2008-08-27 No. 984). Major escapes have shown that efforts made close to the escape site are usually not very successful, with a catch rate of a few per cent of the total number of escapees (Anfinsen, 2005; Buvik, 2005). Escaped farmed salmon are usually reported from the fishery for wild salmon, which is seasonal and dominated by bag nets in the sea and angling in the rivers (Fiske et al., 2001). A late autumn/ early winter sea fishery that particularly targets escaped farmed salmonids was opened in several Norwegian counties during the late 1990s. A positive relationship between the catch per unit effort (cpue) of gillnetted salmon in the fishery and reported local escape incidents indicates that a bigger proportion of the fish can be recaptured if the fishery covers a larger geographical area, e.g. a whole fjord (Skilbrei and Wennevik, 2006). Farmed salmon that are not captured may find their way to the Norwegian Sea (Hansen et al., 1999), then return as maturing adults to the Norwegian coast and enter rivers over a broad geographical area (Skilbrei and Holm, 1998; Hansen, 2006). There is a need to develop effective recapture methods capable of being employed at all times of the year. The efficiency of fishing gears depends on the swimming behaviour and movements of the fish that they target. There is little information in the literature on the behaviour in the sea of newly escaped farmed salmon, particularly concerning the movements of fish after they escape or are released from the cage. The behaviour of cultured smolts released from cages for sea-ranching purposes has been monitored by underwater video equipment (Skilbrei et al., 1994a, b), and sonic tagging has been used to study the behaviour of simulated escapes of larger farmed salmon. Furevik et al. (1990) followed adult salmon after release from a farm in western Norway at different times of year. They used a hand-held receiver to track the fish, which they were able to follow for a couple of days. Whoriskey et al. (2006) used stationary receiver units to locate acoustically tagged # 2009 International Council for the Exploration of the Sea. Published by Oxford Journals. All rights reserved. For Permissions, please email: [email protected] 279 Modelling the movements of escaped salmon in a Norwegian fjord salmon set free from a cage site in the Bay of Fundy, North America, and were able to track some fish over longer periods of time and longer distances. None of these studies reported on the vertical distribution of the tracked fish. Knowledge of the swimming depth of escaped fish is a key to target recapture. We studied the swimming behaviour and movements of escaped, farmed salmon by releasing fish tagged with depth-sensitive, acoustic transmitters at different times of the year from a commercial farm in a large Norwegian fjord. Here, we report on their vertical distribution. Our goals were to monitor: (i) the immediate reaction of the fish to being released, (ii) the swimming depth of the fish during the following four weeks, (iii) their diurnal activity, and (iv) to determine whether the depth preferences of the “escapees” changed with the season of the year. Material and methods Study area The Hardanger Fjord was selected as the study area (Figure 1). The hydrographic conditions in the middle of the fjord, the area around Varaldsøy, are typical of fjords. There are major freshwater sources in the near and inner region which tend to produce a distinct upper brackish water layer at a depth of 5 –10 m and a surface salinity of ,15 –20 from spring until late autumn. Water flow in this layer is out of the fjord on both sides of Varaldsøy. The temperature of the brackish layer depends on the temperature of the discharged fresh water, how well it is mixed on its way out of the fjord, and solar heating. The surface temperature rises to 158C during summer. In winter, when the upper brackish layer is normally absent, surface-layer temperatures are similar to or colder than those of the deeper water. Below 20 –30 m, the variability in salinity and temperature throughout the year is less, with typical values ranging between 34 and 35, and 7.58C and 8.58C, respectively. There is a large salmon-farming industry in the fjord. The total production of salmon from 50 locations was close to 40 000 t in 2003, and a fishery targeting escaped salmon is prosecuted during autumn and winter (Skilbrei and Wennevik, 2006). There is a reason for concern about the negative effects of escaped farmed salmon on the genetic make-up of several wild salmon stocks that inhabit rivers that drain to the fjord (Skaala et al., 2006). Tagging and release In all, 132 farmed salmon were tagged with V13P-1L-S256 coded pingers carrying a depth sensor (45-mm long, weight in water 6 g, min/max of time 40/120 s; Vemco Ltd, Shad Bay, Nova Scotia) at two commercial salmon farms located close to the island of Varaldsøy in the middle of the Hardanger Fjord in western Norway (Figure 1), and released on five dates in the course of a year in groups of 19 –30 fish (releases 1, 2, 3, 4, and 5; Table 1). The fish that were tagged showed no external signs of maturity. The mean fork lengths of the released groups ranged from 54 to 70 cm (Table 1). The fish were anaesthetized with a mix of benzocaine and metomidate. The tags were sterilized and treated with Terramycinwvet and inserted surgically in the abdomen of the fish. The wounds were closed with three stitches, and the gills were irrigated occasionally during the surgery. After surgery Figure 1. The locations of the acoustic receivers (black stars), the two cage sites (arrows and numbers) close to the island of Varaldsøy, from which farmed salmon were tagged and released in the Hardanger Fjord, and the location of the three weather stations in Bergen, Kvam, and Ullensvang. 