1. Department of Forest Resources and Environmental Conservation
Recombination rate variation, hitchhiking, and central-peripheral structure shape deleterious load in black cottonwood
Jason Holliday
Department of Forest Resources and Environmental Conservation
Virginia Tech

2. Understanding the adaptive past to predict the adaptive future
The rate of adaptation to anthropogenic climate change depends on:
Number of loci involved
Distribution of effect sizes
Distribution of adaptive alleles in the genome
New mutation vs standing variation
Many non-adaptive processes also play a role:
Historical demography (e.g., bottlenecks)
Migration rate relative to selection
Deleterious alleles

3. Populus trichocarpa as model to understand adaptive potential and constraint
Compact (~480mb), sequenced genome
Clonal propagation
Wide latitudinal and altitudinal range = strong differentiation for adaptive traits
High gene flow via wind polination = low background population structure
Postglacial history = possbile adaptive constraint

4. Sampling and phenotyping
449 samples spanning the latitudinal and altitudinal range of the species
Outgroups: P. tremula, P. tremuloides, P. deltoides
Re-sequence most of the ‘gene space’
Sequence capture
Replicated gardens in Virginia and British Columbia
Phenotyped for growth, bud phenology, cold hardiness

5. Genotyping by sequence capture

6. Sequence capture recovers regions of interest in a mostly repeatable way

7. Coverage is tightly linked to bait locations

8. Genomic sampling
~ 2 million single nucleotide polymorphisms passed quality filters

9. Adaptation is not just about having the ‘right’ alleles; it’s also about not having the ‘wrong’ alleles
Many tree species suffer substantial inbreeding depression, while hybrid vigor is common
Likely due in part to accumulation of deleterious alleles
Wang et al 2004

10. Deleterious alleles
Coding SNPs often simply categorized as synonymous (same amino acid) or non-synonymous (different amino acid)
But not all non-synonymous SNPs are equivalent ≠
Ala
Gly
Ala
Cys

11. Factors that may govern accumulation of deleterious allele
Genomic
Recombination
Hitchhiking
Direct positive selection – are some of these alleles conditionally advantageous?
Demographic
Population history (bottlenecks, etc)
Population size (efficiency of selection)

12. Sorting Intolerant From Tolerant (SIFT)
Uses multiple alignments from related species to classify SNPs as tolerated or damaging
Examples:
If no amino acid substitution is found at that site, any non-syn change is deleterious
If only hydrophobic residues found, changes to other amino acid classes deleterious
No information about protein structure, but performs similarly to algorithms that account for this
Advantage: can be used on just about any sequence, not just those with known structure information

13. Deleterious SNPs segregate at lower frequency than tolerated SNPs
Derived allele frequency

14. Deleterious SNPs segregate at lower frequency than tolerated SNPs
Percent
Deleterious : tolerated ratio
Derived allele frequency

15. Determinants of deleterious frequency: recombination
More generally – recombination rate variation correlated with deleterious ratio

17. Determinants of deleterious frequency: hitchhiking
iHS = measure of incomplete hitchhiking events
Deleterious ratio higher in top 1% of iHS windows
iHS higher for windows enriched for deleterious SNPs
Signal of direct or linked selection?

18. Any evidence for direct selection?
FST outliers identified across three sampling transects (two altitudinal and one latitudinal)
Deleterious alleles underrepresented among FST outliers
Suggests deleterious alleles mostly not direct targets of positive selection

19. Determinants of deleterious frequency: historical demography
Like many northern species, poplar experienced bottlenecks associated with Pleistocene glaciation
Increased drift in leading edge populations
More generally, variation in Ne across the range may lead to differential efficiency of purifying selection

20. Determinants of deleterious frequency: historical demography

21. Why does this matter? Deleterious alleles affect fitness
BC Garden
VA Garden

22. Can we detect the signature of deleterious alleles at individual SNP loci?
Sort of – deleterious alleles overrepresented among associated genes, but underrepresented among associated SNPs
Low frequency deleterious alleles generating ‘synthetic’ associations?

23. Conclusions
Deleterious alleles impact performance, and may explain some phenotypic associations for tolerated SNPs
Accumulate as a result of recombination rate variation, hitchhiking, and demographic history
More common in peripheral populations
Response to selection may be weaker in these populations, which suggests less adaptive potential under climate change

24. Acknowledgements Mandy Zhang, Lecong Zhou, Haktan Suren, Rajesh Bawa
Kyle Peer, Clay Sawyers, Debbie Byrd (VA Garden)
Mike Carlson (BC Forest Service) and Cees Van Oosten (BC Garden)
Carl Douglas and BC Forest Service – access to some BC samples