No abstract

BackgroundMicrosatellite markers are widely used for estimating genetic diversity within and differentiation among populations. However, it has rarely been tested whether such estimates are useful proxies for genome-wide patterns of variation and differentiation. Here, we compared microsatellite variation with genome-wide single nucleotide polymorphisms (SNPs) to assess and quantify potential marker-specific biases and derive recommendations for future studies. Overall, we genotyped 180 Arabidopsis halleri individuals from nine populations using 20 microsatellite markers. Twelve of these markers were originally developed for Arabidopsis thaliana (cross-species markers) and eight for A. halleri (species-specific markers). We further characterized 2 million SNPs across the genome with a pooled whole-genome re-sequencing approach (Pool-Seq).ResultsOur analyses revealed that estimates of genetic diversity and differentiation derived from cross-species and species-specific microsatellites differed substantially and that expected microsatellite heterozygosity (SSR-H e) was not significantly correlated with genome-wide SNP diversity estimates (SNP-H e and θ Watterson) in A. halleri. Instead, microsatellite allelic richness (A r) was a better proxy for genome-wide SNP diversity. Estimates of genetic differentiation among populations (F ST) based on both marker types were correlated, but microsatellite-based estimates were significantly larger than those from SNPs. Possible causes include the limited number of microsatellite markers used, marker ascertainment bias, as well as the high variance in microsatellite-derived estimates. In contrast, genome-wide SNP data provided unbiased estimates of genetic diversity independent of whether genome- or only exome-wide SNPs were used. Further, we inferred that a few thousand random SNPs are sufficient to reliably estimate genome-wide diversity and to distinguish among populations differing in genetic variation.ConclusionsWe recommend that future analyses of genetic diversity within and differentiation among populations use randomly selected high-throughput sequencing-based SNP data to draw conclusions on genome-wide diversity patterns. In species comparable to A. halleri, a few thousand SNPs are sufficient to achieve this goal.Electronic supplementary materialThe online version of this article (doi:10.1186/s12864-016-3459-7) contains supplementary material, which is available to authorized users.

Adaptation to adverse environmental conditions such as high altitude requires physiological and/or morphological changes. Genome scans provide a means to identify the genetic basis of such adaptations without previous knowledge about the particular genetic variants or traits under selection. In this study, we scanned 3027 amplified fragment length polymorphisms (AFLP) in four populations of the common vole Microtus arvalis for loci associated with local adaptation and high altitude. We investigated voles from two populations at high elevation (~2000 m a.s.l.) representing the upper limit of the altitudinal distribution of the species and two geographically close low-altitude populations (<600 m a.s.l.). Statistical analysis incorporated a new Bayesian F(ST) outlier approach specifically developed for AFLP markers, which considers the intensity of AFLP bands instead of mere presence/absence and allows to derive population-based estimates of allele frequencies and F(IS) values. Computer simulations showed that this approach increases the statistical power of the detection of AFLP markers under selection almost to the power of single nucleotide polymorphism (SNP) data without compromising specificity. Our enhanced genome scan resulted in 20 prime candidate markers for positive selection, which show mostly extremely high allele frequency differences between the low- and high-altitude populations. The comparison of global- and pairwise-enhanced genome scans demonstrated further that very strong selective signatures may also be associated with single populations suggesting the importance of local adaptation in alpine populations of common voles.

Natural genetic variation is essential for the adaptation of organisms to their local environment and to changing environmental conditions. Here, we examine genomewide patterns of nucleotide variation in natural populations of the outcrossing herb Arabidopsis halleri and associations with climatic variation among populations in the Alps. Using a pooled population sequencing (Pool-Seq) approach, we discovered more than two million SNPs in five natural populations and identified highly differentiated genomic regions and SNPs using FST-based analyses. We tested only the most strongly differentiated SNPs for associations with a nonredundant set of environmental factors using partial Mantel tests to identify topo-climatic factors that may underlie the observed footprints of selection. Possible functions of genes showing signatures of selection were identified by Gene Ontology analysis. We found 175 genes to be highly associated with one or more of the five tested topo-climatic factors. Of these, 23.4% had unknown functions. Genetic variation in four candidate genes was strongly associated with site water balance and solar radiation, and functional annotations were congruent with these environmental factors. Our results provide a genomewide perspective on the distribution of adaptive genetic variation in natural plant populations from a highly diverse and heterogeneous alpine environment.