280 O. T. Skilbrei et al. Table 1. Description of releases 1, 2, 3, 4, and 5: dates of tagging and release of the fish at release site 1 or 2, mean size of fish, numbers tagged and recorded at receivers for the four weeks post-release. Description Date tagged Date released Release site Mean length (cm; s.d.) Mean weight, (kg; s.d.) Number released Number of fish detected Days 2 –8 Days 9 –15 Days 16– 22 Days 23– 29 Duration (h:min) Day of days 2, 29 Night of days 2, 29 1—July 24 June 2005 1 July 2005 1 63.7 (3.5) – 19 2—August/September 23 August 2005 28 August 2005 1 72.4 (5.6) – 24 3—December/January 8 December 2005 15 December 2005 2 54.0 (3.1) 2.8 (0.5) 30 4—March/April 8 March 2006 15 March 2006 2 60.5 (3.3) 3.0 (0.5) 30 5—June 1 June 2006 8 June 2006 2 70.4 (3.9) 4.3 (0.8) 29 18 11 8 4 24 21 13 10 16 10 10 10 29 20 11 11 29 22 14 8 18:42, 17:01 2:04, 4:56 14:21, 11:54 6:11, 9:06 5:51, 6:34 14:19, 13:48 11:44, 14:25 9:11, 5:59 18:45, 18:34 1:52, 2:25 The calculated duration of day and night are shown for days 2 and 29 post-release. (2– 3 min per tag), the fish were placed in a small tank until they recovered from the anaesthesia, then transferred to a cage 4 –5 m deep and 8 m in diameter. They were fed by hand for the next 5 –7 d, until release. This recovery period was chosen because it has been documented that salmonids show stress responses that last for several days at least (Olsen et al., 2002, 2005). All but one of the tags functioned satisfactorily. The experiment and the tagging procedure were approved by the Norwegian committee for the use of animals in scientific experiments (FDU). Acoustic receivers In all, 24 VR2 receiver units (Vemco Ltd) were moored at 24 different sites in the fjord, covering a total range of 74 km from the innermost to the outermost location (Figure 1). Release site 1 was between, and within the range of, the two receivers moored northeast of Varaldsøy (Figure 1). One receiver was located directly at release site 2, and an additional receiver was added to this site in May 2006. The depth below the fish farm at release site 1 increased from 180 to 350 m from the inner to the outer part of the structure, with a depth of 200 m below the cage used for the releases. At release site 2, the depth increased rapidly from 60 to 210 m beneath the fish farm; the depth directly below the cage was 200 m. The receivers were moored to a weight at the bottom, which was also moored to land, and held at a depth of 10– 15 m with a float, except for the two fastened to the floating structure of the cages at release site 2. The water depth at the receiver sites ranged from 25 to 110 m. Within a circular detection area with a horizontal radius of up to 300 m, depths typically ranged from 0 to 200 –500 m, owing to the steep slope of the bottom in this area. Comparisons between vertical distribution and light conditions Solar-radiation data from Bergen were provided by the University of Bergen (Olseth et al., 2006, 2007), and data from the stations at Kvam and Ullensvang close to the Hardanger Fjord were extracted from the AgroMetBase web service of the Norwegian Institute for Agricultural and Environmental Research (http://lmt.bioforsk. no/agrometbase/getweatherdata.php; see locations in Figure 1). The daily sums of solar radiation were available from all three stations. The daily weather was categorized according to the duration of sunshine recorded at the stations in Bergen and Ullensvang. If the sunshine totalled ,6 min in the course of a day, it was denoted “cloudy”, whereas days with a minimum 6 h of sunshine were defined as “bright”. Some 70% of the days classified into these two categories satisfied the criteria at both stations. Owing to variability in weather within the region, minor deviations from the criteria were accepted at one of the two stations. It was also confirmed that the daily sum of radiation recorded at the station in Kvam agreed with the expected range for “cloudy” and “bright” weather. The mean swimming depth during daylight for specific days was calculated if there were data from at least five salmon and a minimum of five recordings per fish, except for release 3 (December), where three salmon per day was taken as acceptable. Four dives to .50 m were regarded as statistical outliers in relation to the intention of studying the finer scaled patterns close to surface, so these were not included in the comparisons between cloudy and bright days. For each release group of fish, a two-tailed Student’s t-test (Statsoft, Inc., 2008) was used to investigate whether the mean swimming depth during daylight differed between bright and cloudy days. The diurnal cycle was divided into daylight and night as follows: daylight, the period from sunrise to sunset; night, the period from the onset of nautical twilight in the evening to the end of nautical twilight in the morning. Civil