Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Apr 21;11(1):1928.
doi: 10.1038/s41467-020-15657-3.

Genomic Release-Recapture Experiment in the Wild Reveals Within-Generation Polygenic Selection in Stickleback Fish

Free PMC article

Genomic Release-Recapture Experiment in the Wild Reveals Within-Generation Polygenic Selection in Stickleback Fish

Telma G Laurentino et al. Nat Commun. .
Free PMC article


How rapidly natural selection sorts genome-wide standing genetic variation during adaptation remains largely unstudied experimentally. Here, we present a genomic release-recapture experiment using paired threespine stickleback fish populations adapted to selectively different lake and stream habitats. First, we use pooled whole-genome sequence data from the original populations to identify hundreds of candidate genome regions likely under divergent selection between these habitats. Next, we generate F2 hybrids from the same lake-stream population pair in the laboratory and release thousands of juveniles into a natural stream habitat. Comparing the individuals surviving one year of stream selection to a reference sample of F2 hybrids allows us to detect frequency shifts across the candidate regions toward the genetic variants typical of the stream population-an experimental outcome consistent with polygenic directional selection. Our study reveals that adaptation in nature can be detected as a genome-wide signal over just a single generation.

Conflict of interest statement

The authors declare no competing interests.


Fig. 1
Fig. 1. Design and sampling locations for the release–recapture experiment.
The study focuses on a natural lake and stream population pair sampled at the sites indicated by the blue (lake) and green (stream) dots in the map of the Lake Constance basin, Central Europe (map created by the authors based on data from Google Earth, Landsat/Copernicus). Representative females and males from these populations are depicted at the same scale in the top and bottom row (Photo credit: D.B.). The breeding scheme describes how F2 hybrids were derived from the pure-bred populations under laboratory conditions, with numbers specifying how many replicate families founded the subsequent laboratory population. The F2 hybrids represented a mix of lake and stream genomic ancestry, as visualized by the color-coded bars symbolizing diploid genotypes for a single chromosome. For the field experiment, a sample of 3000 juvenile F2 hybrid individuals was released into a natural stream (photo from summer 2017, credit: T.G.L.), and the survivors recaptured one year later. An additional sample of 510 individuals served as a reference to quantify allele frequencies at the time point of the release.
Fig. 2
Fig. 2. Genomic differentiation between the natural lake and stream populations.
Profiles of the absolute allele frequency difference (AFD) along two exemplary chromosomes, with the gray dots representing individual single-nucleotide polymorphisms (SNPs) and the solid black lines indicating chromosome-specific median values. The blue curves show differentiation smoothed by averaging AFD across SNPs for 40 kb sliding windows with 20 kb overlap. The dashed red lines give the upper 0.1 percentile of the chromosome-specific AFD distribution used as a threshold for delimiting candidate genome regions under divergent lake-stream selection. The triangles identify a subset of the 126 total genome-wide SNPs eventually used as experimental target SNPs representing the independent candidate regions. Note that some high-differentiation regions are not tagged by a target SNP, or the target SNP is not the one with the highest AFD value, because SNPs in these regions failed to satisfy quality filters in the reference-survivors comparison.
Fig. 3
Fig. 3. Allele frequency shifts at the target SNPs within a single generation.
a Frequency of the stream allele in the natural populations (lake and stream), in the F2 hybrid reference sample, and in the pool of the F2 hybrids surviving one year in the stream (survivors). Each point represents one of the 126 total genome-wide target SNPs, and the horizontal bars indicate the median allele frequency in each sample (lake and stream: 0.219, 0.816; reference: 0.523; survivors: 0.563). b Distribution of the frequency shift of the stream allele at the target SNPs during the experiment, revealing a trend toward positive shifts (i.e., the stream allele increased in frequency). The median shift of 2.5% across all target SNPs is shown by the dashed red line. The blue curve visualizes the distribution kernel density-smoothed (bandwidth: 0.04).
Fig. 4
Fig. 4. Observed allele frequency shift at the target SNPs compared to neutral SNPs.
Distribution of the median reference-survivors frequency shift of the stream allele across 126 neutral SNPs, based on 9999 replicate resampling iterations. The neutral SNPs were required to lie within an AFD range of 0–0.1 in the comparison of the lake and stream population, and were further constrained to a minor allele frequency (MAF) spectrum matching the one observed across the target SNPs by applying a MAF threshold of 0.35 in the reference sample. The dashed blue line indicates the grand median random shift. The dashed red line gives the median reference-survivors shift of the stream allele observed across the 126 target SNPs during the field experiment.
Fig. 5
Fig. 5. Allele frequency shifts in simulations of polygenic directional selection.
a Response to viability selection on 100 loci over a single generation for a range of per-locus selection coefficients and two different fitness schemes. For each of the 10,000 replicate simulations, the median shift of the selectively favored allele across the loci was recorded. The blue curves represent the median of these values across the replicates, and the blue bands give the associated 95 percentiles. The dashed red line indicates the median frequency shift of the stream alleles observed across the 126 target SNPs in the field experiment. b Response to selection on 100 loci over multiple generations, for the same two fitness schemes as in a. For each fitness scheme, a single selection coefficient compatible with the experimentally observed median allele frequency shift was chosen (see a; 0.1 and 0.01 for multiplicative and additive fitness). The blue bands display the full range across the 40 replicate simulations of the median frequency of the favorable allele across all selected loci. The black lines indicate the grand median allele frequency of the focal allele across 20 neutral loci and replicate simulations (the lines are not labeled because they largely overlap between the two fitness schemes). In both a and b, the initial frequencies of the favorable alleles at the selected loci were drawn at random from the distribution observed empirically at the target SNPs.

Similar articles

See all similar articles


    1. Reznick DN, Ghalambor CK. The population ecology of contemporary adaptations: what empirical studies reveal about the conditions that promote adaptive evolution. Genetica. 2001;112–113:183–198. doi: 10.1023/A:1013352109042. - DOI - PubMed
    1. Hendry AP, Kinnison MT. Perspective: the pace of modern life: measuring rates of contemporary microevolution. Evolution. 1999;53:1637. doi: 10.1111/j.1558-5646.1999.tb04550.x. - DOI - PubMed
    1. Delong JP, et al. How fast is fast? Eco-evolutionary dynamics and rates of change in populations and phenotypes. Ecol. Evol. 2016;6:573–581. doi: 10.1002/ece3.1899. - DOI - PMC - PubMed
    1. Lescak EA, et al. Evolution of stickleback in 50 years on earthquake-uplifted islands. Proc. Natl Acad. Sci. USA. 2015;112:E7204–E7212. doi: 10.1073/pnas.1512020112. - DOI - PMC - PubMed
    1. Bell, M. A., Aguirre, W. E. & Buck, N. J. Twelve years of contemporary armor evolution in a threespine stickleback population. Evolution58, 814–824 (2004). - PubMed