Science of salmon stocking: report

3. Risks

Whist the aims when stocking fish are to conserve, mitigate or enhance natural populations, there is a large body of research which illustrates the risks associated with such strategies (Table 1, Appendix 1). Such risks fall within three broad categories: those associated with collection of broodstock and production of individuals in the hatchery, environmental impacts when fish are stocked into the wild, and anthropogenic impacts due to changes in behaviour of those who utilise and manage the resource.

3.1 Hatchery issues

Fish produced in hatcheries have been shown to be less fit than wild conspecifics when released into the natural environment (Araki et al., 2008; Christie et al., 2014), even after a single generation in captivity and/or when produced from wild-caught broodstock (Christie et al., 2012a; Milot et al., 2013). Fitness reduction is a result of a number of physiological and genetic processes (Table 1) and introductions of such hatchery fish can result insignificant deleterious impacts on wild genetic integrity (Garcia de Leaniz et al., 2007; Naish et al., 2007; Blanchet et al., 2008; Araki and Schmid, 2010). In turn, and especially over the long term, this impact may result in entire populations of fish that are maladapted to their natural environment (Araki et al., 2007; Henderson and Letcher, 2011) and, as a result could lead to populations that have reduced genetic variability and associated loss of resilience to environmental changes (McGinnity et al., 2009; Sgrò et al., 2011)

Wild salmon are structured into a hierarchy of populations, from the continental to the tributary level (Verspoor et al., 2005; Cauwelier et al., 2018b; Jeffery et al., 2018) and this structure is associated with local adaptive genetic variation (Garcia de Leaniz et al., 2007). Any disruption to local adaptation in the wild stock can lead to an associated loss of fitness and potentially result in an extinction vortex in the wild fish (McGinnity et al., 2003). Mixing of fish from different populations, and the use of non-native fish, both may result in the homogenisation of population structure (Vasemagi et al., 2005; Williamson and May, 2005; Östergren et al., 2021) and associated loss of genetic adaptation (McGinnity et al., 2009). Where broodstock are collected from the wild, numbers used will often be lower than that found naturally. This will result in a loss of genetic diversity, even if all fish are from native populations, which again may negatively impact the diversity and resilience of the recipient populations (Christie et al., 2012b). Mating is also artificially achieved in the hatchery, with disruption of natural mate choice and reproductive timing (Neff et al., 2008; Tillotson et al., 2019) and associated risks to natural genetic variation.

In many instances escapees from aquaculture facilities are found in rivers together with wild fish (Youngson et al., 1997; Green et al., 2012; Wringe et al., 2018; Glover et al., 2019). Aquaculture stocks have been selected over many generations for traits of interest to the farming industry (Gjedrem, 2000; Gjedrem, 2010), have undergone domestication selection to their artificial environment (Vasemagi et al., 2012; López et al., 2018) and may be from areas far distant from the wild populations surrounding them (e.g. Norwegian origin farm stocks outside Norway). As such, aquaculture stocks are genetically and phenotypically very different to wild fish (Teletchea and Fontaine, 2014; Glover et al., 2017) and the fitness of escapees and farm/wild hybrids is much reduced compared to wild stocks (McGinnity et al., 2003; Glover et al., 2012; Skaala et al., 2012; Diserud et al., 2017; Glover et al., 2017; Skaala et al., 2019). Inadvertent inclusion of such fish in hatchery production will again result in loss of local adaption and fitness (Glover et al., 2017; Glover et al., 2020).

Fish produced and/or reared in a hatchery will face selection pressures to adapt to their hatchery environment, a process termed domestication, which can occur in as little as a single generation of captivity (Fleming and Einum, 1997; Fraser, 2008; Milot et al., 2013; Christie et al., 2016). In order to flourish in a hatchery environment a wide variety of different processes and behaviours (e.g. aggression, food conversion, predator avoidance, immune response etc) will be under different selection pressures than in the wild. Fish which do well in such conditions will tend to dominate, and selection will tend to drive the stocks away from the wild ideal (Araki et al., 2007; see also Appendix 1). Together with this direct genetic impact of domestication selection on genotype and associated fitness, there is also evidence for epigenetic modifications (changes that affect the genomic structure and regulate gene expression) induced by hatchery rearing, which also provide a potential explanatory mechanism for the reduced fitness of early generation hatchery-reared salmon (Le Luyer et al., 2017; Rodriguez Barreto et al., 2019). Genetic changes associated with domestication have been shown to both reduce the fitness of hatchery fish when stocked into the wild (Fleming and Einum, 1997; McGinnity et al., 2003; Araki et al., 2007; Hutchings and Fraser, 2008; McLean et al., 2008; Araki et al., 2009; Thériault et al., 2010; Thériault et al., 2011), and also reduce the fitness of the recipient wild stock (Araki et al., 2009; McGinnity et al., 2009).

