Comparisons between QST and FST—how wrong have we been?

The comparison between quantitative genetic divergence (Q(ST) ) and neutral genetic divergence (F(ST) ) among populations has become the standard test for historical signatures of selection on quantitative traits. However, when the mutation rate of neutral markers is relatively high in comparison with gene flow, estimates of F(ST) will decrease, resulting in upwardly biased comparisons of Q(ST) vs. F(ST) . Reviewing empirical studies, the difference between Q(ST) and F(ST) is positively related to marker heterozygosity. After refuting alternative explanations for this pattern, we conclude that marker mutation rate indeed has had a biasing effect on published Q(ST) -F(ST) comparisons. Hence, it is no longer clear that populations have commonly diverged in response to divergent selection. We present and discuss potential solutions to this bias. Comparing Q(ST) with recent indices of neutral divergence that statistically correct for marker heterozygosity (Hedrick's G'st and Jost's D) is not advised, because these indices are not theoretically equivalent to Q(ST) . One valid solution is to estimate F(ST) from neutral markers with mutation rates comparable to those of the loci underlying quantitative traits (e.g. SNPs). Q(ST) can also be compared to Φ(ST) (Phi(ST) ) of amova, as long as the genetic distance among allelic variants used to estimate Φ(ST) reflects evolutionary history: in that case, neutral divergence is independent of mutation rate. In contrast to their common usage in comparisons of Q(ST) and F(ST) , microsatellites typically have high mutation rates and do not evolve according to a simple evolutionary model, so are best avoided in Q(ST) -F(ST) comparisons.     

If FST does not measure neutral genetic differentiation,then comparing it with QST is misleading. Or is it?

The comparison between neutral genetic differentiation (F(ST) ) and quantitative genetic differentiation (Q(ST) ) is commonly used to test for signatures of selection in population divergence. However, there is an ongoing discussion about what F(ST) actually measures, even resulting in some alternative metrics to express neutral genetic differentiation. If there is a problem with F(ST) , this could have repercussions for its comparison with Q(ST) as well. We show that as the mutation rate of the neutral marker increases, F(ST) decreases: a higher within-population heterozygosity (He) yields a lower F(ST) value. However, the same is true for Q(ST) : a higher mutation rate for the underlying QTL also results in a lower Q(ST) estimate. The effect of mutation rate is equivalent in Q(ST) and F(ST) . Hence, the comparison between Q(ST) and F(ST) remains valid, if one uses neutral markers whose mutation rates are not too high compared to those of quantitative traits. Usage of highly variable neutral markers such as hypervariable microsatellites can lead to serious biases and the incorrect inference that divergent selection has acted on populations. Much of the discussion on F(ST) seems to stem from the misunderstanding that it measures the differentiation of populations, whereas it actually measures the fixation of alleles. In their capacity as measures of population differentiation, Hedrick's G'(ST) and Jost's D reach their maximum value of 1 when populations do not share alleles even when there remains variation within populations, which invalidates them for comparisons with Q(ST) .     

Does habitat fragmentation reduce fitness and adaptability? A case study of the common frog (Rana temporaria)

Studies examining the effects of anthropogenic habitat fragmentation on both neutral and adaptive genetic variability are still scarce. We compared tadpole fitness-related traits (viz. survival probability and body size) among populations of the common frog (Rana temporaria) from fragmented (F) and continuous (C) habitats that differed significantly in population sizes (C > F) and genetic diversity (C > F) in neutral genetic markers. Using data from common garden experiments, we found a significant positive relationship between the mean values of the fitness related traits and the amount of microsatellite variation in a given population. While genetic differentiation in neutral marker loci (F(ST)) tended to be more pronounced in the fragmented than in the continuous habitat, genetic differentiation in quantitative traits (Q(ST)) exceeded that in neutral marker traits in the continuous habitat (i.e. Q(ST) > F(ST)), but not in the fragmented habitat (i.e. Q(ST) approximately F(ST)). These results suggest that the impact of random genetic drift relative to natural selection was higher in the fragmented landscape where populations were small, and had lower genetic diversity and fitness as compared to populations in the more continuous landscape. The findings highlight the potential importance of habitat fragmentation in impairing future adaptive potential of natural populations.     

