Fjord migration and survival of wild and hatchery-reared Atlantic salmon and wild brown trout post-smolts
- Cite this article as:
- Thorstad, E.B., Økland, F., Finstad, B. et al. Hydrobiologia (2007) 582: 99. doi:10.1007/s10750-006-0548-7
- 274 Views
The behaviour of wild (n = 43, mean LT = 152 mm) and hatchery-reared (n = 71, mean LT = 198 mm) Atlantic salmon and wild anadromous brown trout (n = 34, mean LT = 171 mm) post-smolts with acoustic transmitters was compared in a Norwegian fjord system. There was no difference in survival between wild and hatchery reared salmon from release in the river mouth to passing receiver sites 9.5 km and 37.0 km from the release site. Mortality approached 65% during the first 37 km of the marine migration for both groups. There was no difference between wild and hatchery-reared salmon either in time from release to first recording at 9.5 km (mean 135 and 80 h), or in the rate of movement through the fjord (mean 0.53 and 0.56 bl s−1). Hatchery-reared salmon reached the 37 km site sooner after release than the wild salmon (mean 168 and 450 h), but rate of movement in terms of body lengths per second did not differ (mean 0.56 and 0.77 bl s−1). The brown trout remained a longer period in the inner part of the fjord system, with much slower rates of movement during the first 9.5 km (mean 0.06 bl s−1).
KeywordsAcoustic telemetryFjord migrationSwimming speedSalmo truttaSalmo salar
Survival and behaviour during the early stages after the critical transition from freshwater to seawater are important factors for anadromous fish. At the end of the freshwater stage, Atlantic salmon Salmo salar L. and anadromous brown trout Salmo trutta L. juveniles transform physiologically and morphologically and migrate to sea as smolts (Wedemeyer et al., 1980; Hoar, 1988; Finstad & Jonsson, 2001). Atlantic salmon migrate to the open ocean for feeding, whereas the brown trout remain feeding in the fjord and coastal areas (Jonsson, 1985; Hansen et al., 2003; Klemetsen et al., 2003; Rikardsen, 2004). The understanding of the early marine phase of the Atlantic salmon and the environmental factors that may influence their behaviour and distribution in the sea is limited (Moore et al., 2000), and even less is known about the brown trout. This lack of information is particularly critical because the heaviest mortality of salmon in the sea apparently takes place during the first months after the smolts leave fresh water (Hansen et al., 2003).
The first stages of the marine migration of post-smolts can be monitored using telemetry techniques (Moore et al., 2000; Lacroix & Voegeli, 2000). For Atlantic salmon, the method has been limited by the relatively large size of acoustic transmitters, thus most studies have investigated behaviour of larger hatchery-reared posts-molts (e.g. LaBar et al., 1978; Lacroix & McCurdy, 1996; Finstad et al., 2005; Thorstad et al., 2004). However, the behaviour of hatchery-reared post-smolts may differ from that of wild post-smolts (e.g. Jonsson et al., 1991), and as pointed out by Lacroix & McCurdy (1996), there is a general need to obtain information on the movements of normally sized wild Atlantic salmon post-smolts.
Smaller acoustic transmitters are now available, allowing tagging of wild Atlantic salmon post-smolts. The objective of this study was to compare the rate of movement and survival of wild Atlantic salmon post-smolts with hatchery-reared Atlantic salmon and wild brown trout post-smolts in a Norwegian fjord system. It was hypothesised that the rate of movement of Atlantic salmon post-smolts through the fjord system was not dependent on their origin (as suggested by Lacroix & McCurdy, 1996; Voegeli et al., 1998; Lacroix et al., 2004). Furthermore, it was hypothesised that survival did not differ between wild and hatchery-reared Atlantic salmon (as suggested by Hvidsten & Lund, 1988). It was also hypothesised that the sea trout post-smolts remained in the inner part of the fjord system to a larger extent than the wild Atlantic salmon post-smolts (as suggested by Finstad et al., 2005).