Selection in the hatchery can cause direct genetic responses, however, hatchery conditioning can also cause plastic phenotypic impacts on fish raised in such environments. Processes such as growth rates, morphology, behaviour and life-history traits can all be influenced by such rearing environments (Chittenden et al., 2010). Although not a direct genetic effect, such changes can impact the genetic composition of the wild recipient population in an indirect way through competitive interactions between the hatchery and wild fish. Again then, genetic disruption of the wild population can result with associated fitness implications.

The various genetic impacts of hatchery production will often tend to reduce the genetic variability of a population. While this may have immediate impacts due to loss of fitness of individuals, there is also a longer-term risk associated with a populations and/or group of populations resilience to environmental change. The ability to adapt to such change is reliant on the inherent genetic variation both across (the portfolio effect: Schindler et al., 2015) and within populations (Bernatchez, 2016). Loss of such variation means that a population has no genetic resources upon which to call on in times of environmental change, and as such represents a long-term risk to population viability (McGinnity et al., 2009).

There has been a significant amount of both theoretical and more practical based analysis of techniques to reduce the negative genetic effects associated with hatchery rearing (reviewed in Fisch et al., 2015). Such techniques focus on broodstock selection (e.g. collect from single locally adapted populations; collect enough fish to maximise founder genetic diversity; minimise generations in captivity; minimise domestication selection; screen for escaped farm fish), broodstock spawning (e.g. develop optimal mating scheme such as random, factorial or free mate choice; inbreeding avoidance through various molecular and/or pedigree techniques), and rearing and release (e.g. enriched environments; equalisation of family sizes). Research is still needed however to evaluate the long-term effectiveness of such approaches or to see whether the benefit to the population justifies the cost (Fraser, 2008; Fisch et al., 2015). However, as techniques and technology improve both the theoretical and practical application of such approaches mean some of the negative genetic impacts of hatchery rearing may start to be able to be negated.

3.2 Environmental issues

Stocking of fish into areas of a river where natural production is taking place to try to boost fishery numbers has historically been the most common stocking practice, with the first commercial hatchery being developed on the Rhine in 1852 (Harris, 1978). However, from the early days, doubts about the efficiency of such programmes were expressed (Kennedy, 1988). What little scientific assessment that had been undertaken by the 1970's suggested that, although 38 organisations were operating hatcheries, their input contributed to less than 2.5% of commercial catches (Harris, 1978). Notwithstanding the significant developments in hatchery practices, scientific knowledge, ecosystem understanding and monitoring programmes in the 50 years since that date, it is interesting to note that in a recent study of enhancement hatchery inputs to the River Spey in Scotland, just 0% – 1.8% of the rod catch was identified as originating from the hatchery between 2018 – 2012 (Coulson et al., 2013). Such findings are perhaps to be expected. In the absence of external stressors, the production of a particular river section is limited by its carrying capacity and availability of broodstock. Fish stocked at or above carrying capacity will lead to increased detrimental competition with wild fish. In addition, when broodstock are taken which would otherwise have gone on to spawn naturally their contribution to natural spawning is removed. As such, it is unsurprising that there should often be little or no increase in fish production (Saltveit, 2006), and that this scenario has been realised across many different attempted enhancement programmes (e.g. Saltveit, 1993; Fjellheim et al., 1995; Saltveit, 1998; Einum and Fleming, 2001; Fjellheim and Johnsen, 2001 and references therein; Borgstrom et al., 2002; Araki and Schmid, 2010).

