The marine survival and growth of wild and hatchery-reared Atlantic salmon


Nina Jonsson. Tel.: +47 23355115. Fax: +47 23355101. E-mail:


  • 1Catches of Atlantic salmon Salmo salar L. decreased in the 1980s and 1990s over its entire area in the North Atlantic and smolts were often released for stock enhancement. However, there are questions about their survival and performance relative to fully wild fish. This paper reports on the survival and sea growth of River Imsa salmon released from 1981 to 1999 as 1- and 2-year-old hatchery and wild smolts.
  • 2Survival was significantly higher for wild than hatchery fish. Hatchery salmon released as 2-year-old smolts had lower survival, were captured more in coastal than freshwaters, grew more slowly and attained maturity younger than corresponding 1-year-old smolts.
  • 3The survival rate of hatchery fish released as 2-year-old smolts, but not 1-year-olds and wild smolts, decreased during the 1980s and 1990s. Growth rates at sea, adult size and the proportion of multi-sea-winter fish of all three groups also decreased over time.
  • 4Catches in coastal relative to freshwaters were higher for two- than one-sea-winter fish. Salmon captured in coastal water were greater in length than those captured in rivers. Mean specific growth rate at sea was similar for wild and hatchery salmon released as 1-year-old smolts, and higher than in hatchery fish released as 2-year-olds.
  • 5The proportion of two-sea-winter salmon correlated positively with the specific growth rate in the first year at sea. Total capture of wild adult salmon in rivers and Norwegian home waters each year correlated positively with the specific growth rate in the first year at sea. The same correlation held for hatchery fish released as 2- but not 1-year-old smolts.
  • 6Synthesis and applications. The coastal fishery was size-selective in reducing the size and age of salmon. Releases of 1-year-old smolts were financially more profitable than those of 2-year-olds. Decreasing production of River Imsa salmon since 1981 was chiefly caused by reduced sea-age at maturity and growth rate at sea of both hatchery and wild fish. A counteracting measure would be to reduce the size selectivity of the salmon fisheries.


Catches of Atlantic salmon Salmo salar L. have declined during the last two decades on both sides of the North Atlantic (e.g. Mather, Parrish & Fold 1998; Prévost & Chaput 2001; Klemetsen et al. 2003) for unknown reasons (Friedland et al. 2000). Possibly, decreased survival rates are involved (Mills 2000), but the growth rate and age at maturity may be also reduced (Schaffer & Elson 1975; Dempson et al. 2001). A concurrent decrease in the survival of Pacific salmon Oncorhynchus spp. has been linked to climatic variation (Johnson 1988; Beamish & Bouillon 1993), and the same may be true for Atlantic salmon (Friedland et al. 2000).

Rearing of juveniles and subsequent release into the wild is commonly used to increase fish populations (Moksness, Støle & van der Meeren 1998), and Atlantic salmon have been reared and released for more than 150 years (Jonsson & Fleming 1993). However, few studies have investigated the influence of hatchery rearing on survival and growth (Crozier & Kennedy 1993).

Atlantic salmon exhibit an anadromous life history, i.e. the adults spawn in rivers where the young grow for one or more years until smolting (Thorpe 1988). As seaward migrating smolts, the fish are usually between 10 and 20 cm long (Hutchings & Jones 1998; Jonsson, Jonsson & Hansen 1998). Juveniles feed in the North Atlantic Ocean for one or more years until the attainment of sexual maturity upon which they mostly return to their natal river for spawning (Hansen, Jonsson & Jonsson 1993). Returning adults are harvested at sea, along the coast and in freshwater (Dempson et al. 2001).

The freshwater habitat is limiting for juvenile production of wild salmon (Jonsson et al. 1998), where the number that can be supported (i.e. carrying capacity) is largely determined by the fluctuations in physical and chemical conditions, the frequency of extreme events such as droughts and spates, availability of food, the density of other fish species and the density of different life-stages of the salmon (Grant et al. 1998; Elliott 2001). To by-pass this bottleneck, hatchery-reared fish are often released as 1- or 2-year-old smolts. They leave freshwater shortly after their release in spring (Hansen & Jonsson 1985), and the survivors return one or more years later to reproduce (Jonsson, Jonsson & Hansen 1990), when they can be harvested. It is uncertain whether hatchery rearing for 1 or 2 years provides the best results in terms of yield, although large smolts are known to survive better than smaller ones (Salminen, Kuikka & Erkamo 1995).

