- Research article
- Open Access
Molecular pedigree reconstruction and estimation of evolutionary parameters in a wild Atlantic salmon river system with incomplete sampling: a power analysis
© Aykanat et al.; licensee BioMed Central Ltd. 2014
- Received: 18 October 2013
- Accepted: 24 March 2014
- Published: 31 March 2014
Pedigree reconstruction using genetic analysis provides a useful means to estimate fundamental population biology parameters relating to population demography, trait heritability and individual fitness when combined with other sources of data. However, there remain limitations to pedigree reconstruction in wild populations, particularly in systems where parent-offspring relationships cannot be directly observed, there is incomplete sampling of individuals, or molecular parentage inference relies on low quality DNA from archived material. While much can still be inferred from incomplete or sparse pedigrees, it is crucial to evaluate the quality and power of available genetic information a priori to testing specific biological hypotheses. Here, we used microsatellite markers to reconstruct a multi-generation pedigree of wild Atlantic salmon (Salmo salar L.) using archived scale samples collected with a total trapping system within a river over a 10 year period. Using a simulation-based approach, we determined the optimal microsatellite marker number for accurate parentage assignment, and evaluated the power of the resulting partial pedigree to investigate important evolutionary and quantitative genetic characteristics of salmon in the system.
We show that at least 20 microsatellites (ave. 12 alleles/locus) are required to maximise parentage assignment and to improve the power to estimate reproductive success and heritability in this study system. We also show that 1.5 fold differences can be detected between groups simulated to have differing reproductive success, and that it is possible to detect moderate heritability values for continuous traits (h2 ~ 0.40) with more than 80% power when using 28 moderately to highly polymorphic markers.
The methodologies and work flow described provide a robust approach for evaluating archived samples for pedigree-based research, even where only a proportion of the total population is sampled. The results demonstrate the feasibility of pedigree-based studies to address challenging ecological and evolutionary questions in free-living populations, where genealogies can be traced only using molecular tools, and that significant increases in pedigree assignment power can be achieved by using higher numbers of markers.
- Atlantic salmon
- Incomplete sampling
- Parentage assignment
- Power analysis
- Reproductive success
Pedigree reconstruction provides a robust framework for the study of population evolutionary dynamics in the wild [1, 2]. By inferring the degree of relatedness among individuals within and between generations, it is possible to infer the survival and reproductive success of individuals, which, in turn, allows for the testing of a wide range of hypotheses relating to the demographic and evolutionary trajectories of populations and species over several generations. Furthermore, when used in combination with specific mixed-effects model statistical approaches [3–5], pedigree information allows for the estimation of relevant quantitative genetic parameters, including additive genetic and environmental variances and covariances for traits, as well as their underlying and associated quantitative trait loci (QTL). Thus, many empirical studies have successfully utilised pedigree information derived from physical tagging, genetic data or a combination of both approaches, to test a wide range of evolutionary and ecological hypotheses in the wild for various taxa [2, 6]. Such studies have provided insights on the long term response of populations to environmental change , the effect of inbreeding depression in small or introduced populations [8, 9], the elucidation of mating systems [10, 11], the estimation of the variance components of life history traits [12, 13], and the detection of QTLs [14, 15]. However, in species with external fertilisation and limited post-hatching parental care, the construction of pedigrees through observational means is limited by the inability to physically mark individuals and/or monitor parent-offspring associations. In addition, in cases where direct observations are possible, they may be subject to additional error such as those resulting from extra pair paternity as commonly shown in birds (e.g. see ). Therefore, the use of genetic information to reconstruct pedigrees emerges as a viable alternative to extract genealogical information within wild populations.
Atlantic salmon (Salmo salar) is an economically important species, with many wild populations subject to long-term population monitoring and biological sample collection spanning many decades. In principle, given the high level of homing to their natal rivers prior to reproduction, retrieving pedigree information in salmon is more achievable compared to many other fish species. Indeed, pedigree reconstruction in salmon and other salmonids based on DNA genotyping has been shown to be both feasible and valuable for conservation biology and evolutionary research [17–23]. Thus, pedigree-based approaches have been successfully used to address important questions in conservation biology, particularly in determining the efficacy of supplementing wild populations with captive bred individuals ; a number of these studies have demonstrated that captive bred salmon may have reduced reproductive success in the wild e.g. [19, 21–23, 25, 26], but see . Pedigree reconstruction in salmon is also important for estimating quantitative genetic parameters in wild populations. However, to date, the majority of quantitative genetic data in salmonids have been derived from common garden experiments in artificial environmental settings (for a review, see ). Whilst clearly providing valuable information of relevance to salmonid biology, the insights into population evolutionary dynamic processes provided by these data may be only of limited value when extrapolating to wild populations, due to the effect of variable environments on both additive genetic and environmental components of trait variance [6, 28–30]. To date, only a handful of studies have estimated quantitative genetic parameters of fitness-related traits in the wild e.g. [17, 31–33]. This indicates that additional efforts to reconstruct pedigrees in wild salmonid populations are required to fill this critical knowledge gap.
Despite the great potential for pedigree-based studies in wild salmonid populations, there are a number of possible problems that need to be considered. First, the sampling of potential parents in any given year is often incomplete. Partial sampling regimes may be due to several factors, such as logistical constraints (e.g. very large and/or complex aggregations of fish) or issues associated with the handling of wild fish, where possible impacts on behaviour or subsequent survival must be considered. Second, precocious male parr (i.e. males that mature in fresh water before going to sea and potentially, if they survive, returning to spawn second time) can be important components of breeding populations [34, 35] which may remain unsampled if only sea migrating fish that have not matured in freshwater previously are sampled. Finally, long-term datasets that may allow multi-generation pedigree reconstruction are often based on archived material. Depending on the storage conditions, DNA extracted from archived material is likely to be of varying quality, with older samples often more problematic as DNA quality tends to deteriorate with sample age e.g. . Given these many variables, it is imperative that power analysis type methods e.g.  are conducted in the early stages of project development. In the context of pedigree reconstruction, power analysis allows for the exploration of the best approaches to obtain reasonably accurate, unbiased parentage assignments and parameter estimates within the given logistical and budget constraints of a particular study.
Sampling statistics of the S. salar cohorts included in the study
Number of wild returning fish1
Number of sampled fish2
Total number of analysed fish3
% of sampled fish analysed
Overall % of samples analysed
In this study, we have evaluated the feasibility of a pedigree-based approach for exploring population biology and stock dynamics in the Burrishoole River Atlantic salmon. First, we performed a simulation and sensitivity analysis to detect the number of microsatellites required for maximising successful parentage assignment, given the available molecular markers, incomplete sampling regime and the variable quality of DNA in the samples available. Second, we estimated the power to detect variation in reproductive success among different groups of breeders. Finally, we estimated the detection power and confidence intervals of heritability estimates using simulations based on the empirical sampling and pedigree structure, and estimated the power gain in heritability and reproductive success detection when using higher number of markers.