Molecular and quantitative genetic divergence among populations of house mice with known evolutionary histories

Evolutionary biologists have long been interested in the processes influencing population differentiation, but separating the effects of neutral and adaptive evolution has been an obstacle for studies of population subdivision. A recently developed method allows tests of whether disruptive (ie, spatially variable) or stabilizing (ie, spatially uniform) selection is influencing phenotypic differentiation among subpopulations. This method, referred to as the F(ST) vs Q(ST) comparison, separates the total additive genetic variance into within- and among-population components and evaluates this level of differentiation against a neutral hypothesis. Thus, levels of neutral molecular (F(ST)) and quantitative genetic (Q(ST)) divergence are compared to evaluate the effects of selection and genetic drift on phenotypic differentiation. Although the utility of such comparisons appears great, its accuracy has not yet been evaluated in populations with known evolutionary histories. In this study, F(ST) vs Q(ST) comparisons were evaluated using laboratory populations of house mice with known evolutionary histories. In this model system, the F(ST) vs Q(ST) comparisons between the selection groups should reveal quantitative trait differentiation consistent with disruptive selection, while the F(ST) vs Q(ST) comparisons among lines within the selection groups should suggest quantitative trait differentiation in agreement with drift. We find that F(ST) vs Q(ST) comparisons generally produce the correct evolutionary inference at each level in the population hierarchy. Additionally, we demonstrate that when strong selection is applied between populations Q(ST) increases relative to Q(ST) among populations diverging by drift. Finally, we show that the statistical properties of Q(ST), a variance component ratio, need further investigation.     

Comparative studies of quantitative trait and neutral marker divergence: a meta-analysis

Comparative studies of quantitative genetic and neutral marker differentiation have provided means for assessing the relative roles of natural selection and random genetic drift in explaining among-population divergence. This information can be useful for our fundamental understanding of population differentiation, as well as for identifying management units in conservation biology. Here, we provide comprehensive review and meta-analysis of the empirical studies that have compared quantitative genetic (Q(ST)) and neutral marker (F(ST)) differentiation among natural populations. Our analyses confirm the conclusion from previous reviews - based on ca. 100% more data - that the Q(ST) values are on average higher than F(ST) values [mean difference 0.12 (SD 0.27)] suggesting a predominant role for natural selection as a cause of differentiation in quantitative traits. However, although the influence of trait (life history, morphological and behavioural) and marker type (e.g. microsatellites and allozymes) on the variance of the difference between Q(ST) and F(ST) is small, there is much heterogeneity in the data attributable to variation between specific studies and traits. The latter is understandable as there is no reason to expect that natural selection would be acting in similar fashion on all populations and traits (except for fitness itself). We also found evidence to suggest that Q(ST) and F(ST) values across studies are positively correlated, but the significance of this finding remains unclear. We discuss these results in the context of utility of the Q(ST)-F(ST) comparisons as a tool for inferring natural selection, as well as associated methodological and interpretational problems involved with individual and meta-analytic studies.     

Q(ST) in a hierarchically structured population

Under some circumstances, selection and drift can be disentangled with comparisons of the Q(ST) of a quantitative trait and the F(ST) of putatively neutral loci. Most previous comparisons of F(ST) and Q(ST) are carried out at a single spatial scale. We derive a hierarchical Q(ST) appropriate for study across varying levels of spatial structure.     

Divergent selection as revealed by P(ST) and QTL-based F(ST) in three-spined stickleback (Gasterosteus aculeatus) populations along a coastal-inland gradient

Three measures of divergence, estimated at nine putatively neutral microsatellite markers, 14 quantitative traits, and seven quantitative trait loci (QTL) were compared in eight populations of the three-spined stickleback (Gasterosteus aculeatus L.) living in the Scheldt river basin (Belgium). Lowland estuarine and polder populations were polymorphic for the number of lateral plates, whereas upland freshwater populations were low-plated. The number of short gill rakers and the length of dorsal and pelvic spines gradually declined along a coastal-inland gradient. Plate number, short gill rakers and spine length showed moderate to strong signals of divergent selection between lowland and upland populations in comparison between P(ST) (a phenotypic alternative for Q(ST)) and neutral F(ST). However, such comparisons rely on the unrealistic assumption that phenotypic variance equals additive genetic variance, and that nonadditive genetic effects and environmental effects can be minimized. In order to verify this assumption and to confirm the phenotypic signals of divergence, we tested for divergent selection at the underlying QTL. For plate number, strong genetic evidence for divergent selection between lowland and upland populations was obtained based on an intron marker of the Eda gene, of which the genotype was highly congruent with plate morph. Genetic evidence for divergent selection on short gill rakers was limited to some population pairs where F(ST) at only one of two QTL was detected as an outlier, although F(ST) at both loci correlated significantly with P(ST). No genetic confirmation was obtained for divergent selection on dorsal spine length, as no outlier F(ST)s were detected at dorsal spine QTL, and no significant correlations with P(ST) were observed.     