twilight was used in June and July because it is too bright then for nautical twilight at this latitude (608100 ) in midsummer. The computations of sunrise, sunset, and twilight times were made by the Online-Photoperiod Calculator V 1.94 L (http://www.sci.fi/ ~benefon/sol.html). Only salmon with .15 detections were included in the calculation of depth-frequency distributions by day or night. To compare fine-scale differences in the vertical distribution of salmon close to the surface between day and night, we employed two-way analysis of variance [general linear model (GLM) procedure; Statsoft, Inc., 2008]; it used individual depth data from depths of 0– 15 m. The two-way analysis of variance tested the effects of both time of day (day and night) and the variability between fish. The assumptions for the GLM were checked by inspecting the values of homogeneity of variance (Levene’s test 281 Modelling the movements of escaped salmon in a Norwegian fjord for homogeneity of variances), normal p-plots, and plots of means against standard deviations (Statsoft, Inc., 2008). To comply with the assumptions, the data were arcsine-transformed. This was because the surface (0 m) is a boundary for vertical movements of the fish, and an arcsine transformation moves very low values towards the centre, giving them more theoretical freedom to vary (Sokal and Rohlf, 1981). Spearman0 s rank order correlation (Statsoft, Inc., 2008) was applied to test whether differences in size could explain individual variability in the deep-diving activity. For each release, the Spearman0 s R-statistic was calculated to determine whether the proportional time spent below 14 m was correlated with the size of the fish. Hydrography The mean temperature and salinity conditions during the four weeks following each of the five releases were constructed based on records from a combination of sources. First, the Institute of Marine Research (IMR) has conducted regular hydrographic surveys along fixed transects in the Hardanger Fjord, and these provide reasonable temporal and spatial coverage, with profiles of high vertical resolution from the surface down to 50 m. A CTD-profiler (SAIV SD204) was used. Second, salinity and temperature measurements were made weekly at depths of 1, 3, and 8 m at a number of fish farms in the Hardanger Fjord. These observations are provided by Hardanger Fiskehelsenettverk (http:// www.fom-as.no/), a cooperative of farmers established to synchronize the delousing of salmon farms. The equipment used is not known, but the data have been validated against IMR data, and no systematic deviations attributable to faulty instruments or sampling methods have been found. The oceanographic conditions may vary spatially within the study area and in time within the study period of each release. The stratifications presented were compiled as the mean values in time and space during the first four weeks after each release date. Results Hydrographic conditions Physical conditions .30 m were stable during the whole experimental period, from July 2005 to late June 2006, with a water temperature of 8 –98C and a salinity of 33 –35 (Figure 2). During the first release, in July 2005, a typical summer situation was present, with a brackish, warm, upper layer 10 m deep. Surface temperature was 168C and salinity ,15. At the second release, in September 2005, the mean conditions were much the same as during the release in July, but with a slight cooling of the surface water and a rise in salinity; temperatures were still .148C in the upper 5 m. At the beginning of winter when the third group was released (in December), the water had become homogeneous in temperature, at 88C. The salinity still created a relatively weak stratification, with salinity .25 down to 10 – 15 m. During the fourth release at the end of winter (March), the water had become homogeneous in salinity (surface value .30), and a further cooling of the upper layer had taken place. The upper 20 m was ,68C and the surface value was ,48C. During the fifth release (June 2006), summer conditions had been re-established with a distinct 5 –10 m upper layer of warm, fresher water. The temperatures were .168C, but the salinity was still relatively high, at .25 (Figure 2). Figure 2. Mean temperature (top) and salinity (bottom) conditions in the study area during the four weeks following the releases in July, August, December, March, and June (releases 1, 2, 3, 4, and 5). Vertical swimming behaviour Days 0 –1 Many fish dived immediately on release. Diving activity was more intense in autumn and winter than in summer (Figures 3 and 4). Six of the 19 “escapees” dived deeper than 30 m immediately following their summer release in July 2005 (release 1; five of them are shown in Figure 3), and some reached depths of 60 –80 m. The mean percentage of detections .14 m deep during the first 2 d after release summed to just more than 10% of the total, a level comparable with the release in June 2006 (7%; release 5; Figure 4). The diving response was more extensive on 28 August and 15 December (releases 2 and 3), when many of the fish moved deeper than 10 m during the first hours after release, and several repeated their diving to .20 m until the next day (shown for the 28 August release in Figure 3). More than 40 and 20% of the time, respectively, was spent at 15 m or deeper during the first 2 d after the releases in late August and mid-December (releases 2 and 3; Figure 4). Diving peaked following the release on 15 March (release 4, Figure 3). The fish then spent some 50% of their time .14 m deep during the first 2 d (Figure 4), and almost 25% of the time during the next 28 d (Figure 4). Following release 5 in midsummer, on 8 June 2006, fish did not dive immediately. Instead they moved closer to the surface on day 0 than the other release groups. Only one of the 65 detections received during the first 4 h post-release was made from .14 m, and 7% of the time was spent .14 m during the first 2 d post-release (Figure 4). 