Cast away in the Adriatic: Low degree of parallel genetic differentiation in three-spined sticklebacks

The interplay between epigenomic and transcriptomic variation during ecotype divergence in stickleback

BACKGROUND

Populations colonizing contrasting environments are likely to undergo adaptive divergence and evolve ecotypes with locally adapted phenotypes. While diverse molecular mechanisms underlying ecotype divergence have been identified, less is known about their interplay and degree of divergence.

RESULTS

Here we integrated epigenomic and transcriptomic data to explore the interactions among gene expression, alternative splicing, DNA methylation, and microRNA expression to gauge the extent to which patterns of divergence at the four molecular levels are aligned in a case of postglacial divergence between marine and freshwater ecotypes of nine-spined sticklebacks (Pungitius pungitius). Despite significant genome-wide associations between epigenomic and transcriptomic variation, we found largely non-parallel patterns of ecotype divergence across epigenomic and transcriptomic levels, with predominantly nonoverlapping (ranging from 43.40 to 87.98%) sets of differentially expressed, spliced and methylated genes, and candidate genes targeted by differentially expressed miRNA between the ecotypes. Furthermore, we found significant variation in the extent of ecotype divergence across different molecular mechanisms, with differential methylation and differential splicing showing the highest and lowest extent of divergence between ecotypes, respectively. Finally, we found a significant enrichment of genes associated with ecotype divergence in differential methylation.

CONCLUSIONS

Our results suggest a nuanced relationship between epigenomic and transcriptomic processes, with alignment at the genome-wide level masking relatively independent effects of different molecular mechanisms on ecotype divergence at the gene level.

Alternative splicing in parallel evolution and the evolutionary potential in sticklebacks

Repeatability of adaptation to similar environments provides opportunity to evaluate the predictability of natural selection. While many studies have investigated gene expression differences between populations adapted to contrasting environments, the role of post-transcriptional processes such as alternative splicing has rarely been evaluated in the context of parallel adaptation. To address the aforementioned knowledge gap, we reanalysed transcriptomic data from three pairs of threespine stickleback (Gasterosteus aculeatus) ecotypes adapted to marine or freshwater environment. First, we identified genes with repeated expression or splicing divergence across ecotype pairs, and compared the genetic architecture and biological processes between parallelly expressed and parallelly spliced loci. Second, we analysed the extent to which parallel adaptation was reflected at gene expression and alternative splicing levels. Finally, we tested how the two axes of transcriptional variation differed in their potential for evolutionary change. Although both repeated differential splicing and differential expression across ecotype pairs showed tendency for parallel divergence, the degree of parallelism was lower for splicing than expression. Furthermore, parallel divergences in splicing and expression were likely to be associated with distinct cis-regulatory genetic variants and functionally unique set of genes. Finally, we found that parallelly spliced genes showed higher nucleotide diversity than parallelly expressed genes, indicating splicing is less susceptible to genetic variation erosion during parallel adaptation. Our results provide novel insight into the role of splicing in parallel adaptation, and underscore the contribution of splicing to the evolutionary potential of wild populations under environmental change.

Heterogeneous genomic architecture of skeletal armour traits in sticklebacks

Whether populations adapt to similar selection pressures using the same underlying genetic variants depends on population history and the distribution of standing genetic variation at the metapopulation level. Studies of sticklebacks provide a case in point: when colonizing and adapting to freshwater habitats, three-spined sticklebacks (Gasterosteus aculeatus) with high gene flow tend to fix the same adaptive alleles in the same major loci, whereas nine-spined sticklebacks (Pungitius pungitius) with limited gene flow tend to utilize a more heterogeneous set of loci. In accordance with this, we report results of quantitative trait locus (QTL) analyses using a backcross design showing that lateral plate number variation in the western European nine-spined sticklebacks mapped to 3 moderate-effect QTL, contrary to the major-effect QTL in three-spined sticklebacks and different from the 4 QTL previously identified in the eastern European nine-spined sticklebacks. Furthermore, several QTL were identified associated with variation in lateral plate size, and 3 moderate-effect QTL with body size. Together, these findings indicate more heterogenous and polygenic genetic underpinnings of skeletal armour variation in nine-spined than three-spined sticklebacks, indicating limited genetic parallelism underlying armour trait evolution in the family Gasterostidae.