A genomic scan for divergent selection in a secondary contact zone between Atlantic and Gulf of Mexico oysters, Crassostrea virginica

The degree of population structure within species often varies considerably among loci. This makes it difficult to determine whether observed variance reflects neutral-drift stochasticity or locus-specific selection at one or more loci. This uncertainty is exacerbated when evolutionary equilibrium cannot be assumed and/or admixture potentially inflates genomic variance. Thus, the value of a 'genome scan', where locus-specific summary statistics are compared with a simulated neutral distribution among loci, may be limited in secondary contact zones if the null distribution is sensitive to the timing of secondary contact. Of particular interest here is the wide variance previously observed in locus-specific patterns of population structure between Atlantic and Gulf of Mexico populations of eastern oyster, Crassostrea virginica. To test the robustness of an equilibrium null model, we compared among-locus distributions of F(ST) simulated under migration-drift equilibrium and several nonequilibrium secondary contact histories. We then tested for evidence of divergent selection between two oyster populations on either side of a secondary contact zone using 215 amplified fragment length polymorphism (AFLP) loci. Constant-migration equilibrium and nonequilibrium secondary-contact simulations produced equivalent distributions of F(ST) when anchored by the global mean F(ST) observed in oysters, 0.0917. The 99th quantile of simulated neutral F(ST) encompassed most of the variation among oyster loci. Three AFLP loci exhibited F(ST) values higher than this threshold. Although no locus was significant after correcting for multiple tests, our results show in geographically clinal organisms: AFLPs can efficiently characterize the genomic distribution of F(ST); equilibrium models can be used to evaluate outliers; these procedures help focus research on genomic regions of interest.     

The effect of neutral nonadditive gene action on the quantitative index of population divergence

For neutral additive genes, the quantitative index of population divergence (Q(ST)) is equivalent to Wright's fixation index (F(ST)). Thus, divergent or convergent selection is usually invoked, respectively, as a cause of the observed increase (Q(ST) > F(ST)) or decrease (Q(ST) < F(ST)) of Q(ST) from its neutral expectation (Q(ST) = F(ST)). However, neutral nonadditive gene action can mimic the additive expectations under selection. We have studied theoretically the effect of consecutive population bottlenecks on the difference F(ST) - Q(ST) for two neutral biallelic epistatic loci, covering all types of marginal gene action. With simple dominance, Q(ST) < F(ST) for only low to moderate frequencies of the recessive alleles; otherwise, Q(ST) > F(ST). Additional epistasis extends the condition Q(ST) < F(ST) to a broader range of frequencies. Irrespective of the type of nonadditive action, Q(ST) < F(ST) generally implies an increase of both the within-line additive variance after bottlenecks over its ancestral value (V(A)) and the between-line variance over its additive expectation (2F(ST)V(A)). Thus, both the redistribution of the genetic variance after bottlenecks and the F(ST) - Q(ST) value are governed largely by the marginal properties of single loci. The results indicate that the use of the F(ST) - Q(ST) criterion to investigate the relative importance of drift and selection in population differentiation should be restricted to pure additive traits.     

Patterns of differentiation in a colour polymorphism and in neutral markers reveal rapid genetic changes in natural damselfly populations

The existence and mode of selection operating on heritable adaptive traits can be inferred by comparing population differentiation in neutral genetic variation between populations (often using F(ST) values) with the corresponding estimates for adaptive traits. Such comparisons indicate if selection acts in a diversifying way between populations, in which case differentiation in selected traits is expected to exceed differentiation in neutral markers [F(ST )(selected) > F(ST )(neutral)], or if negative frequency-dependent selection maintains genetic polymorphisms and pulls populations towards a common stable equilibrium [F(ST) (selected) < F(ST) (neutral)]. Here, we compared F(ST) values for putatively neutral data (obtained using amplified fragment length polymorphism) with estimates of differentiation in morph frequencies in the colour-polymorphic damselfly Ischnura elegans. We found that in the first year (2000), population differentiation in morph frequencies was significantly greater than differentiation in neutral loci, while in 2002 (only 2 years and 2 generations later), population differentiation in morph frequencies had decreased to a level significantly lower than differentiation in neutral loci. Genetic drift as an explanation for population differentiation in morph frequencies could thus be rejected in both years. These results indicate that the type and/or strength of selection on morph frequencies in this system can change substantially between years. We suggest that an approach to a common equilibrium morph frequency across all populations, driven by negative frequency-dependent selection, is the cause of these temporal changes. We conclude that inferences about selection obtained by comparing F(ST) values from neutral and adaptive genetic variation are most useful when spatial and temporal data are available from several populations and time points and when such information is combined with other ecological sources of data.     

