10 Scopus citations


Population demographic history may be learned from contemporary genetic variation data. Methods based on aggregating the statistics of many single loci into an allele frequency spectrum (AFS) have proven powerful, but such methods ignore potentially informative patterns of linkage disequilibrium (LD) between neighboring loci. To leverage such patterns, we developed a composite-likelihood framework for inferring demographic history from aggregated statistics of pairs of loci. Using this framework, we show that two-locus statistics are more sensitive to demographic history than single-locus statistics such as the AFS. In particular, two-locus statistics escape the notorious confounding of depth and duration of a bottleneck, and they provide a means to estimate effective population size based on the recombination rather than mutation rate. We applied our approach to a Zambian population of Drosophila melanogaster. Notably, using both single- and two-locus statistics, we inferred a substantially lower ancestral effective population size than previous works and did not infer a bottleneck history. Together, our results demonstrate the broad potential for two-locus statistics to enable powerful population genetic inference.

Original languageEnglish (US)
Pages (from-to)1037-1048
Number of pages12
Issue number2
StatePublished - Jun 2017


  • Allele frequencies
  • Demographic inference
  • Diffusion approximation
  • Linkage disequilibrium

ASJC Scopus subject areas

  • Genetics


Dive into the research topics of 'Inferring demographic history using two-locus statistics'. Together they form a unique fingerprint.

Cite this