282 O. T. Skilbrei et al. Figure 3. Swimming depth of ten individually tagged salmon from (a) release 1 on 1 July, (b) release 2 on 28 August, and (c) release 4 on 15 March, during days 0 – 1 post-release. The fish with the most post-release detections during days 0 and 1 were selected for depiction. Values with the same symbols are records for the same individual at different detection times. The time of release is shown by the arrow. Days 2 –29 During the next four weeks the salmon spent most of their time close to the surface, notably in the upper 4 m (Figure 4). This general distribution was evident throughout the year, but was less pronounced during daylight in winter, when the fish were more widely spread throughout the water column. A significant diurnal cycle could be observed after all releases except the midsummer releases (releases from July to April; releases 1, 2, 3, and 4), from which fish were more abundant in the upper layers at night (Figures 4 and 5, Table 2). Differences observed between mean swimming depth by day and by night ranged from fish staying 6 m closer to the surface at night to some that displayed the reverse relationship between day and night (Figure 5). The latter alternative was observed most frequently after the midsummer release in June (Release 5), when about half the fish stayed either lower or deeper at night (Figure 5), and there was no significant effect of time of day (Table 2). Generally speaking, the effect of the diurnal cycle was most evident after the July, August/September, and March/April releases (releases 1, 2, and 4), when most of the fish remained closer to surface at night, but not so clear for the December/ January release in midwinter (Figure 5). These relationships are also reflected in higher F-values for the diurnal effect following the July, August/September, and March/April releases (Table 2). Modelling the movements of escaped salmon in a Norwegian fjord 283 Figure 4. Mean proportion of detections of individual fish at 1-m depth intervals from each of the five releases. The first period from release until the next day (days 0 –1), and the following four weeks (days 2 – 29) are presented separately. The second period is split further into two vertical columns according to the time of day: day and night. See text for the description of the duration and calculation of the periods defined as day and night. 284 O. T. Skilbrei et al. Figure 5. Mean fish swimming depth (m) by day (white squares) and night (black squares) from days 2 to 29 following all five releases within the upper 15-m depth range. The fish are arranged according to decreasing difference between day and night measurements, with fish ascending to the surface at night to the left and fish descending at night to the right of each panel. Lines are drawn only for visualization purposes, and 95% confidence intervals are shown. Table 2. Results from a two-way analysis of variance (GLM, see text for details) test of the effect of day and night (diurnal) and individual variability on swimming depths from 0 to 15 m from days 2 to 29 following releases 1, 2, 3, 4, and 5. Whole model Release 1—July 2—August/September 3—December/January 4—March/April 5—June R2 0.17 0.37 0.35 0.25 0.28 F 77 286 54 44 68 Diurnal p * * * * * F 155 360 51 387 1 Individual 3 Diurnal Individuals p * * * * n.s. Data are arcsine-transformed. R 2, squared correlation; F, F-value; p, significance level, where *¼p , 0.001 and n.s.¼not significant. F 74 382 37 53 45 p * * * * * F 8 63 26 37 12 p * * * * * 285 Modelling the movements of escaped salmon in a Norwegian fjord Discussion Figure 6. The proportion of time spent .14 m deep during days 2 –29 following all five releases plotted against fork length at tagging for individual salmon. Fish with fewer than 20 detections have been excluded. See Table 1 for specific dates. Deep diving (.15 m) was occasional for all groups. The frequency of detections .14 m deep was lower for the first two releases in July and August/September (0 –7%) than for the other three releases, December/January, March/April, and June (releases 3, 4, and 5; 10 –25%; Figure 4). Time spent deeper than 14 m also varied widely among fish, from 0 to 90% (Figure 6). A large proportion of the fish released during summer in June and July (releases 1 and 5) was never recorded .14 m, whereas most of the fish released during winter (releases 3 and 4; the December/January and March/April releases) spent at least 10% of their time .14 m. There were no significant correlations between the tendency to dive deep and fish size (Spearman0 s R-statistic; not significant). The effect of weather conditions on vertical