A polygenic basis for birth weight in a wild population of red deer (Cervus elaphus)

Abstract

The genetic architecture of traits under selection has important consequences for the response to selection and potentially for population viability. Early QTL mapping studies in wild populations have reported loci with large effect on trait variation. However, these results are contradicted by more recent genome-wide association analyses, which strongly support the idea that most quantitative traits have a polygenic basis. This study aims to re-evaluate the genetic architecture of a key morphological trait, birth weight, in a wild population of red deer (Cervus elaphus), using genomic approaches. A previous study using 93 microsatellite and allozyme markers and linkage mapping on a kindred of 364 deer detected a pronounced QTL on chromosome 21 explaining 29% of the variance in birth weight, suggesting that this trait is partly controlled by genes with large effects. Here, we used data for more than 2,300 calves genotyped at >39,000 SNP markers and two approaches to characterise the genetic architecture of birth weight. First, we performed a genome-wide association (GWA) analysis, using a genomic relatedness matrix to account for population structure. We found no SNPs significantly associated with birth weight. Second, we used genomic prediction to estimate the proportion of variance explained by each SNP and chromosome. This analysis confirmed that most genetic variance in birth weight was explained by loci with very small effect sizes. Third, we found that the proportion of variance explained by each chromosome was slightly positively correlated with its size. These three findings highlight a highly polygenic architecture for birth weight, which contradicts the previous QTL study. These results are probably explained by the differences in how associations are modelled between QTL mapping and GWA. Our study suggests that models of polygenic adaptation are the most appropriate to study the evolutionary trajectory of this trait.

Keywords: Cervus elaphus; GenPred‌; genome-wide association study; genomic prediction; genomic relatedness; heritability; maternal effects.

Fig. 1.

Manhattan plot for the association between birth weight and SNPs. Top dashed line: significance threshold equivalent to α = 0.05. Points are coloured by chromosomes (blue: odd numbers; red: even numbers). We only show results for the SNPs with known map positions. Previous potential QTLs found by Slate et al. (2002) were on chromosomes 12, 14 and 21 but not mapped with sufficiently good precision to be represented here.

Fig. 2.

Quantile-quantile (Q-Q) plot of GWAS P-values for birth weight (shown in the Manhattan plot). This is a graphical representation of the deviation of the observed P-values from the null hypothesis. We found no P-values larger than expected under the null hypothesis (there are not points above the 1:1 diagonal).

Fig. 3.

Chromosome partitioning from the genomic prediction analysis. The proportion of additive genetic variance explained by each chromosome is slightly positively correlated with their size. The genomic prediction model has a high accuracy to predict the breeding values for birth weight, with a mean accuracy of 0.71 (using h²_SNP = 0.38). The previous QTLs reported by Slate et al. (2002) were found on chromosomes 21, 12 and 14 (in red), which respectively explain here 1.5, 5.1 and 4.6% of the genetic variance.