Age at maturity varies among years within populations (Jonsson, Jonsson & Hansen 1991a, 1998). Atlantic salmon maturing after one winter at sea (grilse) range from 1 kg to 3 kg in total body mass. Those maturing as two-sea-winter fish are usually 4–7 kg, whereas three-sea-winter fish usually weigh above 7 kg (Jonsson, Hansen & Jonsson 1991b; Hutchings & Jones 1998). For wild fish, discharge volume in natural rivers may be the main factor determining the among population variation in fish size for rivers where mean annual water discharge is < 40 m3 s−1 (Schaffer & Elson 1975; Scarnecchia 1983; Jonsson et al. 1991b). Within populations, temperature may be a main factor influencing temporal differences in growth rate at sea through its influence on the production of food resources and the growth potential of the fish (Friedland, Hansen & Dunkley 1998; Friedland et al. 2000; Jonsson et al. 2001a).

The River Imsa, south-western Norway, supports a wild Atlantic salmon population. In this river, downstream migrating smolts and returning adults are monitored (Jonsson et al. 1998), and hatchery-reared salmon smolts have been released at the river outlet each year since 1981. We used data collected from these fish to estimate the recapture rate (as a surrogate for survival rate) and tested if there was a decline in recapture rate, age at maturity, adult size, or growth rate during the study period (1981–99). Furthermore, we tested if there were differences in marine growth and adult recapture rates (estimate of survival) between wild salmon and hatchery-reared salmon smolting at either 1 or 2 years of age. We also investigated relationships between smolt age, marine growth and sea-age at maturity, as well as evidence for any differential size-selection of salmon caught in coastal fisheries vs. those which survive to enter the river for spawning.

Materials and methods

Atlantic salmon spawn in the River Imsa, south-western Norway (58°50′ N, 6° E), and the young (parr) use the river as a nursery before smolting. On average, fish smolt at 1 (14%), 2 (78%) or ≥ 3 (8%) years of age and most (c. 80%) mature sexually after one winter at sea (one-sea-winter fish or grilse). The rest mature as multi-sea-winter fish (i.e. chiefly two winters at sea; Jonsson et al. 1991b, 1998).

From 1981 to 1999, 20 655 wild smolts migrated from the river to the sea (Fig. 1). During the same period, 133 875 1- and 2-year-old hatchery-reared smolts were released in May when the majority of wild smolts leave the river (Jonsson & Ruud-Hansen 1985).

Figure 1.

Number of (a) wild Atlantic salmon smolts and (b) hatchery-reared 1-year-old (black columns) and 2-year-old (open columns) hatchery-reared smolts tagged and released downstream of the fish traps in the River Imsa, Norway, 1981–99.

The hatchery fish were reared to the smolt stage at the NINA Research Station at Ims before being released at the mouth of the River Imsa. The brood stock was taken from wild parents of the Imsa stock captured in a fixed box trap with no selection for either small or large adults, except that mature parr males were never used as parental fish. The rearing procedure is described by Fleming, Jonsson & Gross (1994).

Fish traps were situated 100 m above the estuary of the river. A Wolf trap (Wolf 1951; apertures 10 mm, inclination 1 : 10) caught all descending fish larger than c. 10 cm, whereas all ascending fish were taken in a fixed box trap. The traps were emptied twice each day during the study period and any salmon caught were measured. Natural tip lengths (mm, i.e. total length of fish with naturally distended caudal fin, Ricker 1979) and body weight (g) were recorded. All smolts of wild and hatchery-reared fish were individually tagged with numbered Carlin tags (Carlin 1955) before release and subsequent migration to sea. All hatchery smolts were tagged at least 2 weeks before release. During the spawning migration back to freshwater, adults were caught at sea, in fjords and coastal waters, and in freshwater. Recaptures of adult salmon at sea and in rivers other than the River Imsa were reported by fishermen, giving the tag number, length and weight of the fish, and time and place of recapture. Both wild and hatchery-reared fish ascending the River Imsa were registered in the trap at the river mouth.

Specific growth rate (G) at sea was calculated from recaptured adults as:

(ln adult body length (mm) − ln smolt body length (mm))/number of growth seasons at sea.

We estimated growth rate for one- and two-sea-winter salmon separately, and we distinguished between 1- and 2-year-old smolts for hatchery salmon.

We used recapture rate as an index of survival for each cohort estimated as the number of adults recaptured divided by the number of smolts released. The figures were not adjusted for any possible mortality effect of tags or tagging. However, Hansen (1988) found total recapture rates of adult Atlantic salmon at 7·7% for unmarked, 4·1% for adipose fin-clipped and 3·1% for Carlin tagged smolts, indicating that the survival rate was more than two times higher for unmarked than Carlin-tagged smolts. In our estimates, we have not controlled for tagging mortality, unreported tags or possible tag losses because the difference among smolt groups was probably small although large smolts are known to tolerate tagging better than small ones (Strand et al. 2001). We did not intend to measure the absolute survival rate of untagged fish.