swimming behaviour There was a tendency for fish to double their mean distance to the surface on bright days compared with cloudy days. This was significant for two of the releases, and close to significance for two others (Table 3). There were no bright days during the observation period of the December/January release (release 3). This may have contributed to the slightly smaller differences between day and night swimming depth for this release group than for the August/September and March/April releases (described above; Table 2, Figure 5). Our study has demonstrated that the immediate response of adult salmon escaping from cages is most probably a dive to depths of 20 –80 m. Subsequently, however, the escapees stayed close to the surface most of the time for the next four weeks. This observation was almost independent of season and the high annual variability in temperature above the pycnocline of the fjord. There was a clear diurnal cycle that affected the abundance of fish within the upper layers, fish tending to move deeper by day, especially in bright weather. Farmed fish kept in cages have not had the opportunity to swim into the full range of depths available in the fjords before they escape. Therefore, it is a dramatic environmental change for a salmon that has escaped from a fish farm, where it was held in an enclosure under crowded conditions, to be able to move freely in all directions outside the cage. Our observations suggest that there is a great probability that this change may stimulate immediate diving. Fish habituated to a cage environment will experience a sudden loss of such physical stimuli as the presence of a cage floor and interaction with other fish in the cage. The reason for diving may simply be the sudden absence of physical obstructions, or it may be a genetically based behavioural response of wild salmon. Four hypotheses have been advanced to explain wild-salmon diving behaviour: (i) an orientation during homing migration, (ii) a need to feed in deeper layers, (iii) to control body temperature, and (iv) predator avoidance (Wada and Ueno, 1999; Reddin et al., 2004). For our escapees, diving during the first hours and days of freedom may be a way of mapping the new environment, perhaps like orientation diving during migration. Moreover, if salmon dive to escape predators and potentially harmful conditions, another possibility would be that immediate deep diving is an escape response triggered by being released into an unknown environment. If diving is related to avoidance of danger, then the darkness of the depths may be sensed as a protecting environment compared with the brighter surface layers. It is less likely that fish migrate to deep water to feed so soon after release, because they have become habituated to surface feeding in culture and do not have any experience of feeding in a new, perhaps deeper, environment. The fact that diving was observed under quite different oceanographic conditions throughout the year and that the rate declined a few days post-release suggests that temperature regulation was not its main reason. However, because farmed salmon seem to prefer higher temperatures at this latitude (Oppedal et al., 2007), the extent and the duration of diving following release at different times of the year may have been influenced by temperature stratification. Pre-release diving activity was least obvious in June and July, when surface waters are warmer than the underlying water Table 3. Mean depth (m) and (s.d.) of fish during daylight under bright and cloudy weather conditions (see text for detail), and the significance levels of a t-test comparing the differences between cloudy and bright days. Cloudy Release 1 2 3 4 5 Period 6 –22 July 2005 31 August–13 September 2005 23 December –11 January 2006 23 March–9 April 2006 14 June –5 July 2006 Depth (s.d.) 2.9 (0.9) 2.2 (0.9) 5.6 (3.9) 4.8 (0.7) 3.2 (1.4) Bright Days 4 5 5 6 5 Depth (s.d.) 4.3 (1.0) 5.1 (2.8) – 10.6 (4.7) 7.4 (3.9) Days 6 5 0 6 9 t-test (p-value) 0.07 0.06 – ,0.05 ,0.05 286 masses, and most pronounced in March/April, when the fish dived from cold surface temperature through the thermocline to warmer water below. Despite the annual surface-temperature range of almost 158C, similar to the natural temperature range of Atlantic salmon, fish released at various times of the year spent most of their first four weeks of freedom close to the surface. Temperature would therefore seem to be of minor importance in stimulating vertical migration of newly escaped salmon. On the other hand, temperature gradients may be one reason for attracting salmon to the mixed surface layer. The depth of travel of adult sockeye salmon (Oncorhynchus nerka) is apparently controlled by a general preference for the surface, avoidance of warm (also low salinity) water, and orientation to the thermocline (Quinn et al., 1989). Westerberg (1982) observed Atlantic salmon making vertical excursions of 5 –15 m and concluded that they tended to follow fine-structure gradient layers in the quasi-mixed surface layer or in the thermocline, possibly for orientation during their homing migration. The pycnocline in the Hardanger Fjord at a depth of 5 –10 m is a strong barrier to vertical water mixing, perhaps isolating surface water from deep water. Johansson et al. (2007) showed in a study of the