It is evident that positive effects of stocking on production can only be realised in a particular location if that location is below carrying capacity and has available spawners to be used as broodstock (Aprahamian et al., 2003). As rivers are highly dynamic systems, this carrying capacity will vary in both time and space (Armstrong et al., 2003) and act with different intensity on different life-history stages in different areas of a system (Malcolm et al., 2019). As such, in many, if not most situations, the limiting factors or bottlenecks for fish production are not well documented (Cowx, 1994b; Saltveit, 2006). In the absence of such knowledge and if accommodation cannot be made for the age-specific capacity of a system, such stocking can result in catastrophic consequences (Saltveit, 2006) for both the hatchery and wild stock, with competitive interactions resulting in reductions in both fitness and juvenile numbers of both types in the system (see examples in Fjellheim and Johnsen, 2001; McGinnity et al., 2009; Araki and Schmid, 2010) and displacement of wild gene frequencies/stocks (Altukhov, 1981; Hindar et al., 1991; Marzano et al., 2003).

A potential stocking enhancement alternative in systems already at or near carrying capacity is to stock fish at life history stages that avoid the age-specific density-dependent bottlenecks that limit production. Such stocking could involve placing ova at uniform densities to reduce local density-dependent mortality; stocking fish into areas that have limited spawning habitat but more extensive juvenile habitat and, what is recently perhaps the most common approach, to capture broodstock and raise offspring to the smolt stage before release.

A significant limitation to the carrying capacity of a system is the availability and distribution of spawning habitat (Taylor et al., 2017; Glover et al., 2018). A combination of factors such as river bed slope, flow, hydraulic and sedimentary variables all interact to define the suitability and, hence, availability this habitat (Moir et al., 1998; Louhi et al., 2008). Thus, in many, if not all, river systems, the distribution of optimal spawning habitat (Louhi et al., 2008) is not uniform (Crisp and Carling, 1989), but is rather a patchy distribution (Moir et al., 1998; Glover et al., 2018). In order to maximise egg to smolt production, hatchery eggs can, in theory, be planted in uniform densities, with the aim of reducing density-dependent mortalities. Quantifying the success of such a process is, however, not straightforward, as outcome is usually measured by smolt and/or adult counts (Glover et al., 2019), which are influenced by numerous factors unrelated to egg distribution. Where it has been possible to study each life history stage in detail, the artificial stocking of ova in a uniform distribution was found to increase juvenile numbers of stocked over wild fish up to the fry stage (Glover et al., 2019). However, strong density-dependent mortalities meant that this population size increase was not transmitted to later parr production (Glover et al., 2019) or smolt output (Bacon et al., 2015). It was concluded that such stocking fails to increase Atlantic salmon production where wild fish populations and suitable habitat remain (Bacon et al., 2015).

An alternative production enhancement strategy is to stock areas of habitat that are below juvenile carrying capacity. Apart from situations where spawning habitat is completely absent and so there is a complete lack of juvenile fish (see below), it is difficult to disentangle the various factors that may result in juvenile habitats being below carrying capacity. The intense density-dependent mortality seen in juvenile salmon (Vincenzi et al., 2012; Walters et al., 2013), especially at the earliest life history stages (Einum et al., 2006), means that sub-optimal juvenile numbers may be not be the result of a lack of spawning opportunities but rather from some other extrinsic stressor. If so, stocking may achieve little benefit (ICES, 2017). The added difficulty of accurately quantifying the juvenile carrying capacity of a system (Uusitalo et al., 2005) also raises the danger of overstocking, increased competition and associated negative consequences to both stocked and wild fish (Cowx, 1994b).

In order to bypass restrictions that a river imposes on production, either from natural (such as intrinsic carrying capacity) or anthropogenic (through a variety of stressors) sources, fish can be reared and released into the environment as smolts (Isaksson et al., 1997; Moberg and Salvanes, 2019). The goals of such programmes are to avoid early age class competitive interactions and increase captures in commercial and/or recreational fisheries by boosting production above that which could occur naturally (Isaksson, 1988; Mustafa et al., 2003). Such ranching programmes have been utilised extensively in an effort to boost fisheries of several salmonid species, including various Pacific salmon species in North America and Japan (Mustafa et al., 2003; Moberg and Salvanes, 2019). It is estimated that around 40% of the salmon in the Pacific Ocean are of hatchery origin (Ruggerone and Irvine, 2018). As with many hatchery programmes, however, and despite the long history and large scale of such hatchery production in the Pacific, their efficiency as a tool for increasing production has rarely been rigorously demonstrated (Naish et al., 2007; Amoroso et al., 2017). Whilst salmon numbers in the Pacific have increased significantly over the period that hatchery inputs have been operating; there has, at the same time, been a major change in productivity in the North Pacific boosting natural production (Amoroso et al., 2017). Disentangling the influence of hatcheries from that of natural variation is crucial to understanding the outcome of hatchery intervention. Where this has been attempted, the findings suggest that positive enhancement effects of the ranching are relatively minor (Morita et al., 2006; Scheuerell et al., 2015; Amoroso et al., 2017) and context-dependent (Kaev, 2012). Further, even if there may be some small enhancement benefit to the fishery, ranching programmes have the potential to negatively impact the wild stocks that they interact with and actually reduce productivity in the wild stocks (Hilborn and Eggers, 2000; Amoroso et al., 2017) through mechanisms, such as replacement (Hilborn and Eggers, 2000), together with enhanced straying rates (Brenner et al., 2012) and associated negative ecological and genetic interactions (Jasper et al., 2013).

