Genomic Characterisation of the Indigenous Irish Kerry Cattle Breed

Abstract

Kerry cattle are an endangered landrace heritage breed of cultural importance to Ireland. In the present study we have used genome-wide SNP array data to evaluate genomic diversity within the Kerry population and between Kerry cattle and other European breeds. Patterns of genetic differentiation and gene flow among breeds using phylogenetic trees with ancestry graphs highlighted historical gene flow from the British Shorthorn breed into the ancestral population of modern Kerry cattle. Principal component analysis (PCA) and genetic clustering emphasised the genetic distinctiveness of Kerry cattle relative to comparator British and European cattle breeds. Modelling of genetic effective population size (N_e) revealed a demographic trend of diminishing N_e over time and that recent estimated N_e values for the Kerry breed may be less than the threshold for sustainable genetic conservation. In addition, analysis of genome-wide autozygosity (F_ROH) showed that genomic inbreeding has increased significantly during the 20 years between 1992 and 2012. Finally, signatures of selection revealed genomic regions subject to natural and artificial selection as Kerry cattle adapted to the climate, physical geography and agro-ecology of southwest Ireland.

Keywords: cattle; conservation genomics; endangered breed; genetic diversity; inbreeding; population genomics; selection signature; single nucleotide polymorphism.

Figure 1

Photograph of a Kerry cow and locations of Kerry cattle herd DNA sampling in southern Ireland. The area of each circle corresponds to the size of each population sample. Dark green = animals sampled during 1991/92 (KY92); light green = animals sampled during 2011/12 (KY12). Kerry cow image is copyright of the Kerry Cattle Society Ltd.

Figure 2

Maximum likelihood (ML) phylogenetic tree network graph with five migration edges (m = 5) generated for genome-wide SNP data (36,000 autosomal SNPs) from European cattle breeds (EU data set). The West African taurine N'Dama breed sampled in Guinea is included as a population outgroup. Coloured lines and arrows show migration edges that model gene flow between lineages with different migration weights represented by the colour gradient.

Figure 3

Maximum likelihood (ML) phylogenetic tree network graph with five migration edges (m = 5) generated for genome-wide SNP data (37,490 autosomal SNPs) from cattle breeds of British and Irish origin (BI data set). The West African taurine N'Dama breed sampled in Guinea is included as a population outgroup. Coloured lines and arrows show migration edges that model gene flow between lineages with different migration weights represented by the colour gradient.

Figure 4

Principal component analysis plot constructed for PC1 and PC2 from genome-wide SNP data (36,621 autosomal SNPs) for the EU data set of 605 individual animals. The smaller histogram plot shows the relative variance contributions for the first 10 PCs and PC1 and PC2 account for 18.2% and 16.8% of the total variation for PC1–10, respectively.

Figure 5

Principal component analysis plot constructed for PC1 and PC2 from genome-wide SNP data (37,395 autosomal SNPs) for the BI data set of 351 individual animals. The smaller histogram plot shows the relative variance contributions for the first 10 PCs and PC1 and PC2 account for 23.7% and 22.7% of the total variation for PC1–10, respectively.

Figure 6

Hierarchical clustering of individual animals using genome-wide SNP data (36,621 autosomal SNPs) for the EU data set of 605 individual animals. Results are shown for modelled ancestral populations K = 2–14. The cluster numbers corresponding to the likely number of ancestral populations are highlighted with a light red overlay and the two outlier Kerry samples (KY12_06 and KY12_58) are indicated with red arrows.

Figure 7

Hierarchical clustering of individual animals using genome-wide SNP data (37,395 autosomal SNPs) for the BI data set of 351 individual animals. Results are shown for modelled ancestral populations K = 2–9. The cluster numbers corresponding to the likely number of ancestral populations are highlighted with a light red overlay and the two outlier Kerry samples (KY12_06 and KY12_58) are indicated with red arrows.

Figure 8

Genetic effective population size (N_e) trends modelled using genome-wide SNP data. Results for the KY92 and KY12 populations are shown with seven comparator heritage and production cattle breeds.

Figure 9

Tukey box plots showing the distributions of F_ROH values estimated with genome-wide SNP data for the KY92 and KY12 populations and nine comparator heritage and production cattle breeds.

Figure 10

Manhattan plots of composite selection signal (CSS) results for Kerry cattle (n = 98) contrasted with EU cattle (n = 102). (A) Unsmoothed results. (B) Smoothed results obtained by averaging CSS of SNPs within each 1 Mb window. Red dotted line on each plot denotes the genome-wide 0.1% threshold for the empirical CSS scores. Red vertical arrows indicate selection peaks detected on BTA09, BTA12, BTA16, BTA17, BTA19, and BTA28.