Tests of significant differences in recapture rate between the wild and hatchery-reared salmon released as 1- and 2-year-old smolts, and between the two groups of hatchery salmon each year, were based on the normal approximation to the binomial distribution (Siegel 1956), and we used the yearly number of fish in each group as the input data, and tested with separate t-tests (Sokal & Rohlf 1981). The overall difference in survival between wild and hatchery-reared fish across all years was tested with a t-test using the averages of the paired differences. Furthermore, we used General Linear Models (GLM) to test for significant effects on adult length (mm) and specific growth rate of smolt-year-classes, whether the fish was captured in coastal sea or river, or whether there was a significant interaction between smolt-year-class and habitat of recapture. The change in adult length and specific growth rate of salmon emigrating as smolts from 1981 to 1999 were tested by regression analysis using smolt year class as the independent variable. We also used regression analyses to test for covariation in growth-rate at sea between one- and two-sea-winter salmon.

Auto-correlation in recapture rates was tested to see whether or not the variation was cyclic (Moran 1952). We correlated the recapture of each cohort with those of succeeding years at increasing time intervals. In cyclic time series, high correlations occur when the intervals in years match corresponding phases in the cycle.


survival estimate

The annual recapture rate of wild adult Atlantic salmon, as an index of survival, was higher than in hatchery-reared fish (Fig. 2). Of the total number of smolts released, 8·9% of the wild fish were recaptured. The corresponding values for hatchery-reared fish released as 1- and 2-year-old smolts were 3·3% and 2·9%, respectively (Table 1), and these differences were highly significant (Table 2). However, the recapture rates were highest for hatchery fish released as 2-year-old smolts in 1982 and 1-year-old smolts in 1999. The capture rates between wild and hatchery-reared salmon released as 2-year-old smolts were not significantly different in 1988 and 1996 (Table 2). The recapture rates of the two hatchery-reared groups were not significantly different in 6 of 17 years (Table 2). In 4 years recapture was highest for fish released as 1-year-old smolts and in 7 years it was highest for those released as 2-year-old smolts.

Figure 2.

Recapture rate (%) of wild (solid line) and hatchery-reared Atlantic salmon released as 1- (broken line) and 2-year-old smolts (dotted line) from 1981 through 1999 in the River Imsa and recaptured as one or two-sea-winter fish.

Table 1.  Recapture rates of one- and two-sea-winter wild and hatchery-reared salmon released as either 1- or 2-year-old smolts
Pooled smolt ages1-year-old smolts2-year-old smolts
Percentage recapture of total no. of smolts released1 7·1 2·9 2·7
2 1·8 0·4 0·2
Percentage recapture in freshwater of total no. of smolts released1 3·2 1·7 0·08
2 0·5 1·1 0·06
Percentage of adult recapture at sea1544259
Percentage two-sea-winter fish of total adult recapture 2112·5 7·5
Table 2.  Test statistics of significant differences (ti) between recapture rates of wild Atlantic salmon and hatchery-reared fish released as 1-year-old (smolt-age-1) and 2-year-old (smolt-age-2) smolts, and between the two hatchery-reared groups for each year and as an average of all annual rates (tmean). * P < 0·05, ** P < 0·01, *** P < 0·001, otherwise P > 0·05
Smolt yearti wild vs. hatchery smolt-age-1ti wild vs. hatchery smolt-age-2Ti hatchery smolt- age-1 vs. age-2
1981 2·40*
1982 4·20***−3·28**−7·81***
1984 7·98*** 7·10***−0·70
1985 4·02*** 5·94*** 3·12**
1986 4·92*** 4·55*** 0·75
1987 5·91*** 7·09*** 1·76
1988 0·83
1989 5·80*** 8·42*** 2·95**
1990 2·19* 5·90*** 4·82***
1991 9·47*** 9·79*** 1·75
1992 5·89*** 4·82***−2·23*
1993 5·41*** 8·75*** 5·19***
1994 7·61***
1995 5·02*** 3·85***−1·66
1996 3·14** 0·32−2·83**
1997 2·70** 2·25*−0·08
1998 2·23* 7·36*** 7·01***
1999−4·59*** 3·79***14·98***
Tmean 3·21** 3·66*** 0·42

The annual recapture rates of wild salmon (Y %) correlated positively with that of hatchery fish released as 2-year-old smolts (X %; Y = 0·45X − 0·001; r2 = 0·36, d.f. = 16, P = 0·01), but not with those of fish released as 1-year-olds (r2 = 0·13, d.f. = 14, P > 0·05). The main reason why the latter correlation was not significant was the high recapture rate of the 1999 cohort of hatchery fish. The recapture rate (Y %) decreased with year of smolt migration (X year) in hatchery-reared salmon released as 2-year-old smolts: Y = 4·0537 − 0·002X, r2 = 0·22, d.f. = 17, P < 0·05, but not in wild salmon (r2 = 0·17, d.f. = 16, P > 0·05), or hatchery fish released as 1-year-old smolts (r2 = 0·07, d.f. = 15, P > 0·05). There was no significant auto-correlation in the time series of recapture rates with wild and hatchery-reared salmon (Fig. 2). The best correlation was found between the recapture rates of wild salmon every 7th year (r2 = 0·11, d.f. = 11, P > 0·05). Thus, we could not detect any significant cyclicity in the variable recapture rates of the salmon studied, although there was approximately 7 years between each abundant year-class.