In the North Atlantic, apart from some limited and often experimental programmes, Iceland has seen the most significant salmon ranching programme (Isaksson et al., 1997; ICES, 2019). Ranching for recapture at the river of release was initiated by the Icelandic government in 1961 (Arnason, 2001), with a number of large-scale facilities operating in the 1980's and 90's (Isaksson et al., 1997). However, due to issues surrounding significant straying, illegal fishing, and especially poor economic results (Arnason, 2001), the activity has now virtually ceased, is restricted to two rivers and productivity has decreased from a high of 499 tonnes in 1993 to just 28 tonnes in 2017 (ICES, 2019).

A final mechanism of commercial stock enhancement which avoids completely juvenile competitive interactions is to stock fish outside their natural ranges. Such an approach covers stocking both within river systems and in regions/oceans outside the species' native range. Some rivers already harbour wild populations of salmon, but include natural barriers, such as waterfalls, which mean that parts of the system have always been inaccessible to wild fish. Fish can be stocked in these inaccessible areas in order to maximise production for the river system as a whole beyond that which could occur naturally. Assuming the suitability of habitat and the ability for downstream passage of the barrier; such a strategy would undoubtedly increase the numbers of migrating smolts, due to the increased amount of productive habitat available. However, such a programme will also have introduced changes to the natural ecosystems in stocked areas, especially through competitive interactions with other fish species (Kennedy, 1982; Hearn, 1987; Berg et al., 2014). Such interactions have been shown to have the potential to change the distribution and depress the natural production of the native species following stocking with salmon (Kennedy and Strange, 1980; Kennedy and Strange, 1986) . Such stocking practices may also result in negative impacts on wild fish populations naturally spawning in areas below the barrier, as any returning spawners will be unable to migrate past the barrier and if all are not collected may stray into neighbouring populations and introduce restricted/novel genotypes into these populations (e.g. Östergren et al., 2021).

An extreme example of stocking fish outside their natural range is the trans-oceanic stocking of Pacific pink salmon (Oncorhynchus gorbuscha) leading to the establishment of self-sustaining populations in the Atlantic White and Barents Sea areas (Gordeeva and Salmenkova, 2011), resulting in a significant commercial fishery (reviewed in Niemelä et al., 2016). Hand-in-hand with this commercial 'success' came the potential and realised negative impacts on native salmonid stocks of other species, especially Atlantic salmon. Significant straying of pink salmon has been observed, especially into areas immediately surrounding the stocked region (e.g. Norway: Mo et al., 2018) but also throughout the whole North Atlantic (e.g. England, Scotland, Ireland, Iceland: Bartlett, 2017; Armstrong et al., 2018; Millane et al., 2019; Sandlund et al., 2019). Associated and ongoing risks of competitive interactions, pathogen and parasite transfer, unbalanced nutrient inputs and economic impacts on recreational fisheries (Mo et al., 2018) across the North Atlantic means such stocking practices have resulted in one of the most significant risks yet seen with any such enhancement programme.