Study system and sampling
Post-spawn wild individuals (i.e. kelts) were sampled on their downstream migration (March – May). Sampled individuals were sexed and measured for length (± 0.25 cm) and three to five scales were removed and archived in individual vanilla paper scale envelopes. Sex was determined on the basis of secondary sexual characteristics (principally the distinct elongation of the lower jaw in male fish known as a kype), which are readily discerned in post-spawned individuals. The current study utilised archived scales from ten consecutive cohorts collected between 1977 and 1986 (Table 1). This encompassed six discrete sets of parent-offspring cohort pairs, enabling family relationships and associated life history patterns to be established between linked cohorts (1977–1981, 1978–1982, 1979–1983, 1980–1984, 1981–1985 and 1982–1986), including two three-generational inter-related pedigrees (1977-1981-1985 and 1978-1982-1986; Figure 3b).
A total of 1640 fish were sampled (Figure 2), of which 1515 (92.5%) were successfully genotyped for at least seven microsatellite loci (Table 1, Figure 2), with the proportion of analysed individuals in each cohort ranging from 12.6% to 50.6% (Table 1, Figure 2). Assuming a four-year generation time between cohorts, the average reproductive success per cohort, calculated as the ratio of progeny to parental census size, also varied across years (Table 1). The sampling was highly skewed towards females as a result of higher male mortality associated with spawning, with female and male individuals comprising 81.2% and 17.8% of sampled proportion per cohort, respectively (Figure 2). The Burrishoole river is the national (Government of Ireland) index river for research into Atlantic salmon and operates under licence (Fisheries Acts 1959–2003) from the Department of Agriculture, Food and Marine and by permission of the Minister of Agriculture, Food and Marine. All sampling was carried out under authorisation (Sec. 4) of the Fisheries Acts, 1959 to 2003.
DNA genotyping and genetic diversity estimation
DNA was extracted from scale samples (two scales per fish) using a QIAamp mini kit (Qiagen Inc. Valencia, CA, USA) and 5 μl of proteinase K (0.5 mg/ml final concentration). The extraction was performed following the manufacturer’s instructions for DNA extraction from tissues, except that the elution volume was reduced to 100 μl instead of the suggested 200 μl in order to increase the DNA concentration. Each fish was genotyped at 14 microsatellite markers (ave. 12 alleles per locus, ave. Ho = 0.70) known to amplify reliably in archived scale samples . Further, they can be electrophoresed in a single semi-automated sequencer column (denoted as Panel 1 markers; Additional file 1) and amplified in two multiplexed PCR reactions (MP1 and MP2; Additional file 1). Four cohorts (1977, 1979, 1981 and, 1983) were further amplified at an additional 15 markers (Panel 2; Additional file 1), to be used in the heritability simulation and to further assess potential improvement in power through the use of higher marker number (see below for details). The sets of loci used for PCR multiplexing have earlier been optimised for use with scale-extracted DNA in a highly efficient manner, which yields high genotyping success with low error rates  and therefore provide an empirical basis for assessing optimal microsatellite number. The details of all primers, PCR reactions and genotyping information are given in Additional file 1. For fragment analysis, 10 μl of HiDi and 0.1 μl of GSL600 LIZ size standard (Life Technologies. Norwalk, CT, USA) were mixed with 2 μl of the pooled PCR multiplex dilution (see details in Additional file 1), and microsatellite fragments were separated using an ABI 3130xl Prism Genetic Analyzer. Genotyping was carried out with Genemarker 2.2 (Softgenetics) software using custom-made allele bins, and the allele scorings were visually inspected and edited.
Unless otherwise stated, all statistical analyses were performed using R software version 2.15.2 . Genetic diversity indices, including allelic richness, number of alleles, observed and expected heterozygosities, and the polymorphism information content (PIC) within cohorts, were calculated using the PopGenKit package 1.0 . Within-sample deviations from Hardy-Weinberg equilibrium (HWE) were estimated using the HWE.test.genind function (with 100000 permutations) of the adEgenet package 1.3-6 . Null allele frequencies were estimated using cervus 3.0 []. Allele frequencies were estimated for every combined parent-offspring cohort pair, both in parentage inference as well as when simulating genotypes, whereby no differentiation was assumed between parental and offspring cohorts. Therefore, we assessed temporal stability of the Burrishoole Atlantic salmon using genetic differentiation statistics θ ST  and Jost’s D  with the diveRsity package 1.5.0  and by bootstrapping loci 10000 times. For Jost’s D, we measured differentiation on a per locus basis (i.e. single locus estimate) as suggested in , since it is sensitive to mutation rate differences across loci .
where pji is the allele frequency of the ith allele of locus j. Although exclusion probability is not fully informative in models with genotyping error rate, it still provides a rough estimate of parentage assignment power on a per-locus basis and across all loci.
Parentage analysis and evaluation of marker number by simulations
We used a Bayesian approach in constructing the pedigrees as implemented in the MASTERBAYES package 2.50 . MASTERBAYES incorporates the estimation of genotyping error rate and the number of unsampled individuals when inferring pedigree structure, which provides better assignment rates especially if unsampled individuals represent a considerable proportion of the potential parent set [56, 57]. In the pedigree constructions, we assumed a four year interval between parent and offspring cohorts, such that an individual within a cohort was defined as a potential parent for any individual sampled four years later (Figure 3b).
Summary of descriptive statistics of the microsatellite marker panels used in the study
Panel 1 markers (N = 14)
Panel 2 markers (N = 15)
Total number of individuals
Total number of genotyped individuals
% Mean genotyping success
Mean allele number
Mean allelic richness
Mean H O
Mean H E
Null allele frequency
Combined exclusion probability
MASTERBAYES was also used to estimate the number of unsampled parents within a Bayesian framework [56, 57, 59]. We first estimated the priors for the number of unsampled parents based on census size information (Table 1) and the sex ratio. As females generally make up 55 to 70% of the salmon in the system, the prior used for unsampled females was 55 to 70% of the census size sampled from a uniform distribution. Likewise, the standard deviation for the number of unsampled females was estimated from the above distribution, but with a value three times larger to cover a broader parameter space in the random walk. Within males, precocious male parr may account for up to 30% of breeding males , which we accounted for in the unsampled male prior. We specified a mean of 17% mature male parr within all breeding males in the prior estimate with a high sampling variance of distribution. In the simulations, the priors were selected for a gamma distribution with shape and scale parameters of 20 and 0.8, respectively. (Scale parameter 20 is equivalent to 17% contribution of precocious males to the male breeding pool, while the shape parameter 0.8 makes the distribution slightly skewed to the left.) The standard deviation for the unsampled male number was estimated based on the above distributions, and a standard deviation ten times higher was used in the analysis to cover a much broader parameter space in the random walk and account for the higher uncertainty associated with the precocious male parr contribution.
