Environmental Biology of Fishes

, Volume 94, Issue 1, pp 325–342

Understanding the adaptive consequences of hatchery-wild interactions in Alaska salmon


DOI: 10.1007/s10641-011-9929-5

Cite this article as:
Grant, W.S. Environ Biol Fish (2012) 94: 325. doi:10.1007/s10641-011-9929-5


About 31% of salmon harvested in Alaska comes from the hatchery production of hundreds of millions of pink and chum salmon and smaller numbers of sockeye, Chinook, and coho salmon. The numbers of hatchery-reared juveniles released in some areas are greater than the numbers of juveniles from wild populations. However, virtually nothing is known about the effects of hatchery fish on wild populations in Alaska. Possible effects of these interactions can be inferred from studies of salmonids in other areas, from studies of other animals, and from theory. Numerous studies show a complex relationship between the genetic architecture of a population and its environment. Adaptive responses to nature and anthropogenic selection can be influenced by variation at a single gene, or more often, by the additive effects of several genes. Studies of salmonids in other areas show that hatchery practices can lead to the loss of genetic diversity, to shifts in adult run timing and earlier maturity, to increases in parasite load, to increases in straying, to altered levels of boldness and dominance, to shifts in juvenile out-migration timing, and to changes in growth. Controlled experiments across generations show, and theory predicts, that the loss of adaptive fitness in hatchery salmon, relative to fitness in wild salmon, can occur on a remarkably short time scale. All of these changes can influence survival and impose selective regimes that influence genetically based adaptive traits. The preservation of adaptive potential in wild populations is an important buffer against diseases and climate variability and, hence, should be considered in planning hatchery production levels and release locations. The protection of wild populations is the foundation for achieving sustained harvests of salmon in Alaska.


Adaptive potentialAlaskaClimate variabilityDomesticationGenetic diversityHatchery-wild interactionsMating behaviorPacific salmonStraying


The enhancement of Pacific salmon populations in Alaska with hatchery-reared fish plays an important role in smoothing out population fluctuations induced by ocean-climate variability in the North Pacific (Beamish et al. 1997; Mantua et al. 1997; Royer et al. 2001). A large hatchery program was initiated by the State of Alaska in 1971 to increase salmon production after harvests declined precipitously in the 1960s (Smoker and Heard 2007). In 1974, the State Legislature implemented a system of private-non-profit hatchery associations, and together with state-owned and federal facilities currently encompass 36 hatcheries, which produced 30.8% of the state’s harvest of salmon in 2009 (White 2010). While the State of Alaska encourages the prudent development of its biological resources to benefit its residents, the State’s constitution recognizes the value of the sustained-yield principle for conserving renewable resources [State Constitution Article VIII, Section (4) (1956); Alaska Statute 16.05.730 (1992)]. The sustained-yield concept implies that wild populations of Pacific salmon should remain robust by protecting their potentials to change adaptively in an ever-changing North Pacific environment. The health of wild populations in Alaska is a fundamental concern in the light of large increases in hatchery production since the 1970s.

Stray hatchery fish can influence wild populations in several ways. Generally, the first effects of encounters between hatchery and wild populations are ecological, and these interactions potentially influence the abundances of wild fish through competition for food, spawning sites, and mates. One consequence of these encounters may be small wild population sizes, which can lead to the loss of genetic variation. Hatchery-wild encounters can also directly influence wild populations through hybridizations, which can lower the adaptive potential of natural populations. Adaptation is the matching of a population with its environment through natural selection and can be defined either by ‘the inherent design of a character’ or by ‘fitness’, as measured by relative reproductive successes among individuals that differ from one another (Ridley 2004). The present review focuses on the fitness concept of adaptation, because natural selection induced by hatchery-wild encounters can produce spatial and temporal shifts in adaptive landscapes, and because fitness may often be correlated with variables affecting reproductive success and population abundance. Individuals in a ‘locally-adapted’ population exhibit higher fitness than do migrants from other locations (Kawecki and Ebert 2004). A meta-analysis indicated that local salmon populations often have a 1.2 times average fitness over a non-local population (Fraser et al. 2011). Adaptive differences can be expressed, for example, by differences in tolerances to water temperatures among streams (Eliason et al. 2011).

Several hatchery practices can alter the fitness of cultured individuals. For example, environmental adaptations can be disrupted when eggs, harvested en masse, are cultured in a hatchery setting, so that the overlay of hatchery production on natural dynamics potentially introduces additional challenges to wild populations. Fry compete with hatchery fish for food in freshwater, and juveniles compete with hatchery fish in estuarine areas and the open sea (Cooney and Brodeur 1998; Ruggerone et al. 2003). Mature hatchery adults compete with wild fish for mates and spawning sites, when hatchery fish stray into natural habitats. Competition with hatchery fish may diminish wild production by reducing the numbers of adults returning to freshwater to spawn, or by reducing the spawning potential of wild fish. The magnitudes of these effects depend not only on the extent of the changes in a hatchery, but also on the rate and volume of straying of hatchery fish into natural habitats supporting wild populations.

