Stock-Specific Growth and Length Frequency Bimodality in Brown Trout

Introduction

Like other salmonids, brown trout Salmo trutta exhibit a range of phenotypically variable traits, such as coloration (Skaala and Jørstad 1987), growth (L'Abee-Lund et al. 1989), the age and size of a smolt at first entry into seawater (Borgstrøm and Heggenes 1988; Titus and Mosegaard 1989; Økland et al. 1993), and morphological characteristics (Yevsin 1976). Further, genetic variation within the brown trout (as estimated from isoenzyme loci) is extensive (Taggart et al. 1981; Ferguson 1989). Compared with other salmonid species, a greater proportion of the total genetic variation within the brown trout is distributed between as opposed to within populations (Ryman 1983). In addition, examples of genetic influence on phenotypic variation in brown trout can be found for characters such as a specific pigmentation pattern (Skaala and Jørstad 1988), smoltification (Soivio et al. 1989), life history and behavioral variation (Palm and Ryman 1999), susceptibility to salmon lice infection (Glover et al. 2001), and morphology (Pakkasmaa and Piironen 2001).

In general, the genetic effects on growth are well documented in salmonid fish (see Jørstad and Nævdal 1996). In Atlantic salmon S. salar, for example, heritability (h²) for growth rate has been estimated between 0.2 and 0.3 (see Gjedrem 2000). In addition, genetic differences in growth between wild stocks under controlled conditions have been observed in Atlantic salmon (Gunnes and Gjedrem 1978) and chinook salmon Oncorhynchus tshawytscha (Withler et al. 1987), as well as among rainbow trout O. mykiss strains used in aquaculture (Linder et al. 1983). However, despite the fact that growth has been shown to vary greatly among brown trout populations in their natural habitats (Edwards et al. 1979; L'Abee-Lund et al. 1989; Elliott 1994 and references within), studies of the genetic influence on stock-specific growth rates are limited. Elliott (1989) observed no genotypic differences in mean growth rates between two brown trout populations reared in a laboratory. However, in that particular study, only two sibling groups were reared for each stock, making it difficult to estimate the degree of influence family variation may have had on the result. Accounting for variation between families is important in stock comparisons, especially when studies are based on low numbers of families. In Atlantic salmon, for example, the variation among families within strains may be nearly as great as variation between strains (Gunnes and Gjedrem 1978). Näslund et al. (1992) observed genetic differences in growth rate among six stocks of brown trout. However, since these stocks had very different histories of artificial culture, the observed divergence in growth rates may have been confounded by domestication effects.

In Atlantic salmon reared in culture, length frequency growth curves for a cohort of fish are often bimodal in the first winter of life. This is thought to be the first observable sign of the preparatory changes associated with the migration to seawater in a process known as smoltification (Simpson and Thorpe 1976; Bailey et al. 1980; Metcalfe et al. 1988). Although the proportion of fish entering the upper and lower modes is not fixed, it is size dependant and related to the growth opportunities of individuals (Villarreal and Thorpe 1985). When day length decreases (signaling the onset of winter), individuals larger than the threshold size of approximately 7–8 cm, maintain a high winter growth rate compared with individuals below the threshold size (Skilbrei 1991; Skilbrei et al. 1997). Upper-mode individuals complete the process of smoltification the next spring and migrate to the sea, whereas lower-mode individuals delay migration. This phenomenon is well studied in Atlantic salmon, and has been observed in several other salmonid species (Clarke and Shelbourne 1986; Hirata et al. 1988; Yamamoto and Nakano 1996) and in wild populations of salmonids (Nicieza et al. 1991; Debowski 1997). However, there is little documentation of length frequency bimodality in brown trout (Elliott 1989; Tanguy et al. 1994; Debowski 1997). In addition, other changes associated with the process of smoltification (e.g., silvering and peak of gill Na⁺,K⁺-ATPase activity) appear to be limited and highly variable in anadromous brown trout (Soivio et al. 1989; Tanguy et al. 1994; Pirhonen and Forsman 1998).

