Panel of experts: Dr. Ruth Withler (chair), Dr. Craig Busack, Mr. Richard Carmichael, Dr. Kenneth Currens, Prof. Tony Gharrett, Prof. Michael Gilpin, Dr. Stewart Grant (rapporteur), Prof. Michael Lynch, Prof. Thomas Quinn, Prof. Nils Ryman, Prof. Dolph Schluter, Prof. Eric Taylor

A panel of biologists with expertise in population genetics and related fields met on the second day of the workshop to summarize the information presented by the speakers, and to evaluate, as best they could in a day, the genetic effects of the straying of hatchery fish into natural populations. Panelists agreed firstly that one-way gene flow from genetically distinct, non-native fish into a set of local populations decreases the levels of diversity among populations. Secondly, such one-way gene flow also accelerates the loss of both neutral and selectively advantageous genetic diversity within populations. The loss of diversity within populations results in the potential for decreased fitness and, hence, reduced productivity in a short time frame (50 years and less), whereas the loss of diversity among populations decreases the flexibility to adapt to changing environmental conditions and, hence, decreases productivity of local populations in a longer time frame (50+ years). It is relatively easy to identify the demographic parameters affecting genetic diversity and fitness, and to determine the general direction of impact from straying as the parameter values change (Table 1 [below]). However, it is difficult to predict the magnitude of the overall impact because of our limited ability to quantify the effects of gene flow, natural selection, and other processes on natural salmon populations.

Forces that bring about genetic change in populations are mutation, migration, natural selection, and random genetic drift. Most of the panel's discussion focused on the relative importance of genetic drift and natural selection on the fate of genetic variation introduced by mutation or migration under a variety of population structures. Particular consideration was given to the effects of drift and natural selection on adaptive genetic variation in small populations. Many of the assumptions made in population genetic models, such as constant population size and discrete generations, are not met in salmonid populations, and the panel attempted to outline the consequences that violation of these assumptions would have on predicting the effects of non-native straying. Before specific questions about straying are addressed, the following brief descriptions of processes bringing about genetic change and demographic factors influencing the type and extent of change in salmonid populations are outlined.

Table 1. Parameters that influence the genetic effects that straying of non-native fish has on natural populations. Values on the right generally result in larger effects on natural populations.

Natural selection

The strength of natural selection for or against a trait (phenotype) is measured by the selection coefficient, s, which is the difference in fitness between the phenotype in question and an alternative form. Estimates of s for adaptive traits are between 0.0 and 0.5. However, reported values are most likely higher than the average value in nature, because the effects of small selection coefficients are difficult to detect. Moreover, for traits influenced by many genes, the quantity that affects the maintenance or loss of alleles at each locus is the strength of selection on a locus, not the strength of selection on the expressed trait.

Random genetic drift

Even in the absence of natural selection, not all spawners contribute equally to the next generation. Moreover, for any given spawner, both alleles at each locus may not be equally represented in the next generation, because of the chance segregation of alleles into gametes and because of the chance union of some gametes over others. Genetic drift is the random change in allele frequencies that occurs in finite populations due to the sampling of gametes between generations. The loss of alleles by genetic drift is unpredictable, but may be substantial in small populations, in which even beneficial alleles may be lost due to the effects of drift. A theoretical measure of the genetically effective population size is N_e, the size of a hypothetical population of equal sexes and random variation in family size that experiences the same amount of random drift as a natural population with census size N. In salmon, N_e for a population is the effective number of spawners each year times the generation length (in years).

Stray rate and gene flow

From the perspective of natural populations, the key parameter for measuring gene flow is the proportion of non-native fish actually spawning in the receiving local population(s). Only strays (immigrants) that actually reproduce in a locality result in gene flow into a local population. The observed number of strays in a local population is often used to calculate the proportion of immigrants, m, but stray fish may not reproduce, or may have reduced reproductive successes relative to local fish. Thus, m is the genetically effective rate of straying that represents actual gene flow into a population. The observed proportion of strays in a local population theoretically provides an upper limit for estimating m, assuming that strays have, at best, reproductive success equal to wild fish.

Just as the observed proportion of strays may overestimate the genetically effective proportion of strays, the census size, N, of a population typically overestimates the effective population size, N_e, that determines the genetic consequences of gene flow into a population. Studies of other organisms indicate that N_e is usually much less than N, so that N_e/N is often between 0.33 and 0.10. Since natural gene flow may increase the effective size of local populations, spawner counts alone may underestimate the effective sizes of geographically or temporally defined stocks. Such stocks represent partially isolated local subpopulations of a much larger metapopulation, and the much larger N_e of the metapopulation determines the impact of gene flow on the total population. However, when the N_e of the immigrant population is small, gene flow can reduce the N_e of a local population.