Recapture rates of the hatchery-reared fish in freshwater were significantly higher for adults released as 1-year-old smolts, than 2-year-old smolts when pooled over all year-classes (χ2 = 47·0, d.f. = 1, P < 0·001; Table 1). The recoveries at sea, relative to those in rivers, were significantly higher for adults released as 2-year-old smolts (61%) than for those released as 1-year-olds (46%; χ2 = 85·7, d.f. = 1, P < 0·001). When comparing similar sea-age groups, this trend was only significant for one-sea-winter fish (P < 0·001).

sea-age at maturity

More fish attained maturity after one than two winters at sea, and the percentage of two-sea-winter salmon (Y) decreased with increasing year of smolt migration for wild (Y = 1550·4 − 0·771X; r2 = 0·21, d.f. = 16, P = 0·05), and hatchery-reared fish released as 1-year-old (Y = 3064·4 − 1·532X; r2 = 0·70, d.f. = 14, P < 0·001) and 2-year-old smolts (Y = 2000 − 0·997X; r2 = 0·44, d.f. = 17, P < 0·01; Fig. 3). The decrease differed significantly between wild salmon and hatchery fish released as 2- (F1,32 = 32·48, P < 0·001) but not 1-year-old smolts (F1,30 = 2·91, P > 0·05). The youngest age-group of hatchery smolts produced more two-sea-winter fish than those being 1 year older at release, i.e. 1-year-old smolts produced 12·5% two-sea-winter returns whereas 2-year-old smolts gave 7·5% two-sea-winter-old adults (χ2 = 27·00, d.f. = 1, P < 0·001; Table 1). Wild smolts produced even more two-sea-winter salmon than hatchery salmon released as 1-year-old smolts (21%; χ2 = 50·0, d.f. = 1, P < 0·001).

Figure 3.

Percentage of two-sea-winter salmon among adult wild (solid line) and hatchery-reared fish of the River Imsa stock released as 1- (broken line) and 2-year-old smolts (dotted line).

body size and growth

Mean body length (± SD) of the wild smolts, recaptured as adults, was 165 ± 17·7 mm (n = 648). Hatchery-reared 1-year-old smolts were similar in size to the wild smolts with a mean body length of 161 ± 27·0 mm (n = 1153), whereas reared 2-year-old smolts were larger with a mean body length of 227 ± 81·1 mm (n = 685).

Mean body length of wild one-sea-winter salmon varied significantly among year-classes (Fig. 4a, Table 3), but not between those captured in rivers and coastal waters. The decrease in size with increasing year-class was significant. There was no similar significant decline with time for wild two-sea-winter salmon. In hatchery salmon, mean body length of one-sea-winter fish varied with both smolt year-class and whether they were captured at sea or in rivers (Fig. 4b, Table 3; 1-year-old smolts: 593 mm in sea vs. 584 mm in rivers; 2-year-old smolts: 628 mm in sea vs. 616 mm in rivers).

Figure 4.

Mean lengths of adult Atlantic salmon leaving the River Imsa as smolts between 1981 and 1999. (a) Wild fish captured as one- (solid line) and two-sea-winter fish (broken line), and (b) hatchery-reared fish released as 1-year-old smolts and recaptured as one-sea-winter (solid line) and two-sea-winter fish (dotted line), and hatchery fish released as 2-year-old smolts and recaptured as one-sea-winter fish (broken line).

Table 3.  Significant effects (F-values) on adult size and specific growth rate of smolt-year-class, whether the fish was captured in sea or river, or whether there was a significant interaction between smolt-year-class and habitat of capture, tested with univariate analysis of variance. The fish was either wild, or hatchery-reared of the River Imsa stock, released as 1- or 2-year-old smolts and recaptured as either one- or two-sea-winter-fish. *P < 0·05, **P < 0·01, ***P < 0·001, otherwise P > 0·05
TypeSea-ageInterceptSmolt yearSea-riverSmolt year × sea-river
Adult size
Wild131888*** 8·40*** 0·821·73*
2 5198*** 0·95 0·421·17
1-year old130138***19·78*** 4·07*2·95***
Smolts2 8561*** 1·80* 0·0021·17
2-year-old137600***18·02*** 6·65**1·39
Smolts2 4583*** 4·00*** 0·031·09
Growth rate
Wild134907***13·28*** 2·711·44
2 7331*** 0·91 0·0010·40
1-year-old134373***54·33*** 9·17**1·47
Smolts212646*** 5·19*** 1·041·19
2-year-old1 7218***38·39***18·54***0·63
Smolts2 1083 3·35*** 0·420·48

There was significant negative correlation between body length and smolt year-class of wild and hatchery-reared fish released as 1-year-old smolts, whereas the decrease was only marginally significant for those released as 2-year-olds (Tables 4, P = 0·06). There was no similar decrease in size for two-sea-winter fish.