Protecting wild populations of Pacific salmon in Alaska from the detrimental effects of wild-hatchery interactions is important for several reasons. First, wild populations represent a store of genetic variability that is available for the development of future hatchery broodstock. However, not all wild populations at a particular moment can be used as hatchery broodstock, because such traits as run timing, survival, growth, and resistance to pathogens may not be suitable for hatchery production. The preservation of genetic variability among wild populations, and not just within populations, is important to the survival of regional populations and future broodstock availability (Schindler et al. 2010). Second, genetic variability underpins physiological and behavioral responses to environmental variables that challenge fry in freshwater habitats and juveniles and adults in the marine realm. Selection from these challenges changes constantly because of human activities and swings in climate (Steele 1998; Mantua and Hare 2002). Climate influences patterns of precipitation, which affect salmon fry production in freshwater and ocean nutrient cycles, which influence growth and survival in marine waters (Mantua et al. 1997; Crozier et al. 2008). Third, adaptive variation among populations may be essential for the regional persistence of populations (Hilborn et al. 2003; Schindler et al. 2010). The abundances of local populations may change considerably, even though regional abundances remain constant.

The concern over hatchery-wild interactions is centered on the effects that hatchery-reared strays have on wild populations. These concerns are focused on ecological mechanisms that ultimately reduce the abundances of wild populations when fish compete for mates or for nesting sites. Most interactions between wild and hatchery-reared fish have the potential to influence the demography or genetics of wild populations. First, competition from hatchery fish can lead to the loss of genetic diversity in a wild population by reducing its effective population size (Weber and Fausch 2003; Huusko and Vehanen 2011). Second, mating between hatchery and wild fish can also lead to the loss of adaptive fitness in wild populations, especially when hatchery fish come from cultured populations that have diverged substantially from wild populations. First-generation hybrids between hatchery and wild fish may represent a loss of productivity in the short term, and genetic introgression can produce a shift in adaptive potential that reduces the long term viability of a wild population.

The goal of this review is to evaluate the genetic implications of hatchery-wild interactions and their relevance to salmon in Alaska, an area in which few experimental studies of salmon have been made. Alaskan rivers support the largest populations of Pacific salmon in North America, and it is important that the history of mismanagement in other areas be avoided in Alaska (Mobrand et al. 2005; Naish et al. 2008). The objective of this review is to outline the mechanisms that produce undesirable genetic changes in wild populations. Hatchery-wild interactions in Alaska salmon are part of a larger, world-wide problem of compromising the genetic integrities of wild populations with large-scale releases of cultured, or invasive, individuals (Laikre et al. 2010). This review complements other reviews on the biology of salmon (Hindar et al. 1991; Taylor 1991; Waples 1991; Quinn 1993; Thomas and Mathisen 1993; Campton 1995; Jonsson 1997; Utter 1998; Jonsson and Jonsson 2006; Araki et al. 2008; Naish and Hard 2008; Naish et al. 2008; Waples et al. 2008; Kostow 2009; Nielsen and Pavey 2010; Fraser et al. 2011).

Assessing effects of hatchery-wild interactions

The potential for adaptive change is generally related to the amount of genetic variation in a population (Wang et al. 2002; Reed and Frankham 2003), and to interactions between genes (Karasov et al. 2010). Genetic variation arises from mutations in a gene, from changes in regulatory genes, or from gene re-arrangements. On short time scales of tens or hundreds of generations, the amount of genetic variation in or among populations is influenced by effective population size (Ne) and gene flow between populations. In contemporary populations, Ne and dispersal between populations is shaped by both natural and human-mediated events: decadal cycles in rainfall influence population abundances (Mantua and Hare 2002), harvests can deplete wild populations, and hatchery strays can influence wild population sizes and the distribution of genetic variation among wild populations.

When hatchery strays differ from local wild fish, spawning between them may disrupt ecological and adaptive processes of wild populations (Gharrett and Smoker 1993). The adaptive consequences of hatchery-wild interactions can be understood in three ways: 1) observations and experiments on Pacific salmon, 2) results for other salmonids or other organisms, and 3) theory (Hey et al. 2005). Adaptive responses to environmental challenges during the various life-history stages of salmon must ultimately be assessed by measuring the reproductive success of offspring from a particular mating (Puurtinen et al. 2009). While controlled experimental matings and generational monitoring of offspring are the best means of measuring adaptive fitness (e.g. Kostow 2004; Araki et al. 2007a, 2008), short-term assessments can still give insights into the mechanisms conferring an adaptive advantage to individuals (e.g. Einum and Fleming 1999). Most studies of adaptation are focused on adult traits such as morphology, run-timing, and spawning behavior, and on the effects these traits have on growth and survival to reproductive maturity. This focus derives from the ready access to adults in spawning areas, the relative ease of observing egg and fry development, and the ability to monitor juvenile growth and behavior before outmigration. Egg size, hatching time, and fry behavior, among many other variables, are under diverse selective pressures during early life history stages in freshwater habitats (Table 1). However, various forms of environmental selection confront individuals at all life-history stages.
Table 1

Examples of adaptations and the effects of hatchery practices and hatchery strays on Pacific salmon and other salmonids. The literature on this topic is abundant, and the information on these topics in this table is by no means a comprehensive





