Enhanced growth reduces precocial male maturation in Atlantic salmon


Correspondence author. E-mail: dmoreau@mun.ca


1. Understanding the proximate and ultimate mechanisms shaping the expression of alternative reproductive phenotypes is a fundamental question in life-history evolution. Precocial maturation in fishes, one such alternative phenotype, has been thought to reflect rapid growth and/or energy accumulation; however, mechanistically linking these specific traits to discrete life-history patterns is complex and poorly understood.

2. Here, we use growth hormone (GH) transgenic Atlantic salmon to elucidate the effects of intrinsically fast growth on precocial male maturation as parr (freshwater life stage). Despite facilitating growth to sizes typical of mature wild-type parr, transgenesis did not influence maturation in the first year of life. In the second year, the number of maturing transgenic parr was only half that of non-transgenic individuals.

3. By manipulating intrinsic growth and controlling for both environment and genetic background, this study provides direct empirical evidence suggesting that the physiological mechanisms promoting growth do not play a causative role in precocial male maturation in fishes.

4. In addition, this study provides the first empirical data on the relative incidence of precocial male maturation in GH transgenic and non-transgenic Atlantic salmon and, therefore, provides valuable information for the ecological risk assessment process.


Phenotypic variation within populations can be expressed in discrete life-history forms that evolve through disruptive selection. A well-documented category of discrete, alternative phenotypes are those associated with reproductive competition, where conspecific (usually male) individuals adopt discrete traits or tactics that are believed to be part of a single conditional strategy maintained by both frequency- and condition-dependent selection (Hutchings & Myers 1994; Gross 1996; Calsbeek et al. 2002). Many of these alternative phenotypes are controlled by threshold switches, which are genetically determined, internal triggers that direct the expression of one phenotype over another and are responsive to the environmental conditions experienced during ontogeny (Shuster & Wade 2003; Oliveira, Taborsky & Brockmann 2008). Threshold switches are thought to initiate distinct internal resource allocation pathways leading to alternative phenotypes; however, the underlying proximate mechanisms are not well documented.

Atlantic salmon, Salmo salar L., provide a model system with which to study the underlying mechanisms influencing alternative reproductive phenotypes (Fig. 1). The breeding system includes anadromous (older, ocean migratory) and precocially maturing male parr (younger, non-migratory) phenotypes that compete for fertilizations with spawning females. Anadromous males, which are large and display specialized secondary sexual characters, fight for access to breeding females, while mature male parr adopt a breeding tactic reliant on small size (often weighing two orders of magnitude less than anadromous males) and crypsis to sneak fertilization attempts (Fleming & Reynolds 2004).

Figure 1.

 Mature male Atlantic salmon parr (Salmo salar L.) blend with a characteristically rocky stream bed habitat. Photograph: Anders Lamberg.

An intriguing feature of this life-history dichotomy involves the proximate and ultimate mechanisms with which stream-dwelling male parr follow a developmental trajectory towards precocial maturation or anadromy. Current evidence suggests that precocious maturation is initiated by parr reaching a polygenic performance threshold that, within populations, correlates with fast growth/size at age (Hutchings & Myers 1994; Aubin-Horth & Dodson 2004; Piché, Hutchings & Blanchard 2008; Páez, Bernatchez & Dodson 2011). Proximate causation is thought to be associated with the available energy (lipid) reserves, and the rate of energy reserve allocation, an individual has attained prior to the onset of environmental triggers initiating sexual development (Thorpe et al. 1998; Letcher & Gries 2003; Mangel & Satterthwaite 2008). A similar proximate causation, however, is thought to underlie an alternative life-history trajectory that involves phenotypic transformation for migration to sea (i.e. smoltification), whereby the largest fish from within a cohort are those most likely to exceed the performance threshold to become sea-migrating smolts. Thus, a developmental conflict may exist between precocial maturation and smoltification.

