Genetic architecture and lifetime dynamics of inbreeding depression in a wild mammal

Abstract

Inbreeding depression is ubiquitous, but we still know little about its genetic architecture and precise effects in wild populations. Here, we combine long-term life-history data with 417 K imputed SNP genotypes for 5952 wild Soay sheep to explore inbreeding depression on a key fitness component, annual survival. Inbreeding manifests in long runs of homozygosity (ROH), which make up nearly half of the genome in the most inbred individuals. The ROH landscape varies widely across the genome, with islands where up to 87% and deserts where only 4% of individuals have ROH. The fitness consequences of inbreeding are severe; a 10% increase in individual inbreeding F_ROH is associated with a 60% reduction in the odds of survival in lambs, though inbreeding depression decreases with age. Finally, a genome-wide association scan on ROH shows that many loci with small effects and five loci with larger effects contribute to inbreeding depression in survival.

Fig. 1. Runs of homozygosity (ROH) variation among individuals and across the genome.

a ROH longer than 5 Mb in the seven individuals with the highest inbreeding coefficients F_ROH in the seven top rows and the seven individuals with the lowest F_ROH in the seven bottom rows. b Distribution of ROH among different length classes. Each data point represents the proportion of ROH of a certain length class within an individual’s autosomal genome. ROH length classes were categorised by their expected average physical length when the underlying haplotypes had a most recent common ancestor (MRCA) 2–32 generations (g) ago. c Genome-wide ROH density among all 5952 individuals in non-overlapping 500 Kb windows. The colour gradient has been scaled according to the ROH density, which is shown in the figure legend.

Fig. 2. Correlates of ROH density variation across the genome.

Runs of homozygosity (ROH) density, recombination rate and heterozygosity were quantified in non-overlapping 500 Kb windows, with each point representing one window. The top 0.5% of windows with the highest and lowest ROH density in the population, termed ROH islands (n = 24) and deserts (n = 24), are coloured in purple and yellow respectively in all four plots. a Relationship between ROH density and recombination rate. b Relationship of ROH density and SNP heterozygosity. c Recombination rate within ROH islands and deserts. d Heterozygosity within ROH islands and deserts. Solid lines in a and b are linear regression lines and dashed lines in c and d are genome-wide means. Boxplots show the median as a centre line with the bounds of the box as 25th and 75th percentiles and upper and lower whiskers as largest and smallest value but no further than 1.5 * inter-quartile range from the hinge. Source data for this figure are also provided as a source data file.

Fig. 3. Inbreeding depression in annual survival.

a Distributions of inbreeding coefficients F_ROH in Soay sheep age classes ranging from 0 to 9 years. b Proportion of surviving individuals per year in four different life stages and among different F_ROH classes. As highly inbred individuals are relatively rare, the last class spans a wider range of inbreeding coefficients. Source data for this figure are also provided as a source data file. c Predicted survival probability and 95% credible intervals over the range of inbreeding coefficients F_ROH for each life stage while holding sex and twin constant at 1 (male) and 0 (no twin). The predictions for the later life stages classes exceed the range of the data but are shown across the full range for comparability.

Fig. 4. GWAS of SNP-wise ROH status effects on annual survival.

Regional inbreeding depression was conceptualised and tested using two binary ROH status predictors. One of the predictors quantified the ROH status of allele A (in ROH = 1, not in ROH = 0), while the other quantified the ROH status of allele B. a Distribution of effect sizes for SNP-wise ROH status effects. b Distribution of p-values for SNP-wise ROH status effects. The yellow histograms showing positive effects are superimposed on top of the purple histograms showing negative effects to highlight a substantially larger proportion of negative ROH status effects than expected by chance. c Manhattan plot of the ROH status p-values across the genome. The dotted line marks the genome-wide significance threshold for a Bonferroni correction which was based on the effective number of tests when accounting for linkage disequilibrium.