Experimental evidence for rapid genomic adaptation to a new niche in an adaptive radiation

Abstract

A substantial part of biodiversity is thought to have arisen from adaptive radiations in which one lineage rapidly diversified into multiple lineages specialized to many different niches. However, selection and drift reduce genetic variation during adaptation to new niches and may thus prevent or slow down further niche shifts. We tested whether rapid adaptation is still possible from a highly derived ecotype in the adaptive radiation of threespine stickleback on the Haida Gwaii archipelago, Western Canada. In a 19-year selection experiment, we let giant sticklebacks from a large blackwater lake evolve in a small clearwater pond without vertebrate predators. A total of 56 whole genomes from the experiment and 26 natural populations revealed that adaptive genomic change was rapid in many small genomic regions and encompassed 75% of the change between 12,000-year-old ecotypes. Genomic change was as fast as phenotypic change in defence and trophic morphology, and both were largely parallel between the short-term selection experiment and long-term natural adaptive radiation. Our results show that functionally relevant standing genetic variation can persist in derived radiation members, allowing adaptive radiations to unfold very rapidly.

Fig. 1. Phenotypic and genomic change in the selection experiment

a Summary of the phenotypic change observed in the selection experiment, as reported in Leaver and Reimchen, with colours indicating trait increase or decrease and asterisks indicating significant change. Phenotypic change in six bony predator defence traits (FSL: first dorsal spine length, SSL: second dorsal spine length, PSL: pelvic spine length, # plates: number of lateral plates, LP3H: lateral plate 3 height, LP2: lateral plate 2 frequency), four feeding morphology traits (LJL: lower jaw length, # rakers: number of gill rakers, GRL: gill raker length, GRS: gill raker spacing) and eye diameter (ED) was in the expected direction, i.e. parallel, given the shift from vertebrate- to invertebrate-dominated predation and zooplankton- to invertebrate-dominated diet and observed phenotypic divergence between large lake and small pond populations in the adaptive radiation on Haida Gwaii. SL: standard length. b Transplant of 100 adult giant threespine stickleback from Mayer Lake into Roadside Pond and evolution for 13 generations led to moderate genomic differentiation (F_ST), a minor reduction nucleotide diversity (π) and to a positive shift in the Tajima’s D (T_D) distribution. c Even though several rare alleles were fixed, allele frequencies (AF) did not change much over 13 generations. MAF: minor allele frequency.

Fig. 2. Comparison of the extent of phenotypic and genomic evolution in the 19 years selection experiment with the ~12,000 years old adaptive radiation

Phenotypic divergence and adaptive genomic differentiation arose rapidly in the selection experiment, comparable in extent to postglacial divergence between large lake and pond or stream ecotypes. a Population means and distributions for six phenotypic traits in Mayer Lake (source population), Roadside Pond (transplant population, orange), postglacial stream (green) and pond (blue) ecotype populations. b Absolute divergence (D_XY), c relative differentiation (F_ST) and d adaptive differentiation (F_ST) between postglacial large lake and stream (green) or pond (blue) ecotype populations as well as in the selection experiment (orange). F_ST estimates for lake-stream and large lake vs. small pond comparisons are based on SNP chip data from a previous study. Adaptive differentiation SNPs are outlier SNPs in this previous study and top 5% F_ST SNPs in each outlier window for the selection experiment. Numbers are mean percentages of phenotypic or genomic change in the selection experiment compared to postglacial lake vs. stream divergence (upper) and large lake vs. small pond divergence (lower number).

Fig. 3. Genomic footprints of divergent selection are widespread across the genome

a Absolute allele frequency change (|ΔAF|) at the top 0.1% strongest |ΔAF|-SNPs, with black points highlighting SNPs for which the rarer allele in Mayer Lake went to fixation in Roadside Pond. Grey vertical bars highlight 77 outlier regions across 15 of the 21 stickleback chromosomes (roman numerals) with overlapping top 1% outlier 10kb windows against neutral demographic expectations highlighted with larger, darker points, for the statistics b high differentiation (F_ST), c change in diversity (Δπ) and d Tajima’s D (ΔT_D) and haplotype-based selection statistics e w-iHS and f w-XPEHH (see Methods).

Fig. 4. Local signatures of divergent selection in the genome

For five of the 77 outlier regions in the genome (grey shading), patterns of differentiation (F_ST), allele frequency change (|ΔAF|), nucleotide diversity (π), Tajima’s D (T_D) and iHS, H12 and XPEHH are shown. Outlier region I.b is centred on a two genes, trappc6bl and bloc1s3, controlling pigmentation in the retina, II.a contains the pigmentation gene mc1r, XVII.j the blue-sensitive colour vision gene opnsw2, likely targets of divergent selection on pigmentation and visual perception. See Supplementary Figs. 5-19 for further outlier regions. The top panel colour code indicates quantiles in the |ΔAF| distribution, black horizontal bars show significant non-overlapping 10kb outlier windows against neutral expectations, gene exons are shown on top (gene names ‘ENSG0000000012345’ shortened to ‘e12345’). Genomic coordinates refer to an improved version of the reference genome. Lines are sliding-window estimates for 10kb sliding windows with 2.5kb step size, dots are single SNP estimates (|ΔAF|, iHS, XPEHH) or 81-SNP windows (H12).

Fig. 5. Outlier regions and overlapping QTL, candidate genes and genome-phenotype/ecology associations across the adaptive radiation

a Distribution of overlapping quantitative trait loci (QTL). Circles indicate QTL peak markers, horizontal bars confidence intervals and colour codes the effect sizes: major (percentage variance explained, PVE > 25%), intermediate (25% > PVE > 5%) and minor effect QTL (PVE < 5%). b Genome vs. environment and phenotype associations (GE/GP assoc.) and directionality of phenotypic and genomic change between selection experiment and Haida Gwaii adaptive radiation. Predictors retained in generalized linear model for each outlier region are shown in coloured squares, with blue boxes representing parallel genomic and phenotypic/ecological change and red boxes non-parallel change (except for genomic PC1, for which directionality cannot be inferred). The colour code shows the relative effect sizes (β). c List of candidate genes centred on divergent selection patterns in outlier regions.