Deciphering the genomic signatures of speciation and introgression

As species diverge, so do their genomes. However, this genetic divergence, which can be measured directly by comparing sequences, has a large random component. Firstly, genetic diversity present within an ancestral species is passed on to its descendants at random and can survive for a long time. Secondly, there may be genetic exchange after the initial split, which further increases the variance in genetic divergence. Being able to model both processes matters in many ways: Firstly, researchers are often want to make inferences about the history of closely related species and ideally identify the factors that caused them to diverge from their genome sequence. For example, it would be interesting to know whether our own ancestors hybridized with Neanderthals in places where they met. Since speciation happens over evolutionary timescales, such indirect inferences from genomic data are often the only way to study the process. Secondly, there is great interest in identifying regions of the genome that may have been under selection both between and within species. In the case of recently diverged species this can tell us which genes were involved in divergence and provides clues about the mechanisms of speciation and the roles of genome structure and geographic separation. Finally, gene flow or introgression between nascent species may be adaptive itself. For example, resistance to insecticide can be acquired more rapidly from close relatives that have already evolved this trait than through independent evolution. To have any hope to identify introgression or selection involving particular genes, we need to know exactly what effects different scenarios of divergence between species have on patterns of neutral diversity in the genome. However, distinguishing between even the simplest histories requires large amounts of data and has proven notoriously difficult. With the rapid advances in sequencing technology, our ability to draw inferences about species divergence is now limited by theory rather than data. I will develop new statistical methods that allow us to estimate key parameters of the divergence between two species, in particular their splitting time and the amount and direction of gene flow between them directly and exactly from genomic data. These methods will be applied to genomic data from three island species of Drosophila: D sechellia, D. mauritiana and D. santomea which have split from their mainland ancestors relatively recently (250-500,000 years ago) and can be seen as independent replicates of speciation in very similar geographic and ecological settings. These species are unique because lab studies have revealed much about the genetic basis of a large range of traits involved in reproductive isolation, including genes that cause hybrids to be infertile and those underlying more complicated traits such as mating behaviour. However, their actual history of divergence is poorly known because current inference methods are limited to small datasets. In particular, we do not know how important gene flow during or after divergence has been. The new statistical method will allow me to compare the history of speciation in these three model systems in a number of ways. Firstly, I will ask whether species on islands closer to the mainland have experienced more genetic exchange with their mainland relatives than those further away. Secondly, I will test if divergence differs between different regions of the genome as predicted by current models of speciation and whether these differences are species-specific or universal. Finally, I will investigate whether genes known to be responsible for reproductive isolation can be distinguished against the genomic background by their different patterns of divergence alone. This would make it possible to identify such genes in other organisms where functional knowledge is not available through simple genomic scans.