Genotypes were simulated to generate hypothetical, but realistic, pedigrees in order to assess optimal marker number as well as for use in heritability power analyses. We simulated genotypes based on empirical data from the Burrishoole system using the simgenotype function in the MASTERBAYES package, based on the empirical allele frequency distributions of Panel 1 markers, and by defining the relationships among individuals based on the empirical parentage assignment using 14 markers (see below) with a low probability threshold (p = 0.50) to define a higher number of parental links. The genotyping error rate and missing genotype frequency were also parameterised based on the empirical results estimated above; for every simulated genotype, error rates were sampled from the posterior probability distribution. Six different marker numbers (5, 10, 14, 20, 28 and 50 loci per individual) were generated, and each pedigree associated with each of the six separate parent-offspring cohort pairs was simulated 100 times. The markers were generated randomly (with replacement) based on the empirical allele frequency distributions of Panel 1 markers, assuming independent assortment.
Parentage analysis of simulated genotypes
The parentage assignment in the simulated parent-offspring cohort pairs was conducted in MASTERBAYES with 13,000 iterations, 3,000 burn-ins and a thinning interval of 10. The mismatch tolerance for the initial data filtering was adjusted to two loci for all simulations except with 50 loci, where the mismatch tolerance was adjusted to three. Other parameters (e.g. priors for the unsampled male and female numbers, error rates) were the same as above. Metropolis Hasting acceptance rates were in the range of 25% - 35%, after tuning scaling constants to 0.5 and 0.1 for unsampled dam and unsampled sire, respectively, using the tunePed function in MASTERBAYES. Likewise, autocorrelation of Markov Chain was monitored in all analyses and was reasonable (r2 < 0.06) in all runs at respective thinning intervals. Next, most potential parents were accepted using one of five Bayesian posterior probability thresholds (p = 0.2, 0.4, 0.6, 0.8, and 0.95), and pedigrees were constricted based on the inferred parent-offspring relations . For every pedigree generated, we compared it to the simulated (true) pedigree and calculated the number of true and false assignments. Further, we explored the effect of parameterising missing genotypic information and genotyping error rates in the simulation on the accuracy of parentage assignment efficiency. For that purpose, we sequentially excluded these parameters in simulating genotypes, such that no missing genotypes were assumed or both error rates (ER1 and ER2) were assumed to be zero during the simulations (MASTERBAYES, simgenotype function), and the parentage analysis was otherwise performed as above. Finally, we attempted to use colony2 software for parentage and halfsib inference , but the initial results suggested prohibitively long run times would be required to resolve parentage. Likewise, accurate half-sib inference appeared to be unlikely, probably due to run time limitations or as the result of smaller half sib groupings.
Empirical pedigree reconstruction
We assigned parentage in six parent-offspring cohort pairs (with 14 Panel 1 markers) using the MASTERBAYES package (130,000 iterations, 30,000 burn-ins and a thinning interval of 50, mismatch tolerance = 2) and by using highly stringent (0.95) and less stringent (0.8) Bayesian posterior probability thresholds. The higher threshold is a commonly used threshold in the literature for the MASTERBAYES package e.g. [56, 57], while the lower threshold was selected arbitrarily.
Additionally, we used both Panel 1 and Panel 2 (i.e. 29 loci in total) in assigning parentage in a subset of parent-offspring cohort pairs (1977–1981 and 1979–1983). This further analysis was used to determine the detection power of heritability estimations (see below) and to compare empirical and simulated parentage efficiency.
Estimating the power to detect differences in relative reproductive success (RSS) among groups of individuals
Within a cohort, we simulated two hypothetical groups with differential reproductive success and estimated the detection power of this difference based on the demographics of Burrishoole salmon (i.e. census size and reproductive success of the cohort; Figure 2 and Table 1) and the successfully genotyped proportions of the parental and offspring cohorts (Table 1). For Burrishoole salmon, such groups could represent several life history traits with an apparent (i.e. observable) dichotomy in the population, such as the duration of the ocean migration (1 year sea-winter vs. multiyear sea-winter), early maturation in males (mature male parr vs. sea-migrating males), or origin of stock (wild vs. hatchery). For the analysis, we first arbitrarily categorised the parental individuals within a cohort into one of the two ontological groups that differ in reproductive success (males and females separately). We define reproductive success of an individual as the number of offspring of an individual that return to reproduce four years later. Then, we modelled the reproductive success of every individual within the parental cohort using a negative binomial distribution, with dispersion parameters of 0.25 and 0.75 for males and females, respectively (Additional file 4).
The dispersion parameters within each sex were selected arbitrarily to reflect a realistic distribution of reproductive success in salmon, such that male reproductive success variance is higher than that of females (i.e., Bateman’s principle ), and more males than females fail to reproduce (Additional file 4). The reproductive success of parental individuals summed up to the numbers of offspring in the empirical Burrishoole cohorts (i.e. empirical reproductive success of the cohort; Table 1), reflects the reproductive success difference between the two groups. The ratio of the parental group with higher reproductive success within the cohort was set to be 0.01, 0.05, 0.15, 0.3, 0.5, 0.7, 0.85, 0.95 or 0.99. Next, the individuals in the offspring cohort were allocated to a parent pair based on the above conditions. Finally, the parental and offspring cohorts were sampled from the population pool based on the empirical Burrishoole sampling proportions (Figure 2), and the observed RRS between parental groups was inferred from the ratio of progeny in the offspring cohort using Fisher’s exact test. To estimate power, we simulated the above routine 1,000 times for all six parent-offspring cohort pairs. We also included in the model the parentage assignment success rate, by parameterising assignment rate of simulated genotypes for 14 and 28 loci (see Parentage analysis of simulated genotypes section above), and contrasted the power when using relatively low (i.e. 14) and high (i.e. 28) number of markers.
