heritability
Hansen TF, Pélabon C & Houle D 2011 Heritability is not evolvability. Evol Biol 38:258-277.
- the correlation between heritability and evolvability is essentially zero
- this is likely due to inherent positive correlations between the additive variance and other components of phenotypic variance
- scaling always involves non-trivial assumptions
- a lack of awareness of these assumptions constitutes a systemic error in the field of evolutionary biology
- there is an increasing understanding of how properties of the genotype-phenotype map such as pleiotropy and epistasis may influence the evolution of additive effects and thus evolvability
- the breeder's equation entails the assumption that the selection differential and the heritability are quasi-independent units in the sense that we expect the response to selection to be proportional to their values taken in isolation
- this assumption is problematic
- both the selection differential and the heritability depend on the environmental variance, which causes a negative correlation between them
- e = R / β
- eσ = VA / VP = h2
- eμ = VA / m2 = IA
- IA is equal to the square of the coefficient of additive genetic variance, CVA
- VAA = εe2 VA2 / 2
- εe2 is an average of squared pairwise epistasis coefficients
- the intuition behind variance scaling may rest on the assumption that environmental variance is independent of the genetic variances
- this assumption, embedded in theoretical population genetics, where environmental variance is routinely treated as an invariant parameter, and where predictions about additive variance and heritability are not distinguished, is, however, highly problematic
- there are many biological reasons why genetic and environmental variation should be correlated
- in theoretical population genetics the environmental variance is almost always modeled as a single invariant parameter
- predictions about additive genetic variance are often phrased as predictions about heritabilities
- the "Lande—Turelli" debate on the maintenance of additive genetic variance was usually phrased in terms of explaining observed heritabilities
- the assumption of a constant environmental variance unrelated to the additive variance is empirically unjustified
- additive genetic variances should increase with mutation rate and mutational target size, and decrease with strength of stabilizing selection
- none of these predictions are justified for heritabilities
- the typical strong directional selection observed in nature is perfectly capable of generating large "qualitative" changes in the traits on less than a geological time scale
- a lack of evolvability would rarely be a constraint on macroevolution
- this conclusion may, however, be premature
- the concept of heritability is commonly used in human genetics
- estimates from highly standardized sub-population samples are sometimes naively extrapolated to the population at large, or even to argue that differences between social, ethnic or temporal groups are genetically based
- such extrapolations have been criticized
- observing a high heritability does not imply a high level of genetic determination, or that the trait is insensitive to environmental factors
- it could simply mean that the environment in which the heritability was measured was relatively stable