Mutation

Mutational rates differ from one part of the genome to another, and individual mutations vary in their effect on the organism in which they occur. In large populations, natural selection will effectively limit the frequency of deleterious alleles through the elimination of individuals homozygous for them. However, the impact of individual mutations on an organism are often low, so that in small populations genetic drift may lead to an accumulation of deleterious alleles in spite of selection against them.

Heterosis or hybrid vigor

Heterosis is an increase in fitness, primarily in the first (F₁) generation after hybridization, that results after mating between individuals from genetically different populations. It can be caused by the masking of deleterious recessive genes in inbred populations, and by balancing selection (heterozygote advantage) at some loci.

Outbreeding and outbreeding depression

Outbreeding is the mating of individuals from genetically divergent populations. If the genetic differences are great enough, the result can be a reduction in fitness, termed outbreeding depression. Two types of outbreeding depression may occur.

Type 1: Reduced hybrid fitness. Mating with individuals possessing traits that are maladaptive in the local environment will result in the production of hybrid progeny with reduced fitness in that environment. This effect will occur in the first (F₁) generation after hybridization and in subsequent generations if any of the hybrid progeny survive to spawn. However, a reduction in fitness from Type 1 outbreeding depression may be masked by an increase in fitness from heterosis, particularly in the F₁ generation.

Type 2: Breakup of coadapted gene complexes. Some combinations of alleles at different loci within a population may function as a unit, or a coadapted gene complex, to confer a selective advantage upon individuals. Matings between local and non-native individuals can lead to the disruption of these gene complexes and produce a reduction in fitness. Because F₁ hybrid progeny receive one complete set of chromosomes from each parent, gene complexes generally remain intact until chromosome reassortment and recombination occurs during reproduction. Thus, Type 2 outbreeding depression will typically not be apparent until F₂ or later generations.

Estimates of effective population size and gene flow can be used to predict the rate of loss of neutral genetic diversity within and among populations and, to a lesser degree, the loss in fitness within populations that can result from particular levels of one-way gene flow (straying) into these populations. For populations in which the selection coefficient, s, is less than the effective stray rate, m, the time in generations for which the proportion, P, of native genes remains in a local population is

For example, the time to a 50% loss of local, neutral genes (P = 0.5) is

With high levels of gene flow, the loss of 50% of alleles can occur fairly rapidly:

69 generations for m = 0.01 (1.0% gene flow)
25 generations for m = 0.025 (1.5% gene flow)
12 generations for m = 0.05 (5.0% gene flow).

Two important consequences follow from Equation (1): First, as m increases, the fraction of alleles lost also increases, and alleles are lost more rapidly. As the fraction of locally beneficial alleles that can be maintained in the population is reduced, the frequencies of nonlocal deleterious alleles will increase. The loss of alleles of adaptive importance at high values of m leads to reduced average fitness within the local populations in the short term, if the strays have lower fitness. Second, the proportion of stray spawners in a natural population, not simply population size, determines the rate at which local alleles are replaced by hatchery alleles. However, at small effective population sizes (N_e << 1,000), the loss of adaptive genetic diversity will be greatly accelerated by genetic drift. Deleterious alleles with effects less than the reciprocal of twice the effective population size (s < 1/2N_e) will not be eliminated by selection, and beneficial alleles with effects less 1/2N_e will be lost due to random drift.

The panelists attempted to predict the magnitude of genetic effects of non-native strays on local populations that results from altering the population structure and demographic factors listed in Table 1.

Genetic distance between hatchery and natural populations

In general, measures of genetic distance among salmon populations are based on biochemical and molecular markers that are assumed to be largely neutral to selection. Thus, biochemical and molecular genetic distances may provide better estimates of the time that populations have been separated, or of the magnitude of gene flow, than the measurement of adaptive traits subject to natural selection. However, the longer two populations have been isolated from each other, the more likely it is that they have diverged genetically, even for adaptive phenotypes shared by the populations. Similar phenotypes in two reproductively isolated populations may be due to convergence in which different genotypes produce the same phenotype through different genetic mechanisms. Thus, although genetic distance based on neutral traits may not be a linear indicator of the type and extent of adaptive differences between populations, the consequences of straying on adaptive traits are likely to increase with increasing genetic distance between populations. For neutral alleles, gene flow from a hatchery population may result in the replacement of local alleles with non-native alleles regardless of the genetic distance between the populations (Equation 1). The loss of genetic diversity within a population will be modified by any population substructuring (e.g., metapopulation structure), and the loss of diversity among populations will be determined by the number of populations receiving strays from the same hatchery stock.