Proximate models suggest that parr maturation occurs if an energetic threshold is reached and maintained for a year prior to the fall breeding season, while smoltification occurs if a growth rate/size threshold is surpassed 7 months prior to the spring seaward migration (Metcalfe 1998; Thorpe et al. 1998; Mangel & Satterthwaite 2008). Under these models, smolt transformation is viewed as an alternative developmental pathway for individuals that have not obtained enough lipid reserves to produce sufficient gametic tissue by the breeding season. However, the allocation of energy between storage tissues, which may support maturation, and structural tissues (growth), which may promote smoltification, remains poorly understood.

Growth hormone (GH) transgenesis allows a unique opportunity to empirically test this proximate life-history framework by permitting the manipulation of intrinsic growth (Du et al. 1992; Cook et al. 2000a), while controlling for other genetic effects not associated directly with the transgene. Based on observed covariation between size at age, fast growth, and lipid investment with parr maturation, GH transgenesis may allow precocial maturation thresholds to be reached faster and in greater proportion relative to non-transgenic individuals, as previously proposed (Valosaari, Aikio & Kaitala 2008). Fast-growing transgenic parr, however, have greater metabolic requirements and preferentially invest energy into structural rather than storage tissues relative to non-transgenic parr (Stevens, Sutterlin & Cook 1998; Cook, McNiven & Sutterlin 2000b; Cook et al. 2000a). Such physiological differences may alter the internal triggers and conditional requirements necessary to reach maturation thresholds, potentially towards life-history trajectories favouring smolt transformation (Metcalfe 1998; Saunders, Fletcher & Hew 1998; Thorpe & Metcalfe 1998; Letcher & Gries 2003; Páez et al. 2011). Thus, GH transgenesis may provide empirical insight, allowing the separation of proximate mechanisms responsible for parr maturation from characteristics that covary.

Using mixed populations of GH transgenic and non-transgenic Atlantic salmon siblings, we test the effect of growth on precocious maturation. If growth has a causative role in parr maturation, then GH transgenesis should increase the likelihood of early male maturation over the first and second years of life. However, if energy accumulation causes maturation, then GH transgenesis may reduce the incidence of parr maturation owing to higher routine metabolic rates and preferential investment in structural tissues.

Materials and methods

During September 2006, wild adult Atlantic salmon females were collected from the Exploits River, Newfoundland, Canada, and transferred to the Ocean Science Centre, Memorial University of Newfoundland. Upon egg ripening (21 November–12 December), male gametes of hemizygous GH transgenic Atlantic salmon [Gene construct: opAFP-GHc2; transgene: EO-1α (Yaskowiak et al. 2006)] were crossed with the wild females to produce eight single family crosses. The background genome of this transgenic strain is derived largely from Saint John River (NB, Canada) salmon. True to Mendelian inheritance patterns, such hemizygous crosses result in c. half the offspring in each family inheriting the GH transgene (Shears et al. 1992). This allows comparison of full siblings, facilitating the control of general genetic background and maternal effects.

During early ontogeny, all families were reared separately in Heath incubation trays. At first feeding (28 May 2007), families were moved into separate, randomly assigned individual compartments of two rearing troughs (261 × 24·5 × 10 cm) and fed 4–8 times daily with a combination of Artemia spp. and a salmonid starter dry feed (Corey Feed Mills, Fredericton, NB, Canada). Temperature (3–16 °C) and photoperiod were kept at ambient conditions throughout the lives of these animals.

On 30 July 2007, 32 fry from each of eight families were haphazardly assigned to one of the six 1-m2 circular rearing tanks (n = 256 per tank) and subsequently fed dry feed from automated feeders every 30 min throughout each day. Initially, the tank replicates were split into high (8% tank biomass per day) and low feed (2% tank biomass per day) treatments. In October 2007, both high and low feed treatments were decreased to accommodate reduced feeding levels (i.e. to 4% and 1% tank biomass per day, respectively). In January 2008, high and low feed treatments were discontinued and feed levels for all six tanks were reduced to 1–3 hand feedings daily. Previous observations of salmon from the same source wild population indicate that such conditions produce high male maturation rates (c. 50%) in the second year of life (1+; D. T. R. Moreau, unpublished).