Environmental adaptations in salmonids

 Population structure

Neutral genetic markers generally show significant divergence between populations, which reflects homing, local adaptation, and ancestral divergence


e.g. Seeb et al. (1999); Waples et al. (2001); and many others

 Local adaptation

Hybrids between distant populations had reduced survival over controlled matings within populations. Demonstrates outbreeding depression from disruption of epistatic gene interactions


Gilk et al. (2004)

Adaptive differences in morphology, growth, and timing of migration in juveniles response to water flow and temperature


Riddell and Leggett (1981)


Inherited differences in agonistic behavior between wild juveniles from different streams


Rosenau and McPhail (1987)

Innate ability to migrate appropriately upstream or downstream to feeding areas after emergence from gravel


Raleigh (1971)

 Temperature adaptation

Abnormal embryo development in thermally stressed eggs


Campbell et al. (1998)

Genotype-temperature interactions influences juvenile traits in pink and chum salmon


Beacham (1988)

 Run timing

Complex interaction between arrival in spawning area and size; early arrival allowed choice of nest


Dickerson et al. (2002, 2005)

Seasonal temperature cycles influence migration into freshwater


Hodgson and Quinn (2002); McGregor et al. (1998); Smoker et al. (1998)

Hatchery practices

 Loss of genetic diversity

Sperm competition from mixing of milt leads to smaller effective broodstock size for a given number of fish


Withler (1988); Hoysak et al. (2004); Campton (2004); Wedekind et al. (2007)

Small effective broodstock size


Allendorf and Phelps (1980); Ryman and Ståhl (1980); Verspoor (1988); Norris et al. (1999)

Inbred fish showed less resistance to pathogens than individuals


Arkush et al. (2002)

 Broodstock selection

Reduced differentiation among hatcheries, relative to wild populations


Garcia-Marin et al. (1991)

 Artificial mating

Artificial random mating prevents mate choice and to loss of fitness


Berejikian et al. (2000); Hankin et al. (2009)

Random hatchery matings select for earlier maturity in Chinook salmon


Hankin et al. (2009)

Artificially bred Atlantic salmon had four times higher parasite loads than fish from naturally matings


Consuegra and Garcia de Leaniz (2008)

 Timing of egg take

Selection of eggs from one part of run can shift run timing of adults


Quinn et al. (2002); Ford et al. (2006)

 Timing of fry release

Smots released in winter tended to stray the most


Hansen and Jonsson (1991)


Small 2.2% advantage of wild fry over first-generation hatchery fry in avoidance of predation


Fritts et al. (2007)

Hatchery-reared fish released into the wild

 Growth rate

Lower growth rates in hatchery-reared fish released into the wild


Reisenbichler and McIntyre (1977); Lachance and Magnan (1990); Finstad and Heggberget (1993); Hesthagen et al. (1999)

Hatchery supplementation led to smaller naturally spawning males


Unwin and Glova (1997)


Hatchery juveniles failed to establish feeding territories, fed less and used less efficient feeding strategies than wild fish


Bachman (1984)

Hatchery juveniles were more aggressive, spent less time foraging and more time in fast flowing water than wild fish


Mesa (1991)

Captive or hatchery-bred salmon were more aggressive than wild-bred fish


Swain and Riddell (1990); Blanchet et al. (2008)

Hatchery domestication promotes boldness in brown trout


Sunderström et al. (2004)


Lower survival in hatchery-reared fish released into the wild


Reisenbichler and McIntyre (1977); Reisenbichler and Rubin (1999)

Decline in fitness relative to wild fish the longer a broodstock is cultivated in a hatchery


Araki et al. (2007b)

Offspring of hatchery spawners produced only 10-20% surviving offspring of wild spawners


Chilcote et al. (1986); Campton et al. (1991)

 Age at maturity

Hatchery-reared fish mature more rapidly in the wild


Leider et al. (1986)

Hatchery supplementation led to earlier male maturity (jacks)


Unwin and Glova (1997)


Homing to hatchery of origin, rather than to location of release site


Hayes et al. (2004)

Hatchery fish strayed more than wild fish


Jonsson et al. (1991); McIsaac and Quinn (1988); Quinn et al. (1991)

No difference in stray rate between hatchery and wild salmon


Labelle (1992)

 Reproductive success

Hatchery fish reproducing in the wild had lower lifetime reproductive success than wild fish


Thériault et al. (2011)

 Run timing

Shift of run timing in hatchery and straying led to shift to earlier run timing in natural population


Ford et al. (2006)

Wild fish returned to freshwater earlier than hatchery fish


Jonsson et al. (1991)

 Wild population abundance

Hatchery supplementation led to depressed recruitment of replacement of large proportion of wild populations with fish of hatchery origin


Unwin and Glova (1997); McGinnity et al. (2009)

Hatchery-reared and wild fish studied in common hatchery environment

 Growth rate

Hatchery fish grow more rapidly than wild fish reared in a hatchery


Vincent (1960); Dwyer and Piper (1984); Fleming et al. (2002)

 Egg size

Wild fish had larger eggs


McDermid et al. (2010)