Table 4.  Regressions of mean fish length and mean specific growth rates over smolt year-classes and specific growth rate in the first over the second year at sea for wild and hatchery salmon of the River Imsa stock. The hatchery fish were released at the mouth of the River Imsa as 1- and 2-year-old smolts. *P < 0·05, **P < 0·01, ***P < 0·001, NS = not significant (P = 0·06)
Salmon typeDependent variableIndependent variableRegression coefficientInterceptCoefficient of determinationd.f.Significance level
Wild one-sea-winterFish lengthSmolt year-class−4·749100440·6816***
Hatchery smolt-age-1 one-sea-winterFish lengthSmolt year-class−3·515 75840·4515**
Hatchery smolt-age-2one-sea-winterFish lengthSmolt year-class−3·346 72770·2016NS
Wild one-sea-winterSpecific growth-rateSmolt year-class−0·0068   14·750·2616*
Hatchery smolt-age-1 one-sea-winterSpecific growth-rateSmolt year-class−0·0152   31·660·6315**
Hatchery smolt-age-2 two-sea-winterSpecific growth-rateSmolt year-class−0·0154   31·640·2216*
WildSpecific growth-rate one-sea-winterSpecific growth-rate two-sea-winter 1·988  −0·2410·3716**
Hatchery smolt-age-1Specific growth-rate one-sea-winterSpecific growth-rate two-sea-winter 1·456    0·1740·6015**
Hatchery smolt-age-2Specific growth-rate one-sea-winterSpecific growth-rate two-sea-winter 1·324    0·2010·71 9**

Mean specific growth rate of wild one-sea-winter salmon varied significantly with smolt year-class (Table 3). There was no significant effect of place of recapture. Fish captured in rivers grew at the same rate as those captured in fjords and coastal waters. There was no significant variation with year of seaward migration or place of recapture for two-sea-winter salmon.

Mean specific growth rate of hatchery fish varied significantly among smolt year-classes (Fig. 5, Table 3). The growth-rate decreased with year of release (Table 4). For one-sea-winter fish, there was also significant variation with place of release (Table 3). Fish recaptured in coastal waters had grown faster than those recaptured in freshwater. There was no significant interaction between smolt year-class and place of recapture. Two-sea-winter fish produced from 2-year-old smolts were only present in 11 cohorts and this group was therefore omitted from Fig. 5. Furthermore, the growth-rate differed significantly among the three groups released (P < 0·001), and was highest for those released as 1-year-old smolts and in wild fish. Averaged over all cohorts, mean specific growth rates ± SD of hatchery-reared salmon released as 1-year-old smolts and recaptured as one- and two-sea-winter fish were 1·31 ± 0·14 and 0·79 ± 0·09, respectively. For 2-year-old smolts they were 1·10 ± 0·27 and 0·70 ± 0·11, and in wild fish 1·30 ± 0·13 and 0·78 ± 0·07, respectively. The growth rate of one-sea-winter fish correlated positively with that of two-sea-winter fish from the same smolt cohorts (Table 4). This held for both wild and hatchery-reared salmon.

Figure 5.

Mean specific growth rate at sea of Atlantic salmon from the River Imsa. (a) Wild fish maturing as one- (broken line) and two-sea-winter fish (solid line). (b) Hatchery-reared fish released as 1-year-old smolts and recaptured as one- (solid line) and two-sea-winter fish (broken line), and 2-year-old smolts recaptured as one- (broken and dotted line) and two-sea-winter fish (dotted line).

recapture and production

More wild two- than one-sea-winter fish were caught in fjords and coastal waters than in rivers (χ2 = 40·9, d.f. = 1, P < 0·001); in total 73% of the two-sea-winter and 54% of the one-sea-winter fish were from the sea. For hatchery fish released as 1-year-old smolts, the corresponding results were 80% and 42%. For 2-year-old smolts, it was 72% and 59% (χ2 = 91·6, d.f. = 1, P < 0·001).