Estimating the power to measure heritability of a continuous trait
We evaluated the power to detect the heritability of a continuous trait in two parent-offspring cohort pairs for which both panels of microsatellite markers (29 in total) had been genotyped (1977–1981, and 1979–1983), so that the pedigree structure was maximised (see Results for optimum marker number); we then compared the power to estimate heritability when using 14 vs. 29 markers. A continuous trait with heritability ranging from 0 to 1 with 0.05 increments was generated using the pedantics package 1.04  based on the cohort pedigree generated using MASTERBAYES (130,000 iterations, 30,000 burn-ins and a thinning interval of 50, probability threshold; p = 0.95). The unsampled male and female numbers and error rates for the MASTERBAYES routine were the same as above. Then, phenotypic data were generated 100 times for every heritability value, using the pedigree information. Next, heritability was estimated from the simulated phenotypic data with an animal model using the PEDIGREEMM package 0.2-4  after a slight modification as in . Finally, the range of estimated heritabilities and the ratio of significant estimates (p < 0.05) were calculated for every expected heritability value, using the likelihood ratio test as described in the RLRSIM package 2.0-12 .
The mean genotyping success over 29 loci (Panel 1 and Panel 2 combined) was 88.0%, ranging from 75.8% to 94.3% per locus (Table 2, Additional file 2). Within the 10 year sampling period, genotyping success was significantly dependent on the age of the specimen (Table 1; adjusted R 2 = 0.50, F (1,8) = 10.11, p = 0.013). The mean allelic richness among cohorts was between 10.2 and 11.4, with locus-specific levels ranging from 3.3 to 26.0 (Table 2, Additional file 2). The mean observed heterozygosity within a marker was similar among cohorts (mean of coefficient of variation = 0.08, SD = 0.05), but spanned a large interval among markers (0.145 to 0.872). Similarly, the polymorphism information content (PIC) ranged between 0.15 and 0.90. The markers generally did not deviate from HWE; within the Panel 1 markers, at most only one cohort out of 10 deviated from HWE per marker, and no cohort was consistently out of HWE (Additional file 2). Likewise, 10 out of 15 Panel 2 markers were in HWE in all four cohorts, but two markers (Sleel53b, SSD30b) departed from HWE in two out of four cohorts. Similar to the Panel 1 markers, no cohorts showed consistent deviations from HWE across the Panel 2 markers (Additional file 2). The estimated null allele frequencies were generally low, ranging from 0.001 to 0.051 and from -0.016 to 0.095 for the Panel 1 and Panel 2 markers, respectively (Table 2, Additional file 2). Pairwise genetic differentiation between temporal samples was very low with Weir and Cockerham’s θ ST ranging between 0.002 and 0.008, suggesting a high level of temporal stability among cohorts within the ten year sampling period (Additional file 5). Likewise, Jost’s D metric among loci was generally low such that only 6% of loci per cohort exceeded 0.1 (Additional file 5).
Prior and posterior estimates of number of unsampled parents
Posterior (95% CI)
Posterior (95% CI)
Parentage assignment simulations
We tested five different probability thresholds for parentage assignment in the simulations. At all thresholds, the number of correctly and incorrectly assigned individuals was similar when using 20 or more markers. At lower numbers of markers, lower probability thresholds had higher number of correct links resolved, but at the same time, the number of incorrectly assigned individuals also increased (Figure 4b).
Empirical parentage assignments
Empirical pedigree statistics
‘77-‘81 (28 loci)
‘79-’83 (28 loci)
Across all cohorts3
Samples available 1
Offspring with parent(s) identified
Maternal links identified
Paternal links identified
Total (at least one parent)
Total (both parents)
Maternal half sib families 4
Total family number
Mean family size
Maximum family size
Paternal half sib families 4
Total family number
Mean family size
Maximum family size
Two generational links
Adding Panel 2 markers (29 markers in total) increased the overall parentage assignment success (Table 4, Additional file 7), such that 1.3 and 1.6 times more individuals were assigned to parent-offspring cohort pairs at the 95% threshold with 29 markers than with 14 markers in 1977–1981 and 1979–1983, respectively (Table 4, Additional file 7). These ratios are not significantly different from the simulated expectations (chi-square test; χ2 = 0.417, d.f. = 1, p = 0.52, and χ2 = 0.111, d.f. = 1, p = 0.74 for the 1977–1981 and 1979–1983 parent-offspring cohorts, respectively).
Power to estimate relative reproductive success (RRS) among individuals
Detecting power to estimate heritability of a continuous trait
Many studies have successfully identified parentage in wild fish populations with incomplete parental sampling using 15 or less microsatellite markers e.g. [27, 35, 57, 66]. However, while explicit statistical methods are presented within parentage assignment packages to minimise false assignment rates at a low number of loci (i.e., type I errors) [1, 67], generally, no empirical assessments are available to identify the proportion of non-assigned true parents in the parental pool (i.e., type II errors) . Maximising the number of parental links that can be identified is crucial in achieving adequate power for testing the various hypotheses and questions that long-term population studies like this aspire to address, which in turn makes it important to maximise and evaluate the quality of parentage assignments, especially within the context of incomplete sampling. On the other hand, a trade-off exists between the occurrence of type I vs. type II errors. A strategy overlooking pedigree error, in the interest of maximising pedigree links, may not be optimal as pedigree errors may also be associated with low power in upstream analyses (i.e. estimation of heritability ). It is possible that general assumptions about parentage assignment efficiency can be demonstrated within a theoretical framework i.e. [59, 67]; however, empirical approaches, as we have demonstrated in this study, provide system-specific parameters (i.e., population size, sampling ratio, genotyping error rate, missing genotype information, empirical marker characteristics) to be utilised in the simulations, and thus allows for the use the minimum number of loci for parentage assignment analysis without excluding true parents, to be determined (see Figure 8). In the Burrishoole system, we concluded that genotyping with at least 20 (and preferably 28) markers (i) maximised parentage assignments without leaving substantial numbers of true parents unassigned (Figure 4); and (ii) improved the biological insights in both RRS and heritability analyses. Therefore, we highly recommend an evaluation of the power of a specific study by explicitly modelling the properties of the DNA markers used in the study as well as the demographical parameters.
