Detecting loci under selection in a hierarchically structured population

Abstract

Patterns of genetic diversity between populations are often used to detect loci under selection in genome scans. Indeed, loci involved in local adaptations should show high F ST values, whereas loci under balancing selection should rather show low F ST values. Most tests of selection based on F ST use a null distribution generated under a simple island model of population differentiation. Although this model has been shown to be robust, many species have a more complex genetic structure, with some populations sharing a recent ancestry or due to the presence of barriers to gene flow between different parts of a species range. In this paper, we propose the use of a hierarchical island model, in which demes exchange more migrants within groups than between groups, to generate the joint distribution of genetic diversity within and between populations. We show that tests not accounting for a hierarchical structure, when it exists, do generate a large excess of false positive loci, whereas the hierarchical island model is robust to uncertainties about the exact number of groups and demes per group in the system. Our approach also explicitly takes into account the mutational process, and does not just rely on allele frequencies, which is important for short tandem repeat (STR) data. An application to human and stickleback STR data sets reveals a much lower number of significant loci than previously obtained under a non-hierarchical model. The elimination of false positive loci from genome scans should allow us to better determine on which specific class of genes selection is operating.

Population genomics of the introduced and cultivated Pacific kelp Undaria pinnatifida: Marinas—not farms—drive regional connectivity and establishment in natural rocky reefs

“…First, we employed a Bayesian method that uses a logistic regression model to partition F ST coefficients into a population‐specific component (beta) and a locus‐specific component (alpha), implemented in BayeScan 2.1 (Foll & Gaggiotti, 2008). Two other outlier tests were then carried out in Arlequin: the first is the default island model of Beaumont and Nichols (1996), whereas the second test utilizes a hierarchical island model that reduces the number of false‐positive outliers by accounting for population structure (Excoffier, Hofer, & Foll, 2009). For sample set 1, the individuals were grouped according to the genetic clusters identified by fastSTRUCTURE (see Results), and for sample set 2, the individuals were grouped by habitat type.…”

Discovering variation of secondary metabolite diversity and its relationship with disease resistance inCornus floridaL.

“…We used genotype by sequencing (GBS; Peterson et al., 2012) of two Illumina Hiseq libraries, de novo assembly into 90‐bp GBS tags with STACKS (Catchen, Amores, Hohenlohe, Cresko, & Postlethwait, 2011), latent factor mixed modeling [a genotype–environment association (GEA) method; Frichot, Schoville, Bouchard, & François, 2013], and two F ST outlier methods (Excoffier, Hofer, & Foll, 2009; Foll & Gaggiotti, 2008) to classify putatively neutral SNPS and SNPs exhibiting varying support for being under selection (Pais et al., 2016). Putatively neutral reference SNPs were used to calculate marker‐based inbreeding coefficients ( F ; Keller, Visscher, & Goddard, 2011) and identity‐by‐state matrices using PLINK (Purcell et al., 2007).…”

A genome scan for selection signatures comparing farmed Atlantic salmon with two wild populations: Testing colocalization among outlier markers, candidate genes, and quantitative trait loci for production traits

“…Methods of detecting F ST outliers resulting from “soft sweeps” suitable for low‐density SNP chip datasets include (i) FDIST2, which compares the F ST observed for a locus to the F ST expected under neutrality relative to its observed heterozygosity assuming an “n‐island” model where the effective population size of all population is constant as is the pairwise population migration rate (Antao, Lopes, Lopes, Beja‐Pereira, & Luikart, 2008; Beaumont & Nichols, 1996), (ii) Arlequin v.3.5, which implements FDIST2 methodology with the addition of a hierarchical island option that assumes two different constant migration rates with the lower rate among regional groups of populations and the higher rate among populations within the same region (Excoffier et al., 2009), and (iii) BayeScan, which separately estimates the posterior probability that each locus is under selection without assuming that all effective population sizes and migration rates are equal (Beaumont & Balding, 2004; Foll & Gaggiotti, 2008). In this study, Arlequin 3.5 and BayeScan 2.1 were chosen for the outlier analysis to be a good combination to reduce type I (false positive) and type II (false negative) error rates, which have been evaluated using simulation (Lotterhos & Whitlock, 2014; Narum & Hess, 2011; Vilas, Perez‐Figueroa, & Caballero, 2012).…”

“…Method 2 was composed of six separate pairwise analyses using the nonhierarchical (FDIST2) option of Arlequin 3.5, each comparing one of the AQUA populations with the TOB_WILD population. Method 3 used the hierarchical island version of the outlier module in Arlequin 3.5 that implements a hierarchical FDIST2 (Excoffier et al., 2009). This module has a hierarchical island model that allows for lower migration rates among different “island” groups than among populations within groups.…”