Materials and methods
Handling, tagging and release of fish
Atlantic salmon (wild and hatchery) and anadromous brown trout (wild) post-smolts tagged with acoustic transmitters and released at the mouth of the River Eira in the Romsdalsfjord system in 2004
Mean body length (LT), mm (range, SE)
Mean body mass, g (range, SE)
Mean transmitter mass in water to fish mass ratio (%) (range, SE)
152 (136–173, 1.2)
25 (18–38, 0.66)
6.0 (4.0–8.0, 0.17)
7, 8, 9, 12, 20 and 22 May
198 (160–245, 2.6)
70 (37–124, 2.7)
3.1 (1.6–5.4, 0.12)
7, 8, 9, 12, 20 and 31 May
171 (143–242, 4.6)
42 (23–114, 3.8)
5.6 (1.8–8.7, 0.31)
7, 8, 9, 12, 20, 22, 26 and 31 May, 6 June
The wild Atlantic salmon and sea trout were caught in a trap close to the river mouth during their downstream migration in the River Eira, transported to the Statkraft hatchery in Eresfjord in a tank with oxygenated water and tagged zero to five days later (this variation was due to practical reasons). The hatchery salmon were 2-year-old smolts from the Statkraft hatchery with wild salmon from the River Eira as parents. The fish were tagged during 6 May–4 June 2004, depending on capture rates of wild smolts, using methods described in Finstad et al. (2005). Water temperature in the hatchery was 2.7–7.0°C. A seawater tolerance test (Blackburn & Clarke, 1987) performed on the hatchery salmon on 6 May revealed plasma chloride levels at 147 mM, indicative that the smolts were ready to be released into seawater (Sigholt & Finstad, 1990).
Groups of tagged smolts were transported in water-filled plastic bags to a cage (volume 0.5 m³) in the fjord at the mouth of the River Eira 0–2 days after tagging. The fish were kept in the cage for 2–21 h prior to release (Fig. 1, Finstad et al., 2005). Fish during the first release were held in the cage from one day to the next, but appeared stressed by being held in the cage. The procedure with keeping the fish in the cage was maintained to adjust the fish to the marine environment, but the time was reduced. All fish appeared to be in good condition at release. Schools of wild and Carlin-tagged Atlantic salmon and brown trout post-smolts were frequently seen at the release site during the study period. All fish were released during high tide in daylight (from 3 h before to 1 h after the highest tide).
Recording of fish after release
Fish were recorded when passing 19 individual receiver/data loggers (VEMCO model VR2) moored at sites 9.5, 37.0 and 65.0 km from the fish release site at the estuary of the River Eira (Fig. 1). At 9.5 km from the release site, where the fjord is 1.48 km wide, four moorings were evenly distributed across the fjord. Receivers were attached to each mooring at 3 m depth. An additional receiver was attached to two of the moorings at 10 m depth (site 1 and 3, Fig. 1). At 37.0 km from the release site, where the fjord is 2.60 km wide, seven moorings were evenly distributed across the fjord. Receivers were attached to each mooring at 5 m depth. An additional receiver was attached to four of the moorings at 10 m depth (site 1, 3, 5 and 7, Fig. 1). At 65.0 km from the release site, where the fjord is 2.46 km wide, one receiver was attached on each side of the fjord at 5 m depth. The receiver at site 1, 65 km distant (Fig. 1), disappeared before data were downloaded.
The sea depth at receiver moorings was 24–288 m. Receiver range for detecting signals from transmitters was typically a radius of 200–260 m for transmitters at 0.5–3.0 m depth, but varied considerably (radius of 45–620 m during range tests) with factors such as wave action and salinity. The receivers collected data until 24 or 25 July 2004. Manual tracking from a boat using a VEMCO VR60 receiver was carried out in the area around the mouth of the River Eira on 27 July 2004.