Knowledge of the numbers of unsampled individuals is critical for efficient parentage assignments in the likelihood-based categorical allocation approaches (e.g. in cervus and MASTERBAYES), such that the probability of genotypes to be present among unsampled parents is included in confidence interval estimations of assigned parents [57, 59]. High error rates (i.e., incorrect assignments) are expected in likelihood-based parentage analyses (e.g. cervus) if the number of unsampled parents is incorrectly estimated in the model . Although estimating unsampled breeders is relatively easy in closed populations or well-established model systems, it is challenging in partially sampled populations and particularly problematic in salmon, where mature male parr may contribute substantially to the breeding pool. Furthermore, in the case of the Burrishoole system, hatchery origin individuals might also potentially contribute to the spawning population in the wild. Therefore, the simultaneous estimation of unsampled parents and parent-offspring linkages within a Bayesian framework is preferable for the Burrishoole system and for other systems where there is some doubt as to the number of unsampled potential parents. Indeed, in the Burrishoole system, the unsampled male number greatly exceeds the total enumerated number of returning males that have migrated to sea, which suggests a significant parr contribution to the breeding pool (Table 3) that may exceed the previous estimates of 30% of the total breeding pool  (but also see  for a review). However, posterior distribution based estimates of unsampled females are lower than the prior estimates in five out of six parent cohorts, which is expected, as not all unsampled females in the system are expected to breed or to have sampled offspring.
An important aspect of parentage analysis in the wild is that it enables comparisons of the reproductive success between different groups of fish to be undertaken e.g. [19, 59, 68, 69]. Groups of individuals with recorded differences in life histories (e.g. sea-age of maturity, male parr maturation) as well as salmon with different wild and hatchery provenance can be compared using family reconstructions. The latter has been of particular interest in this system due to its potential impact on the genetic makeup and productivity of the Burrishoole wild recipient population [25, 38, 41]. We showed using population parameters based on the Burrishoole system that differences in reproductive success of 2.5 fold can be detected successfully between different groups of fish (Figure 6a and b, Additional file 8), while 1.5 fold differences can be detected when cohorts are combined (Figure 6c and d, Additional file 8). These differences are much smaller than, for example, the empirical estimation of marine mortality rates between hatchery origin and wild individuals from the Burrishoole system (i.e., ranched individuals are five times less likely to survive ), though more recent estimates, based on the five year average from 2007 to 2011, show that Burrishoole hatchery origin fish were only two times less likely to survive . Unlike females, however, the low proportion of males sampled results in no power to detect the RRS among males in different groups (Additional file 8). We suggest that estimating the RRS among wild-born fish that may have been descended from hatchery or wild-origin fish might be one of the most tractable applications of the pedigree reconstruction approach in the Burrishoole system. Using 28 loci rather than just 14 improved the power in RRS estimation substantially (Figure 6a, b), which suggests that in the Burrishoole system, a difference in reproductive success within a cohort is more likely to be captured when genotyping is performed with 28 markers. This gain in power is clearer only when group proportions are overly skewed in the population (i.e. proportion of high reproductive success group <0.15, see Figure 6c, d). This scenario is quite likely, given that many dichotomies in life history or phenotype may be present with skewed proportions, such as sea-winterism or ranched proportions in the Burrishoole system [25, 42, 44].
Among Burrishoole salmon in this study, only moderate heritability values for a continuous trait can be detected with sufficient power. The values obtained are comparable to the heritability estimates of life history and morphological traits for salmonids, most of which have been reported under artificial environmental settings (for a review, see: ). The standard deviations of the simulated heritability estimates of this study (i.e., 0.23 and 0.16 for 1977–1981 and 1979–1983 parent-offspring cohorts, respectively, and 0.14 for both cohorts combined) are well within the range of distribution of standard errors reported in , where the mean of the standard errors of 2,108 heritability estimates is 0.15. At present, there are only a handful of heritability estimates for salmonids in the wild e.g. [17, 31–33], illustrating the value of being able to accurately estimate trait heritability in this and other systems. Using 29 loci instead of 14 improved the detection power markedly, especially in the heritability ranges where most of the salmonid values were observed (see inset in Figure 7, ). Furthermore, the estimates become more precise with higher number of loci (Additional file 9); when using 29 loci, the power increase in 79′-83′ parent offspring cohort is as much as the power gained when the two parent-offspring cohort information is combined, but using 14 loci (Figure 7). When using cohort information separately in heritability estimates it is important to measure evolutionary dynamics over time, especially upon changing environmental conditions [7, 71, 72], or when phenotypic variation is introduced by means of supplementation.
Another factor that should be considered is the potential influence of maternal effects on the accuracy of heritability estimates, particularly given that the Burrishoole pedigree is currently heavily reliant on maternal connections (Figures 2 and 4). This situation would have been problematic if the traits of interest were measured in young fish. However, as the maternal effect (environmental) in salmonids rarely persists into maturity (i.e., >9 months) ; it is likely that only a negligible portion of the estimated variance component can be attributed to the phenotype of the mother.
This study demonstrates the feasibility of pedigree-based research to estimate key population parameters, which in turn can be used to resolve difficult ecological and evolutionary questions in wild populations. The approaches described above provide a robust methodological framework for evaluating historically sampled populations for pedigree-based research and to retrieve valuable quantitative genetic information, even where sampling represents only a proportion of the total census population.
Philip McGinnity and Craig R Primmer shared senior authorship.
We would like to thank the staff of the Marine Institute, Burrishoole, for many years of sampling and data collection and Lee Hancox, Tony Holmes, Brian Clarke and Trevor Stafford for sorting the archival scales and fish details for this study. Katja Salminen, Terhi Pajula, Pavel Matos Maravi, and Janne Sulku are thanked for laboratory assistance. Three anonymous reviewers are thanked for their constructive comments on an earlier version of the manuscript. Part of the analysis was performed using computer resources of the IT Center for Science Ltd. (CSC) of the Ministry of Education, Science and Culture, Finland. This study was supported by the Finnish Academy (grants 141231 and 137710 to CRP). PMG, TR were supported by the Beaufort Marine Research Award in Fish Population Genetics funded by the Irish Government under the Sea Change Programme.
