1 Brown Trout Salmo trutta L. life-history traits were studied in 17 coastal streams with mean annual water discharges ≤1 m3 s−1, and with anadromous and freshwater resident trout present in all streams.
2 Mean smolt length and age were positively correlated with mean annual water discharge and latitude. The variation (CV) in smolt length and age was negatively correlated with water discharge. Mean length and age at maturity were also positively correlated with mean annual water discharge.
3 Freshwater resident females were sparse (mean: 3·7%) but males were abundant in all the investigated populations (mean 48·9%). Mean body length at maturity of resident males increased with discharge in parallel to that of the anadromous males. Mean length at maturity increased with smolt length in anadromous males, but not females. Mean sea-age at maturity in females, but not males, was negatively correlated with smolt age.
4 It is concluded that several life-history traits of Brown Trout are strongly influenced by habitat variables associated with the size of small nursery streams.
Body size influences almost all other life-history traits (Roff 1992) and is probably the most important life-history trait influencing animal fitness (Miles & Dunham 1992). For instance, longevity increases with body size (Bonner 1965; Pauly 1980; B. Jonsson et al. 1991), and large individuals often invest more energy in reproduction and are more fecund than smaller conspecifics (Sibly & Calow 1986). At least for Atlantic Salmon Salmo salar L. this results in a longer recovery time between breeding for large than small adults (N. Jonsson et al. 1991).
Intraspecific variation in size may partly be a proximate response to environmental conditions. This is especially known from animals with indeterminate growth such as fishes, amphibians and reptiles (Stearns 1992; Smith & Skúlason 1996). In salmonids, there are differences in adult size among freshwater resident and sea run migratory morphs from the same populations (Gross 1987; Jonsson & Jonsson 1993), and in Arctic Charr Salvelinus alpinus (L.), body size varies among phenotypes exploiting different habitats of the same lake, as a consequence of food abundance and habitat constraints (Jonsson et al. 1988; Jónasson, Jonsson & Sandlund 1998).
Studies on intraspecific variation in body size have identified environmental correlates of variation by using comparisons among populations (Myers et al. 1986; Rhymer 1992; Tracy 1999). In Mosquito Fish Gambusia affinis (Baird & Girard) the water level has been found to influence body size of adults (Stearns 1983a,b), and in Atlantic Salmon there is an increase in mean fish size with mean river water discharge from 1 to 20 m3 s−1, but not for larger rivers (N. Jonsson et al. 1991). In the closely related Brown Trout Salmo trutta L. no similar positive correlation has yet been identified (L’Abée-Lund et al. 1989; L’Abée-Lund 1991). However, for this latter species, most of the investigated streams were large and this may have influenced the result.
Brown Trout are polymorphic. Often, the populations consist of small resident and larger migratory individuals (Jonsson 1989). In coastal streams, the migrants are anadromous (migrating between fresh and saltwater) whereas the residents stay in fresh water throughout life (Jonsson 1985). After spending one or more years in fresh water, the immature fish, called parr, transform through a smolting process and migrate as smolts to sea in spring. They live in coastal waters where they feed during summer (Jensen 1968). In autumn, adults return to their home river for spawning (Elliott 1994). In large rivers, adults often return early in the season, whereas in small streams, they tend to return later, shortly before they are ready to spawn. Food is limiting for growth in fresh water, but less so, if at all, at sea (Jonsson 1985; Jonsson & Jonsson 1998).
Brown Trout smolt size increases with latitude, probably as an effect of decreasing water temperature towards the north (L’Abée-Lund et al. 1989; Jonsson & L’Abée-Lund 1993). Less is known about smolt size variation among populations that are exposed to similar water temperatures at sea. Borgstrøm & Heggenes (1988), however, found particularly small smolts in one south Norwegian stream and hypothesized that this was an adaptation to the low water discharge in the stream, as the water levels of the smallest trout streams may be so low that the feeding opportunities for large presmolts might be reduced relative to smaller conspecifics. If so, it might be advantageous to smolt young and leave the stream already at a small body size. This was supported by Titus & Mosegaard (1989, 1992) and Järvi et al. (1996) who have documented small smolts in some small Swedish streams draining into the Baltic Sea. The salinity of the Baltic Sea is 5–15‰ salt, and less is known how this would be along the east Atlantic coast where the salinity generally is much higher (≈35‰ salt). From amphibians, it is well documented that age and size at metamorphosis vary according to water level and duration of the pond (Newman 1992; Denver et al. 1998). Similarly, one might predict that Brown Trout from small streams that may almost dry out during summers would smolt at a young age and small size.