Data collected did not meet the assumptions for parametric tests, and therefore, non-parametric tests were employed. Rates of movement in terms of body lengths per second are based on time from release to first recording at the different receiver sites. This must not be regarded as actual swimming speeds, but as minimum speeds, since the fish most likely did not follow the shortest migration routes (Thorstad et al., 2004). For analysis of where in the fjord the fish passed, only recordings by receivers at 3 m depth were used 9.5 km from the release site, and at 5 m depth 37.0 km from the release site, for standardisation.
Water temperature and salinity
Water temperature and salinity were measured at eight sites in the middle of the fjord 1–60 km from the river mouth on 4, 14 and 30 May and 11 June at 1 m depth. Water temperature varied between 7.5 and 13.8°C (mean 10.8). Salinity varied between 7.7 and 31.9 ppt (mean 25.5).
Proportions of fish recorded after release
Number and proportion of Atlantic salmon and anadromous brown trout post smolts recorded at receiver sites 9.5 km, 37.0 and 65.0 km from the release site in the Romsdalsfjord system in 2004
Number (%) recorded 9.5 km from the release site
Number (%) recorded 37.0 km from the release site
Number (%) recorded 65.0 km from the release site
Only one fish recorded 37.0 km from the release site was not recorded when passing the receivers 9.5 km from the release site (a wild salmon). The three hatchery salmon recorded 65.0 km from the release site (Table 2) were all recorded when passing receivers both 9.5 and 37.0 km from the receiver site.
On 27 July, transmitters from 6 wild salmon (14%), 6 hatchery salmon (8%) and 7 trout (21%) were located in the area close to the river mouth by manual tracking. One transmitter that had belonged to a hatchery salmon (body size 171 mm, 43 g) was found in a saithe Pollachius virens L. that was caught by an angler in this area.
Rates of movement
Time and rate of movement for tagged Atlantic salmon and anadromous brown trout post-smolts from release to first recording at receiver sites 9.5 km, 37.0 and 65.0 km from the release site in the Romsdalsfjord system in 2004
Mean time (h) from release to first recording 9.5 km from release site (range, SE)
Mean rate of movement (bl s−1) from release to first recording 9.5 km from release site (range, SE)
Mean time (h) from release to first recording 37.0 km from release site (range, SE)
Mean rate of movement (bl s−1) from release to first recording 37.0 km from release site (range, SE)
Mean time (h) from release to first recording 65.0 km from release site (h) (range, SE)
Mean rate of movement (bl s−1) from release to first recording 65.0 km from release site (range, SE)
135 (9–667, 41)
0.53 (0.03–1.88, 0.10)
450 (35–1493, 133)
0.56 (0.04–1.89, 0.14)
80 (7–949, 27)
0.56 (0.01–1.62, 0.07)
168 (26–670, 40)
0.77 (0.08–1.85, 0.12)
154 (82–201, 37)
0.64 (0.40–1.00, 0.18)
350 (134–646, 70)
0.06 (0.02–0.12, 0.01)
1309 (1199–1528, 109)
0.05 (0.04–0.06, 0.01)
The trout took a longer time from release to first recording at 9.5 km than the salmon (wild and hatchery salmon combined, Mann–Whitney U-test, P < 0.001, Table 3). Also the rate of movement in terms of body lengths per second was slower for the trout than for the salmon (Mann–Whitney U-test, P < 0.001, Table 3). At 37.0 km from the release site, only three trout were recorded (Table 3).
Migration routes and patterns
Most of the salmon followed a one-way route out of the fjord, with little evidence of animals reversing their migration route. The exceptions were 1 wild salmon and 5 hatchery salmon (no difference in the proportion between groups, Fisher’s exact test, P = 0.38) subsequently recorded 9.5 km from the release site after they had been recorded 37.0 km from the release site. None of the three trout recorded at 37.0 km returned later to inner site.
Based on range tests it is likely that nearly all fish were recorded when they passed receiver sites at 9.5 and 37.0 km, since 185 m was the longest distance from a receiver possible to pass when passing the receiver lines. This is supported by the fact that only one of 43 fish (2%) that was recorded in the outer areas had not been recorded when passing receiver sites further in. Furthermore, the fish passing were in nearly all cases recorded by several of the receivers and with a high number of recordings per fish, indicating a good coverage by the receivers.