- Jones AG, Ardren WR: Methods of parentage analysis in natural populations. Mol Ecol. 2003, 12: 2511-2523. 10.1046/j.1365-294X.2003.01928.x.PubMedView ArticleGoogle Scholar
- Kruuk LE, Hill WG: Introduction. Evolutionary dynamics of wild populations: the use of long-term pedigree data. Proc Biol Sci. 2008, 275 (1635): 593-596. 10.1098/rspb.2007.1689.PubMedPubMed CentralView ArticleGoogle Scholar
- Kruuk LEB: Estimating genetic parameters in natural populations using the ‘animal model’. Philos Trans R Soc B. 2004, 359: 873-890. 10.1098/rstb.2003.1437.View ArticleGoogle Scholar
- George AW, Visscher PM, Haley CS: Mapping quantitative trait loci in complex pedigrees: a two-step variance component approach. Genetics. 2000, 156 (4): 2081-2092.PubMedPubMed CentralGoogle Scholar
- Almasy L, Blangero J: Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet. 1998, 62 (5): 1198-1211. 10.1086/301844.PubMedPubMed CentralView ArticleGoogle Scholar
- Pemberton JM: Wild pedigrees: the way forward. Proc Biol Sci. 2008, 275 (1635): 613-621. 10.1098/rspb.2007.1531.PubMedPubMed CentralView ArticleGoogle Scholar
- Ozgul A, Tuljapurkar S, Benton TG, Pemberton JM, Clutton-Brock TH, Coulson T: The dynamics of phenotypic change and the shrinking sheep of St. Kilda. Science. 2009, 325 (5939): 464-467. 10.1126/science.1173668.PubMedView ArticleGoogle Scholar
- Bensch S, Hasselquist D, Vonschantz T: Genetic similarity between parents predicts hatching failure - nonincestuous inbreeding in the Great Reed Warbler. Evolution. 1994, 48 (2): 317-326. 10.2307/2410095.View ArticleGoogle Scholar
- Slate J, Kruuk LE, Marshall TC, Pemberton JM, Clutton-Brock TH: Inbreeding depression influences lifetime breeding success in a wild population of red deer (Cervus elaphus). Proc Biol Sci. 2000, 267 (1453): 1657-1662. 10.1098/rspb.2000.1192.PubMedPubMed CentralView ArticleGoogle Scholar
- Primmer CR, Moller AP, Ellegren H: Resolving genetic relationships with microsatellite markers: a parentage testing system for the swallow Hirundo rustica. Mol Ecol. 1995, 4 (4): 493-498. 10.1111/j.1365-294X.1995.tb00243.x.PubMedView ArticleGoogle Scholar
- DiBattista JD, Feldheim KA, Thibert-Plante X, Gruber SH, Hendry AP: A genetic assessment of polyandry and breeding-site fidelity in lemon sharks. Mol Ecol. 2008, 17 (14): 3337-3351. 10.1111/j.1365-294X.2008.03833.x.PubMedView ArticleGoogle Scholar
- Garant D, Kruuk LEB, Wilkin TA, McCleery RH, Sheldon BC: Evolution driven by differential dispersal within a wild bird population. Nature. 2005, 433: 60-65. 10.1038/nature03051.PubMedView ArticleGoogle Scholar
- Dibattista JD, Feldheim KA, Garant D, Gruber SH, Hendry AP: Evolutionary potential of a large marine vertebrate: quantitative genetic parameters in a wild population. Evolution. 2009, 63 (4): 1051-1067. 10.1111/j.1558-5646.2008.00605.x.PubMedView ArticleGoogle Scholar
- Johnston SE, Beraldi D, McRae AF, Pemberton JM, Slate J: Horn type and horn length genes map to the same chromosomal region in Soay sheep. Heredity (Edinb). 2010, 104 (2): 196-205. 10.1038/hdy.2009.109.View ArticleGoogle Scholar
- Slate J, Visscher PM, MacGregor S, Stevens D, Tate ML, Pemberton JM: A genome scan for quantitative trait loci in a wild population of red deer (Cervus elaphus). Genetics. 2002, 162 (4): 1863-1873.PubMedPubMed CentralGoogle Scholar
- Charmantier A, Reale D: How do misassigned paternities affect the estimation of heritability in the wild?. Mol Ecol. 2005, 14 (9): 2839-2850. 10.1111/j.1365-294X.2005.02619.x.PubMedView ArticleGoogle Scholar
- Dickerson BR, Willson MF, Bentzen P, Quinn TP: Heritability of life history and morphological traits in a wild pink salmon population assessed by DNA parentage analysis. Trans Am Fish Soc. 2005, 134 (5): 1323-1328. 10.1577/T04-006.1.View ArticleGoogle Scholar
- Herbinger CM, O’Reilly PT, Verspoor E: Unravelling first-generation pedigrees in wild endangered salmon populations using molecular genetic markers. Mol Ecol. 2006, 15 (8): 2261-2275. 10.1111/j.1365-294X.2006.02923.x.PubMedView ArticleGoogle Scholar
- Araki H, Cooper B, Blouin MS: Genetic effects of captive breeding cause a rapid, cumulative fitness decline in the wild. Science. 2007, 318: 100-103. 10.1126/science.1145621.PubMedView ArticleGoogle Scholar
- Araki H, Cooper B, Blouin MS: Carry-over effect of captive breeding reduces reproductive fitness of wild-born descendants in the wild. Biol Lett. 2009, 5 (5): 621-624. 10.1098/rsbl.2009.0315.PubMedPubMed CentralView ArticleGoogle Scholar
- Serbezov D, Bernatchez L, Olsen EM, Vollestad LA: Mating patterns and determinants of individual reproductive success in brown trout (Salmo trutta) revealed by parentage analysis of an entire stream living population. Mol Ecol. 2010, 19 (15): 3193-3205. 10.1111/j.1365-294X.2010.04744.x.PubMedView ArticleGoogle Scholar
- Christie MR, Marine ML, French RA, Blouin MS: Genetic adaptation to captivity can occur in a single generation. Proc Natl Acad Sci U S A. 2012, 109 (1): 238-242. 10.1073/pnas.1111073109.PubMedPubMed CentralView ArticleGoogle Scholar
- Anderson JH, Faulds PL, Atlas WI, Quinn TP: Reproductive success of captively bred and naturally spawned Chinook salmon colonizing newly accessible habitat. Evol Appl. 2013, 6 (2): 165-179. 10.1111/j.1752-4571.2012.00271.x.PubMedPubMed CentralView ArticleGoogle Scholar
- Laikre L, Schwartz MK, Waples RS, Ryman N, Ge MWG: Compromising genetic diversity in the wild: unmonitored large-scale release of plants and animals. Trends Ecol Evol. 2010, 25 (9): 520-529. 10.1016/j.tree.2010.06.013.PubMedView ArticleGoogle Scholar