Hybridisation of hatchery fish with wild conspecifics is a further potential risk associated with stocking programmes. If stocked fish breed at either the parr stage or as returning adult spawners again this can negatively impact the fitness of the wild stocks. Such outcomes are again the result of the disruption of natural genetic population structure (Östergren et al., 2021) and loss of genetic diversity (Marzano et al., 2003) associated with the hatchery inputs. Impacts can be severe, and cumulative, and the depressed recruitment and disruption in the capacity of natural populations to adapt to environmental change raises risks to the long term viability of such populations (McGinnity et al., 2009). Together with the problem of inter-specific hatchery/wild hybridisation also comes the risk of intra-species hybridisation. In areas which have seen hatchery stocking, especially where native wild populations are depressed (Garcia de Leaniz and Verspoor, 1989), or stocking takes part outside the native range (Verspoor, 1988), enhanced rates of intra-specific hybridisation may occur with associated potential future depression of recruitment.

Introduction of fish conditioned or selected in the hatchery can cause changes in their behaviours in the wild on top of the direct competitive interactions. Hatchery reared individuals are more aggressive and less risk adverse than wild counterparts (Johnsson et al., 1996). Such behaviours, taken together with an increased resource due to stocking, has been shown to risk the attraction of predators (Collis et al., 1995). In turn, this may lead to enhanced predation not only on the stocked fish but also on the wild with associated negative impacts on survival (Kennedy and Greek, 1988; Shively et al., 1996; Walter et al., 2005).

There is always a natural level of straying of a proportion of wild fish away from their natal spawning grounds (Keefer and Caudill, 2014). Such straying may be enhanced in hatchery fish (Jonsson et al., 1991a), and/or stocked fish displaced from their rearing site for release (Quinn, 1993). If fish are stocked as mitigation for a barrier in the river, there is also the risk that such fish will stray into populations below the barrier on return and again disrupt local genetic structure. Stocking thus carries risks not only to the enhanced population, but also those in surrounding areas.

A final environmental risk associated with stocking is the inadvertent transfer of parasites and/or disease from the hatchery to the wild. Diseases which become problematic in the hatchery may then be enhanced in the wild as has been seen with Bacterial Kidney Disease (BKD) caused by Renibacterium Salmoninarum in hatcheries stocking chinook salmon and steelhead in the Columbia river system (Elliott et al., 1997). By far the biggest such impact has been the introduction and spread of the ectoparasite Gyrodactylus salaris through the stocking of infected hatchery fish in Norwegian salmon populations, and which have devastated a number of rivers (Johnsen and Jensen, 1986). Extreme care careful monitoring, and a disease-free certification process should thus be carried out in any hatchery planning to introduce fish into the wild.

3.3 Anthropogenic impacts

The hatchery environment can not only change the behaviour of the stocked fish, but it can also change the behaviour of those exploiting the resource. Increasing a stock of fish such that it may be target for fishery and/or angling exploitation risks impacts not only on the wild remnants of the stocked population, especially if they are not easily distinguished through some sort of tagging, but also on neighbouring populations. In both the marine and freshwater environments fish exists in mixed river and/or population stocks. Exploitation of one stock can thus risk many stocks through by-catch as exploitation is increased targeted at stocked individuals (Hilborn, 1985; Beamish et al., 1997; Unwin and Glova, 1997). Such exploitation may be offset to a degree through marking of hatchery fish using mechanisms such as adipose clipping allowing hatchery fish to be identified and wild avoided (Saltveit, 2006; Bronte et al., 2012; ICES, 2018; WDFW, 2019).

Changes in behaviour can also be associated with changes in perception of both the resource and the environment in which it exists. Many people view wilderness as "the one place in our increasingly human-dominated world that is specifically designated to be left alone and not manipulated for human desires" (Landres et al., 2001). Any act which compromises this status can negatively influence perception of the wildness of both the environment and the fish resource. Catching a large stocked fish is, in many people's opinion, not the same as catching a truly wild individual. This may lead to lower catches if stocking for enhancement is not undertaken, and while this may suit those who view wilderness in this way, it may be sub-optimal to those who simply want to enjoy the environment while catching fish of any source.

There is a further risk that the operation of a hatchery becomes an end in itself due to the perception, often with no evidence, that stocking is bringing benefits to a system through the addition of fish. In the absence of such evidence, the rationale behind the programme rests on perception of success, tradition and the need to preserve employment opportunities. In reality stocking carries risks, as have been outlined, and may in fact be negatively impacting recipient populations. Even if such risks are not realised, in the absence of any proved benefits resources (and indeed employment) may be diverted from ecosystem management strategies which may result in better outcomes (Burton and Tegner, 2000; Carr et al., 2015).