Genomics of adaptation revealed in threespine stickleback

Natural selection is the ultimate, but not only force underlying organismal diversity. Despite this general biological insight, our understanding of how selection targets and shapes the genome during adaptation remains incomplete and is the central quest of this thesis. My main model organism is the threespine stickleback fish (Gasterosteus aculeatus). Stickleback provide an outstanding opportunity to study adaptive evolutionary change, because marine ancestors have repeatedly colonized and adapted to different freshwater environments all over the northern hemisphere since the last glacial retreat about 12,000 years ago. Besides wild populations, I also make use of lab-raised stickleback hybrids from controlled crosses for this thesis work.
Thousands of genome-wide genetic polymorphisms (i.e., genetic markers) called in marine, but predominantly in distinct lake and stream stickleback populations from different geographic locations allow me to decipher the number and position of genomic targets of selection in the early phase of adaptive divergence. I find that selection acts on many loci distributed widely across the genome. On a genomic scale, the recombination landscape along chromosomes proves to be - in concert with selection - an important factor in driving heterogeneous genetic differentiation among populations.
To investigate the rate of recombination across the stickleback's genome in more detail, I use an artificially crossed second-generation (F2) population. This reveals constraints in the frequency and location of detectable recombination events (i.e., cross-overs) within the genome. For example, cross-overs prove to be more frequent in chromosome peripheries than centers. This, together with selection, results in decreased within-population genetic diversity and increased between-population differentiation in the centers of chromosomes as opposed to the peripheries. Furthermore, I show that the cessation of recombination between the heterogametic sex chromosomes occurred in independent bouts. As a consequence, I find extended genomic regions distinct in their degree of degeneration between the X and Y chromosome, so called evolutionary strata. Finally, recombination reveals to be an important determinant of other aspects of a genome, such as its nucleotide composition.
Integrating theoretical modeling with targeted and genome-wide sequencing, my research further demonstrates that the inference and interpretation of genomic regions exhibiting particularly high and low population differentiation is not as straightforward as commonly believed. This is because the type of genetic variation available to selection (i.e., pre-existing vs. de novo variation) as well as the mode of adaptation (i.e., divergent vs. parallel adaptation) influence the way neutral variation is shaped by selection across the genome. I demonstrate that a genomic region of high differentiation may not necessarily be indicative of divergent selection when populations adapt in parallel to similar environments from a shared pool of genetic variation.
Based on several hundreds of F2 specimens reared under standardized conditions in the laboratory, I also link variation in heritable phenotypic traits to genetic variation, a research program generally referred to as quantitative trait locus (QTL) mapping. Corroborating with the results from my genome scans within and between wild populations (indicating that adaptive divergence involves many loci widespread across the genome), QTL mapping reveals that most phenotypic traits are controlled by numerous genetic loci. In general, each of these loci explains a small fraction of the entire phenotypic trait variation.
I also use high resolution SNP data to infer the demographic history of several lake and stream stickleback populations from the Lake Constance watershed (Central Europe) and demonstrate that the repeated occurrence of similar stream phenotypes are, in this particular system, better explained by an evolutionary scenario of 'ecological vicariance' rather than repeated parallel divergence. I then show how selection has shaped local and broad-scale linkage, diversity and differentiation across the genome in these populations. Interestingly, I find evidence for strong divergent selection acting on large chromosomal rearrangements I had previously detected to be important for marine vs. freshwater adaptation. This finding provides a strong case for the re-use of pre-existing genetic variation in stickleback and demonstrates that the same genomic regions can be involved in adaptive divergence between disparate ecotype pairs.
Overall, I come to conclude that signatures of selection are - at various physical scales - frequent within the stickleback genome. Stickleback repeatedly use pre-existing genetic variation, shared across various geographic ranges, to adapt to similar or disparate environments. Yet, there is a substantial degree of genetic non-parallelism - at least at the level of neutral markers - that goes along with parallel phenotypic evolution. My thesis emphasizes that the reliable detection and interpretation of genomic signatures of selection requires integrating many replicate study populations within a clear-cut ecological and demographic framework, as well as complementary analytical approaches. Controlled crossing experiments and theoretical modeling are key to deriving predictions about the genomics of adaptation in the wild and to facilitate our understanding of complex biological processes and patterns.