Natural selection drives clinal life history patterns in the perennial sunflower species, Helianthus maximiliani

In plants, ecologically important life history traits often display clinal patterns of population divergence. Such patterns can provide strong evidence for spatially varying selection across environmental gradients but also may result from nonselective processes, such as genetic drift, population bottlenecks and spatially restricted gene flow. Comparison of population differentiation in quantitative traits (measured as Q(ST) ) with neutral molecular markers (measured as F(ST) ) provides a useful tool for understanding the relative importance of adaptive and nonadaptive processes in the formation and maintenance of clinal variation. Here, we demonstrate the existence of geographic variation in key life history traits in the diploid perennial sunflower species Helianthus maximiliani across a broad latitudinal transect in North America. Strong population differentiation was found for days to flowering, growth rate and multiple size-related traits. Differentiation in these traits greatly exceeds neutral predictions, as determined both by partial Mantel tests and by comparisons of global Q(ST) values with theoretical F(ST) distributions. These findings indicate that clinal variation in these life history traits likely results from local adaptation driven by spatially heterogeneous environments.     

The effects of dominance, regular inbreeding and sampling design on Q(ST), an estimator of population differentiation for quantitative traits

To test whether quantitative traits are under directional or homogenizing selection, it is common practice to compare population differentiation estimates at molecular markers (F(ST)) and quantitative traits (Q(ST)). If the trait is neutral and its determinism is additive, then theory predicts that Q(ST) = F(ST), while Q(ST) > F(ST) is predicted under directional selection for different local optima, and Q(ST) < F(ST) is predicted under homogenizing selection. However, nonadditive effects can alter these predictions. Here, we investigate the influence of dominance on the relation between Q(ST) and F(ST) for neutral traits. Using analytical results and computer simulations, we show that dominance generally deflates Q(ST) relative to F(ST). Under inbreeding, the effect of dominance vanishes, and we show that for selfing species, a better estimate of Q(ST) is obtained from selfed families than from half-sib families. We also compare several sampling designs and find that it is always best to sample many populations (>20) with few families (five) rather than few populations with many families. Provided that estimates of Q(ST) are derived from individuals originating from many populations, we conclude that the pattern Q(ST) > F(ST), and hence the inference of directional selection for different local optima, is robust to the effect of nonadditive gene actions.     

Comparing Bayesian estimates of genetic differentiation of molecular markers and quantitative traits: an application to Pinus sylvestris

Comparison of the level of differentiation at neutral molecular markers (estimated as F(ST) or G(ST)) with the level of differentiation at quantitative traits (estimated as Q(ST)) has become a standard tool for inferring that there is differential selection between populations. We estimated Q(ST) of timing of bud set from a latitudinal cline of Pinus sylvestris with a Bayesian hierarchical variance component method utilizing the information on the pre-estimated population structure from neutral molecular markers. Unfortunately, the between-family variances differed substantially between populations that resulted in a bimodal posterior of Q(ST) that could not be compared in any sensible way with the unimodal posterior of the microsatellite F(ST). In order to avoid publishing studies with flawed Q(ST) estimates, we recommend that future studies should present heritability estimates for each trait and population. Moreover, to detect variance heterogeneity in frequentist methods (ANOVA and REML), it is of essential importance to check also that the residuals are normally distributed and do not follow any systematically deviating trends.     