In February 2008, the number of individuals in each tank was reduced to 100 to accommodate expected biomass increases in the following spring/summer growing season. To prevent sampling bias in both family and transgenic composition, a mass frequency-based selection process (5-g intervals) was used whereby the individuals were haphazardly removed in a manner consistent with maintaining the mass distribution within each tank. To assess the ratio of transgenic to non-transgenic fish, fin clips from a representative sample of fish (n = 40 per tank, across the size distribution of fish present) were placed in microcentrifuge tubes containing 99% ethanol and subsequently screened for the transgene using a previously described polymerase chain reaction (PCR) protocol (Deitch et al. 2006).

In 2007 and 2008, male maturation was assessed once weekly between mid-October and the end of December, by gently squeezing the belly along the length of the body and looking for the presence of sperm at the genital papilla. The mass (g) and fork length (mm) of each mature parr were documented. Subsequently, a tissue sample of each mature parr was collected for transgenic identification, and the animals were euthanized with MS-222 (Western Chemical Inc., Ferndale, WA, USA) prior to being frozen whole (−20 °C). Frozen gonadal and whole body mass (g) were later collected to determine the gonadal-somatic investments of transgenic and non-transgenic mature parr.

Data Analyses

Logistic regressions with binomial error (LR) were used to test for tank effects on proportions of transgenic parr and to evaluate the proportion of mature parr with respect to feed level during the early growth phase (July–December 2007) and genotype (transgenic or non-transgenic). General linear models (GLM) were used to compare fork length, body mass, gonad mass and body condition [mass as the response variable and length as a covariate (García-Berthou 2001)] with respect to genotype and early feed level. Data fit with GLM’s that did not meet the requirements of normality was either fit to a gamma distribution (link: inverse) or natural log transformed prior to insertion into a linear model. All data were analysed using the R statistical software application (version: R-2.10.1.; http://www.r-project.org).


In 2007, the first year of life (0+), 1·3% of the total population consisted of mature male parr (Fig. 2). While more transgenic (n = 11) than non-transgenic (n = 8) parr matured, there was no significant difference in the rates of maturity (LR; n = 19; χ2 = 0·90; P = 0·35). Similarly, feed level did not influence the rates of early maturity (LR; n = 19; χ2 = 0·18; P = 0·67). Among 0+ mature parr, total body mass and fork length were greater in high than in low feed tanks (GLM; n = 19; mass: χ2 = 14·26, P < 0·01; length: χ2 = 13·10, P < 0·01) and transgenic fish were larger than non-transgenics (GLM; n = 19; mass: χ2 = 102·05, P < 0·01; length: χ2 = 48·82, P < 0·01; Fig. 3a,b). Testing for differences in body condition (length-adjusted mass) between 0+ transgenic and non-transgenic mature parr indicated a three-way interaction between length, feed level and genotype (n = 19; P < 0·001; Fig. 4a). Qualitatively, the transgenic parr tended to have a higher mass for a given length than the non-transgenic parr.

Figure 2.

 The incidence (%) of mature male transgenic and non-transgenic Atlantic salmon parr (Salmo salar) during the first (0+) and second (1+) years of life. High and low feed levels were applied only during the first year of life. The error bars represent the 95% confidence intervals around the mean.

Figure 3.

 The mean (a) wet mass (g) and (b) fork length (mm) of transgenic and non-transgenic precocious male Atlantic salmon (Salmo salar) during the first (0+) and second (1+) years of life. High and low feed levels were applied only during the first year of life. The error bars represent the 95% confidence intervals around the mean.

Figure 4.

 The length–mass relationship of transgenic and non-transgenic precocious male Atlantic salmon (Salmo salar) during the first (a) and second (b) years of life. High and low feed levels were applied only during the first year of life.