Here, we studied size and age at smolting and sexual maturity of Brown Trout spawning in streams with a mean annual discharge ≤1 m3 s−1 in south and mid-Norway. We hypothesized that the sizes and ages at smolting and maturity would increase with mean annual water discharge of the streams.
Materials and methods
A total of 774 anadromous and 797 freshwater resident Brown Trout were sampled on their spawning grounds in 12 streams in south Norway (numbers 1–12 in Table 1) and five streams in mid-Norway (numbers 13–17 in Table 1). The fish were sampled in the spawning area of each stream with electric fishing gear (1600 V DC unloaded) (Bohlin et al. 1989) during the first part of October 1992–1997, except for trout from the River Dyrvo, which were sampled in October 1977. All streams were small and located to different catchments with annual mean water discharges between 0·04 and 1·0 m3 s−1. For each fish, natural tip length (mm) (Ricker 1979) was measured, and sex and degree of sexual maturation were determined according to Lagler (1978). Scale samples were used (1) for age determination, (2) to distinguish between anadromous and resident trout, and (3) for back-calculation of smolt length for anadromous trout by use of linear regression (Jonsson & Stenseth 1976). Sea-age at maturity is given as the number of growth seasons the anadromous trout have spent at sea from smolting to maturation.
Table 1. Streams sampled, mean annual water discharge and total number of anadromous females and males and freshwater resident males sampled in each stream
No of anadromous
Water discharge (m3 s−1)
No of resident males
The mean annual water discharge (Q m3 s−1) of the various streams was estimated from the size of the catchment (a km2) and the mean annual run-off (q l s−1 km−2): Q = aq/1000. The drainage area of each stream was marked on 1 : 50 000 maps, and measured by use of a planimeter. Mean annual water runoff was determined from runoff maps (NVE 1987).
There were few resident females in all streams, with 3·7% or 3·5 fish as the average number on the spawning grounds in the autumn. Sexually mature, resident males constituted 48·9% of all adults, whereas anadromous females and males made up 20·0% and 27·5%, respectively. Owing to the small sample of resident females, they were not used in the later analyses. No consistent trend was found in the proportion of resident vs anadromous fish in the populations relative to the size of the streams (Spearman rank correlation, r = 0·32, P = 0·2).
Mean smolt length of the anadromous spawners increased with mean annual water discharge in the smallest nursery streams, but this trend levelled off at water discharges of approximately 0·1 m3 s−1 (Fig. 1a). There was no sexual difference in mean smolt length or age within any of the streams (anova, all P > 0·05). The sexes were therefore treated together. In two of the smallest streams, Grefstad (0·04 m3 s−1) and Sæveli (0·05 m3 s−1), mean smolt lengths ± SD were particularly low, only 84 ± 36·8 mm and 67 ± 34·3 mm, respectively. There was variation in smolt length among similar sized streams, and the largest smolts were found in the Slimestad (185 ± 40·1 mm) with a mean annual water discharge of 0·43 m3 s−1.
Mean smolt length correlated significantly with smolt age (all data log-transformed, r2 = 0·75, P < 0·001). The longer time the parr stay in fresh water, the larger they become, independent of latitude or mean water flow. This was seen by use of multiple regression analysis using smolt length as the dependent variable and smolt age (P < 0·001), water flow (P > 0·9) and latitude (P > 0·9) as independent variables. Because of this, mean smolt age of the population correlated significantly with mean annual water flow (Fig. 1c). The fit, however, was less good than for smolt length (r2 = 0·49, P = 0·002). The two smallest streams, Grefstad and Sæveli in south Norway, had particularly young smolts, 1·6 and 1·2 years, respectively.