In hatchery fish, the proportion of two-sea-winter salmon produced (Y %) was positively correlated with specific growth rate during the first year at sea of the cohort (X). Thus, high growth rate during the first season of marine growth appeared to augment the proportion of two-sea-winter fish from a year-class (Fig. 6). This correlation held significantly for both smolt ages of hatchery fish (1-year-old smolts: Y = 65·45X1 −71·51, r2 = 0·51, F1,14 = 14·57, P < 0·002; 2-year-old smolts: Y = 9·47X2 − 5·93, r2 = 0·30, F1,15 = 6·51, P < 0·02). Smolt length as the second independent variable did not add significantly to these regressions, but the model was significant for 1-year-old smolts (r2 = 0·59, P < 0·003). In wild fish there was no similar significant correlation (r2 = 0·09, P > 0·05).

Figure 6.

The correlation between the proportion of adult Atlantic salmon (%) attaining sexual maturity after two winters at sea and specific growth rate in the first year at sea of hatchery salmon released as (a) 1-year-old smolts and (b) 2-year-old-smolts.

Annual recapture rate (Y %) increased with adult size of wild one-sea-winter salmon (r2 = 0·29, d.f. = 16, P < 0·05), and it correlated positively with specific growth rate during the year at sea (X1; Y = 23·31X1 −22·79, r2 = 0·28, F1,15 = 5·85, P = 0·03; Fig. 7). Thus, relatively more fish tended to survive and return as one-sea-winter fish in years with good first year growth. By adding mean smolt length (X2 mm) as a second independent variable, the coefficient of determination increased to 0·50 (Y = 44·95X1 + 0·39X2 − 112·57; r2 = 0·50, F2,14 = 6·86, P = 0·008). This means that the recapture rate increased with specific growth rate the first year at sea and fish length at smolting. Both independent variables were significant in the model (P < 0·05). In hatchery salmon, the similar regression was significant for 2-year-old (Y = 1·16X1 − 0·022X2+ 7·12; r2 = 0·36, F2,15 = 4·14, P = 0·04), but not 1-year-old smolts (P > 0·05). In this case, the recapture rate increased with specific growth rate but decreased with smolt length at release.

Figure 7.

The correlation between annual recapture of wild adult salmon in rivers and Norwegian home waters (%) and specific growth rate the first year at sea.

Hatchery-reared 1- and 2-year-old smolts produced similar yields of adults, c. 74 kg salmon per 1000 smolts released (Table 5). One-year-old smolts, however, were smaller at release and gave a relatively higher river return than 2 years old smolts.

Table 5.  Recaptures (kg) in rivers, and coastal waters of hatchery-reared Atlantic salmon per 1000 smolts released at the mouth of the River Imsa
Smolt ageWeight releasedRecaptured
RiversFjords and coastal seaTotal
1 year 41·04031·67342·93574·608
2 years179·35024·83749·26374·100



The use of recapture rate as an index for survival assumes that the probability of being recaptured is the same for all three groups of salmon tested. This assumption holds if the homing to the River Imsa is perfect and all fish not caught at sea will be recovered in the home river. However, no group exhibits perfect homing, and the observed straying rate in wild and hatchery-reared Atlantic salmon are 6% and 15%, respectively (Jonsson, Jonsson & Hansen 2003). Thus, hatchery fish stray more than wild fish. The observed straying rates are minimum estimates because all adult salmon are monitored in the River Imsa, but not in any other river the fish might enter, where the figures are based on recoveries from anglers. We therefore feel that the straying may be about twice as high as that recorded (Jonsson et al. 2003). Therefore, the survival rate of hatchery fish may be somewhat underestimated relative to that of wild salmon, making the real difference smaller than the 60% indicated by the recaptures.

We did not correct for tag losses, increased mortality due to tagging, or unreported tags from fish captured in fisheries. We know that the Carlin tags are more harmful to small than large smolts (Strand et al. 2001) and that the effect of the anaesthesia may last for several days (Hansen & Jonsson 1988), meaning that 2-year-old smolts might return to the River Imsa in relatively higher proportions than the younger and smaller smolts because of lower tagging mortality, and that wild smolts might experience relatively high tagging mortality because of a shorter recovery period after anaesthetization and tagging. Furthermore, the handling might be more harmful to wild than hatchery smolts as hatchery fish are domesticated and quite used to human handling. Because of these differences between the groups tested, we may only with care use recoveries as an index of survival.


The survival rate appeared to be higher for wild than released hatchery-reared salmon. This result was significant despite handling and tagging potentially causing more harm to wild than hatchery salmon. Whatever additional stress handling and tagging may have imposed on wild fish (Lepage et al. 2000), the effect was smaller than the negative survival effect of juvenile hatchery rearing.