Life-history similarity between the non-native hatchery and local populations

As indicated above, salmon populations that appear to be phenotypically similar for adaptive traits may be genetically different. Conversely, life-history differences between populations may reflect environmental rather than genetic differentiation. Thus, the level of phenotypic similarity exhibited by hatchery and local populations is not a reliable estimator of the amount of outbreeding depression (and hence loss of productivity) that can follow hybridization. Outbreeding depression is experimentally difficult to demonstrate because hybridization and reciprocal translocation experiments over several generations are required. The type and extent of outbreeding depression in salmonids is unknown, although it probably does occur.

Magnitude of straying

Persistent one-way straying at any level will eventually lead to the loss of effectively neutral genes in a local population (Equation 1), except when selection against F₁ hybrids is absolute. Even genes under positive selection in a local population will be replaced if the proportion of effective strays, m, is greater than the coefficient of selection, s, for local advantageous alleles. Natural selection, however, is expected to maintain genes with high fitnesses in local populations in spite of immigration, except when population sizes are very small. There are few reliable estimates of selection coefficients associated with alleles producing adaptive phenotypes in salmonid populations. The number of genes influencing the variability of a morphological or life-history trait is generally unknown, so the selective value of the trait itself provides an upper limit to the selective value of any one locus influencing the trait. Traits may be selectively advantageous at one life-history stage in one set of environmental circumstances, and selectively disadvantageous at another life-history stage or in another set of environmental circumstances. Thus, the differential in fitness between alternative phenotypes must be evaluated over the salmonid life cycle, and even then may vary over time depending on environmental conditions. It is experimentally more difficult to detect small selection coefficients than large ones, and difficult to measure the adaptive value of a trait over time. Thus, published selection coefficients are higher, on average, than values for most adaptive alleles, which likely have selection coefficients of less than 0.05 (5%). Such alleles would be lost from local populations experiencing consistent levels of gene flow of 5% and higher.

Reproductive success of hatchery strays

The reproductive success of hatchery strays is one factor that affects the magnitude of gene flow resulting from straying, as discussed above. If the reproductive successes of strays are low, the effective stray rate, m, and the rate of replacement of local genes are lower than would be estimated by simply calculating the proportion of strays in the local population. However, implications for the genetic diversity and productivity of local populations differ if selection operates on the migrants themselves rather than on their hybrid progeny. If hatchery migrants fail to compete for mates and do not otherwise disrupt normal spawning in the local population, the rate of introgression will be low, and productivity of the local population may be little affected. However, if migrants hybridize extensively with local fish but fail to produce viable progeny, much of the local, as well as the non-native, contribution to the next generation may be lost by selection against hybrid progeny. Thus, at high rates of immigration and hybridization, losses in productivity and genetic diversity in local populations may be substantial even if the reproductive success of hatchery migrants, as measured by surviving progeny, is low.

Duration of hatchery straying

The proportion of effectively neutral native alleles lost in a local population as a result of continuous one-way straying by non-native fish increases asymptotically with time (Equation 1). Long-term straying will lead to the replacement of local alleles with non-native alleles at effectively neutral loci (s < m). If migration from a hatchery population occurs for a short period of time (1-2 generations), or occurs only sporadically, natural selection may eliminate much of the hatchery contribution to the local population. Outbreeding depression resulting from this selection may be concentrated in the first generation after hybridization or may occur over several generations, depending on the nature of the outbreeding depression.

Local population size and structure

The proportion of migrant genes that are incorportated into the local population, not the absolute size of the local population alone, determines the effect of gene flow on the genetic composition of the local population. Populations with an N_e of 1,000 or more tend to act like populations of infinite size, so that little genetic diversity is lost through random drift. However, even in these large populations, the replacement of local alleles with hatchery alleles will proceed through gene flow (Equation 1).