swimming behaviour of cultured salmon in cages that the occurrence of a pycnocline influenced both the distribution of the fish and the oxygen depth profile. The water above the pycnocline was characterized by low salinity and varying temperature, with high levels of oxygen, whereas the salinity and the temperature of the water below the pycnocline were more stable, with lower concentrations of oxygen. In the present study, the fish spent most of their time in the water above the pycnocline at depths of 0 –5 m. Perhaps the salmon preferred the physical parameter(s) typical of this water, such as the greater oxygen content. The vertical distribution of the released adult salmon appeared to be similar to those observed for wild Atlantic salmon during migration in the sea. According to a study by Sturlaugsson (1995) in Iceland, most homing Atlantic salmon spend most of their time in the upper 3 m. Similar results were observed in another Icelandic study, with as much as 91% of the period of migration spent in the upper 2 m (Sturlaugsson and Thorisson, 1997). In a study in the Baltic Sea, almost all the salmon remained close to the surface, with a median swimming depth of 2 m (Karlsson et al., 1996). Diving was rapid in all three of these studies. Occasional dives of wild salmon similar to those observed for our “escaped” farmed fish were made down through the thermocline to depths of 20– 40 m (Karlsson et al., 1996), and even down to .100 m (Sturlaugsson, 1995; Sturlaugsson and Thorisson, 1997). Our escaped fish moved closer to the surface at night or, alternatively, tended to move away from the upper layer by day. The fact that the fish were closer to the surface in cloudy than in bright weather supports the notion that light intensity is a factor that influences the depth preference of escaped salmon. Salmon in cages apparently follow a similar diel migration by descending at dawn and moving closer to the surface at dusk (Juell and Westerberg, 1993; Oppedal et al., 2001; Juell and Fosseidengen, 2004; Johansson et al., 2006). By adjusting the levels of artificial light, Oppedal et al. (2007) concluded that farmed salmon positioned themselves vertically in a cage according to a behavioural trade-off between a preference for the highest temperature available and an attraction to a preferred light intensity. The data on wild salmon are not conclusive. Reddin et al. (2004) found O. T. Skilbrei et al. evidence in two of three Newfoundland stocks that kelts spend more time in warm water close to the surface at night. Similar observations have been made during post-smolt seaward migrations, and it has been suggested that post-smolts avoid bird predators during the hours of daylight by locating themselves deeper in the water column (Reddin et al., 2006). However, Sturlaugsson and Thorisson (1997) observed that spawning migrants along the coast of Iceland were closest to the surface around noon, and Karlsson et al. (1996) failed to find any diurnal pattern in the mean depth preference of salmon during the spawning migration in the Baltic Sea. It is an open question whether the small-scale, vertical diurnal cycle of our simulated escapees is a behaviour learned during their life in cages or whether it resembles the behaviour of wild salmon. Post-release diving behaviour is disadvantageous to fish farming for several reasons. If farmed salmon are accidentally lost from a cage, there may be no sudden surface activity by the fish in the water around the cages to notify the farmer of an escape. It may also be a difficult task to estimate the number of fish that have escaped, especially if few fish escape from large production units holding tens of thousands. Moreover, the recapture of the escaped salmon will be difficult, and only marginally effective. There seems to be a great risk that the fish will disperse downwards in the water column, so avoiding many types of fishing gear. If the fish are distributed widely in the water column, they will occupy a larger volume, requiring much larger and more costly fishing effort to recapture them. The fact that escaped salmon seem to have a preference for remaining close to the surface during at least their first four weeks at liberty, almost irrespective of the time of the year, makes it highly likely that their catchability will be enhanced, with traditional salmon-fishing gear such as the gillnets and bag nets deployed by anglers the most efficient at catching them. This behaviour may explain the observations of Skilbrei and Wennevik (2006) that escape incidents were followed by a regional increase in the cpue of gillnetted, escaped farmed salmon, of duration 4 – 5 weeks. However, the pronounced variability in the depth distribution of the salmon we observed suggests that catchability will vary substantially at an individual level, especially during winter, when the fish appear to be more widely distributed in the water column. If some fish prefer to stay in deeper water, then the likelihood of their recapture will be less. The clear tendency of the fish to remain close to the surface at night may affect the capture efficiency of the different types of fishing gear. Such behaviour should favour the use of floating gillnets, because it will increase the abundance of the fish at the setting depth of those nets. The proportion of the catch of