Gene flow favors clustering of adaptive alleles in a globally distributed species

Transcript

1. Gene flow favors clustering of adaptive alleles in a globally distributed species Kieran Samuk Schluter Lab University of British Columbia
3. Hundreds of studies have found signatures of selection in the genome
4. We know that signatures of selection are ubiquitous in the genome Seehausen et al. Nat Rev. Gen. 2014
5. Why do we see adaptation in some regions of the genome and not others? ? ? Seehausen et al. Nat Rev. Gen. 2014
6. Are there general constraints on where adaptation can occur in the genome? …or can it happen anywhere that adaptive variation is available?
7. Recent theory suggests that gene flow can constrain where adaptation occurs in the genome. Bürger & Akerman 2011 (Theor Popul Biol) Yeaman & Whitlock 2011 (Evolution)
8. Scenario: divergent selection in two populations with a shared ancestor
9. Selection builds up sets of co-selected alleles
10. Gene flow + recombination breaks apart sets of co-selected alleles
11. Tight linkage can prevent allele sets from being broken apart
12. Tight linkage can prevent allele sets from being broken apart
13. Maintained! Broken! Tight linkage can prevent allele sets from being broken apart
14. Highest fitness Tight linkage can prevent allele sets from being broken apart
15. Goal: Test this prediction! Divergent selection + gene flow favors tightly linked clusters of locally-adapted alleles Bürger & Akerman 2011 (Theor Popul Biol) Yeaman & Whitlock 2011 (Evolution)
16. Goal: Test this prediction! Divergent selection + gene flow favors tightly linked clusters of locally-adapted alleles Bürger & Akerman 2011 (Theor Popul Biol) Yeaman & Whitlock 2011 (Evolution)
17. “DS-GF” = Divergent selection with gene flow
18. Problem: Other processes cause “clustered” signatures of selection
19. Gene density and mutation rate variation also cause clustering
20. Hitchhiking causes “clustered” signatures of selection
21. Statistical controls Compare DS+GF to selection and gene flow alone There are solutions to both these issues:
22. Approach • Classify pairs of populations based on presence of selection and/or gene flow • Find loci that are adaptively differentiated between each pair of populations • Quantify the degree of clustering of adaptive loci in the genome, controlling for mutation + gene density
23. Sticklebacks are an excellent system for applying this approach!
25. Approach • Classify pairs of populations based on presence of selection and/or gene flow • Find loci that are adaptively differentiated between each pair of populations • Quantify the degree of clustering of adaptive loci in the genome, controlling for mutation + gene density
26. Parallel Selection Divergent Selection Gene flow Allopatry Idea: compare clustering among different selection and gene flow regimes
27. We used morphological differences as a proxy for divergent selection Divergent Parallel Benthic vs. Limnetic Alaska Marine vs. Oregon Marine
28. We used geographic distance + barriers as a proxy for gene flow 1. No physical barriers 2. < 500 km apart All pairs of stickleback populations with evidence of recent gene flow
29. Approach • Classify pairs of populations based on presence of selection and/or gene flow • Find loci that are adaptively differentiated between each pair of populations • Quantify the degree of clustering of adaptive loci in the genome, controlling for mutation + gene density
30. We found signatures of selection using FST outliers between pairs of populations
31. FST outliers = enriched for adaptive loci
32. Parallel Selection Divergent Selection Gene flow Allopatry We performed FST scans for selection for population pairs in each category
33. Example: one chromosome in one pair of populations per regime
34. Example: one chromosome in one pair of populations per regime
35. Approach • Classify pairs of populations based on presence of selection and/or gene flow • Find loci that are adaptively differentiated between each pair of populations • Quantify the degree of clustering of adaptive loci in the genome, controlling for mutation + gene density
36. DS+GF favors tightly linked clusters of locally-adapted alleles Bürger & Akerman 2011 (Theor Popul Biol) Yeaman & Whitlock 2011 (Evolution)
37. We measured “nearest neighbor” genetic distances between all SNPs vs. outliers All SNPs Outliers only
38. Outliers from DS-GF comparisons were unusually close together in the genome
39. Outliers from DS-GF comparisons were unusually close together in the genome DS+GF
40. Outliers from DS-GF comparisons were unusually close together in the genome *** Permutation test p = 0.0015
41. As predicted, adaptive alleles are more tightly linked in DS-GF comparisons
42. As predicted, adaptive alleles are more tightly linked in DS-GF comparisons What does this actually look like in the genome?
44. Recombination rate
45. Low recombination
46. DS+GF favors tightly linked clusters of locally-adapted alleles
47. DS+GF favors tightly linked clusters of locally-adapted alleles …via concentration of these alleles in regions of low recombination
48. Bürger & Akerman 2011 (Theor Popul Biol) Yeaman & Whitlock 2011 (Evolution)
49. Implications
50. Gene flow may make the usable area of the genome effectively “smaller”
51. Gene flow may make the usable area of the genome effectively “smaller”
52. Smaller set of genes = More re-use of same genes DS+GF = more parallelism? Less variation available overall Novel effect of gene flow?
53. Summary • Gene flow favors the clustering of adaptive alleles in regions of low recombination • May be an important and general constraint on adaptive evolution
54. Acknowledgements Kate Ostevik Marius Roesti Sam Yeaman Armando Geraldes Mark Ravinets Anne-Laure Ferchaud Jun Kitano Thor Veen Seth Rudman Baocheng Guo Semiahmoo First Nation Waycobah First Nation Funding Committee Dolph Schluter Mike Whitlock Trish Schulte Loren Rieseberg
55. Dxy trends in the same direction, but difference is not significant
56. Heterozygosity vs. recombination does not vary among regimes
57. DS+GF remains more clustered when we control for genome-wide FST
58. Results are qualitatively similar when we use continuous distance