 Egg development

Egg survival better in wild fish


McDermid et al. (2010)

 Fry growth and development early growth rate

Fewer deformities in wild fish


McDermid et al. (2010)

Wild fish grew more rapidly


McDermid et al. (2010)


Hatchery and hatchery × wild hybrids showed higher levels of agonistic behavior than wild fish


Wessel et al. (2006)

Hatchery juveniles dominant over wild juveniles


Metcalfe et al. (2003)

 Life-time survival

Reduced survival of hatchery offspring relative to offspring from wild parents


Kostow (2004)

aSpecies: 1 = pink salmon, 2 = chum salmon, 3 = coho salmon, 4 = Chinook salmon, 5 = sockeye salmon, 6 = masu salmon, 7 = rainbow/steelhead trout, 8 = cutthroat trout, 9 = Atlantic salmon, 10 = brown trout, 11 = lake trout, 12 = brook trout

Two kinds of molecular markers can provide insights into ecological and behavioral variables influencing reproductive success. The first kind includes selectively neutral, and nearly neutral, markers. These markers can be used to estimate population processes, such as migration between populations, or can be used to assess parentage to evaluate the successes of individual matings (Araki and Blouin 2005; Funk et al. 2005; Hauser et al. 2006). The second kind of molecular marker is influenced by selection. The strength and mode of selection on these markers varies from weak selection on particular alleles (e.g. slightly deleterious alleles), to lethal effects of some mutants.

Modes of selection shaping adaptive traits

Genetic mechanisms influencing a phenotype can be complex. In some cases, a single gene confers a selective advantage in a particular environment (e.g. Powers et al. 1991; Miller et al. 2001). In most cases, several genes, each with small effects, interact with one another to produce advantageous phenotypes (quantitative genetic traits). Selective breeding of quantitative traits is used in aquaculture to improve phenotypic traits, or to enhance the production of cultured populations (Tave 1993). Environmental selection in the wild also acts in the same way to produce morphological and behavioral shifts in natural populations.

Populations respond to selection in several ways. Directional (positive) selection leads to the differential survival of some phenotypes over other phenotypes, so that the genes shaping the phenotype increase in frequency. This mode of selection can lead to local adaptation and to a landscape with genetically different populations. Balancing selection, on the other hand, can produce genetic homogeneity among populations. This form of selection operates on an individual locus (e.g. heterozygote advantage), or on the entire genomes of different life-history forms (e.g. eggs, alevins, juveniles, adults), or on the same population in different years.

Selection occurs during the various life-history stages, which exploit resources in contrasting habitats. Adults spawn in freshwater streams and lakes, or along brackish shorelines (pink and chum salmon); eggs develop in permeable gravel or cobble substrates; alevins emerge from the gravel and juveniles spend a year or more in freshwater streams (Chinook, coho salmon), lakes (sockeye salmon), or migrate directly to sea (chum and pink salmon); adults home to natal streams and lakes to spawn, after maturing during long migrations in the sea. Numerous opportunities for selection occur during each of these life-history stages. A fundamental premise of this review is that wild salmon populations in Alaska are adapted to environmental conditions in spawning streams, juvenile rearing areas, and ocean environments. A few studies of Alaskan populations demonstrate the same kinds of adaptations found in salmonid population elsewhere (Table 1).

Effects of hatchery culture on fitness

Hatchery practices impose several intentional, or inadvertent, forms of selection. One common response to the hatchery environment is domestication, which can arise in three ways (McDermid et al. 2010). First, selection against traits that enhance ecological fitness in natural environments reduces the adaptive fitness of hatchery fish. For example, the collection of broodstock during only a portion of a run may shift the return-timing of the hatchery population relative to optimal run-timings of nearby wild populations (e.g. Ford et al. 2006). When run-timing in wild populations is under natural selection because of seasonal patterns of precipitation, hybrids between hatchery strays and wild fish may not return at an appropriate time. Second, selection may favor traits that enhance survival in a hatchery setting, but that are maladaptive in the wild. For example, warmer hatchery incubation temperatures may hasten development, but accelerated development may lead to an environmental mismatch with food availability at the time of release. Third, natural selection may be relaxed so that non-adaptive phenotypes increase in frequency by chance. Artificial mating in a hatchery eliminates opportunities for mate choice, which may be important for conferring adaptive benefits (Neff and Pitcher 2005; Hankin et al. 2009). When selective pressures on hatchery populations are not managed, or when hatchery fish hybridize with wild fish, adaptive transformations of wild populations can potentially lead to declines in abundance (Lynch and O’Hely 2001).

A growing body of theoretical and empirical research indicates that artificially cultured stocks almost always change genetically. Hatchery practices can produce life-history shifts in fitness that affect fry survival (Fritts et al. 2007), growth rate (Knudsen et al. 2006, 2008), juvenile behavior (Pearsons et al. 2007), adult run-timing (Quinn et al. 2002; Dickerson et al. 2005), and mating behavior (Fleming and Gross 1992, 1993; Berejikian et al. 1997; Turner et al. 2009). When these altered hatchery fish interact with wild fish several ecological and genetic mechanisms can lead to declines in wild fish abundance, or to reduced adaptive potential.

