We’ve just had a paper published in Methods in Ecology and Evolution on “Simultaneously estimating evolutionary history and repeated traits phylogenetic signal”. This work, lead by Bram Vrancken, Philippe Lemey and Marc Suchard, estimates phylogenetic signal in a fully Bayesian fashion, integrating over uncertainty in evolutionary history. This method provides an estimate of the extent to which variance of a continuous trait (like viral load) can be explained by evolutionary relatedness and thus gives an estimate of the trait’s heritability.

Of particular interest to me, is that this method gives a heritability estimate of influenza antigenic phenotype of ~0.73, which it quite high and supports a strong genetic basis to antigenic phenotype. Here, total variance in a phenotype trait stems from a combination of variance due to genotype and variance due to the environment $V_P = V_G + V_E$, so that heritability is measured as the proportion of the total variance explained by genotype $H^2 = V_G / V_P$. In the case of influenza antigenic phenotype we expect noise in the hemagglutination inhibition (HI) assay to contribute to total variance, so H^2 = ~0.73 should represent a lower bound to actual heritability.