Parallel evolution despite low genetic diversity in three-spined sticklebacks

When populations repeatedly adapt to similar environments they can evolve similar phenotypes based on shared genetic mechanisms (parallel evolution). The likelihood of parallel evolution is affected by demographic history, as it depends on the standing genetic variation of the source population. The three-spined stickleback (Gasterosteus aculeatus) repeatedly colonized and adapted to brackish and freshwater. Most parallel evolution studies in G. aculeatus were conducted at high latitudes, where freshwater populations maintain connectivity to the source marine populations. Here, we analysed southern and northern European marine and freshwater populations to test two hypotheses. First, that southern European freshwater populations (which currently lack connection to marine populations) lost genetic diversity due to bottlenecks and inbreeding compared to their northern counterparts. Second, that the degree of genetic parallelism is higher among northern than southern European freshwater populations, as the latter have been subjected to strong drift due to isolation. The results show that southern populations exhibit lower genetic diversity but a higher degree of genetic parallelism than northern populations. Hence, they confirm the hypothesis that southern populations have lost genetic diversity, but this loss probably happened after they had already adapted to freshwater conditions, explaining the high degree of genetic parallelism in the south.

De Novo Mutation Rates in Sticklebacks

Mutation rate is a fundamental parameter in population genetics. Apart from being an important scaling parameter for demographic and phylogenetic inference, it allows one to understand at what rate new genetic diversity is generated and what the expected level of genetic diversity is in a population at equilibrium. However, except for well-established model organisms, accurate estimates of de novo mutation rates are available for a very limited number of organisms from the wild. We estimated mutation rates (µ) in two marine populations of the nine-spined stickleback (Pungitius pungitius) with the aid of several 2- and 3-generational family pedigrees, deep (>50×) whole-genome resequences and a high-quality reference genome. After stringent filtering, we discovered 308 germline mutations in 106 offspring translating to µ = 4.83 × 10-9 and µ = 4.29 × 10-9 per base per generation in the two populations, respectively. Up to 20% of the mutations were shared by full-sibs showing that the level of parental mosaicism was relatively high. Since the estimated µ was 3.1 times smaller than the commonly used substitution rate, recalibration with µ led to substantial increase in estimated divergence times between different stickleback species. Our estimates of the de novo mutation rate should provide a useful resource for research focused on fish population genetics and that of sticklebacks in particular.

The genetics of adaptation in freshwater Eurasian shad (Alosa)

Studying the genetics of phenotypic convergence can yield important insights into adaptive evolution. Here, we conducted a comparative genomic study of four lineages (species and subspecies) of anadromous shad (Alosa) that have independently evolved life cycles entirely completed in freshwater. Three naturally diverged (A. fallax lacustris, A. f. killarnensis, and A. macedonica), and the fourth (A. alosa) was artificially landlocked during the last century. To conduct this analysis, we assembled and annotated a draft of the A. alosa genome and generated whole‐genome sequencing for 16 anadromous and freshwater populations of shad. Widespread evidence for parallel genetic changes in freshwater populations within lineages was found. In freshwater A. alosa, which have only been diverging for tens of generations, this shows that parallel adaptive evolution can rapidly occur. However, parallel genetic changes across lineages were comparatively rare. The degree of genetic parallelism was not strongly related to the number of shared polymorphisms between lineages, thus suggesting that other factors such as divergence among ancestral populations or environmental variation may influence genetic parallelism across these lineages. These overall patterns were exemplified by genetic differentiation involving a paralog of ATPase‐α1 that appears to be under selection in just two of the more distantly related lineages studied, A. f. lacustris and A. alosa. Our findings provide insights into the genetic architecture of adaptation and parallel evolution along a continuum of population divergence.