There are few estimates of effective population size, N_e, for salmon populations. Many populations are currently of such small size that if N_e were only one-third to one-tenth the number of spawners, even summing numbers of spawners over the generation time yields estimates of N_e less than 50. This indicates that as few as 1 or 2 migrants spawning in the populations each year would have a large impact within 10 generations. While many salmonid populations are at historical low levels of abundance, some species (e.g., rainbow trout, steelhead, coho salmon) appear to have persisted in small populations over time without obvious signs of inbreeding depression. Salmonids may therefore form metapopulations consisting of small, partially isolated subpopulations with some natural level of reproductively effective straying between them. Straying may not be continuous or symmetrical among subpopulations and may occur only when triggered by particular environmental or demographic conditions.

Metapopulation structure can affect the rate of introgression of hatchery alleles into local subpopulations in two ways. First, the N_e of importance in determining what proportion of the population hatchery spawners represent is the N_e of the entire metapopulation. If hatchery strays enter only one or a few local subpopulations, they constitute a much smaller proportion of the metapopulation than the local populations to which they strayed. Second, if hatchery strays spawn in only a few of the local subpopulations, hatchery alleles replace local alleles in those subpopulations at a slow rate because strays from other subpopulations replenish native genes. The flow of hatchery alleles into subpopulations not directly receiving strays would also occur, but might be greatly slowed if natural straying between the subpopulations followed a stepping-stone model of migration.

Number of populations affected

Genetic differentiation among populations can decrease, if non-native fish stray into several local populations. This is true whether the populations are isolated from one another or whether they are subunits of a larger metapopulation. When hatchery straying occurs, hatchery alleles ultimately enter all the subpopulations of a metapopulation, but the replacement of local alleles is slower in subpopulations not directly receiving hatchery migrants. Therefore, local adaptations shared among subpopulations are less likely to be lost from the metapopulation as a whole if straying occurs into only one or a few local populations.

Salmon have evolved so that genetic differences, both neutral and adaptive, exist between populations in the presence of natural levels of gene flow. However, we do not know which of the following is the reason that observed population structure is maintained:

• Natural levels of gene flow are very low. Low levels of gene flow may be due to strong homing behavior, or to the lack of reproductive success of stray fish in local populations, or to a combination of both factors.
  
• Natural gene flow occurs only sporadically and therefore has much less potential to swamp genetic diversity within or between natural populations.
  
• Strays tend to be exchanged among geographically nearby populations, and this stepping stone mode of gene flow to more distant populations results in a) efficient natural selection against disadvantageous genes, and b) retardation of the introgression of neutral alleles into distant populations. This latter factor leads to large-scale regional differentiation among populations, even for neutral traits such as allozymes.
  

Without knowing whether salmon populations, finely tuned to their environments, are structured to withstand high levels of gene flow, or whether the natural level of gene flow is low, it is difficult to predict the consequences of increased amounts of straying from genetically dissimilar populations. If the reproductive success of strays is low, the direct genetic consequences of increased straying from non-native populations may be relatively small, although indirect effects may still be important. On the other hand, natural strays may be generally reproductively successful but would not destroy population structure, either because they are from nearby genetically similar populations or because straying is sporadic. Under these circumstances, hatchery strays, which are genetically dissimilar to natural populations and which are genetically or environmentally predisposed to straying, will have a greater detrimental effect on both the diversity and the fitness of a natural population.

The expected loss of genetic diversity from gene flow is based on populations that behave as if they were infinite, with effective sizes greater than 1,000 fish. The replacement of local genes by non-native genes for neutral traits follows the predictions of Equation 1, on average, and genes with selective coefficients greater than the immigration rate will be replaced. The suggestion that large populations, because of their size, can withstand the loss of productivity from outbreeding depression is based on the assumption that local populations are currently well adapted to their environments and are currently productive enough to seed the environment to carrying capacity even with diminished fry or smolt production, or both. However, the assumptions of well-adapted populations and productivity may not be true for many populations. Human activity and natural events have changed the habitats of many salmon populations so rapidly in recent decades that populations may not be as well adapted to their environments as was historically the case. Salmon populations have already experienced a loss of productivity, because of natural selection against some genotypes that occur naturally at high frequencies in the population. Heavily exploited populations may, at least in some years, possess fewer spawners than necessary to produce optimal numbers of juvenile fish. Current populations of salmon may not be able to maintain adequate levels of fitness, because they are smaller than they were historically, and because of rapidly changing environmental conditions. Gene flow from non-native fish is an additional challenge which will affect the ability of salmon to adapt to future changes and which can greatly decrease productivity.

What are the appropriate parameters to consider in evaluating the effects of straying?