- McGinnity P, Jennings E, DeEyto E, Allott N, Samuelsson P, Rogan G, Whelan K, Cross T: Impact of naturally spawning captive-bred Atlantic salmon on wild populations: depressed recruitment and increased risk of climate-mediated extinction. Proc Biol Sci. 2009, 276 (1673): 3601-3610. 10.1098/rspb.2009.0799.PubMedPubMed CentralView ArticleGoogle Scholar
- Milot E, Perrier C, Papillon L, Dodson JJ, Bernatchez L: Reduced fitness of Atlantic salmon released in the wild after one generation of captive breeding. Evol Appl. 2013, 6 (3): 472-485. 10.1111/eva.12028.PubMedPubMed CentralView ArticleGoogle Scholar
- Hess MA, Rabe CD, Vogel JL, Stephenson JJ, Nelson DD, Narum SR: Supportive breeding boosts natural population abundance with minimal negative impacts on fitness of a wild population of Chinook salmon. Mol Ecol. 2012, 21 (21): 5236-5250. 10.1111/mec.12046.PubMedPubMed CentralView ArticleGoogle Scholar
- Carlson SM, Seamons TR: A review of quantitative genetic components of fitness in salmonids: implications for adaptation to future change. Evol Appl. 2008, 1 (2): 222-238. 10.1111/j.1752-4571.2008.00025.x.PubMedPubMed CentralView ArticleGoogle Scholar
- Roff DA: Evolutionary Quantitative Genetics. 1997, New York, NY: Chapman & HallView ArticleGoogle Scholar
- Wilson AJ, Pemberton JM, Pilkington JG, Coltman DW, Mifsud DV, Clutton-Brock TH, Kruuk LE: Environmental coupling of selection and heritability limits evolution. PLoS Biol. 2006, 4 (7): e216-10.1371/journal.pbio.0040216.PubMedPubMed CentralView ArticleGoogle Scholar
- Wilson AJ, Hutchings JA, Ferguson MM: Selective and genetic constraints on the evolution of body size in a stream-dwelling salmonid fish. J Evol Biol. 2003, 16 (4): 584-594. 10.1046/j.1420-9101.2003.00563.x.PubMedView ArticleGoogle Scholar
- Theriault V, Garant D, Bernatchez L, Dodson JJ: Heritability of life-history tactics and genetic correlation with body size in a natural population of brook charr (Salvelinus fontinalis). J Evol Biol. 2007, 20 (6): 2266-2277. 10.1111/j.1420-9101.2007.01417.x.PubMedView ArticleGoogle Scholar
- Serbezov D, Bernatchez L, Olsen EM, Vollestad LA: Quantitative genetic parameters for wild stream-living brown trout: heritability and parental effects. J Evol Biol. 2010, 23: 1631-1641. 10.1111/j.1420-9101.2010.02028.x.PubMedView ArticleGoogle Scholar
- Jordan WC, Fleming IA, Garant D: Mating system and social structure. The Genetics of Atlantic Salmon: Implications for Conservation. Edited by: Verspoor E, Nielsen J, Stradmeyer L. 2007, Oxford: BlackwellGoogle Scholar
- Christie MR, Marine ML, Blouin MS: Who are the missing parents? Grandparentage analysis identifies multiple sources of gene flow into a wild population. Mol Ecol. 2011, 20 (6): 1263-1276. 10.1111/j.1365-294X.2010.04994.x.PubMedView ArticleGoogle Scholar
- Johnston SE, Lindqvist M, Niemela E, Orell P, Erkinaro J, Kent MP, Lien S, Vaha JP, Vasemagi A, Primmer CR: Fish scales and SNP chips: SNP genotyping and allele frequency estimation in individual and pooled DNA from historical samples of Atlantic salmon (Salmo salar). BMC Genomics. 2013, 14 (1): 439-10.1186/1471-2164-14-439.PubMedPubMed CentralView ArticleGoogle Scholar
- Mills CPR, Piggins DJ: The release of Reared Salmon Smolts (Salmo-Salar) into the Burrishoole River System (Western Ireland) and their contribution to the rod and line fishery. Fish Manage. 1983, 14 (4): 165-175.Google Scholar
- Piggins DJ, Mills CPR: Comparative aspects of the biology of naturally produced and hatchery-Reared Atlantic Salmon Smolts (Salmo-Salar L). Aquaculture. 1985, 45: 321-333. 10.1016/0044-8486(85)90278-9.View ArticleGoogle Scholar
- McGinnity P, Prodohl P, Ferguson A, Hynes R, Maoileidigh NO, Baker N, Cotter D, O’Hea B, Cooke D, Rogan G, Taggart J, Cross T: Fitness reduction and potential extinction of wild populations of Atlantic salmon, Salmo salar, as a result of interactions with escaped farm salmon. Proc Biol Sci. 2003, 270 (1532): 2443-2450. 10.1098/rspb.2003.2520.PubMedPubMed CentralView ArticleGoogle Scholar
- Nielsen C, Holdensaard G, Petersen HC, Bjornsson BT, Madsen SS: Genetic differences in physiology, growth hormone levels and migratory behaviour of Atlantic salmon smolts. J Fish Biol. 2001, 59 (1): 28-44. 10.1111/j.1095-8649.2001.tb02336.x.View ArticleGoogle Scholar
- Poole WR, Nolan DT, Wevers T, Dillane M, Cotter D, Tully O: An ecophysiological comparison of wild and hatchery-raised Atlantic salmon (Salmo salar L.) smolts from the Burrishoole system, western Ireland. Aquaculture. 2003, 222 (1–4): 301-314.View ArticleGoogle Scholar
- McGinnity P, Prodohl P, Maoileidigh NO, Hynes R, Cotter D, Baker N, O’Hea B, Ferguson A: Differential lifetime success and performance of native and non-native Atlantic salmon examined under communal natural conditions. J Fish Biol. 2004, 65: 173-187. 10.1111/j.0022-1112.2004.00557.x.View ArticleGoogle Scholar
- Byrne CJ, Poole R, Rogan G, Dillane M, Whelan KF: Temporal and environmental influences on the variation in Atlantic salmon smolt migration in the Burrishoole system 1970–2000. J Fish Biol. 2003, 63: 13-View ArticleGoogle Scholar
- Metcalfe NB, Thorpe JE: Determinants of geographical variation in the age of Seaward Migrating Salmon, Salmo-Salar. J Anim Ecol. 1990, 59 (1): 135-145. 10.2307/5163.View ArticleGoogle Scholar
- Thompson CE, Poole WR, Matthews MA, Ferguson A: Comparison, using minisatellite DNA profiling, of secondary male contribution in the fertilisation of wild and ranched Atlantic salmon (Salmo salar) ova. Can J Fish Aquat Sci. 1998, 55 (9): 2011-2018. 10.1139/f98-076.View ArticleGoogle Scholar
- Vaha JP, Erkinaro J, Niemela E, Primmer CR: Life-history and habitat features influence the within-river genetic structure of Atlantic salmon. Mol Ecol. 2007, 16 (13): 2638-2654. 10.1111/j.1365-294X.2007.03329.x.PubMedView ArticleGoogle Scholar
