Sex differences in traits such as body size, morphology, behaviour, and life history – a pattern called sexual dimorphism – is a feature present in all dioecious species. The evolution of sexual dimorphism is the central theme of my work, more specifically, how sex differences can evolve when there is a shared genome. Most or all of the genome is inherited by offspring of both sexes in the vast majority of dioeceous species – humans only have sex-limited transmission of the Y-chromosome and mitochondria, all other genetic material can go from parent to offspring of either sex. Sexual dimorphism is most commonly driven to evolve by sexually antagonistic selection, selection for different phenotypes in either sex.
The only part of the Drosohpila nuclear genome not present in females is the Y-chromosome (green).
Phenotypic traits are generally determined by a mixture of genetic, environmental, and gene x environment interactions. If we consider height in humans as an example, we can hypothetically state that both male height and female height are each determined by 100 equally influential loci in the genome, and that sexually antagonistic selection is occurring (for large males, small females). If these genes are sex-limited (the 100 loci determining male height do not affect female height and vice versa) then sexually antagonistic selection can easily spread mutations which make males larger and females shorter. Given enough time, mutations, and stable selection, then populations will evolve to reach their sex-specific phenotypic optima. A population where the genes affecting height are entirely sex-limited then the correlation of height between the sexes will be zero, the height of a female has no relation to the height of her father, and the height of a male has no relation to the height of his mother.
A correlation of ~0 means the sexes can evolve independently along any trajectory within the trait space.
Consider now what happens if the 100 loci affecting male height are also the same 100 loci affecting female height, but sexually antagonistic selection remains. The correlation between the sexes, intersexual genetic correlation (rmf), would be one (or minus one if there is an inverse relationship). Assuming an rmf of one, then tall mothers will have tall sons and daughters, so too will tall fathers, and short mothers will have short sons and daughters, again so too will short fathers. With sexually antagonistic selection acting on these shared loci a mutation which makes males larger will also make females larger, this moves males towards their optimal phenotype but also moves the females away from theirs – therefore the mutation will not spread through the population by selection (though it could still drift). The intralocus conflict that occurs causes a fitness depression in the population so sex-specific genetic machinery must evolve to resolve the conflict and allow the optimization of the sexes.
A correlation of 1 means male and female trait values are tightly connected and evolution can only occur along one trajectory.
According to Fisher’s fundamental theorem, the amount of variance in a trait determines the rate at which it can respond to selection. More genetic variance in a trait means a greater rate of evolution can be realized.This is related to the breeders equation where the response = selection x heritability (additive genetic variance). Simply looking at this you can see that variance will determine the rate of response, if heritability is zero then the response is zero, selection will have no effect. The same applies for genetic covariance between traits, such as male and female height. If the genome is entirely shared then there is a high covariance, selection affecting males will produce a large response in males. If a genome is sex-limited then there is no covariance so selection will produce no response in the other sex.
The conclusion of all this theory is that sexual dimorphism should be related to the intersexual genetic correlation, with sexual dimorphism only evolving when the intersexual correlation is not one or minus one, and being more extreme at zero rmf. This has been previously shown using 8 traits in waltzing flies (Bonduriansky & Rowe 2005), a meta-analysis by Poissant et al 2010, using multivariate breeders equations by both Lewis et al 2011 and Gosden et al 2012. My own paper in 2013 used gene expression data from multiple species of Drosophila and thousands of genes to demonstrate the pervasive and long-term constraint of a shared genome on the evolution of sexual dimorphism. A previous post of mine discusses in more detail the G-matrix, developed by Lande as a component of the multivariate breeders equation.