Wild salmon live under natural conditions with predators and keen competition for resources, while hatchery salmon live sheltered in tanks and experience a superabundance of food and very little exercise. At release, they have no prior predator experience and have never eaten natural food items, which may result in lower success under natural conditions (Sundström & Johnsson 2001). Survival from eggs to smolts has been estimated to be more than 20 times higher under hatchery conditions than in the wild (Jonsson & Fleming 1993). Therefore, inferior hatchery fish that would be selected against in nature may survive until smolting in the hatchery tanks. This difference in experience, exercise and selection pressures in the pre-smolt stage may affect their subsequent survival at sea and the return to the River Imsa as spawners.

Hatchery-reared salmon released as 2-year-old smolts survived slightly better than those released as 1-year-olds. Most probably, this is a combined effect of larger smolt size and a higher proportion returning after only 1 year at sea. Large smolts survive better than small smolts as they are less vulnerable to predators, and their larger size decreases the extra burden imposed by the external tag (Salminen et al. 1995). On the other hand, fishing pressure increases with fish size, and the size selective coastal fishery takes a larger part of the grilse produced from 2- rather than 1-year-old smolts. Therefore, a smaller proportion of the fish released as 2-year-old smolts may return to the home river.

There was no clear negative trend in survival rate among years except for fish released as 2-year-old smolts. However, survival at sea was low in most years during the 1990s. Obviously, there was large variation in survival at sea among years, and abundant year-classes were produced with intervals of c. 7 years.

The recapture rate of wild salmon and hatchery fish released as 2-year-old smolts correlated positively. This means that common factors at sea appeared to influence the survival rate. Fish released as 1-year-old smolts did not correlate significantly with the others. In this case, variation in smolt quality appeared to be important for the sea survival, and the quality of the group released in 1999 seemed particularly high.


The growth rate at sea was similar for wild salmon and hatchery salmon released as 1-year-old smolts. This was unexpected because hatchery fish had little or no prior experience with natural food items (Sundström & Johnsson 2001). It is likely that most smolts reacted spontaneously to natural food items when encountered, and little may be gained by exposing hatchery fish to natural food items before release. However, more hatchery than wild fish died after release, suggesting that some individuals found it difficult to locate and handle natural prey, at least during the first days after release (Johnsen & Ugedal 1989).

Salmon released as 2-year-old smolts grew more slowly than those released as 1-year-olds. This may be due to the larger size at release of the 2-year-old smolts. Typically, growth rates decline as the fish grow larger. The physical basis for this is not well understood, but Pauly (1981) attributes the effect to the allometric growth of the gill surface area relative to the mass of the fish. This means that the oxygen consumption becomes gradually more limiting for growth as the fish become larger. It is also possible that the increase in surface area over which food is absorbed becomes limiting (Wootton 1998). A third alternative is that fish smolting as 2-year-olds grew slowly throughout their lives due to a genetic trait, a common feature among salmonids (Allendorf, Knudsen & Leary 1983; Einum, Thorstad & Næsje 2002).

Cohorts with high growth rates during the first year at sea produced more two-sea-winter salmon than cohorts with poor growth during their first year. Similarly, it has been found that niche shifts resulting in increased growth rate such as the change from invertebrate to fish feeding, delay maturation in salmonids (Jonsson, Hindar & Northcote 1984; Jonsson et al. 1999). The same pattern occurs when they smolt and go to sea (Jonsson & Jonsson 1993). Also, Hutchings & Jones (1998) found a positive among–population association between sea-age at maturity and growth rate at sea for Atlantic salmon. Moreover, Norwegian multi-sea-winter populations tend to grow faster than the one-sea-winter populations (Jonsson et al. 1991b). Fish appear to delay maturity when the growth rate stays high, and they attain maturity when the growth rate starts to level off (Jonsson & Jonsson 1993; Hi & Steward 2002). This is opposite to the observation from salmon parr where fast growers mature younger than slow growers (Thorpe 1986).

sea-age at maturity

Hatchery salmon smolting as 1-year-olds produced a higher percentage of two-sea-winter salmon than those smolting as 2-year-olds. This means that the youngest fish released tended to stay longer in the ocean before attaining sexual maturity compared with older smolts. However, the cohort difference in mean sea-age at maturity is smaller than the 1 year difference in smolt age. A similar tendency was found for hatchery-reared Atlantic salmon released in a river in northern Norway (Hansen & Lea 1982), and within and among populations of wild anadromous brown trout (L’Abée-Lund 1994; Jonsson et al. 2001b). Hutchings & Jones (1998), on the other hand, reported that among Atlantic salmon populations early maturation is associated with decreased smolt age and increased smolt length.