- R-Core-Team: R: A language and environment for statistical computing. R Foundation for Statistical Computing. 2012, Vienna, AustriaGoogle Scholar
- Paquette RS: Useful functions for (batch) file conversion and data resampling in microsatellite datasets. R package version 1.0. 2012, http://CRAN.R-project.org/package=PopGenKit,Google Scholar
- Jombart T: adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics. 2008, 24 (11): 1403-1405. 10.1093/bioinformatics/btn129.PubMedView ArticleGoogle Scholar
- Kalinowski ST, Taper ML, Marshall TC: Revising how the computer program CERVUS accommodates genotyping error increases success in paternity assignment. Mol Ecol. 2007, 16 (5): 1099-1106. 10.1111/j.1365-294X.2007.03089.x.PubMedView ArticleGoogle Scholar
- Weir BS, Cockerham CC: Estimating F-statistics for the analysis of population-structure. Evolution. 1984, 38 (6): 1358-1370. 10.2307/2408641.View ArticleGoogle Scholar
- Jost LOU: GSTand its relatives do not measure differentiation. Mol Ecol. 2008, 17 (18): 4015-4026. 10.1111/j.1365-294X.2008.03887.x.PubMedView ArticleGoogle Scholar
- Keenan K, McGinnity P, Cross TF, Crozier WW, Prodöhl PA: diveRsity: An R package for the estimation and exploration of population genetics parameters and their associated errors. Methods Ecol Evol. 2013, 4 (8): 782-788. 10.1111/2041-210X.12067.View ArticleGoogle Scholar
- Meirmans PG, Hedrick PW: Assessing population structure: F(ST) and related measures. Mol Ecol Resour. 2011, 11 (1): 5-18. 10.1111/j.1755-0998.2010.02927.x.PubMedView ArticleGoogle Scholar
- Whitlock MC: G’ST and D do not replace FST. Mol Ecol. 2011, 20 (6): 1083-1091. 10.1111/j.1365-294X.2010.04996.x.PubMedView ArticleGoogle Scholar
- Hadfield JD, Richardson DS, Burke T: Towards unbiased parentage assignment: combining genetic, behavioural and spatial data in a Bayesian framework. Mol Ecol. 2006, 15 (12): 3715-3730. 10.1111/j.1365-294X.2006.03050.x.PubMedView ArticleGoogle Scholar
- Koch M, Hadfield JD, Sefc KM, Sturmbauer C: Pedigree reconstruction in wild cichlid fish populations. Mol Ecol. 2008, 17 (20): 4500-4511. 10.1111/j.1365-294X.2008.03925.x.PubMedView ArticleGoogle Scholar
- Wang J: Sibship reconstruction from genetic data with typing errors. Genetics. 2004, 166 (4): 1963-1979. 10.1534/genetics.166.4.1963.PubMedPubMed CentralView ArticleGoogle Scholar
- Nielsen R, Mattila DK, Clapham PJ, Palsboll PJ: Statistical approaches to paternity analysis in natural populations and applications to the North Atlantic humpback whale. Genetics. 2001, 157 (4): 1673-1682.PubMedPubMed CentralGoogle Scholar
- Wang J: Computationally efficient sibship and parentage assignment from multilocus marker data. Genetics. 2012, 191 (1): 183-194. 10.1534/genetics.111.138149.PubMedPubMed CentralView ArticleGoogle Scholar
- Arnold SJ, Duvall D: Animal mating systems - a synthesis based on selection theory. Am Nat. 1994, 143 (2): 317-348. 10.1086/285606.View ArticleGoogle Scholar
- Morrissey MB, Wilson AJ: pedantics: an r package for pedigree-based genetic simulation and pedigree manipulation, characterization and viewing. Mol Ecol Resour. 2010, 10 (4): 711-719.PubMedView ArticleGoogle Scholar
- Vazquez AI, Bates DM, Rosa GJ, Gianola D, Weigel KA: Technical note: an R package for fitting generalized linear mixed models in animal breeding. J Anim Sci. 2010, 88 (2): 497-504. 10.2527/jas.2009-1952.PubMedView ArticleGoogle Scholar
- Aykanat T, Bryden CA, Heath DD: Sex-biased genetic component distribution among populations: additive genetic and maternal contributions to phenotypic differences among populations of Chinook salmon. J Evol Biol. 2012, 25 (4): 682-690. 10.1111/j.1420-9101.2012.02462.x.PubMedView ArticleGoogle Scholar
- Scheipl F, Greven S, Kuchenhoff H: Size and power of tests for a zero random effect variance or polynomial regression in additive and linear mixed models. Comput Stat Data Anal. 2008, 52: 3283-3299. 10.1016/j.csda.2007.10.022.View ArticleGoogle Scholar
- Ford MJ, Williamson KS: The aunt and uncle effect revisited–the effect of biased parentage assignment on fitness estimation in a supplemented salmon population. J Hered. 2010, 101 (1): 33-41. 10.1093/jhered/esp068.PubMedView ArticleGoogle Scholar
- Harrison HB, Saenz-Agudelo P, Planes S, Jones GP, Berumen ML: Relative accuracy of three common methods of parentage analysis in natural populations. Mol Ecol. 2013, 22 (4): 1158-1170. 10.1111/mec.12138.PubMedView ArticleGoogle Scholar
- Araki H, Blouin MS: Unbiased estimation of relative reproductive success of different groups: evaluation and correction of bias caused by parentage assignment errors. Mol Ecol. 2005, 14 (13): 4097-4109. 10.1111/j.1365-294X.2005.02689.x.PubMedView ArticleGoogle Scholar
- Araki H, Berejikian BA, Ford MJ, Blouin MS: Fitness of hatchery-reared salmonids in the wild. Evol Appl. 2008, 1 (2): 342-355. 10.1111/j.1752-4571.2008.00026.x.PubMedPubMed CentralView ArticleGoogle Scholar
- ICES: Report of the Working Group on North Atlantic Salmon (WGNAS), 3–12 April, 2013. ICES CM 2013/ACOM:09, pp376. 2013, Copenhagen, DenmarkGoogle Scholar
- Morrissey MB: Exploiting natural history variation: looking to fishes for quantitative genetic models of natural populations. Ecol Freshw Fish. 2011, 20 (3): 328-345. 10.1111/j.1600-0633.2010.00445.x.View ArticleGoogle Scholar
- Johnston SE, Gratten J, Berenos C, Pilkington JG, Clutton-Brock TH, Pemberton JM, Slate J: Life history trade-offs at a single locus maintain sexually selected genetic variation. Nature. 2013, 502 (7469): 93-95. 10.1038/nature12489.PubMedView ArticleGoogle Scholar
- Heath DD, Fox CW, Heath JW: Maternal effects on offspring size: variation through early development of chinook salmon. Evolution. 1999, 53: 1605-1611. 10.2307/2640906.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.