Genetic basis of adaptation in Arabidopsis thaliana: local adaptation at the seed dormancy QTL DOG1

Local adaptation provides an opportunity to study the genetic basis of adaptation and investigate the allelic architecture of adaptive genes. We study delay of germination 1 (DOG1), a gene controlling natural variation in seed dormancy in Arabidopsis thaliana and investigate evolution of dormancy in 41 populations distributed in four regions separated by natural barriers. Using F(ST) and Q(ST) comparisons, we compare variation at DOG1 with neutral markers and quantitative variation in seed dormancy. Patterns of genetic differentiation among populations suggest that the gene DOG1 contributes to local adaptation. Although Q(ST) for seed dormancy is not different from F(ST) for neutral markers, a correlation with variation in summer precipitation supports that seed dormancy is adaptive. We characterize dormancy variation in several F(2) -populations and show that a series of functionally distinct alleles segregate at the DOG1 locus. Theoretical models have shown that the number and effect of alleles segregatin at quantitative trait loci (QTL) have important consequences for adaptation. Our results provide support to models postulating a large number of alleles at quantitative trait loci involved in adaptation.     

Contrasting evolutionary forces driving population structure at expressed sequence tag polymorphisms, allozymes and quantitative traits in white spruce

Patterns of variation in quantitative characters and genetic markers were compared among six regional populations of white spruce [Picea glauca (Moench) Voss]. Although some phenotypic characters were correlated with latitude (r = 0.791), longitude (r = -0.796) and precipitation during the growing season (r = 0.789), variability at genetic markers was not correlated with geographical or bioclimatic variables, and followed neutral expectations. Estimates of genetic diversity and population differentiation for 14 allozymes (translated regions of coding genes) were essentially indistinguishable from those observed for 11 expressed sequence tag polymorphisms (ESTPs) from untranslated regions of coding genes. Variation among populations for quantitative traits such as eighth year height (Q(ST) = 0.082), thirteenth year height (Q(ST) = 0.069), total wood density (Q(ST) = 0.102) and date of budset (Q(ST) = 0.246), was greater than for allozymes (G(ST) = 0.014) and ESTPs (G(ST) = 0.019). These trends suggest a strong adaptive response in quantitative traits, contrasting to allozymes and ESTPs where no selective response could be detected and where populations appeared to be essentially in a migration-drift equilibrium.     

A comparison of variability and population structure for major histocompatibility complex and microsatellite loci in California coastal steelhead (Oncorhynchus mykiss Walbaum)

The major histocompatibility complex (MHC) contains genes integral to immune response in vertebrates. MHC genes have been shown to be under selection in a number of vertebrate taxa, making them intriguing for population genetic studies. We have conducted a survey of genetic variation in an MHC class II gene for steelhead trout from 24 sites in coastal California and compared this variation to that observed at 16 presumably neutral microsatellite loci. A high amount of allelic variation was observed at the MHC when compared to previously published studies on other Pacific salmonids. Elevated nonsynonymous substitutions, relative to synonymous substitutions, were detected at the MHC gene, indicating the signature of historical balancing selection. The MHC data were tested for correlations to and deviations from the patterns found with the microsatellite data. Estimates of allelic richness for the MHC gene and for the microsatellites were positively correlated, as were estimates of population differentiation (F(ST)). An analysis for F(ST) outliers indicates that the MHC locus has an elevated F(ST) relative to the neutral expectation, although a significant result was found for only one particular geographical subgroup. Relatively uniform allele frequency distributions were detected in four populations, although this finding may be partially due to recent population bottlenecks. These results indicate that, at the scale studied here, drift and migration play a major role in the observed geographical variability of MHC genes in steelhead, and that contemporary selection is relatively weak and difficult to detect.     

Adaptive population differentiation in phenology across a latitudinal gradient in European aspen (Populus tremula, L.): a comparison of neutral markers, candidate genes and phenotypic traits

A correct timing of growth cessation and dormancy induction represents a critical ecological and evolutionary trade-off between survival and growth in most forest trees (Rehfeldt et al. 1999; Horvath et al. 2003; Howe et al. 2003). We have studied the deciduous tree European Aspen (Populus tremula) across a latitudinal gradient and compared genetic differentiation in phenology traits with molecular markers. Trees from 12 different areas covering 10 latitudinal degrees were cloned and planted in two common gardens. Several phenology traits showed strong genetic differentiation and clinal variation across the latitudinal gradient, with Q(ST) values generally exceeding 0.5. This is in stark contrast to genetic differentiation at several classes of genetic markers (18 neutral SSRs, 7 SSRs located close to phenology candidate genes and 50 SNPs from five phenology candidate genes) that all showed F(ST) values around 0.015. We thus find strong evidence for adaptive divergence in phenology traits across the latitudinal gradient. However, the strong population structure seen at the quantitative traits is not reflected in underlying candidate genes. This result fit theoretical expectations that suggest that genetic differentiation at candidate loci is better described by F(ST) at neutral loci rather than by Q(ST) at the quantitative traits themselves.     