Among the immature 0+ fish, the proportion of transgenic (n = 119) to non-transgenic (n = 120) parr did not differ (Exact Binomial test; χ2 = 0·12, P = 0·73); moreover, this pattern was consistent across tanks (LR; n = 239; χ2 = 0·44, P = 0·51). There was a strong interaction between feed level and genotype on fish mass (P < 0·01), indicating that transgenic and non-transgenic parr respond differently to the feed treatments. Transgenics outgrew non-transgenics, being larger in both high (mean ± SE transgenics: 31·85 ± 1·26 g; non-transgenics 6·39 ± 0·25 g; GLM; n = 119; χ2 = 783·37, P < 0·01) and low feed treatments (transgenics: 18·45 ± 0·71 g; non-transgenics 6·41 ± 0·54; GLM; n = 120; χ2 = 127·43, P < 0·01). However, the size of non-transgenics did not differ between feed levels (GLM; n = 120; χ2 = 0·001, P = 0·98), while that of transgenics did (GLM; n = 119; χ2 = 97·62, P < 0·01).

In 2008, the second year of life (1+), 35% of the total population consisted of mature male parr (Fig. 2), of which non-transgenics (N = 129) were 1·8 times more likely to mature than transgenics (n = 70; LR; n = 199; χ2 = 14·12; P < 0·01). Maturation was not influenced by the feed level in the first year (LR; n = 199; χ2 = 1·55; P = 0·21). Similar to the immature parr, mature parr showed strong interactions between feed level and genotype for all size measures (P < 0·05). Mature transgenic parr outgrew their non-transgenic counterparts, being larger irrespective of whether they spent their first year of life in the high (GLM; n = 106; mass: χ2 = 563·22, P < 0·01; length: χ2 = 542·74, P < 0·01) or low feed treatments (GLM; n = 91; mass: χ2 = 69·52, P < 0·01; length: χ2 = 68·76, P < 0·01; Fig. 3a,b). Under low feed, body condition (length-adjusted mass) did not differ between mature transgenic and non-transgenic parr (GLM; n = 91; χ2 = 2·21, P = 0·130; Fig. 4b). Under high feed, a strong interaction occurred between length and genotype (N = 106; P < 0·001). These data qualitatively suggest that non-transgenic parr had a greater length-adjusted mass than transgenic parr at the larger end of the size distribution, with the opposite pattern occurring among smaller fish. However, the small overlap in size distribution between transgenic and non-transgenic parr makes interpretation difficult. The growth effects of differing feed levels during the first year persisted among transgenic mature parr into the following year, with those from high feed being larger than those from low feed at age 1+ (GLM; mass: n = 70; χ2 = 10·30, P < 0·01; length: n = 70; χ2 = 11·43, P < 0·01; Fig. 3a, b). However, the opposite occurred among non-transgenic parr, with those experiencing low feed during the first year being larger than those experiencing high feed (GLM; mass: n = 129; χ2 = 14·05, P < 0·01; length: n = 129; χ2 = 7·02 P < 0·01).

Comparing size across years, mature 0+ transgenic parr (29·6 ± 3·5 g; mean ± SE) did not differ in total body mass from that of 1+ non-transgenic parr (27·2 ± 1·4 g; GLM; n = 140; χ2 = 0·22, P = 0·64; Fig. 3a). However, 1+ non-transgenic parr (132·4 ± 1·7 mm; mean ± SE) were significantly smaller in length (113·6 ± 9·9 mm; GLM; n = 140; χ2 = 8·58, P < 0·01; Fig. 3b). Thus, mature 0+ transgenic parr were heavier for their length relative to 1+ non-transgenic parr (GLM; n = 140; χ2 = 15·50, P < 0·01).

At age 1+, the absolute gonadal mass of mature transgenic parr was greater than that of non-transgenic parr (GLM; n = 163; F = 15·69, P < 0·01); however, this was mainly because of their larger body size (Fig. 5). For a given body mass, non-transgenic parr actually invested proportionately more in gonadal mass than transgenic parr (GLM, n = 163, slope: F = 0·61, P = 0·55; intercept: F = 224·20, P < 0·01). Unlike the non-transgenic parr, many of the immature transgenic parr, both as 0+ and as 1+, exhibited secondary smolt characteristics, including long, silver bodies with darkened fins.

