Family-based Selection: An Efficient Method for Increasing Phenotypic Variability

Abstract

Persistent idiosyncrasies in behavioral phenotypes have been documented across animal taxa. These individual differences among organisms from the same genotype and reared in identical environments can result in phenotypic variability in the absence of genetic variation. While there is strong evidence to suggest that variability of traits can be heritable and determined by the genotype of an organism, little is known about how selection can specifically shape this heritable variance. Here, we describe a Python-based model of directional artificial selection for increasing the variability of a polygenic trait of interest. Specifically, our model focuses on variability in left-vs-right turn bias in D. melanogaster. While the mean value of turn bias for a genotype is non-heritable and constant across genotypes, the variability of turn bias is a heritable and polygenic trait, varying dramatically among different genetic lines. Using our model, we compare different selection regimes and predict selection dynamics at population and genetic levels. We find that introducing population structure via a family-based selection regime can significantly affect selection response. When selection for increased variability is implemented on the basis of independently measured traits of individuals, the response is slower, but leads to a population with a greater genetic diversity. In contrast, when selection is implemented by measuring traits of families with half or full siblings, the response is faster, albeit with a reduced final genetic diversity in the population. Our model provides a useful starting point to study the effect of different selection regimes on any polygenic trait of interest. We can use this model to predict responses of laboratory-based selection experiments and implement feasible experiments for selection of complex polygenic traits in the lab.