Presentation Detail

Population Genetics and/or Genomics, Theoretical

Harris, Kelley [1], Lorenzen, Eline [2], Fumagalli, Matteo [2], Liu, Shiping [3], Willerslev, Eske [4], Nielsen, Rasmus [5].

The effects of population size fluctuation and coalescent model choice on inference of population structure from shared haplotype tracts.

Despite the increasing quality and quantity of genomic data being produced, it remains a challenging problem to make inferences about population divergence and gene flow. Recently, we described a method for inferring such parameters from the length distribution of tracts of identity by state (IBS), the segments of identical haplotype stretching between polymorphisms in a pairwise alignment between sequences from different populations. We infer such divergence parameters jointly with effective population sizes, which are informed by IBS tracts shared within the same population. The approach allows for joint inference of complex population size changes with divergence and gene flow. In this framework, we show that conclusions about structure and admixture can vary substantially depending on assumptions that are made about effective population size during the period when populations are drifting apart. Assumptions about population size have the potential to bias all inferences about divergence times and gene flow; this effect is not limited to our particular demographic method. Rather, it can partially explain the wide range of demographic histories that have been inferred with different methods for the same populations. Recent estimates of the divergence time between polar bears and brown bears have ranged from 600,000 years ago to 5 million years ago, and we present strong evidence that that these estimates were upwardly biased to different degrees by not accounting for a strong reduction in polar bear population size during the ice age. In addition to exploring how accurate gene flow and divergence time estimates require the accurate joint inference of population size fluctuations, we explore the results of deriving IBS tract length distributions from different coalescent approximations, showing that certain approximations yield more accurate demographic estimates in a composite likelihood framework. Less accurate coalescent approximations can yield biased parameter estimates, especially in the setting of complex demographic histories.

1 - UC Berkeley, Mathematics, Berkeley, CA, USA
2 - UC Berkeley, Integrative Biology
3 - Beijing Genomics Institute
4 - University of Copenhagen
5 - UC Berkeley, Integrative Biology and Statistics

Keywords:
demographic history
coalescent
Recombination
divergence-time estimation
migration
effective population size
identity by state
polar bears
admixture.

Presentation Type: Regular Oral Presentation
Session: 5
Location: Alpine A and B/Snowbird Center
Date: Saturday, June 22nd, 2013
Time: 9:30 AM
Number: 5005
Abstract ID:1155
Candidate for Awards:W.D. Hamilton Award for Outstanding Student Presentation