Figure 5.

 Natural log transformed gonadal and somatic mass (g) of 1+ mature male transgenic and non-transgenic Atlantic salmon (Salmo salar) parr. The dashed and solid lines of best fit represent the non-transgenic and transgenic parr, respectively.


During the first year of life (0+), GH transgenic parr showed accelerated growth, reaching sizes typical of two (1+)- or three (2+)-year-old non-transgenic parr (Hutchings & Jones 1998). Moreover, while transgenics exposed to high feed outgrew those exposed to low feed, the size of non-transgenics in both feed treatments was equal, suggesting that transgenic individuals were limiting the energy consumption of non-transgenics through direct competition. Notwithstanding this fast growth and domination of food resources, precocious maturation was low (1·3%) and not influenced by transgenesis or feed level. With the feed treatment eliminated for the second year of life, the incidence of 1+ non-transgenic mature parr was nearly twice that of the much larger transgenic parr. Moreover, while the absolute gonadal mass of mature 1+ transgenic parr was greater than that of non-transgenic parr, the relative investment for a given somatic mass was less. These results suggest that growth rate and/or size at age are not proximate mechanisms responsible for precocial male maturation in Atlantic salmon and support the idea that energy accumulation thresholds dictate the proximate basis of this life-history decision, a pattern that may be common to organisms with similar alternative life histories.

Prior to the current study, it had been difficult to separate the effects of intrinsic growth and the potential implications of energy accumulation on precocious maturation. Fast growth has been previously found to reduce levels of parr maturation; however, the rapid growth was in response to a period of undernutrition, and thus, it was not possible to differentiate between the effects of rapid growth and the costs associated with the compensatory nature of the rapid growth that occurred (Morgan & Metcalfe 2001). Contemporary thought suggests that, in late summer, large Atlantic salmon parr that have exceeded a threshold level of energy reserves will mature that fall, whilst those large parr lacking such energy reserves may undergo smolt transformation (Metcalfe 1998; Thorpe & Metcalfe 1998; Thorpe et al. 1998; Letcher & Gries 2003). Supporting evidence includes observations that precocious maturation is more common in resource-rich environments (Rowe & Thorpe 1990; Letcher & Terrick 1998). Moreover, within populations, both parr maturation and smoltification correlate with high growth rates (Metcalfe, Huntingford & Thorpe 1988; Kadri et al. 1996; Letcher & Gries 2003). By manipulating intrinsic growth and controlling for both environment and genetic background (transgene excluded), this study provides direct empirical evidence suggesting that the physiological mechanisms promoting growth do not play a causative role in the early maturation of male parr and may even hinder it.

Growth hormone contributes to a wide array of biological processes. The most well-documented effect of GH is growth stimulation, which occurs in part by promoting lipolysis and protein synthesis (Björnsson 1997; Björnsson et al. 2002). High levels of circulating GH, such as that experienced by transgenic fish, have been shown to reduce energy reserves in stream salmonids, which likely reflects the metabolic effects mentioned earlier (Johnsson et al. 1999; Neregard et al. 2008). In addition to stimulating growth, GH is involved in smolt transformation, both directly and as a regulatory factor (McCormick 1996; Pelis & McCormick 2001; Björnsson et al. 2002). GH is also implicated in the maturation of salmonids (Björnsson et al. 1994); however, the exact role remains uncertain (Björnsson 1997; Björnsson et al. 2002). While the extent of physiological changes associated with the GH transgene may not be fully understood, a suite of pleiotropic effects have been observed including higher metabolic demands, increased activity and preferential investment in somatic tissue over energy reserves (Stevens, Sutterlin & Cook 1998; Cook, McNiven & Sutterlin 2000b; Cook, Sutterlin & McNiven 2000c). These physiological differences, coupled with the reduced maturation rates and secondary smolt characteristics observed here, suggest that the transgene may induce physiological pathways towards smoltification preferentially, an observation that is consistent with previous research on GH transgenic salmonids (Devlin et al. 1994a, 2000; Saunders, Fletcher & Hew 1998; Devlin, Biagi & Yesaki 2004b). Collectively, these results suggest that the proximate mechanisms underlying intrinsically fast growth promote life-history shifts towards smolt transformation as opposed to precocial maturation. Further efforts to elucidate the proximate bases underlying alternative reproductive phenotypes should apply integrative research methods that measure genetic, physiological and phenotypic changes simultaneously over the relevant life-history period.

