Search

Hybrid zones are regions where genetically differentiated forms come together and exchange genes through hybrid offspring. The study of characters gradually changing across such zones (clines) can give insight into evolutionary processes, providing exceptionally sensitive estimates of the intensity of selection, and allowing the detection of loci that might be involved in reproductive isolation and speciation. The Alpine grasshopper Podisma pedestris has a hybrid zone in Southern France where two populations meet. They differ in their sex chromosome system, and strong selection against hybrids is observed. These distinct populations likely have split and re-joined several times during the Quaternary glacial cycles. A model explaining the selection observed against hybrids postulates hundreds of loci of small effect spread over two differentiated genomes meeting in secondary contact. Yet, over 50 years of study to-date non have been discovered. However, so far the study of P. pedestris has not made use of high-throughput sequencing data which provides an unprecedented resolution of molecular markers. I am aiming to close the gap with this thesis. I assemble the grasshopper's mitochondrial genome sequence and infer what proportion of its genome is made up by mitochondrial inserts (Numts). Using transcriptome data from two individuals, I then go on to fit demographic models, finding the populations split approximately 400 000 years ago and that the current-day population sizes are considerably smaller than the ancestral one. The final data chapter explores the genetic architecture of the hybrid zone using data from a targeted sequence capture of hundreds of loci covering some 10 000 polymorphic sites. Only two loci under selection are identified, which is surprising given the power of the analysis. Both loci are located on the X chromosome and are subject to weak selection (0.3% and 0.03%). This shows the power of hybrid zone analysis to infer targets of selection. The results are discussed in light of a theoretical chapter on the 'inexorable spread' phenomenon and lead to the proposal for further research into the causes of the reproductive isolation observed between the grasshopper populations.

Comparative genomic studies have identified noncoding regions of the genome which are often more highly conserved between species than protein-coding sequences. One possible explanation for this conservation of non-coding sequences is some form of selective constraint since sequence conservation at great evolutionary depths is a preliminary indication of functional constraint. Here, I consider nearly 2500 putative regulatory elements, termed Conserved Noncoding Elements (CNEs), that are conserved across seven vertebrate species (human, macaque, mouse, chicken, frog, zebrafish and fugu). I distinguish between CNEs that show accelerated rates of evolution and those that have remained more constrained throughout evolution, and identify CNEs that show higher than expected substitution rates in the human lineage that may be potential candidates of adaptive evolution. However, it is not trivial to demonstrate the action of selection on such sequences. It is relatively easier in the case of protein-coding DNA, since selection would be predicted to result in different rates of substitution for synonymous and non-synonymous sites. Hence, I use the same seven species to define phylogenetically invariant positions in CNEs in contrast to those that have at least one substitution and analyse them independently to determine if there is a positive correlation between evolutionary conservation and the strength of purifying selection at individual sites. In the 1000 Genomes, but not the HapMap, data I find a significant excess of rare derived alleles in CNEs relative to coding sequences. This excess of rare alleles can be best explained if selection is relatively consistent across sites, with most mutations resulting in a similar reduction in fitness. Finally, I explore patterns of variation in the allele-frequencies within human populations, however do not detect any significant differences in the underlying distribution of negatively selected variants among human populations.