Smolt lengths were more variable in small than large streams (Fig. 1b). The coefficient of variation (CV) was negatively correlated with mean annual water discharge of the stream (r2 = 0·61, P < 0·01). Latitude (°N) as a second independent variable was not significantly correlated with the variation in smolt length (P > 0·1). Similarly, the coefficient of variation in smolt age decreased with mean water discharge (Fig. 1d). Latitude did not add significantly to the model (P > 0·1).
Mean body length at sexual maturity (L mm) increased significantly with increasing mean annual water discharge of the spawning and nursery stream (Fig. 2). The relationship differed between anadromous males and females (ancova comparison of slopes: F1,29 = 2·75, P > 0·05, comparison of intercepts: F1,30 = 23·11, P < 0·001) and between anadromous and resident males (comparison of slopes: F1,30 = 1·82, P > 0·05, comparison of intercepts: F1,31 = 123·24, P < 0·001). Adding latitude (B°N) and mean smolt length (S mm) as independent variables to mean annual water discharge (Q m3 s−1), the coefficient of determination increased from 0·45 to 0·90 for anadromous males (ln L = 0·0784 ln Q + 2·77 ln B + 0·324 ln S − 7·844; F3,13 = 38·87, P < 0·001; constant and all three independent variables, P < 0·05). For anadromous females, the addition of latitude increased the coefficient of determination from 0·38 to 0·45, whereas mean smolt length did not add significantly to the model (ln L = 0·074 ln Q + 0·013B + 5·28; F2,13 = 5·33, P = 0·02; constant and the two independent variables, P < 0·05). This means that for both sexes, size at sexual maturity of anadromous trout was larger in small brooks in the northern than in the southern sampling area. Moreover, in anadromous males but not in females, size at maturity also increased with smolt age (Fig. 3a). In resident males, latitude did not add significantly to the model. Analysing the streams in southern Norway alone (58–608°N), the water discharge explained 70% of the variance in size at sexual maturity in anadromous males, 61% in anadromous females and 54% in resident males. The five mid-Norwegian streams alone gave no significant correlation between Brown Trout size and water discharge (P > 0·05).
Mean age at maturity (M years) of male anadromous trout increased with the mean water discharge of the stream, but the fit was poorer than for mean length at maturity (males: ln M = 0·122 ln Q + 1·375; r2 = 0·29, F15 = 6·23, P < 0·03). Adding latitude as a second independent variable increased the coefficient of determination to 0·59 (P < 0·01). However, mean smolt age (A years) also correlated positively with mean age at maturity (Fig. 3b), and used in a multiple regression model together with mean water discharge and latitude, water discharge was no longer significantly correlated with mean age at maturity (P > 0·05): ln M = 1·813 ln B + 0·637 ln A − 6·746, r2 = 0·77, F2,14 = 23·38, P < 0·001; constant and the two independent variables, P < 0·05). For anadromous females, neither latitude nor mean smolt age correlated significantly with mean age at sexual maturity in models also having water discharge as the first independent variable: M = 1·018Q + 3·78, r2 = 0·35, F14 = 7·57, P < 0·02). Mean age at maturity of resident males was not significantly correlated with mean annual water discharge (r2 = 0·08, P > 0·05) or latitude (total r2 = 0·12, P > 0·05).
The number of years a trout feeds at sea before maturing sexually may be influenced by the smolt age. Mean sea-age at maturity (R years) was found to correlate negatively with mean smolt age (A years) in females (ln R = 1·40 − 0·558 ln A;r2 = 0·48, F14 = 13·08, P < 0·01), but not in males (r2 = 0·008, P >> 0·05). Latitude as an independent variable explained 36% (P = 0·01) of the variation in sea-age at maturity in males, but not in females (r2 = 0·06, P = 0·4). Furthermore, the mean length increment at sea of first time spawning females (T mm) decreased with increasing smolt length (S mm) (T = 366·4 − 1·17S;r2 = 0·28, F14 = 5·40, P < 0·05), but not for males (r2 = 0·20, P > 0·05). In males, latitude (B°N) explained most of the variance in sea growth (T = 13·96B − 670·44, r2 = 0·65, F15 = 27·61, P < 0·001). For females, no similar significant correlation with latitude was found (r2 = 0·02, P >> 0·05).