Several studies have documented rapid fitness declines in hatchery-reared fish. Araki et al. (2007b) used parentage analysis to identify the experimental offspring of hatchery x wild and wild x wild matings of steelhead on their return to Hood River, Oregon. An adult-to-adult assessment of reproductive success over 3 years indicated that the success of hatchery × wild offspring was only 55% of wild × wild offspring. Even a single generation of culture adversely influenced the fitness of hatchery offspring. The longer a broodstock lineage was cultured in a hatchery, the greater the reduction in their fitness relative to wild populations (Araki et al. 2007a, c).

Run and spawning timing and offspring survival

Since water flow and temperature vary seasonally in salmon streams, run and spawning timing must coincide with environmental windows of opportunity to maximize the survivals of offspring. The return of adults to spawning areas must coincide with water flows that permit upstream migration, and spawning must be timed to synchronize egg development and fry emergence with temperatures and plankton cycles that are conducive to juvenile growth and survival (e.g. Brännäs 1995; Einum and Fleming 2000). Several studies show that run-timing is inherited to some extent, because of its importance to the timing of emergence and fry survival. Taylor (1980) measured times to emergence of fry, survival, and time of adult return in early- and late-run pink salmon spawning in Auke Creek, Alaska. Eggs from wild parents incubated in the hatchery developed more rapidly than naturally spawned eggs in Auke Creek, largely because warmer temperatures in the hatchery promoted faster development. Hatchery fry from late-run parents were larger, on average, than early-run hatchery fry and showed eight times greater survival to maturity (1.46%) than did early-run offspring (0.17%). These results indicate that estuarine temperatures greatly influence fry survival as they enter marine waters.

Artificial spawning in hatcheries can interrupt these finely tuned adaptations. Hatchery practices can shift the return timing of a hatchery population by a non-random selection of broodstock over the period of return to the hatchery. Hatchery managers tend to fill production quotas with early returning fish, because they cannot predict the number of later-returning fish. This practice has generally led to earlier return times to hatcheries. For example, both coho and Chinook broodstock have shown a shift to earlier return dates in several hatcheries in Washington State, despite a local environmental warming trend that should have led to later spawning times (Quinn et al. 2002). Altered return timings have been observed for steelhead trout (Leider et al. 1986; Mackey et al. 2001; McLean et al. 2003), coho salmon (Fleming and Gross 1992; Ford et al. 2006), and Atlantic salmon (McGinnity et al. 1997; Fleming et al. 2000). These shifts in run-timing have the potential to influence wild populations when hatchery-origin fish stray into natural spawning areas and mate with wild fish.

Mating behavior

Artificial mating in hatcheries also influences some forms of selection. Sexual selection in the wild is often based on male–male competition and mate choice, most often by females. Winners of intra-sexual competition are thought to confer superior genetic quality to offspring. Although females may benefit by choosing dominant males in some cases, the cost to a female may outweigh benefits, because the traits promoting dominance may not contribute to fitness of the offspring (e.g. Candolin 2003; Wong 2004; Jacob et al. 2007). Hence, females may not always discriminate among dominant and subordinate males (Foote et al. 1997), or females may prefer subordinates. In some cases, the high reproductive success of dominant males may arise simply as a consequence of dominant males monopolizing access to females and reducing opportunities for female choice.

Male–male competition may favor large body size or high levels of dominance. Mate choice may be based on ‘good genes’ showing additive genetic variance, or on ‘compatible genes’ showing non-additive genetic variance (Bernatchez and Landry 2003; Neff and Pitcher 2005; Pitcher and Neff 2006). For example, microsatellite and MHC markers in offspring of 41 male and 35 female wild Atlantic salmon indicated that free mate choice led to higher levels of MHC heterozygosity than expected with random mating (Landry et al. 2001). In experiments with naturally spawning Chinook salmon grouped in various sex ratios, large males with brighter coloration sired more offspring than smaller, duller males; however, neither these morphological variables, nor sex ratio, influenced female reproductive success (Neff et al. 2008). Additionally, MHC genotypes tended to influence female, but not male, mate choice, resulting in more MHC heterozygous offspring (Neff et al. 2008). In a study of coho salmon, parentage analysis with microsatellite markers showed that the lifetime reproductive success of hatchery fish released as unfed (age 0) or fed (age 1) fry was reduced for males, but less so for females and jacks (Thériault et al. 2011). The reduction in success was attributed to the lack of the sexual selection in a hatchery setting.

Major histocompatibility complex (MHC)

Major histocompatibilty complex (MHC) genes directly affect fitness, because they help to shape immune responses to pathogens (Jeffrey and Bangham 2000; Langefors et al. 2000; Grimholt et al. 2003). The MHC complex encodes a large diversity of immune system proteins that defend an individual against environmental pathogens. While numerous MHC genes are expressed in mammals and some fishes (Malaga-Trillo et al. 1998), only a few of these genes have been documented in salmon (Miller and Withler 1998). Several studies show that MHC variation is shaped by mate selection and environmental variability. For example, Miller and Withler (1996) found an excess of MHC amino acid replacements in seven species of Pacific salmon, indicating positive selection. The selective value of MHC variability is largely due to increased offspring heterozygosity. Infectivity experiments on endangered winter-run Chinook showed that individuals with heterozygous MHC genes survived longer on average than homozygous individuals (Arkush et al. 2002). Further, outbred fish showed less infection severity than inbred fish (Arkush et al. 2002). In Atlantic salmon, offspring of artificially bred fish had higher parasite loads than naturally spawning fish, despite similar levels of MHC diversity in the two groups (Consuegra and Garcia de Leaniz 2008).