Presumably, the investment of energy into structural tissues leads to high growth rates at the expense of investment into storage tissues for other purposes. Thus, from an ultimate perspective, it is unclear why fast growth consistently correlates with precocial maturation in natural populations, if it may reduce available resources for other functions. This may be explained in part by the importance of body size in the breeding success of precocial males, with selection likely stabilizing. Large size affords larger gonads (Fleming 1998) and an ability to behaviourally dominate smaller parr during competition for access to breeding females (Thomaz, Beall & Burke 1997; Koseki & Maekawa 2000), although this advantage appears to decline at high parr densities (Jones & Hutchings 2001, 2002). By contrast, small size may afford crypsis during sneak mating and reduce the likelihood of targeted aggression by anadromous adults. The present study thus suggests that the patterns of fast growth/large size at age with precocious maturation are correlations reflecting limits to plasticity in the partitioning of energy between structural and storage tissues. Neither growth rate nor size at age are proximate mechanisms responsible for precocial male maturation in Atlantic salmon. Rather, energy accumulation thresholds may be dictating the proximate basis of this alternative life-history decision.


The genetic effects associated with interbreeding and introgression are among the greatest concerns associated with the potential entry of transgenic organisms into nature (Muir & Howard 2002; Devlin, Sundström & Muir 2006). Moreover, age at sexual maturity is considered a key fitness-related trait influencing the invasion of foreign genes into wild populations because early maturation reduces generation time and increases the probability of survival to reproduction (Muir & Howard 2002; Garant et al. 2003). This study provides the first empirical data on the relative incidence of precocial male maturation in GH transgenic and non-transgenic Atlantic salmon and, therefore, provides valuable information for the ecological and genetic risk assessment process.

Among farmed Atlantic salmon strains, mature male parr have been identified as a potential means of increasing the pace of farmed gene introgression into wild populations (Garant et al. 2003; Weir et al. 2005). From a demographic perspective, the reduced expression of precocial male maturation among GH transgenic parr relative to non-transgenic parr suggests that the rate of transgene introgression may be limited by the number of maturing parr. However, mature male parr compete with one and other for proximity to nesting females (Fleming 1996); thus, differences in competitive ability could either enhance or reduce the influence of proportional differences between mature male transgenic and non-transgenic parr (Moreau, Conway & Fleming 2011a). While our observations are valuable, caution is required when inferring risk scenarios from these data because the relative incidence and size of mature male transgenic and non-transgenic parr are likely to vary with environment, particularly in nature. Previous efforts have shown that there is strong genotype by environment interactions on juvenile growth in transgenic salmon (Devlin et al. 2004c; Moreau et al. 2011b). Therefore, when used contextually, these data may provide valuable information for decision-makers, assessing the risks of GH transgenic salmonid biotechnologies.


The authors would like to thank Corinne Conway and Danny Ings for assistance with data collection and Aqua Bounty Farms Inc. for providing transgenic gametes. The authors would also like to thank our reviewers for their valued input into earlier versions of this manuscript. All animals were treated in accordance with the guidelines provided by the Canadian Council on Animal Care and the approval of Memorial University’s Institutional Animal Care Committee (AUP 07-03-IF). Support was kindly provided by a collaborative grant led by Dr E. M. Hallerman and funded by the USDA Biotechnology Risk Assessment Research Grants Program. Further financial assistance was provided by a National Sciences and Engineering Research Council of Canada Discovery Grant awarded to I.A.F.