Investigating the genetic architecture of complex traits in Soay sheep

Abstract

The science of quantitative genetics aims to understand how genetics influences variation in traits between individuals, how traits respond to selection, and thus how traits evolve over time. In order to answer these questions, it is important to uncover the genetic architecture of the focal traits. This includes estimating the proportion of phenotypic variation controlled by genetic variance vs non-genetic (i.e. environmental) variance – known as heritability estimation – examining how the underlying genetic variance is composed of additive genetic variance and non-additive genetic variance, and locating, identifying and characterising the causal genetic variants. Having this information about a trait’s genetic architecture allows for a better understanding of the trait itself.

Whilst the bulk of quantitative genetic research is performed in humans and agricultural and livestock populations, studies in wild populations also benefit from these types of analyses. Studying quantitative genetics in wild populations can help us to gain insights into how these traits evolve in natural settings, which are less controlled environments than those experienced by non-wild populations. For example, we can learn how natural selection shapes the genetic variation and adaptation of these traits, how gene-environment interactions influence the expression and plasticity of these traits, and how genetic diversity and inbreeding affect the fitness and survival of these traits.

Studies of wild populations often have more obstacles to overcome than those in human and domestic populations: wild pedigrees are more likely to be short, incomplete and contain errors; large genotyping arrays yielding genomic relatedness can be prohibitively expensive; and it is often difficult to reach the appropriate sample sizes of individuals. Therefore, when performing quantitative genetic analyses in a wild population, it is important to assess how well each methodology works in respect to the sample size, genotype density and any erroneous or missing data.

In this thesis, I set out to characterise the genetic architecture of 11 polygenic traits in a wild population of Soay sheep, and later an additional four traits thought to be monogenic.

In Chapter 2, I perform heritability estimation and genome-wide association studies on the 11 polygenic traits using both the lower density genotype data and the higher-density imputed data to examine how the increase in SNP density affects the results of these analyses. Heritability estimates did not differ between the two SNP densities, but the high-density imputed SNP dataset revealed four new SNP-trait associations that were not found with the lower density dataset.

In Chapter 3, I impute the genotype data available for the Soay sheep, increasing SNP density by a factor of 10.

In Chapter 4, I used a method designed to estimate heritability in a population of related individuals on both the polygenic and monogenic traits to assess how the genetic variation underpinning these traits can be partitioned into population-level additive genetic variance and family-level genetic variance. Whilst the partitioned model did not improve model fit over standard animal models for the monogenic traits, it improved the fit for some of the polygenic traits, suggesting that dominance, epistasis and/or common environment not already captured by the non-genetic random effects fitted in my models may influence these traits. I also found evidence to suggest that three of the monogenic traits I focused on may be influenced by genetic variance outside of their known causal gene.

In Chapter 5, I used regional heritability mapping methods on both the polygenic and monogenic traits to determine their suitability for the Soay sheep data and to try and identify any regions containing causal genetic variants not identified in Chapter 2. Whilst not all previously identified associated loci were recovered, I discovered new regions associated with variation in my traits. In addition, I was able identify whether the genetic variance contributed by each significantly associated region was due to individual SNPs or haplotype alleles.

In this thesis, I have applied various quantitative genetic methods to characterise the genetic architecture of 15 traits in a wild population of Soay sheep. These findings contribute to our understanding of how genetics influences trait variation and evolution in wild populations, and highlight the challenges and opportunities of performing quantitative genetic analyses in such populations. Moreover, these methods can benefit other studies of wild populations by providing more accurate estimates of genetic parameters, identifying novel genomic regions associated with traits of interest, and uncovering the complex genetic mechanisms underlying trait variation.