This study is consistent with the results and conclusions from previous studies of behaviour of hatchery-reared Atlantic salmon post-smolts (e.g. LaBar et al., 1978; Lacroix & McCurdy, 1996; Thorstad et al., 2004; Finstad et al., 2005), and suggests that those findings may also be representative of wild Atlantic salmon. The proportions of fish recorded 9.5 and 37.0 km from the river mouth (survival) did not differ between wild and hatchery-reared post-smolts. Similarly, time from release to first recording and rate of movement in terms of body lengths per second from release to 9.5 km from the river mouth did not differ. The only difference found was that time from release to first recording 37.0 km from the river mouth was shorter for hatchery-reared than for wild fish. However, this difference can be attributed to the larger body size of the hatchery-reared salmon, since the rate of movement in terms of body lengths per second was similar. It is therefore concluded that no behavioural differences between wild and hatchery-reared salmon post-smolts were detected, but that time from release to reaching 37.0 km from the release site was dependent on body size.
The rates of movement for the salmon (0.53–0.77 body lengths per second) were similar to those previously recorded for larger hatchery-reared salmon in the same fjord system (mean body length 29.5 cm, rates of movement 0.54–0.69 body lengths per second, Finstad et al., 2005). Higher ground speeds have been recorded for Atlantic salmon post-smolts in estuaries (e.g. LaBar et al., 1978; Fried et al., 1978; Moore et al., 1992, 1995) and in Passamaquoddy Bay in Canada (Lacroix & McCurdy, 1996), probably at least partly due to higher water velocities in these localities. Ground speed depends on water velocities and by which methods the rates of movement are recorded, and may therefore be difficult to compare among studies.
Swimming speeds have been recorded in the same fjord system as the present study for hatchery-reared Atlantic salmon post-smolts by continuous manual tracking, with post-smolts swimming at a mean ground speed of 1.27 body lengths per second over 10-min periods (Thorstad et al., 2004). If these ground speeds are representative, it is likely that the salmon post-smolts did not follow the straightest route out of the fjord in the present study, but had actually swum approximately twice this distance. However, it must be pointed out that the individual variation in the rate of movement, and thereby likely in the migration pattern, was large. Also, a few individuals returned to 9.5 km from the release site after they had been recorded 37.0 km away, indicating that some salmon post-smolts may move around in the fjord system for some time. Thorstad et al. (2004) found that post-smolts were actively swimming (at a speed of 1.32 body lengths per second when corrected for the speed and direction of the water current) and not simply drifting passively with the current.
Trout had a slower rate of outward movement than the salmon, and a smaller proportion of the brown trout post-smolts was recorded reaching both the 9.5 and 37.0 km receiver lines. Thus, the trout remained longer in the inner fjord than the wild and hatchery-reared Atlantic salmon, although a few individuals reached the outer fjord areas (37.0 km from the river mouth) within the 2 month study period. The results agree with a previous study on anadromous brown trout (Finstad et al., 2005). The longer time spent in the inner part of the fjord system could potentially expose the trout to a higher predation pressure than the Atlantic salmon (see below).
Both the wild and hatchery-reared Atlantic salmon post-smolts migrated in the middle of the fjord as well as closer to the shore, with a large individual variation. However, the results indicated that there may be some patterns in fjord migration routes, with more fish being recorded on one side of the fjord than on the other side. It is not known whether this corresponds with environmental patterns (such as currents) in the fjord system.