Q(St) meets the G matrix: the dimensionality of adaptive divergence in multiple correlated quantitative traits.

The Q(ST)-F(ST) comparison has become an increasingly common method for inferring adaptive quantitative trait divergence among populations. For cases in which there is divergence in multiple traits, most studies have applied the method by performing multiple univariate Q(ST)-F(ST) comparisons. However, because traits are often genetically correlated, such univariate analyses are likely to paint a simplified picture of adaptive divergence. Here we show how the multivariate analogue of Q(ST), F(STq), which accounts for genetic correlations among traits, can be used to supply a more detailed picture of multitrait divergence. We apply the method to naturally occurring genetic variation for a suite of sexually selected display traits in Drosophila serrata. The analyses suggest the operation of divergent multivariate selection that has influenced multiple independent axes of genetic variance in a sex-specific manner. Finally, we show how a comparison of the components of F(STq), the average within and among population genetic variance-covariance matrices, G(W) and G(B), can be used as an additional test of the null expectation of neutral divergence, and allows for an investigation of whether natural populations have diverged along major or minor axes of genetic variance.     

Genome scanning for interspecific differentiation between two closely related oak species [Quercus robur L. and Q. petraea (Matt.) Liebl.

Interspecific differentiation values (G(ST)) between two closely related oak species (Quercus petraea and Q. robur) were compiled across different studies with the aim to explore the distribution of differentiation at the genome level. The study was based on a total set of 389 markers (isozymes, AFLPs, SCARs, microsatellites, and SNPs) for which allelic frequencies were estimated in pairs of populations sampled throughout the sympatric distribution of the two species. The overall distribution of G(ST) values followed an L-shaped curve with most markers exhibiting low species differentiation (G(ST) < 0.01) and only a few loci reaching >10% levels. Twelve percent of the loci exhibited significant G(ST) deviations to neutral expectations, suggesting that selection contributed to species divergence. Coding regions expressed higher differentiation than noncoding regions. Among the 389 markers, 158 could be mapped on the 12 linkage groups of the existing Q. robur genetic map. Outlier loci with large G(ST) values were distributed over 9 linkage groups. One cluster of three outlier loci was found within 0.51 cM; but significant autocorrelation of G(ST) was observed at distances <2 cM. The size and distribution of genomic regions involved in species divergence are discussed in reference to hitchhiking effects and disruptive selection.     

Maternal genetic effects on adaptive divergence between anadromous and resident brook charr during early life history

The importance of directional selection relative to neutral evolution may be determined by comparing quantitative genetic variation in phenotype (Q(ST)) to variation at neutral molecular markers (F(ST)). Quantitative divergence between salmonid life history types is often considerable, but ontogenetic changes in the significance of major sources of genetic variance during post-hatch development suggest that selective differentiation varies by developmental stage. In this study, we tested the hypothesis that maternal genetic differentiation between anadromous and resident brook charr (Salvelinus fontinalis Mitchill) populations for early quantitative traits (embryonic size/growth, survival, egg number and developmental time) would be greater than neutral genetic differentiation, but that the maternal genetic basis for differentiation would be higher for pre-resorption traits than post-resorption traits. Quantitative genetic divergence between anadromous (seawater migratory) and resident Laval River (Québec) brook charr based on maternal genetic variance was high (Q(ST) > 0.4) for embryonic length, yolk sac volume, embryonic growth rate and time to first response to feeding relative to neutral genetic differentiation [F(ST) = 0.153 (0.071-0.214)], with anadromous females having positive genetic coefficients for all of the above characters. However, Q(ST) was essentially zero for all traits post-resorption of the yolk sac. Our results indicate that the observed divergence between resident and anadromous brook charr has been driven by directional selection, and may therefore be adaptive. Moreover, they provide among the first evidence that the relative importance of selective differentiation may be highly context-specific, and varies by genetic contributions to phenotype by parental sex at specific points in offspring ontogeny. This in turn suggests that interpretations of Q(ST)-F(ST) comparisons may be improved by considering the structure of quantitative genetic architecture by age category and the sex of the parent used in estimation.