It is suggested that as the brown trout exhibits a range of life history strategies and extensive phenotypic variation, growth potential and the patterns of growth may differ among populations. By comparing stocks under common domestic environmental conditions, it is possible to minimize environmental noise so that stock-specific genetic differences for these traits can be studied. Therefore, the aim of the present investigation was to study growth patterns in three anadromous brown trout stocks reared under common environmental conditions while controlling for the family-level component of variation within and among stocks. In addition, the aim was to compare the size distribution of the presumptive smolts from one of the stocks reared in the hatchery with wild recruited migrating brown trout smolts originating from the same river system.

Methods

Materials

In autumn 1998, mature anadromous brown trout adults were collected from the following three rivers located on the west coast of Norway: the river Fortun, located in the inner part of the Sogne fjord (61°29″N, 7°35″E), and the rivers Guddal (59°58″N, 6°00″E) and Sima (60°30″N, 7°08″E) located in the middle and inner parts of the Hardanger fjord, respectively. Two of the stocks (Guddal and Sima) have been screened for biochemical genetic variation at enzyme loci, and differences in allele frequencies have been observed between them (Ø. Skaala, unpublished). Adults were mated between 26 October 1998 and 18 November 1998 to produce 19 families (Table 1).

Table 1. Overview of experimental design, including information on length of parents (L), parental crosses, and numbers of offspring and rearing tanks; Gud = Guddal, Sim = Sima, and For = Fortun

Rearing

Fertilized eggs from Guddal, Sima, and Guddal × Sima (hereafter referred to as the GS hybrid) families were incubated in single-family tanks at a hatchery located on the River Sima at approximately 6°C. Eggs from the Fortun families were incubated in single-family tanks at the hatchery located on the River Fortun at 3–4°C. The temperatures chosen reflect the standard incubation temperatures for brown trout eggs used by these hatcheries up until the eyed egg stage of development. When the Fortun eggs reached the eyed stage, they were transported to the Sima hatchery where they were incubated under conditions identical to those of the other stocks. Two days prior to the initiation of start-feeding, the water temperature was raised to 12°C (±1°C) in each 200-L, single-family tank. Start-feeding for the Fortun families commenced on 28 February 1999, 2 weeks later than all the other families (15 February 1999). On 28 April 1999, all families were transferred to individual family, 3000-L tanks. Between 4–7 August 1999, all individuals from each family were fin-clipped with the following combinations: Fortun, adipose; Sima, left ventral; Guddal, anal; and GS hybrid, right ventral. These fish were then mixed (one family from each stock) into five 12,000-L tanks with ambient water temperature at 7–8°C (Table 1). Mixing families in this manner allowed the unambiguous identification of individuals to both stock and family.

The fish were reared under a standard hatchery light regime to encourage the development of age-1 smolts. This consisted of a constant light regime (24 h light per day) during the entire experiment, except between 1 December 1999 and 14 March 2000 when the fish were reared under short day lengths (8 h light : 16 h darkness). Fish were fed continually during the light hours by automatic feeders. Ration was calculated as a function of the biomass of fish in each tank and was readjusted biweekly throughout the experiment.

Sampling and size grading fish

The fish were sampled on 7–8 July 1999, 7–8 December 1999, and 13 March 2000. Individual length, weight, family, and stock were recorded during these and subsequent samples. In addition, all sampled fish were gently squeezed in the abdominal region to check for mature males (ripe males will express sperm). From the data set collected in December and March, length frequency plots were produced for each stock and for all stocks combined (Figure 1).

On 19 March 2000, all fish were size-graded at 12 cm total length. Size grading was carried out in order to divide the fish into presumptive smolts (upper-mode fish) that were to be transferred to salt water later in the spring, and presumptive nonsmolts (lower-mode fish) that were to remain in freshwater for another year before being transferred to salt water. The length used to divide all the fish into two modes was chosen from a visual inspection of the length frequency plots for the stocks combined (see top of Figure 1). It is important to note that the length used to size-sort the fish is different from the stock-specific dividing lengths between the lower and upper modes that were also estimated from Figure 1. The latter were used to compare percentage of lower and upper-mode fish among families and stocks in March, whereas a single length was used to size-grade all fish due to practical considerations.

