Contrasting results from GWAS and QTL mapping on wing length in great reed warblers. (30th April 2018)

• Record Type: Journal Article
• Authors: Hansson, Bengt; Sigeman, Hanna; Stervander, Martin; Tarka, Maja; Ponnikas, Suvi; Strandh, Maria; Westerdahl, Helena; Hasselquist, Dennis
• Is Part Of: Molecular ecology resources. Volume 18:Number 4(2018:Jul.)
• Page Start: 867
• Page End: 876
• Publication Date: 2018-04-30
• Subjects: adaptation -- association -- birds -- genotype -- phenotype -- wing length; Molecular ecology -- Periodicals; 572.8
• DOI: 10.1111/1755-0998.12785
• Languages: English
• ISSNs: 1755-098X

Abstract

Abstract: A major goal in evolutionary biology is to understand the genetic basis of adaptive traits. In migratory birds, wing morphology is such a trait. Our previous work on the great reed warbler ( Acrocephalus arundinaceus ) shows that wing length is highly heritable and under sexually antagonistic selection. Moreover, a quantitative trait locus (QTL) mapping analysis detected a pronounced QTL for wing length on chromosome 2, suggesting that wing morphology is partly controlled by genes with large effects. Here, we re‐evaluate the genetic basis of wing length in great reed warblers using a genomewide association study (GWAS) approach based on restriction site‐associated DNA sequencing (RADseq) data. We use GWAS models that account for relatedness between individuals and include covariates (sex, age and tarsus length). The resulting association landscape was flat with no peaks on chromosome 2 or elsewhere, which is in line with expectations for polygenic traits. Analysis of the distribution of p ‐values did not reveal biases, and the inflation factor was low. Effect sizes were however not uniformly distributed on some chromosomes, and the Z chromosome had weaker associations than autosomes. The level of linkage disequilibrium (LD) in the population decayed to background levels within c. 1 kbp. There could be several reasons to why our QTL study and GWAS gave contrasting results including differences in how associations are modelled (cosegregation in pedigree vs. LD associations), how covariates are accounted for in the models, type of marker used (multi‐ vs. biallelic), difference in power or a combination of these. Our study highlights that the genetic architecture even of highly heritable traits is difficult to characterize in wild populations.