Table 1 Brief description of population genetic methods used in the focal studies

δaδi (Gutenkunst et al. 2009)

Site frequency spectrum (SFS) from one population or the joint frequency spectrum from multiple populations; calculated from neutral polymorphisms

Likelihood of demographic model is calculated using a diffusion approximation

Assumes all SNPs are neutral and unlinked.

Linnen et al. 2013; Laurent et al. 2016; Jones et al. 2018

fastsimcoal2 (Excoffier et al. 2013)

SFS from one population or the joint frequency spectrum from multiple populations; calculated from neutral polymorphisms

Likelihood of demographic model is calculated using coalescent simulations

Assumes all SNPs are neutral and unlinked.

Laurent et al. 2016; Pfeifer et al. 2018

SweepFinder (Nielsen et al. 2005)

SFS from whole-genome data of one population

For each location in the genome, calculates the likelihood of the observed data under a selective sweep to the likelihood under neutrality. Outputs a composite likelihood ratio (CLR) test statistic. To determine significance, neutral simulations are performed under an inferred demographic model.

Calculates a null SFS from the background SFS, rather than from a strictly equilibrium neutral model

Linnen et al. 2013; Poh et al. 2014; Mallarino et al. 2017

SweeD (Pavlidis et al. 2013)

SFS from whole genome data of one population

Computationally improved version of SweepFinder

Same as above

Laurent et al. 2016; Jones et al. 2018

SweepFinder2 (DeGiorgio et al. 2016)

SFS from whole-genome data of one population

Uses statistical framework of SweepFinder, but accounts for local reductions in diversity caused by purifying selection

Same as above

Pfeifer et al. 2018

ω-statistic (Kim and Nielsen 2004)

SNP data

Calculates the likelihood of the observed data under a selective sweep to the likelihood under neutrality, based on patterns of linkage disequilibrium associated with genetic hitchhiking. To determine significance, neutral simulations are performed under an inferred demographic model.

Demography is not directly accounted for - though demographic simulations should be performed to obtain a neutral distribution of the statistic for comparison to data.

Poh et al. 2014

Fst scan

Allele frequency data from at least two populations

Measures genetic differentiation between populations in sliding genomic windows

Implicitly assumes that populations have the same effective population size and are derived independently from the same ancestral population.

Laurent et al. 2016; Jones et al. 2018

HapFLK (Fariello et al. 2013)

Genome-wide SNPs from at least two populations and a population kinship matrix

Calculates a global measure of differentiation for each SNP (FLK) or inferred haplotype (HapFLK) after rescaling frequencies using a population kinship matrix. Genetic differentiation around a selected variant is expected to be larger than expected under drift alone.

Incorporates hierarchical population structure; demographic simulations should be performed to obtain a neutral distribution of the statistic for comparison to data

Pfeifer et al. 2018