Estimation of number and size of QTL effects in forest tree traits

Abstract

Mapping the genetic architecture of forest tree traits is important in order to understand the evolutionary forces that have shaped these traits and to facilitate the development of genomic-based breeding strategies. We examined the number, size, and distribution of allelic effects influencing eight types of traits using 30 published mapping studies (linkage and association mapping) in forest trees. The sizes of allelic effects, measured as the phenotypic variance explained, generally showed a severely right-skewed distribution. We estimated the numbers of underlying causal effects (n qtl ) for different trait categories by improving a method previously developed by Otto and Jones (Genetics 156:2093(Genetics 156: -2107(Genetics 156: , 2000. Estimates of n qtl based on association mapping studies were generally higher (median at 643) than those based on linkage mapping (median at 33). Comparisons with simulated linkage and association mapping data suggested that the lower n qtl estimates for the linkage mapping studies could partly be explained by fewer causal loci segregating within the full-sib family populations normally used, but also by the cosegregation of causal loci due to limited recombination. Disease resistance estimates based on linkage mapping studies had the lowest median of four underlying effects, while growth traits based on association mapping had about 580 effects. Theoretically, the capture of 50% of the genetic variation would thus require a population size of about 200 for disease resistance in linkage mapping, while growth traits in association mapping would require about 25,000. The adequacy and reliability of the improved method was successfully verified by applying it to the simulated data.

Genome‐wide association study identified novel candidate loci affecting wood formation in Norway spruce

“…Wood properties have previously been indicated to have a complex genetic architecture, in which association studies that make use of historical recombination represent a method that presents a substantial increase in QTL detection power for such complex traits (Hall et al ., ). In our study, the number of QTLs detected reflected the complex nature of the traits under study, and our experimental design allowed the detection of the largest/most significant QTLs.…”

“…Therefore in our study increasing the sample size from 517 individuals might allow the inclusion of rare alleles, explaining some of the missing heritability (Hamblin et al, 2011;De La Torre et al, 2019). The detection of true lowfrequency alleles associated with complex traits is challenging as it requires large and genetically diverse populations (Hall et al, 2016). Variants with low minor allele frequencies are usually discarded due to the potential of genotyping errors.…”

Genomic architecture of complex traits in loblolly pine

“…With the advent of the genomic era, association mapping became a promising tool due to the undomesticated status of conifer trees, outcrossing mating system and rapid decay of linkage disequilibrium (Neale & Kremer, 2011). This made it possible to detect larger numbers of polymorphisms in close proximity to QTL or even QTL themselves, although effect sizes are reported to be smaller than those detected by QTL mapping (Neale & Savolainen, 2004;Neale & Kremer, 2011;Hall et al, 2016;Plomion et al, 2016). With the absence of reference genomes, these studies focused only on candidate genes or specific regions (although see Fuentes-Utrilla et al, 2017;Lu et al, 2017).…”

“…However, the dissection of the genetic architecture of complex traits is still challenging due to the fragmented nature of the reference genomes, incomplete gene annotation, and the absence of physical and high-density linkage maps (De La Torre et al, 2014;Neale et al, 2017). Also, due to the highly polygenic basis of complex traits and rapid decay of linkage disequilibrium, large mapping population sizes are needed to capture a large proportion of the genetic variation, especially when using GWAS instead of QTL mapping (Hall et al, 2016).…”

Accuracy of genomic selection for growth and wood quality traits in two control-pollinated progeny trials using exome capture as genotyping platform in Norway spruce

“…In GS, model adequacy is more related to genetic architecture of the target trait. Based on genome-wide association studies (GWAS), most growth and wood quality traits for conifer species have polygenic inheritance with a gamma or exponential distribution of allelic effects [10,17]. To account for these skewed distributions of a few genes with large effects and most of genes with small effects, Bayes A, B, and Cπ and Bayesian Least Absolute Shrinkage and Selection Operator (BLASSO) were developed to fit the models more accurately, in contrast to Genomic best linear unbiased prediction (GBLUP) model that assumes a normal distribution of allelic effect.…”