A genome wide survey of SNP variation reveals the genetic structure of sheep breeds

Abstract

The genetic structure of sheep reflects their domestication and subsequent formation into discrete breeds. Understanding genetic structure is essential for achieving genetic improvement through genome-wide association studies, genomic selection and the dissection of quantitative traits. After identifying the first genome-wide set of SNP for sheep, we report on levels of genetic variability both within and between a diverse sample of ovine populations. Then, using cluster analysis and the partitioning of genetic variation, we demonstrate sheep are characterised by weak phylogeographic structure, overlapping genetic similarity and generally low differentiation which is consistent with their short evolutionary history. The degree of population substructure was, however, sufficient to cluster individuals based on geographic origin and known breed history. Specifically, African and Asian populations clustered separately from breeds of European origin sampled from Australia, New Zealand, Europe and North America. Furthermore, we demonstrate the presence of stratification within some, but not all, ovine breeds. The results emphasize that careful documentation of genetic structure will be an essential prerequisite when mapping the genetic basis of complex traits. Furthermore, the identification of a subset of SNP able to assign individuals into broad groupings demonstrates even a small panel of markers may be suitable for applications such as traceability.

Figure 1. Ovine idiogram showing the distribution of targets used for re-sequencing to identify SNP.

Orange bars represent genomic locations containing SNP which were formatted for array based genotyping. Blue bars represent genomic targets which yielded SNP during re-sequencing which were not included on the array. Black bars represent targets in which no SNP was identified following re-sequencing.

Figure 2. The distribution of genetic distance (D) between pairs of individuals.

D was plotted separately where pairs were drawn from within the same breed (blue bars) from different breeds (red bars) from domestic sheep and one of the two species of wild sheep (O. canadensis or O. dalli, green bars).

Figure 3. Clustering of animals from 22 breeds and two wild species of sheep, based on multidimensional scaling of genetic distance.

The first (C1) and second (C2) dimensions are plotted. Animals drawn from Australia, New Zealand, Europe and North America (blue), Asia (green) and Africa (red) appear clustered according to geographic origin. Wild sheep are shown in black. Populations are represented using different characters as follows: Dorper (▵), Suffolk (+), Blackface (×), Charollais (⋄), German Mountain Brown (▿), Javanese Thin Tail (⊠), Italian Sarda (), Merino (), Poll_Dorset (), Rambouillet (), Red Masai (), Romney (), Soay (), Sumatran Thin Tail (▪) , Texel () , Tibetan (▴), Finsheep (♦), Katahdin (), Romanov (), Namaqua Afrikaner (x̂), Ronderib Afrikaner (□), Composite (⋄), Bighorn (▵) and Thinhorn (x̂).

Figure 4. Clustering of animals from 22 breeds and two wild species of sheep, based on multidimensional scaling of genetic distance.

Individuals are plotted for the first (C1) and third (C3) dimensions. Animals drawn from Asia (green) appear distinct from those drawn from Africa (red) and other countries (blue). Populations are represented using the same characters used in Figure 3 which are as follows: Dorper (▵), Suffolk (+), Blackface (×), Charollais (⋄), German Mountain Brown (▿), Javanese Thin Tail (⊠), Italian Sarda (), Merino (), Poll_Dorset (), Rambouillet (), Red Masai (), Romney (), Soay (), Sumatran Thin Tail (▪) , Texel (•) , Tibetan (▴), Finsheep (♦), Katahdin (•), Romanov (•), Namaqua Afrikaner (x̂), Ronderib Afrikaner (□), Composite (⋄), Bighorn (▵) and Thinhorn (x̂).

Figure 5. Model based clustering of 392 sheep where 3–6 sub-populations (K) were assumed.

The geographic origin of breeds are indicated above the box plot. Individuals are represented in breed groups which are separated by vertical black lines. The breeds are given below the box plot.

Figure 6. Multidimensional scaling plots for four breeds showing the genetic differentiation between geographically distinct sub-populations.

The position of New Zealand (circles) and American (squares) Texels are shown in the top right panel. Two populations of American Suffolk (triangles, circles) are shown with UK Suffolks (squares) in the top left panel. African (triangles) and American (squares) Dorper are plotted in the bottom left panel while Australian Poll Dorset (triangles) and American Dorsets (squares) are shown in the bottom right panel. Note that the scale differs between panels. F _ST was calculated between each sub-population.

Figure 7. Cluster analysis of individuals from European derived breeds (blue triangles), African and Asian breeds (red squares), the Soay breed (blue patterned boxes) and wild sheep (black asterisks).

Multidimensional scaling was performed using marker panels which contained either 96 (panel 1 and 3) or 384 (panel 2 and 4) SNP which were either the top (panel 1 and 2) or bottom (panel 3 and 4) ranked markers for the informativeness metric I _n .