juvenile coho salmon (Oncorhynchus kisutch) caught at depths of 0 –2 m in stratified gillnet sets fell from 52.5 to 12.1% from the darker to the brighter portion of the day (Pearcy and Fisher, 1988). However, several authors have noted that successful catches of post-smolt coho and Atlantic salmon with pelagic or surface trawls are made almost exclusively by day (Shelton et al., 1997; Holm et al., 2000; Krutzikowsky and Emmet, 2005). Krutzikowsky and Emmet (2005) suggested that the fish missed the surface trawl because the headrope passed below them during night trawling, usually at depths between 0.8 and 1.6 m. The greater abundance of simulated escaped salmon in the top metre of the water column at night, as observed here, therefore suggests that surface trawls will be less efficient at recovering “escapees” by night. Modelling the movements of escaped salmon in a Norwegian fjord In summary, we have presented an analysis of vertical movements of Atlantic salmon released from cages in the sea. The results seemingly show that any fishing effort to recapture escaped farmed salmon should be concentrated in the surface layers. The associated data on horizontal movements from these experiments are currently being analysed and will be documented in a forthcoming paper, which will include a broader discussion that includes consideration of the distribution of the fish in space and time, as well as the effects of these distributions on, for instance, recapture practices. Acknowledgements We thank Marine Harvest AS for their cooperation in the project. We also gratefully acknowledge Per Arne Åkre for helping to organize the tagging and releases at the fish farms, Håkon R. Sæbø for his skills in fish surgery, Hugh Allen for comments on the paper, Fred Whoriskey and Frode Oppedal for improving the work through comments and suggestions during the final stages of drafting of the paper, and Ole Torrissen for playing such a significant role during the initiation of the project. The Norwegian Ministry of Fisheries and Coastal Affairs provided the financial support for the study. References Anfinsen, A. R. 2005. Oppsummeringsrapport etter havari av fiskeoppdrettsanlegg i Storvikja i Tustna kommune i Møre og Romsdal. Etter arbeid i referansegruppen i tilknytting Aquastructures AS sin tekniske rapport (in Norwegian). 11 pp. Buvik, P. 2005. Gjenfangstfond for rømt oppdrettslaks—oppsummering etter drift fra 20.06-04.08.05. Rapport fra Alta Utviklingsselskap (in Norwegian). 7 pp. Ferguson, A., Fleming, I., Hindar, K., Skaala, Ø., McGinnity, P., Cross, T. F., and Prodöhl, P. 2007. Farm escapes. In The Atlantic Salmon: Genetics, Conservation and Management, pp. 357– 398. Ed. by E. Verspoor, L. Stradmeyer, and J. L. Nielsen. Blackwell Publishing, Oxford. Fiske, P., Lund, R. A., and Hansen, L. P. 2001. Rømt oppdrettslaks i sjø- og elvefiskeriet i årene 1989– 2000. NINA Oppdragsmelding, 704 (in Norwegian). 26 pp. Fiske, P., Lund, R. A., and Hansen, L. P. 2006. Relationships between the frequency of farmed Atlantic salmon, Salmo salar L., in wild salmon populations and fish farming activity in Norway, 1989– 2004. ICES Journal of Marine Science, 63: 1182 – 1189. Ford, J. S., and Myers, R. A. 2008. A global assessment of salmon aquaculture impacts on wild salmonids. Public Library of Science Biology, 6: e33. doi:10.1371/journal.pbio.0060033. Furevik, D., Rabben, H., Mikkelsen, K. O., and Fosseidengen, J. E. 1990. Migratory patterns of escaped farm-raised Atlantic salmon. ICES Document CM 1990/F: 55. 19 pp. Hansen, L. P. 2006. Migration and survival of farmed Atlantic salmon (Salmo salar L.) released from two Norwegian fish farms. ICES Journal of Marine Science, 63: 1211 – 1217. Hansen, L. P., Jacobsen, J. A., and Lund, R. A. 1999. The incidence of escaped farmed Atlantic salmon, Salmo salar L., in the Faroese fishery and estimates of catches of wild salmon. ICES Journal of Marine Science, 56: 200– 206. Holm, M., Holst, J. C., and Hansen, L. P. 2000. Spatial and temporal distribution of post-smolts of Atlantic salmon (Salmo salar L.) in the Norwegian Sea and adjacent areas. ICES Journal of Marine Science, 57: 955– 964. ICES. 2008. Report of the Working Group on North Atlantic Salmon (WGNAS), 1 – 10 April 2008, Galway, Ireland. ICES Document CM 2008/ACOM: 18. 233 pp. 287 Johansson, D., Juell, J-E., Oppedal, F., Stiansen, J-E., and Ruohonen, K. 2007. The influence of the pycnocline and cage resistance on current flow, oxygen flux and swimming behaviour of Atlantic salmon (Salmo salar L.) in production cages. Aquaculture, 265: 271– 287. Johansson, D., Ruohonen, K., Kiessling, A., Oppedal, F., Stiansen, J-E., Kelly, M., and Juell, E. 2006. Effect of environmental factors on swimming depth preferences of Atlantic salmon (Salmo salar L.) and temporal and spatial variations in oxygen levels in sea-cages at a fjord site. Aquaculture, 254: 594– 605. Jonsson, B., and Jonsson, N. 2006. Cultured Atlantic salmon in nature: a review of their ecology and interaction with wild fish. ICES Journal of Marine Science, 63: 1162 – 1181. Juell, J-E., and Fosseidengen, J. E. 2004. Use of artificial light to control swimming depth and fish density of Atlantic salmon (Salmo salar) in production cages. Aquaculture, 233: 269 – 282. Juell, J-E., and Westerberg, H. 1993. An ultrasonic telemetric system for automatic positioning of individual fish used to track Atlantic salmon (Salmo salar L.) in a sea cage. Aquacultural Engineering, 12: 1 – 18. Karlsson, L., Ikonen, E., Westerberg, H., and Sturlaugsson, J. 1996. Use of data storage tags to study the spawning migration of Baltic salmon (Salmo salar L.) in the Gulf of Bothnia. ICES Document CM 1996/M: 9. 