Brown Trout exhibited large among-stream variation in life-history traits which correlated with mean water discharge of their spawning and nursery streams. The population genetics of Brown Trout vary among streams and even between neighbouring populations within the same river (Ryman, Allendorf & Ståhl 1979; Ferguson & Taggart 1991; Hindar et al. 1991), and here we show that important production parameters such as mean age and size at smolting and sexual maturity of Brown Trout populations correlate positively with the water flow in small streams, in accordance with our hypothesis.
Smolts were on average larger and older in large than small brooks, whereas the individual variation in size and age decreased with increasing stream size. The more heterogeneous smolt sizes in the smallest streams may reflect a more variable success for large smolts in small than large streams. In some, but not all, years it may be profitable to leave small streams young and small, e.g. because of droughts (Elliott 1994). Large streams have probably enough water to reduce the importance of flow variation for the size when the trout should leave for feeding in the sea, and factors influencing the sea survival may become more important for the size at smolting.
The more variable age at smolting may be a consequence of the correlation between age and size of the fish. However, this may reflect that age is a discrete (rather than continuous) variable. It is a statistical fact that if the smolts in a stream are all aged either 1 or 2, the CV is much larger than if the smolts are all aged 2 or 3. Therefore streams with low discharge, which produce younger smolts, will also have greater values for CV in smolt age.
Within age-group variation in trout size may also be related to variation in fish density. In Black Brows Beck, CVs for mass and length decreased with density of the different year-classes, so the greatest size ranges were found in years with the lowest densities. CVs for length decreased from about 19% at the lowest densities to about 7% at the highest density (Elliott 1994). We do not have precise estimates of parr density enabling us to test if density variation of parr correlated significantly with the variation in size or age at smolting.
Annual mean water discharge explained more than 60% of the among-population variance in smolt size. One reason why this is not higher may be that minimum and not mean water flow is the important limiting factor for trout size. Sæveli and Grefstad streams are particularly small during summer (Ingebrigtsen 1998). Summer droughts are more common in southern than mid-Norwegian streams. The catchments of the smallest streams studied in south Norway have mean annual runoffs of 20–30 l s−1 km−2. The mid-Norwegian streams are in runoff areas of 30–40 l s−1 km−2 (NVE 1987). Unfortunately, however, we have no systematic study of minimum water flow to allow us to analyse how the minimum water discharge of the stream correlates with smolt size and age.
L’Abée-Lund et al. (1989) reported a weak, positive relationship between smolt length and water discharge. In that study, however, only 1 of 34 streams was smaller than 1 m3 s−1 in mean annual water discharge. Thus, as found in the present study, stream size appear to constrain body size of anadromous Brown Trout in small streams, but the effect of stream size is minute (if any) in large spawning rivers. During dry summers, water is found in only a few pools in the two smallest of our study streams, Sæveli and Grefstad (Ingebrigtsen 1998). To survive, large individuals probably have to leave the stream early in the season, and even small ones may suffer from lack of water during dry summers when the fish become concentrated in pools. When there is no lake in the system, we have never found spawning sea trout in streams smaller than those investigated here (B. Jonsson, N. Jonsson, E. Brodtkorb and P. J. Ingebrigtsen, personal observations), and we feel that a annual mean water flow of 0·04 m3 s−1 is close to the lower limit for spawning and nursery streams of anadromous Brown Trout.