Forces that bring about genetic change in populations are migration, mutation, natural selection, and random genetic drift. Mutation rates are typically low for most genes, and mutation was not considered in detail for the time frame of interest (< 100 years) (but see Lynch (in press) for more discussion of the role of mutation). Most of the discussion focused on the relative importance of drift migration, as well as selection under a variety of scenarios. Brief descriptions of some key terms and parameters follow.

• Stray rate: From the perspective of effects on natural populations, the key straying parameter is the proportion of non-native fish actually spawning in a natural population. The same number of strays can represent different rates of migration, depending on the size of the natural population. However, stray fish may not reproduce, and even if they do, they may have reduced reproductive success.
  
• Gene flow: The proportion of non-local genes that actually enter a natural population is known as gene flow and is denoted by m in the formulae presented here. Since gene flow occurs only to the extent that genes from stray fish become integrated into local natural populations, the stray rate is an approximate upper limit to the rate of gene flow.
  
• Local population size: The genetic consequences of straying depend on the effective population size (N_e) more than on census size (N). Studies of a variety of organisms indicate that the ratio N_ee/N is often about 0.1 to 0.33. Since a given year-class of young return over several years for most species of salmon, the effective population size for an entire generation is the average effective size per year times the generation length (for salmon, generation length is average age at spawning).
  
• Natural selection: The strength of natural selection on a genotype or discrete trait can be measured by the selection coefficient, s, which is the reduction in fitness of a trait or genotype compared to one with optimal fitness. For traits influenced by many genes, the quantity that matters most is the strength of selection (s) per locus, not the strength of selection on the trait itself.
  
• Random genetic drift: Genetic drift is the random change in allele frequencies that occurs in all finite populations due to the sampling of gametes between generations. The effects of drift are unpredictable and can be substantial in small populations. Isolated populations with N_e of about 50 or less will lose substantial amounts of genetic variability through drift. Populations with N_e of about 1,000 or more tend to act like an infinitely large population in which drift is not important.
  
• Inbreeding and inbreeding depression: Inbreeding is the mating of related individuals. In small populations, the level of inbreeding increases, because most or all individuals are closely related. A consequence of inbreeding is an overall increase in homozygosity. Homozygosity for recessive deleterious alleles, in turn, can cause a reduction in fitness known as inbreeding depression.
  
• Outbreeding and outbreeding depression: Outbreeding is the mating of genetically divergent individuals. If the genetic differences are large enough, the result can be a reduction in fitness known as outbreeding depression. Outbreeding depression can be caused by either (or both) of two factors: 1) loss of local adaptation, and 2) breakdown of favorable combinations among gene loci. Outbreeding depression due to loss of local adaptation may occur in the first generation after hybridization, but reductions in fitness due to breakdown of gene complexes may not be apparent until the F₂ or later generations. Fitness of the local population before the hybridization event is the reference point to evaluate whether outbreeding depression has occurred.
  

What other parameters are important in determining the genetic effects of straying?

Several other parameters also help to determine the genetic consequences of straying (see Table 1). Some of these are described below.

• Genetic and life-history differences between hatchery and natural populations: In general, larger differences between natural and hatchery populations will increase the effects of straying, but the dynamics of any particular situation can be complex. If differences between natural and hatchery populations are large, substantial reductions in productivity may occur in the short term because of outbreeding depression; however, reduced survival of strays and their hybrids may help to limit the extent of introgression into the native population. More modest genetic differences may not result in such large, short-term reductions in productivity, but persistent gene flow would probably cause the replacement of local genes with non-native ones. Genetic distances derived from molecular genetic data may not reflect adaptive differences between hatchery and natural populations. Conversely, life-history differences may be due to environmental factors as well as genetic differences.
  
• Magnitude of straying and strength of selection: The effects of these two parameters must be evaluated together. Persistent one-way straying at any level will eventually lead to the loss of selectively neutral genes in a local population. The rate of swamping of genes under selection--those that form the basis of local adaptation--depends on the magnitude of straying and the strength of selection. If the migration rate (M) is larger than the selection coefficient (s), migration will swamp local adaptations, so that even low rates of migration will eventually lead to the loss of genes with small selective advantage. Natural selection, however, is expected to maintain genes with large fitness values in a local population in spite of migration (except when population size is very small, in which case the effects of drift would dominate selection). It is not clear, however, what values of s are typical for adaptive traits. Estimates of s are biased toward large values because the use of small sample sizes in many experiments reduces the probability of detecting small but significant values.
  