16 pp. Krutzikowsky, G. K., and Emmet, R. L. 2005. Diel differences in surface trawl fish catches of Oregon and Washington. Fisheries Research, 71: 365– 371. Naylor, R., Hindar, K., Fleming, I. A., Goldburg, R., Williams, S., Volpe, J., Whoriskey, F., et al. 2005. Fugitive salmon: assessing the risks of escaped fish from net-pen aquaculture. BioScience, 55: 427– 437. Olsen, R. E., Sundell, K., Hansen, T., Hemre, G-E., Myklebust, R., Mayhew, T. M., and Ringø, E. 2002. Acute stress alters the intestinal lining of Atlantic salmon, Salmo salar L.: an electron microscopical study. Fish Physiology and Biochemistry, 26: 211 –221. Olsen, R. E., Sundell, K., Mayhew, T. M., Myklebust, R., and Ringø, E. 2005. Acute stress alters intestinal function of rainbow trout, Oncorhynchus mykiss (Walbaum). Aquaculture, 250: 480 – 495. Olseth, J. A., Cleveland, F., and de Lange, T. 2006. The radiation observatory radiation yearbook No. 41. Radiation observations in Bergen, Norway, 2005. Meteorological report series, University of Bergen, Department of Geophysics. 78 pp. Olseth, J. A., Cleveland, F., and de Lange, T. 2007. The radiation observatory radiation yearbook No. 42. Radiation observations in Bergen, Norway, 2006. Meteorological report series, University of Bergen, Department of Geophysics. 78 pp. Oppedal, F., Juell, J-E., and Johansson, D. 2007. Thermo- and photoregulatory swimming behaviour of caged Atlantic salmon: implications for photoperiod management and fish welfare. Aquaculture, 265: 70– 81. Oppedal, F., Juell, J-E., Taranger, G. L., and Hansen, T. 2001. Artificial light and season affect vertical distribution and swimming behaviour of post-smolt Atlantic salmon in sea cages. Journal of Fish Biology, 58: 1570– 1584. Pearcy, W. G., and Fisher, J. P. 1988. Migrations of coho salmon, Oncorhynchus kisutch, during their first summer in the ocean. Fishery Bulletin US, 86: 173– 195. Quinn, T. P., Terhart, B. A., and Groot, C. 1989. Migratory orientation and vertical movements of homing adult sockeye salmon, Oncorhynchus nerka, in coastal waters. Animal Behaviour, 37: 587– 599. Reddin, D. G., Downton, P., and Friedland, K. D. 2006. Diurnal and nocturnal temperatures for Atlantic salmon postsmolts (Salmo salar L.) during their early marine life. Fishery Bulletin US, 104: 415– 428. Reddin, D. G., Friedland, K. D., Downton, P., Dempson, J. B., and Mullins, C. C. 2004. Thermal habitat experienced by Atlantic 288 salmon (Salmo salar L.) kelts in coastal Newfoundland waters. Fisheries Oceanography, 13: 24 – 35. Shelton, R. G. J., Turrell, W. R., and Macdonald, A. 1997. Records of post-smolt Atlantic salmon, Salmo salar L., in the Faroe– Shetland Channel in June 1996. Fisheries Research, 31: 159– 162. Skaala, Ø., Wennevik, V., and Glover, K. A. 2006. Evidence of temporal genetic change in wild Atlantic salmon, Salmo salar L., populations affected by farm escapees. ICES Journal of Marine Science, 63: 1224– 1233. Skilbrei, O. T., and Holm, M. 1998. Effects of long-term exercise on survival, homing and straying of released Atlantic salmon (Salmo salar) smolts. Journal of Fish Biology, 52: 1083 – 1086. Skilbrei, O. T., Holm, M., Jørstad, K., and Handeland, S. O. 1994a. Migration motivation of cultured Atlantic salmon, Salmo salar L., smolts in relation to size, time of release and acclimatization period. Aquaculture and Fisheries Management, 25: 65– 77. Skilbrei, O. T., Jørstad, K., Holm, M., Farestveit, E., Grimnes, A., and Aardal, L. 1994b. A new release system for coastal ranching of salmon and behavioural patterns of released smolts. Nordic Journal of Freshwater Research, 69: 84– 94. Skilbrei, O., and Wennevik, V. 2006. The use of catch statistics to monitor the abundance of escaped farmed Atlantic salmon and rainbow trout in the sea. ICES Journal of Marine Science, 63: 1190– 1200. O. T. Skilbrei et al. Sokal, R. R., and Rohlf, J. F. 1981. Biometry. W. H. Freeman and Co, New York. 859 pp. StatSoft, Inc. 2008. STATISTICA (data analysis software system), version 8.0. www.statsoft.com. Sturlaugsson, J. 1995. Migration studies on homing of Atlantic salmon (Salmo salar L.) in coastal waters W-Iceland—depth movements and sea temperatures recorded at migration routes by data storage tags. ICES Document CM 1995/M: 17. 13 pp. Sturlaugsson, J., and Thorisson, K. 1997. Migratory pattern of homing Atlantic salmon (Salmo salar L.) in coastal waters W-Iceland, recorded by data storage tags. ICES Document CM 1997/CC: 09. 23 pp. Wada, K., and Ueno, Y. 1999. Homing behaviour of chum salmon determined by an archival tag. NPAFC Document, 425. 29 pp. Westerberg, H. 1982. Ultrasonic tracking of Atlantic salmon (Salmo salar L.). 2. Swimming depth and temperature stratification. Drottningholm, 60: 102– 115. Whoriskey, F. G., Brooking, P., Doucette, G., Tinker, S., and Carr, J. W. 2006. Movements and survival of sonically tagged farmed Atlantic salmon released in Cobscook Bay, Maine, USA. ICES Journal of Marine Science, 63: 1218 – 1223. doi:10.1093/icesjms/fsn213
© Copyright 2026 Paperzz