Mean smolt sizes in the smallest streams are smaller than those from Atlantic Brown Trout populations reported earlier (review in Jonsson & L’Abée-Lund 1993), although quite small smolts were reported from one brook in south Norway (Borgstrøm & Heggenes 1988). Moreover, Titus & Mosegaard (1989, 1992) and Järvi et al. (1996) reported very small smolts in south-east Swedish streams flowing to the Baltic Sea. However, as the salinity in the Baltic is low, ionic regulation in sea water should not be problematic for small Brown Trout (Hoar 1976). On the Skagerrak coast, the area used for feeding by the south Norwegian populations, the trout probably have to find a way to avoid the high salt content of the water. Most probably, the small fish stay in brackish near-surface water, not moving far from the river estuary before they are large enough to regulate their salt content in sea water.
We do not know to what extent size and age at smolting are genetically determined or to what extent the fish leave the stream as a plastic response to adversity in the nursery stream (Taylor & Taylor 1977; Thorpe 1987). Refstie, Steine & Gjedrem (1977) found genetic variation for age at smolting among populations of the closely related Atlantic Salmon, and it is reasonable to expect some degree of heritability for this trait also in Brown Trout. Recently, however, Forseth et al. (1999) found that young Brown Trout, which left the nursery stream and migrated into a downstream mountain lake for feeding, had a smaller energetic surplus from feeding in the stream just prior to the descent than those that stayed on. This may indicate that their migration started as a direct response to declining feeding and growth performance. To what extent parr of anadromous Brown Trout respond similarly to adverse feeding conditions in the nursery stream is not known. Investigations in the present streams, however, have revealed that trout lipid energy reserves are minimal at the time of smolting, and similar to those of fish exhausted from spawning (Jonsson & Jonsson 1997, 1998).
In amphibians, the larvae exhibit extreme plasticity in age and size at metamorphosis, which in addition to pond duration is influenced by water temperature (Heyes, Chan & Licht 1993), intra- and interspecific signalling (Werner 1986), food availability (Berven & Chandra 1988), risk of predation (McCollum & Van Buskirk 1996) and density (Scott 1990), making the pond hostile for the larvae. Similarly, factors making the nursery stream hostile for young Brown Trout may also influence size and age at smolting.
Adult size correlated positively with stream size, paralleling findings in Atlantic Salmon where the small streams (<10 m3 s−1) harbour smaller fish than large rivers (N. Jonsson et al. 1991). In Brown Trout, adult size appears not to be correlated with stream size for rivers larger than approximately 1 m3 s−1 (L’Abée-Lund et al. 1989), where other factors probably limit body size such as temperature and osmoregulatory conditions at sea. In addition, large migratory costs in long rivers may select for large adult body size (Schaffer 1979; L’Abée-Lund 1991). The present rivers, however, are short decreasing the importance of stream length.
Anadromous males were smaller than females. The size difference seemed particularly large in the smallest streams. A proximate reason for a smaller size of anadromous males in small rather than large streams may be that their adult size increased with the smolt size, and the smolts were especially small in the smallest streams. Furthermore, females but not males tended to stay longer and grow relatively more at sea the younger and smaller they were as smolts. This should decrease the size dimorphism in large relative to small streams, but the gradual decrease in size dimorphism with increasing stream size was not significant (P > 0·05).
Body size of resident males increased with stream size. This may be because habitat conditions are more adverse for residents in small than larger streams, making it selectively advantageous to mature early when the water level is low. We do not know any other study showing an increase in size of resident male trout with stream size.
Our study shows that Brown Trout can smolt early in life and at a small size, probably as an adaptation to low water level in the nursery stream, as also indicated by the higher variability in smolt size in small than large streams. Also adult size seems to be adapted to stream size as small streams have smaller spawners than larger streams. The size dimorphism between anadromous males and females appeared particularly large in the smallest streams. This is probably connected to different relationships between size, age and sexual maturity between the sexes.
We thank Hans M. Berger, Ørnulf Haraldstad, Pål A. Larsen and Dag Matzow for helpful cooperation during field work, and J. M. Elliott and Neil Metcalfe for commenting critically on drafts of the paper. Financial support was received from the Norwegian Research Council, the Norwegian Directorate for Nature Management and the Norwegian Institute for Nature Research.
Received 14 September 2000; revised 25 October 2000; accepted 7 November 2000