Assuming all passing fish were recorded at the receiver sites, the mortality of the salmon post-smolts was 40–45% during the 9.5 first km (the individual not recorded at 9.5 km but at 37.0 is included), and 36–42% during the next 27.5 km, with a total mortality during the first 37.0 km of the sea migration of 65%. These proportions must be regarded as maximum mortality compared to non-tagged post-smolts, because a few fish may theoretically have passed receivers without being recorded, and because the mortality may be increased due to handling and tagging (see below). However, the results correspond with reports of high predation rates in other Norwegian estuaries by cod Gadus morhua L. and saithe on post-smolts. In the estuary of the River Orkla, cod predation was estimated to be 20% (Hvidsten & Lund, 1988), and in a small area in the estuary of the River Surna to be 25% (Hvidsten & Møkkelgjerd, 1987). A large number of saithe with wild untagged post-smolts in their stomach were caught close to the river mouth during the present study, suggesting a high loss of post-smolts due to predation (Jepsen et al., 2006). The 12 salmon transmitters located in this area in late July were probably from fish being eaten by saithe or cod, and the transmitter later being ejected from the predator. The possibility of some tags being detected while in the stomachs of large predators passing the receiver lines cannot be ruled out. The movement behaviour of saithe and cod in the fjord system is not known, and therefore not the likelihood of these fish to move outwards in the fjord and being detected by the receivers. However, the fast gastric evacuation rates of gadoid predators (Andersen, 1999; 2001) indicate that this is not a large methodological problem in this study.
The present study confirms that the first phase of the marine migration seems to imply high mortality on the Atlantic salmon post-smolts, as suggested by Hansen et al. (2003). However, when comparing these results with the total mortality during the marine phase for Atlantic salmon from the nearby River Orkla (80–98% for one-sea-winter salmon, 89–99% for multi-sea-winter salmon, Hvidsten et al., 2004), it seems that there must also be a high mortality during later stages of the marine migration. The same conclusion was indicated by a study of acoustically tagged steelhead trout Oncorhynchus mykiss, with survival rates of ≥55% in the early ocean phase (Welch et al., 2004).
Hvidsten & Lund (1988) found no difference in predation rates between wild and hatchery-reared salmon smolts and no evidence of selective predation on the smallest wild and hatchery-reared smolts. This corresponds with the results from the present study, except that there was an indication of a higher survival to 37 km from the release site for larger than for smaller wild salmon smolts. Hatchery-reared post-smolts recorded at 37.0 km from the release site had a faster migration to 9.5 km from the release site than those not recorded at 37.0 km, indicating a higher loss among the slowest migrating fish. Thus, the risk of being predated may increase with time spent in the fjord system. An alternative explanation may be that the fish with the lowest migration speeds were the weakest fish, and therefore more likely to be predated.
Handling and tagging may influence the behaviour of the fish. For Atlantic salmon post-smolt studies in the sea, Moore et al. (2000) recommended tags to be less than 5% of fish mass to minimise effects on behaviour and survival. Some individuals had larger tags in the present study, and the wild post-smolts had relatively larger transmitters than the hatchery-reared post-smolts. According to Peake et al. (1997), wild salmon smolts may also be generally more affected by tagging than hatchery-reared smolts. The most likely potential negative effects of transmitters in the present study are higher mortality and lower swimming speeds for the smallest fish (e.g. McCleave & Stred, 1975), although transmitters weighing up to 12% of the body mass did not affect the swimming performance of juvenile rainbow trout (Oncorhynchus mykiss) (Brown et al., 1999). No difference in tag to body mass ratio between fish recorded or not recorded after release was found in the present study, indicating that potential negative effects of the tagging did not affect the conclusions of the study. However, if conclusions were affected it was most likely by overestimating mortality and underestimating swimming speeds for the wild compared to hatchery-reared post-smolts.
We would like to thank the staff at the Statkraft hatchery in Eresfjord, Bjørg Anne Vike, Petter Sira and Torbjørn Utigard, for extensive help and co-operation during the project. Stian R. Almestad, Espen Holthe, Jan Gunnar Jensås, Niels Jepsen and Stig Sandring are also thanked for help during the project The study was partially financed by the European Commission contract no: Q5RS-2002–00730, AquaNet Canada (RSM), the Norwegian Institute for Nature Research and Statkraft Energy AS. The experimental procedures used conformed to Norwegian ethical requirements.