Specific alleles may also confer a selective advantage (Langefors et al. 2000; Garrigan and Hedrick 2001). Selection for local conditions was indicated by strong regional differences in the diversity of sockeye salmon populations in British Columbia (Miller et al. 2001). On a regional scale, MHC allele-frequency differentiation was similar to microsatellite differentiation, but selective forces within river systems tended to produce greater MHC than microsatellite differentiation (Landry and Bernatchez 2001). In Alaska, similar geographic and temporal MHC allele-frequency variability has been observed in populations of sockeye salmon around Bristol Bay (Gomez-Uchida et al. 2011). Together, these studies show the association between genetic and environmental variation.

Mechanisms producing shifts in fitness

The mechanisms leading to rapid declines in fitness in hatchery-reared fish are uncertain and are the focus of current research. Possible mechanisms include the loss of diversity in hatchery broodstock, the accumulation of less fit mutations, domestication selection on single or multiple gene loci, and molecular mechanisms (e.g. DNA methylation) that alter gene expression.

Loss of genetic diversity

Genetic diversity can be lost through random drift with the use of small brood stocks (Ryman and Laikre 1991). Concern for conserving genetic diversity arises, because less genetically diverse populations may be less fit to face environmental challenges from predation, disease, and climatic shifts (Verspoor 1997; Reed and Frankham 2001; Borrell et al. 2004). The loss of genetic variability in a hatchery broodstock has been documented in several salmonids (e.g. Allendorf and Phelps 1980; Ryman and Ståhl 1980; Verspoor 1988; Norris et al. 1999). Although the State of Alaska requires effective broodstock sizes of at least 400 fish (GPRT 1985), most Alaskan production hatcheries use thousands of parents for broodstock, so that little diversity is lost each generation.

Accumulation of deleterious genes

Hatchery culture may lead to elevated frequencies of deleterious alleles that are otherwise kept at low frequencies in wild populations by selection (Lynch and O’Hely 2001). The effects of mutation load are greatest in brood stocks closed to the immigration of wild individuals and in small brood stocks. However, this mechanism is unlikely to produce the rapid shifts in fitness observed in experimental studies, unless salmon generally harbor large numbers of deleterious recessive genes (genetic load) or have large mutation rates (Lynch et al. 1999).

Domestication selection

Selective breeding has been the cornerstone of improving the performances of species used in aquaculture (Gjedrem 1983). While the intent of salmon hatchery culture is generally not to alter the genetic makeup of hatchery-reared fish, the high survival rates of fish inevitably favor traits that are not favored in the wild. In Alaska, most hatcheries represent closed populations, because fish returning to the hatchery are used as brood stock. Hence, selective gains for traits conducive to hatchery culture accumulate each generation. Adaptation to hatchery culture (domestication) can alter a variety of traits, including morphology (Fleming and Einum 1997), behavior (Berejikian et al. 1996; Fleming and Einum 1997; Metcalfe et al. 2003), and reduced competitive ability with wild fish (Berejikian et al. 1997; Fleming and Einum 1997; Weber and Fausch 2003). Most of these traits are controlled by several genes (quantitative trait loci), and changes in these traits may take several generations. Ford (2002) modeled shifts in domestic traits and showed that even small amounts of straying between hatchery-reared and wild fish decreases the fitness of the wild population. While selection for domestic traits undoubtedly produces genetic changes in hatchery populations, domestication is unlikely to cause a rapid decline in fitness in one or two generations.

Modification of gene expression

Alternatively, rapid reductions in fitness of hatchery fish may be due to shifts in gene expression. Heritable epigenetic changes can modify the expression of information encoded in DNA without changing the underlying DNA sequences (Bonasio et al. 2010). Epigenetic events during development, for example, lead to different tissues from identical embryonic germ cells (Feng et al. 2010). These changes are generally mediated by the suppression of regulatory genes. Changes in DNA expression can result from DNA methylation (generally of CpG dinucleotide pairs), from changes in the proteins (histones) that chaperone DNA, and from non-coding RNA activity. Blouin et al. (2010) searched for differences in genomic methylation in hatchery-reared and wild steelhead using methylation-sensitive amplified fragment length polymorphisms, but failed to find evidence of hypermethylation (reduced DNA transcription) or hypomethylation (increased DNA transcription).