After the fish had been size-graded, approximately 50 individuals from the upper-mode fish from each of the 18 families (total of 900 fish) were selected randomly. One family (Guddal G3) was not selected at this stage because it had low numbers of individuals among the upper-mode fish. The selected upper-mode fish were individually Floy tagged and reared together in a single, 2000-L tank for another 7 weeks at 7–8°C. These fish were sampled upon tagging (22–23 March 2000) and 7 weeks later (15 May 2000). On 23 May 2000, the tagged fish were transferred directly to a marine cage where salinity was 32‰ in order to test for tolerance to salinity. Survivorship was observed in these fish for a further 23 d before the experiment was terminated.

From the lower-mode fish that had been placed into a single, 12,000-L tank after size-grading, a random sample was taken from the Guddal (n = 80) and Fortun (n = 69) stocks. These individuals were killed by an overdose of anesthetic and then sexed. The representation of males and females among these lower-mode Guddal and Fortun fish was recorded.

Smolt trap data

A permanent smolt trapping system was built 100 m from the mouth of the River Guddal in the winter of 2000. This system has been monitored daily during the smolt migration in the spring of each year since construction. A range of parameters has been noted from migrating brown trout smolts, as well as from Atlantic salmon smolts. Individual length data from the brown trout smolt migration in 2001 are presented here and are compared with the length frequency distributions of the hatchery-reared fish originating from the River Guddal.

Statistical analyses

Only data from samples taken on 7–8 December 1999, 13 March 2000, and 15 May 2000 are presented. Length frequency plots were used to describe the variation in family and stock lengths at the December and March sample points. These plots were calculated using data transformed from absolute numbers to frequencies so that family contribution to the pattern of stock growth was not biased by the variation in family sample size. The numbers of upper and lower-mode fish for each family and stock were calculated from the sample in March, after setting the cutoff point between lower and upper-mode fish specifically for each stock from inspection of length frequency plots at this time point (Figure 1). A rows × columns G-test of independence was used to examine the differences in relative numbers of upper length category fish between all families. An observed versus expected G-test was used to examine the representation of the sexes in the lower-mode fish for the Guddal and Fortun stocks. The differences in length and weight among families on 16 May 2000 after all fish had been size-sorted were tested by analysis of variance (ANOVA). The differences in length and weight among the stocks (the hybrid was treated as a separate stock) were tested by nested ANOVA (families nested within stocks). These data were then investigated further by the Newman–Keuls post hoc test that comprises specific pairwise comparisons. Tests were performed in Statistica 5.0. A Spearman's rank correlation test was used to investigate potential maternal effects on the length or weight of families.

Results

On 16 May 2000, significant differences in average length (F_3,858 = 72.8, P < 0.001) and average weight (F_3,858 = 72.6, P < 0.001) were observed among the upper-mode fish representing the stocks (three pure and one hybrid; Table 2). Pairwise, post hoc tests revealed that of the pure stocks, the Sima stock was significantly longer and heavier than either the Guddal or Fortun stocks (all P-values <0.001). The Guddal stock was slightly heavier than the Fortun stock (P < 0.015), although not significantly longer (P > 0.9). The hybrid stock was significantly longer and heavier than both its parental stocks as well as the Fortun stock (all P-values <0.001; Table 2).

Table 2. Mean size and percentage of upper-mode fish among stocks (SD values in parentheses). Percentages of upper-mode fish were calculated by averaging the percentages for each family shown in Table 3. One Guddal family (G3) was excluded because of too few upper-mode individuals

Significant differences in length and weight were observed among the family groups in the sample taken in May 2000 (F_17,858 = 17.9, P < 0.001; Table 3). The families reared in this study were created from females ranging widely in size (Table 1). A Spearman's rank correlation test was performed to examine the potential for maternal effects on offspring size. However, no correlation was observed between dam length and percentage of upper-mode fish within families in March (P > 0.7; Figure 2), mean family length in May (P > 0.5; Figure 3), or mean family weight in May (P > 0.3; Figure 4).