Sea-age at maturity is positively associated with growth rate during the first year at sea. The same tendency was reflected by the decreasing proportion of two-sea-winter fish during the study period paralleling the decrease in growth rate. There are two possible reasons for this. First, there appears to be a positive association between poor first year growth at sea and early age at maturity. If the size advantage by staying at sea is small, the salmon mature early. This tendency contrasts with that of early mature parr where fast growers mature younger than more slow growing fish. It also differs from Pacific salmon where slow growth at sea is associated with older age at maturity in those species with multiple ocean age groups. According to Beamish & Mahnken (2001), this may be because these species need to reach a critical size to attain maturity. Secondly, there is heritability for age at sexual maturity (Ricker 1972; Gjerde, Simianer & Refstie 1994), and the proportion of fish with inheritance for late maturity may have decreased during the study period. If so, wild salmon in the River Imsa may originate from two-sea-winter salmon to a larger extent than hatchery fish. Offspring of two-sea-winter fish may survive proportionally better in the river than in the hatchery.


The present results lend only modest support to the hypothesis that the sea-survival of salmon has decreased during recent years. However, growth rate, fish size and sea-age at maturity have decreased gradually since 1981, possibly due to a climatic change as suggested for Pacific salmon (Beamish et al. 1999; Noakes, Beamish & Kent 2000; Finney et al. 2002). Friedland et al. (1998, 2000) indicated that this may be caused by a change in spring temperature. However, the most obvious change during recent years has been in winter conditions (Hoerling, Hurrell & Xu 2001; Benestad 2002; Wallace & Osborn 2002), which may also influence the life history of the fish (Reddin 1988).

Salmon harvested in the coastal and fjord fisheries were on average larger than those which escaped and entered rivers. Two-sea-winter fish were exploited more heavily during the return migration in coastal waters than one-sea-winter fish. For salmon originating from the River Imsa, it has been shown that two-sea-winter salmon return to the coast earlier in the season than one-sea-winter fish (Jonsson et al. 1990). Due to the larger size of the first group and the low waterflow in the small River Imsa during summer, small fish (one-sea-winter fish) will ascend the river earlier in the season than the larger fish. Thus, the multi-sea-winter salmon will be vulnerable to the net fisheries for a longer time period than the grilse.

Salmon released as 2-year-old smolts were more heavily exploited in marine nets than those released as 1-year-olds, and the recapture rate increased with smolt length and growth rate. This reflects the size-selective nature of the nets, which usually have mesh sizes (bar mesh) of between 58 mm and 66 mm. Thus, the net fishery may reduce the proportion of large salmon in the spawning populations. This accords with Schaffer & Elson (1975) who used this argument to explain the increased proportion and density of grilse in North American rivers in the 1960s and early 1970s. Size-selective fishing probably influences population genetics in the long term with a stronger tendency towards small-sized salmon (Ricker 1981; Machiels & Wijsman 1996; Hamon et al. 2000; De Leo & Gatto 2001).

Because growth and sea-age at maturity are partly inherited (Gjerde et al. 1994), the size and age selection pressures may influence the salmon populations differently, with the greatest effect in rivers supporting large-sized fish. These are large rivers with the greatest potential for salmon production and sport fishing. The within-population effect is a selective advantage for small-sized individuals counteracting the selective advantage of large body size during spawning (Fleming et al. 1996). Furthermore, the size-selective fishery may influence the sex ratio in rivers in favour of males, as this sex tend to return more as grilse, and females more as two-sea-winter salmon (Jonsson et al. 1990). Females are probably the sex limiting recruitment, and as many salmon populations now are below their conservation limits, particularly in the southern part of Europe and in eastern North America (ICES 2002), the size-selective fishery may increase the problem of producing enough recruits. To counteract this, most countries supporting salmon populations have regulated their fisheries heavily, resulting in significant reductions in fishing effort, and subsequent catches. However, size-selective fisheries should be reduced even further to counteract the tendency towards younger sea-age at maturity.

In the case of sea ranching (i.e. the release of smolts with the purpose of harvesting the entire crop when they return as adults), 1-year-old smolts appear to be more useful than 2-year-olds, although the total recaptures were similar. In spite of poorer survival, 1-year-old smolts produced more two-sea-winter fish, and, to a greater extent, their grilse proportion escaped the coastal fishery and returned to the home river. Furthermore, the production cost of 2-year-old smolts is higher than that of 1-year-olds. Either way, the return to the river is low, and the amount of fish returning to the home river is lower than the weight of the smolts released.


The staff at the NINA Research Station at Ims, under supervision of the two managers Christofer Senstad and Jon G. Backer, sampled the fish in the traps on a daily basis. They also reared, tagged and released the hatchery fish. A large number of fishermen and anglers provided tag returns of recaptured fish. The tag returns were registered by Berit Larsen. Funding was provided by the Norwegian Directorate for Nature Management and the Norwegian Institute for Nature Research. We are indebted to them all.