Oceanic islands have been a test ground for evolutionary theory, but here, we focus on the possibilities for evolutionary study created by offshore islands. These can be colonized through various means and by a wide range of species, including those with low dispersal capabilities. We use morphology, modern and ancient sequences of cytochrome b (cytb) and microsatellite genotypes to examine colonization history and evolutionary change associated with occupation of the Orkney archipelago by the common vole (Microtus arvalis), a species found in continental Europe but not in Britain. Among possible colonization scenarios, our results are most consistent with human introduction at least 5100 bp (confirmed by radiocarbon dating). We used approximate Bayesian computation of population history to infer the coast of Belgium as the possible source and estimated the evolutionary timescale using a Bayesian coalescent approach. We showed substantial morphological divergence of the island populations, including a size increase presumably driven by selection and reduced microsatellite variation likely reflecting founder events and genetic drift. More surprisingly, our results suggest that a recent and widespread cytb replacement event in the continental source area purged cytb variation there, whereas the ancestral diversity is largely retained in the colonized islands as a genetic ‘ark’. The replacement event in the continental M. arvalis was probably triggered by anthropogenic causes (land‐use change). Our studies illustrate that small offshore islands can act as field laboratories for studying various evolutionary processes over relatively short timescales, informing about the mainland source area as well as the island.

Sequencing of pooled samples (Pool-Seq) using next-generation sequencing technologies has become increasingly popular, because it represents a rapid and cost-effective method to determine allele frequencies for single nucleotide polymorphisms (SNPs) in population pools. Validation of allele frequencies determined by Pool-Seq has been attempted using an individual genotyping approach, but these studies tend to use samples from existing model organism databases or DNA stores, and do not validate a realistic setup for sampling natural populations. Here we used pyrosequencing to validate allele frequencies determined by Pool-Seq in three natural populations of Arabidopsis halleri (Brassicaceae). The allele frequency estimates of the pooled population samples (consisting of 20 individual plant DNA samples) were determined after mapping Illumina reads to (i) the publicly available, high-quality reference genome of a closely related species (Arabidopsis thaliana) and (ii) our own de novo draft genome assembly of A. halleri. We then pyrosequenced nine selected SNPs using the same individuals from each population, resulting in a total of 540 samples. Our results show a highly significant and accurate relationship between pooled and individually determined allele frequencies, irrespective of the reference genome used. Allele frequencies differed on average by less than 4%. There was no tendency that either the Pool-Seq or the individual-based approach resulted in higher or lower estimates of allele frequencies. Moreover, the rather high coverage in the mapping to the two reference genomes, ranging from 55 to 284x, had no significant effect on the accuracy of the Pool-Seq. A resampling analysis showed that only very low coverage values (below 10-20x) would substantially reduce the precision of the method. We therefore conclude that a pooled re-sequencing approach is well suited for analyses of genetic variation in natural populations.

In the last decade, amplified fragment length polymorphisms (AFLPs) have become one of the most widely used molecular markers to study the genetic structure of natural populations. Most of the statistical methods available to study the genetic structure of populations using AFLPs consider these markers as dominant and are thus unable to distinguish between individuals being heterozygous or homozygous for the dominant allele. Some attempts have been made to treat AFLPs as codominant markers by using AFLP band intensities to infer the most likely genotype of each individual. These two approaches have some drawbacks, the former discarding potentially valuable information and the latter being sometimes unable to correctly assign genotypes to individuals. In this study, we propose an alternative likelihood-based approach, which does not attempt at inferring the genotype of each individual, but rather incorporate the uncertainty about genotypes into a Bayesian framework leading to the estimation of population-specific F(IS) and F(ST) coefficients. We show with simulations that the accuracy of our method is much higher than one using AFLP as dominant markers and is generally close to what would be obtained by using the same number of Single-Nucleotide Polymorphism (SNP) markers. The method is applied to a data set of four populations of the common vole (Microtus arvalis) from Grisons in Switzerland, for which we obtained 562 polymorphic AFLP markers. Our approach is very general and has the potential to make AFLP markers as useful as SNP data for nonmodel species.

No abstract