Table 3. Number of observations, average length (cm), and average weight (g) of the families at the three sampling points (SD values in parentheses). Percentages of upper-mode fish in March are given

Differences in growth patterns were observed among the stocks. Length frequency distributions of both Fortun stock and Guddal stocks displayed clear bimodal distributions in March 2000 (Figure 1). For Guddal and Fortun stocks, bimodality was first observed in December while fish were being reared under constant light, but the degree of bimodality increased significantly in the period December 1999 to March 2000 under short day lengths (Figure 1). Length frequency distributions of the Sima and the GS hybrid stocks displayed weak length frequency bimodality in March, with most fish being in the upper-mode group (Figure 1).

The dividing length between the upper and lower modes was estimated separately for each stock in March. The dividing length was 12 cm for Guddal, 11.5 cm for Fortun, 13 cm for Sima, and 13 cm for the GS hybrid stock. Significant differences in the representation of fish in the upper and lower modes were observed among family groups (rows × columns G-test of independence, P < 0.001; Table 3). In addition, differences in the distribution of individuals in the upper and lower modes were observed among stocks (Table 2), although this was not tested statistically because data originated from heterogeneous sources (families).

During winter, no ripe males were detected for any of the stocks. Furthermore, there were no significant differences in the sex ratio in a random sample of the fish below 12 cm in length on 19 March 2000 for either Guddal (observed-versus-expected G-test, 39 females versus 41 males, P > 0.9) or Fortun (observed-versus-expected G-test, 34 females versus 35 males, P > 0.9). Therefore, male maturation and sex-specific growth rates were not implicated in the recorded length frequency bimodality of these two stocks. Furthermore, the mortality of the 900 tagged upper-mode fish which were transferred directly to a marine cage for 23 d was less than 1% (n = 5), indicating that these fish were capable of tolerating salt water.

The length frequency distributions of the naturally recruited brown trout smolt migration (n = 1,091) from the River Guddal between April and June 2001 is presented in Figure 5. It is clear that the majority of individuals migrating to the sea from the River Guddal were between 12 and 18 cm total length (median length, 14.5). The size range of these individuals was in accordance with the size range of the hatchery-reared upper-mode Guddal fish (Figures 1, 5), although it is important to note that the dates of sampling for these two groups were not identical.

Discussion

Three significant observations were made in the present study. First, differences in growth were observed among three pure brown trout stocks. Second, the hybrid stock outgrew both of its two parental stocks. Because stocks were reared in mixed family tanks for the majority of the duration of this study, and no significant maternal effects on offspring size or percentage of upper-mode fish were detected, we conclude that the results probably reflect genetic differences in growth potential between the stocks under the described experimental conditions. Third, these data are the first to show clear evidence of length frequency bimodality development in hatchery-reared brown trout. We suggest that the observed patterns of length frequency bimodality development in the brown trout stocks reared in this study are linked to smoltification, similar to that observed in Atlantic salmon.

The Sima stock was significantly larger than both the Fortun and Guddal stocks, both of which displayed similar lengths but different average weights. The observed difference in growth between individuals from the Sima and Guddal stocks in the present study is in accordance with Borgstrøm and Skaala (1999), who compared the sea growth of naturally recruited individuals from these two rivers. They reported a significantly greater length increase in the Sima as compared with the Guddal trout.

The differences in mean fish size among stocks in the sample taken in May are an underestimate of the actual growth differences between the stocks since this sample included only the upper-mode fish for each stock. The Sima stock not only had the largest upper-mode fish among the pure brown trout stocks, it produced more upper-mode fish (which also reflects growth differences among the stocks). Additionally, the average size of upper-mode fish calculated in May for the Sima and the GS hybrid stocks are underestimates of the actual sizes of the upper-mode fish for these two stocks. This is due to the fact that all stocks were size-sorted at 12 cm total length in March, resulting in the presence of a few lower-mode individuals within these two stocks. Consequently, the average size of the upper-mode fish for these two stocks may have been slightly decreased.

