Unmeasured prior viability selection resolves the paradox of stasis for body size in wild Soay sheep

The estimation of natural selection is used to understand ecological and evolutionary processes in wild populations and is often used to predict change. However, the direct application of quantitative genetic methods, originally developed in animal breeding, has been less successful in the wild; in particular, predictions of evolutionary change are often made that are not observed. This misprediction, known as the "paradox of stasis," can arise due to bias in estimates of selection via nonrandom missing data in phenotypes if viability selection has previously occurred on correlated traits. Here, we check for this bias in a wild population of Soay sheep where estimates of selection suggest that "bigger is better" for adult size, but evolutionary change of the predicted magnitude does not occur. We establish that standard procedures for estimating total lifetime selection are biased by prior viability selection. In particular, while phenotypically large lambs have high first year survival, we also show that lambs that are genetically predisposed to large adult size traits also suffer elevated juvenile mortality. While the phenotypic traits driving early-life selection against large adult body size are unknown, our genetic analysis reveals correlated selection against larger adult sizes that essentially resolves the paradox of stasis for adult body size traits in this wild population. The pattern we reveal is potentially widespread in nature; previous results showing predominantly positive selection of large body size could in general be explained by this kind of antagonistic prior viability selection.