Gene expression can also be modified by transposons (transposable DNA elements), which are virus-like particles whose replication is mediated by RNA (class I) or DNA (class II) (Kramerov and Vassetzky 2005). Autonomous transposons encode the information needed for insertion into the host’s DNA. Nonautonomous transposons do not encode proteins and must rely on the replication machinery of autonomous transposons. Class I elements consist of LTR-transposons (long terminal repeats), LINES (long interspersed elements), or SINES (short interspersed elements). SINES are nonautonomous elements, depending on the enzymatic machinery of LINEs, and generally make up 50% of the transposons in an organism. Unlike other transposons, SINES are inserted irreversibly into the DNA of an organism and are therefore inherited (Takasaki et al. 1997). Some families (elements with related sequences) of these elements are common in salmonids (Hamada et al. 1997; Matveev and Okada 2009), occurring in as many as 15 000 copies in Atlantic salmon (Goodier and Davidson 1994). SINEs have been used in systematic studies of Pacific salmon because their insertion is irreversible (Kido et al. 1991; Murata et al. 1993, 1998). Transposons controlling gene expression have been used to improve aquaculture production (Tafalla et al. 2006). However, the role of transposons in shaping short-term responses to hatchery environments has not yet been researched.


A remarkable life-history feature of Pacific salmon is their ability to return to natal streams to spawn after migrating long distances at sea. Homing to natal areas often leads to substantial genetic differences between populations (e.g. Waples et al. 2001). Nevertheless, some salmon stray into non-natal streams and spawn with fish of other populations. While the degree of straying may differ among species, the incidence of straying varies considerably among populations within a species, depending on stream conditions at the time of return and on the ages of returning fishes (Quinn 1993).

Several life-history and environmental variables influence straying. For example, in some areas, pink salmon may stray, because they spawn in the lower reaches of streams and sometimes in upper intertidal areas and migrate out to sea soon after fry emerge from the gravel (Heard 1991). This short freshwater phase fails to strongly imprint fry on distinctive waters that would help guide them to natal areas in the final stages of migration as mature adults (Cooper et al. 1976; Dittman and Quinn 1996). Molecular markers show that some salmon have the ability to return to the same portion of a stream in which they were spawned (Gharrett and Smoker 1993, Bentzen et al. 2001; Neville et al. 2006).

In Alaska, straying is especially problematic in two regions, because hatcheries release hundreds of millions of juveniles. In Prince William Sound, stream surveys of spawning salmon found marked hatchery fish in some spawning areas in large numbers (Brenner et al. 2011). The proportion of stray hatchery fish ranged from 0% to 98% for pink salmon, 0–63% for chum salmon, and 0–33% for sockeye salmon. Hatchery fish strayed most frequently into streams within 40 km of a hatchery. Overall, a model of these data indicated that more than 10% of pink salmon found in wild-salmon streams was of hatchery origin.

Over 75% of the harvest of pink salmon in Prince William Sound comes from hatchery production (Smoker and Heard 2007). Large numbers of chum salmon are also produced in Southeast Alaska and may contribute to an increase in hatchery-wild interactions. Hilborn and Eggers (2000, 2001) argued that hatchery production of pink salmon in Prince William Sound may not have increased overall production above that expected from wild populations alone. Pink salmon with hatchery ancestry may have displaced wild pink salmon in this area, or may have severely restricted ecological opportunities for wild populations (Hilborn and Eggers 2000). However, Wertheimer et al. (2001) countered that ocean-climate variability in the North Pacific may have had a greater influence on the abundances of wild populations in Prince William Sound than hatchery production and that hatchery production complemented wild production. Even though hatchery production of pink salmon in Prince William Sound may have yielded a net benefit to fishery harvests, hatchery-wild interactions may still ratchet down wild population fitness and abundances. Heard (2003) estimated that as many as 4.5 million wild pink salmon may have been lost annually because of hatchery production.

Hatchery strays can swamp wild populations

A model incorporating the effect of hatchery fish straying on the demographic and genetic risks in wild populations showed that persistent straying can lead to the demographic extinction of a wild population in a short time (Hutchings 1991). Assuming displacement and hybridization, the model showed that the wild segment of a population would decline 50–100% within 4 years with annual stray rates of 20%. Even without hybridization, the wild component of a population can be expected to decline 50% with annual stray rates of 30%. The persistent straying of hatchery fish into wild populations will eventually lead to the complete replacement of wild individuals with hatchery descendents, even at small annual stray rates.

Ryman and Laikre (1991) modeled the demographic effects of straying by considering the effective population sizes of the wild (NW) and hatchery (captive) (NC) fish. The effective population size equals the sum of wild and hatchery adults only when the fraction of hatchery progeny in the natural population is NC/(NC+ NW). Effective wild population sizes at other values of NC and NW are smaller. Importantly, a large amount of straying reduces the total effective size of a wild population and can indirectly lead to a loss of genetic diversity with a consequential drop in adaptive fitness. Waples and Do (1994) explored this effect in more detail for Pacific salmon. They found that the effect of straying depended on the number of parents used in culture and not on the fraction of the wild population used for spawning in the hatchery. The use of hatchery individuals for subsequent broodstock, as in a segregated hatchery, can hasten the decline in the effective population size of a wild population (Box 1).