• Duration of straying: The time in generations over which a particular amount of genetic diversity will be lost can be calculated. For example, the time to a 50% loss of variability for neutral and slightly adaptive genes is 69 generations for 1% gene flow, 25 generations for 2.5% gene flow, but only 12 generations for 5% gene flow. It is important to note that, as m increases, the time to a 50% loss of diversity also decreases, and the fraction of all genes subject to replacement increases.
  
• Number of natural populations affected: Greater reductions in genetic diversity among populations occur if non-native hatchery fish stray into multiple populations rather than into a single population of a metapopulation (with natural gene flow among subpopulations). Even if strays move only into a single population, they will still influence a wide range of populations, but at a much slower rate.
  

Do short- and long-term effects of straying differ?

Yes. A short-term infusion of non-native alleles may lead either to heterosis (increased fitness) or to outbreeding depression (decreased fitness), or both, in local populations. Although outbreeding depression and associated reduced productivity might persist in local populations for several generations after the straying occurred, selection against deleterious non-native alleles could result in the retention of primarily local genetic information. In contrast, long-term straying will eventually replace neutral genes and those with small adaptive effects (s < m). In small populations, the loss of genes with greater adaptive value will be accelerated by genetic drift.

Are the effects of hatchery straying likely to be permanent?

Yes. The changes in genetic structure of local populations resulting from straying are likely to be permanent. If straying stops, local populations may recover lost fitness over time through mutation, but the original genetic composition of the population will not be restored.

Can hatchery straying be beneficial for natural populations?

Theoretically, short-term straying can be beneficial under certain circumstances. The initial introduction of non-native alleles will generally increase genetic diversity in local populations. In well-adapted populations, this may cause a loss of adaptive fitness. However, in small populations experiencing inbreeding depression, the introduction of non-native alleles may mask effects of deleterious recessive genes and act to increase fitness.

Is there any safe level of hatchery straying that is consistent with the conservation of natural populations?

There are no "safe" levels of hatchery straying. Any level of long-term straying will change the structure of local populations. For neutral genes and genes with small adaptive effects, persistent straying at any level will lead to replacement of local alleles. Local alleles with adaptive values greater than migration (s > m) will be maintained, but selection against maladaptive non-native alleles will lead to reductions in productivity.

Can the effects of hatchery straying be predicted with any certainty?

To the extent that straying leads to one-way gene flow, initial changes in allele frequencies are predictable, as discussed above. However, the amount of gene flow resulting from a given level of straying is seldom known, and it is likely highly variable. Moreover, the fitness consequences of altered allele frequencies depend on the adaptive differences between local and non-native populations, which are seldom known. As a result, the effects of straying on average fitness in a local population, and on the long-term ability of a population to persist, are not predictable. Experimentation and long-term monitoring may be required to determine the effects of non-native gene flow into local populations. Both increased fitness from heterosis and decreased fitness from outbreeding depression may occur. Short-term monitoring of the effects of hatchery introgression may be overly optimistic because outbreeding depression may not occur until the second and succeeding generations after hybridization.

What will occur with straying at the 5% level?

As noted in the previous question, the genetic effects of straying at any given level cannot reliably be predicted, but some of the effects of gene flow are predictable. Based on estimates of gene flow from allozyme frequencies in natural populations, a value of 5% gene flow is much higher than that generally occurring between non-local salmon populations. Also based on what is known about the strength of selection in other animals, this amount of gene flow would quickly lead to the replacement not only of neutral genes, but also of locally adapted ones. Most genes in natural populations probably have selection coefficients less than 5% and would thus be subject to loss if gene flow occurred at this level. The panel found no genetic justification for allowing gene flow from non-native fish at levels as high as 5%.

What research should be undertaken to help resolve uncertainties of hatchery straying?

The following topics were identified as particularly important for research:

• The relationship between the rate of hatchery straying and the rate at which gene flow occurs.
• The nature and extent of outbreeding depression in natural salmon populations.
• Rates of straying and gene flow among natural populations.
• Selection intensities on whole traits.
• Genetic attributes of successful populations.
  

Lynch, M. In press. Genetic risks of extinction for Pacific salmon. In: A. D. MacCall and T. C. Wainwright (editors), Assessing extinction risk for West Coast salmonids: Proceedings of the workshop. U.S. Dep. Commer., NOAA Tech. Memo.

Table of Contents