In the present study, the GS hybrid stock significantly outgrew both the pure Sima and pure Guddal stocks and produced a higher percentage of upper-mode fish than either stock. Although the hybrid families were not established according to a strict half-sibling family design, the fact that the final average length and average weight of all the hybrid families were greater than the average lengths and weights of either the Sima or Guddal families strongly suggests hybrid vigor, although this may also reflect outbreeding depression (e.g., Gharrett and Smoker 1991). Hybrid vigor may occur as the result of mating between individuals from genetically distinct strains or populations (e.g., Gjerde 1988). The apparent hybrid vigor between the Sima and Guddal stocks, in conjunction with the differences in growth between them as reported here and in the study of Borgstrøm and Skaala (1999), support the conclusion that these two stocks are genetically distinct in a quantitative genetic trait (i.e., growth rate).

Significant differences in the growth and percentage of upper-mode fish were observed among the families reared in this study. These differences reflect, to a certain degree, genetic differences in the growth potential among the families. However, families were reared first in single-family tanks and then in mixed-family tanks. Consequently, it is not possible to identify the degree to which the differences in growth observed among families were affected by environmental tank effects. Furthermore, although no effect of dam length on family growth was detected in this study, other potential maternal effects (such as egg size or quality) cannot be ruled out as influencing the growth and final size attained by the families.

The observation that two of the stocks reared in this study developed clear length frequency bimodality under controlled conditions is significant. Although length frequency bimodality is well studied in cultured Atlantic salmon and has been observed in other salmonid species (Clarke and Shelbourne 1986; Hirata et al. 1988; Yamamoto and Nakano 1996), length frequency bimodality is not well documented in brown trout. In brown trout, length frequency bimodality has been observed within a natural habitat (Debowski 1997) and multimodality has been observed in fish reared in a hatchery and in the wild (Elliott 1989). However, in a controlled hatchery study, Tanguy et al. (1994) found no evidence of length frequency bimodality in an anadromous and a freshwater resident population of brown trout. Furthermore, these authors suggested that length frequency bimodality had not been previously observed in brown trout.

Data from the present study strongly suggest that the development of length frequency bimodality in the three stocks reared in this study are likely linked to the process of smoltification. The growth patterns of the brown trout stocks reared in this study displayed remarkable similarity to the growth patterns and development of bimodal length frequency distributions in Atlantic salmon (Thorpe et al. 1982; Kristinsson et al. 1985; Skilbrei 1988, 1991; Skilbrei et al. 1997). Fish within the Guddal and Fortun stocks displayed divergent growth rates in the period December 1999 to March 2000, and were separated by 12 cm and 11.5 cm, respectively, at the end of this period. This is similar to the situation in Atlantic salmon, where only the upper-mode fish continue to grow significantly in the winter, in preparation for smoltification. We also observed this pattern of growth within the Sima and the GS hybrid stocks, but higher overall growth within these two stocks lead to a higher percentage of upper-mode fish, and therefore, less clear bimodal distributions. Importantly, there was no evidence to suggest that skewed sex ratio or early maturation promoted bimodality in the Guddal and the Fortun stocks in the present study. The low mortality (<1%) of the 900 upper-mode fish transferred to a marine cage in 32‰ seawater for 23 d indicates a tolerance to salinity and, therefore, their physiological preparedness for entry into seawater. This is supported by the similarity in the size range of the upper-mode Guddal fish in the hatchery to the size range of the naturally recruited brown trout smolts caught migrating to the sea from the River Guddal.

In summary, significant differences in growth, thought to reflect genetic growth potential, were observed among three pure brown trout stocks and among a hybrid brown trout stock and its two parental stocks. The differences in growth among the stocks may reflect adaptive differences to their local environments. In addition, a variable degree of length frequency bimodality was observed among the stocks. It is concluded that the pattern of length frequency bimodality development observed within the brown trout stocks reared in this study is linked to the process of smoltification.

Acknowledgments

We acknowledge the work of the staff at the Statkraft Hatchery in rearing the fish, and thank J. B. Taggart and referees for comments on the manuscript. This project was carried out with funding from the Norwegian Electricity Association (Enfo) and the Norwegian Research Council (NFR).