Examples of the negative influence hatchery releases have on wild populations can be derived from studies of Oregon coastal coho salmon and steelhead trout populations in Washington and Oregon. In an effort to protect production from Oregon coastal wild populations of coho salmon, hatchery releases were cut back from a high of 34 million in 1981 to about 1.6 million during 1998–2002 (Nicholas et al. 2005). These populations were at 3–19% of their historical abundances. A multi-factorial Ricker model incorporating the time series of spawners and recruits showed that the per capita population growth rate was most influenced by density-dependent spawner abundance (Buhle et al. 2009). Hatchery spawners had a strong negative influence on recruitment, so that spawning areas with a large fraction of hatchery fish produced fewer recruits than comparable areas with only wild fish.

Similar results were found for hatchery enhanced populations of steelhead trout. Chilcote (2003) found that the intrinsic growth rate of wild populations in Southwest Washington was negatively correlated with hatchery production. In another study of steelhead, Araki et al. (2007c) found that hatchery adults, produced by a segregated hatchery broodstock for several generations in Hood River, Oregon, depressed the effective number of spawners in the river, but supplemental fish with a history of only a single generation did not influence the effective number of breeders in the river system. In the Clackamas River, Oregon, Kostow and Zhou (2006) used Ricker and Beverton-Holt stock-recruitment models to understand the relationship between hatchery releases and patterns of wild population recruitment in winter-run populations impacted by the introduction of summer-run hatchery steelhead into the river basin. The results showed a 50% decline in the number of recruits per spawner. Hatchery production led to the density-dependent effects from exceeding the carrying capacity of the river. The inverse relationship between hatchery production and wild population abundances, demonstrated in these studies, could have been due to several mechanisms, including poor reproductive success of hatchery spawners in the wild, or reduced success of wild fish because of competition as adults or juvenile for resources with hatchery (Kostow and Zhou 2006).

Hatchery-wild hybridization

In addition to demographic effects of hatchery fish on wild populations, hybridizations and genetic introgression represent a threat to wild populations. Reduced hybrid fitness has been documented experimentally in ‘common garden’ experiments for several anadromous salmonids (e.g., Leary et al. 1985; Hawkins and Foote 1998; McGinnity et al. 2003). A major genetic risk of introgressive hybridization is the disruption of adapted gene complexes and loss of fitness (outbreeding depression) (Gharrett and Smoker 1993; Rhymer and Simberloff 1996). In one form of outbreeding depression, native individuals are better adapted to particular habitat conditions than either the introduced or hybrid individuals. For example, experimental hybrids between even- and odd-year run pink salmon showed much lower survival rates than either of the two control groups, even though these fish spawned in the same stream (Gharrett and Smoker 1991). Odd- and even-year pinks, however, have diverged considerably from each other, and strong outbreeding effects can be expected (Gharrett et al. 1999). However, outbreeding depression can also occur in hybrids between geographically separated groups of the same year-type (Gilk et al. 2004). A second form of outbreeding depression occurs when non-native genes are introduced into the genomes of wild individuals after the first generation of hybridization (introgression). Introgression disrupts the genes influencing a particular adaptation. Depending on the mode of expression of the genes, first generation hybrids may not be affected, but genetic recombination during reproduction separates co-adapted genes on parental chromosomes and reduces fitness in the introgressed individuals.

Future directions

The objective of this review has been to show first that hatchery culture reduces the fitness of hatchery-reared fish, relative to wild fish, and second that straying of hatchery fish into wild populations can lower the fitnesses of these populations. Without question, adaptations to seasonal and annual environmental cycles allow Pacific salmon to successfully complete their life-history cycles in freshwater, estuarine, and marine habitats. Several experimental studies show that numerous ecological and life-history variables, such as run-timing, nest building, mating behavior, egg size, fry emergence, and juvenile behavior, are closely tied to environmental variables by selection, which shape adaptive responses that affect survival and, hence, fitness of wild fish. These finely tuned adaptations can be disrupted by hybridization with genetically altered hatchery fish.

Evidence for understanding the effects of hatchery strays on wild populations in Alaska comes from studies of Pacific salmon and other salmonids in a variety of regional settings and from theoretical considerations. It may be argued that these results are not entirely applicable to Pacific salmon populations in Alaska, because of environmental differences between regions, because of life-history differences between species, and because many models for investigating the effects of straying were constructed to address conservation concerns of threatened populations and not the management of large populations. However, the results of the numerous studies reviewed here show that artificial culture inevitably changes the genetic architectures of hatchery-reared fish. When these fish stray into streams and rivers, they can alter the fitnesses of wild populations through ecological interactions and hybridization. The application of these conclusions for salmon populations in Alaska would be strengthened by controlled, generational experimental studies with Alaskan species.


Comments by David Bedford, Lisa Creelman, Tyler Dann, Marianne Grant, Chris Habicht, Kathyrn Kostow, David Noakes, Fred Utter, Eric Volk and two anonymous reviewers greatly improved the manuscript. This is paper PP-270 of the Commercial Fisheries Division of the Alaska Department of Fish and Game.

Copyright information

© Springer Science+Business Media B.V. 2011

Authors and Affiliations

  1. 1.Commercial Fisheries Division, Alaska Department of Fish and GameAnchorageUSA