Genetic diversity and differentiation of pelt, mutton and wool sheep breeds of South Africa using genome-wide single nucleotide polymorphisms.

Abstract

Sheep, Ovis aries, are a versatile species that has, over hundreds of years, been adapted to South African environmental conditions resulting in more than 40 breeds that are raised for various objectives and production systems and constituting a population of close to 30 million animals. The South African sheep genetic resource presents unique and distinct phenotypes and genotypes that, put together, contribute to the global biodiversity observed in sheep that ought to be conserved and used for improved human livelihoods and economies. South Africa shares its sheep genetics with the global world, through both exportation and importation of germplasm. The broad objective of the study was to profile the genomic architecture of South African sheep populations to provide information for optimal utilization, conservation and improvement. Four hundred South African sheep belonging to 13 breeds of mutton, wool, dual purpose (mutton and wool), pelt and uncharacterised non-descript indigenous sheep were sampled and genotyped. In addition, 623 genotypes from the International Sheep Genomics Consortium representing European, Asian, African sheep breeds were subsampled. A series of statistical genomic analyses were pursued. In Chapter 3, genetic diversity, population genetic structure and divergence between South African sheep breeds was investigated using the OvineSNP50 Beadchip. A total of 400 sheep belonging to 13 breeds representing mutton, pelt and mutton and wool dual-purpose breeds and Nguni sheep as a representative of indigenous non-descript genotypes were genotyped. To gain a clearer understanding of the genetic diversity of South African breeds relative to other breeds, 623 genotypes from six African, two Asian and eight European breeds were included in the analyses. The study demonstrated low genetic diversity (HO ≤ 0.27) in small and geographically restricted populations of Namaqua Afrikaner; Nguni, and Blackhead Persian relative to moderate to high diversity (HO ≥ 0.38) in Merino and Merino-derived commercial breeds (i.e. Dohne Merino, Australian Merino and Chinese Merino). Overall, the African and Asian populations were the most inbred populations with FIS ranging from 0.17 ± 0.05 in Grey Swakara and Ronderib Afrikaner sheep to 0.34 ± 0.07 in the Namaqua Afrikaner. Principal component analysis separated the fat-tailed sheep (i.e. Swakaras, Nguni, Black Head Persian, Ethiopian Menzi, Meatmaster) from the rump-tailed sheep of Merino and Dorset Horn etc., as well as according to breed history and production systems. Similarly, ADMIXTUREbased clustering revealed various sources of within- and amongst-breed genomic variation associated with production purpose, adaptation and history of the breeds. An analysis of FSTv based breed differentiating SNPs suggested selection and population divergence on genomic regions associated with growth, adaptation and reproduction. Overall, the analysis gave insight into the current status of the sheep genetic resources of South Africa relative to the global sheep population highlighting both genetic similarities as well as divergence associated with production system and geographical distribution and local adaptation. The second set of analyses (Chapter 4) focused on assessing the genetic diversity, population structure and breed divergence in 279 animals including the three Merino-derived breeds and five presumed ancestral populations of Merinos and non-Merino founding breeds of Damara, Ronderib Afrikaner and Nguni. Highest genetic diversity values were observed in Dohne Merino with Ho = 0.39 ± 0.01 followed by Meatmaster and South African Merino with Ho = 0.37 ± 0.03. The level of inbreeding ranged from 0.0 ± 0.02 (Dohne Merino) to 0.27 ± 0.05 (Nguni). Analysis of Molecular Variance (AMOVA) showed high within population variance (>80 %) across all population categories. The first Principal Component (PC1) separated the Merino, South African Mutton Merino (SAMM), Dohne Merino and Afrino from the Meatmaster, Damara, Nguni and Ronderib Afrikaner. PC2 aligned each Merino derived breed with its presumed ancestors and separated the SAMM from the Merino and SA Merino. Within population selection based on |iHS| indices yielded selection sweeps across the AFR (12 sweeps), Meatmaster (4 sweeps) and Dohne Merino (29 sweeps). Genes associated with hair/wool traits such as FGF12, metabolic genes of ICA1, NXPH1 and GPR171 and immune response genes of IL22 IL26, IFNAR1 and IL10RB were reported. Other genes included HMGA which was observed as a selection signature in other populations, WNT5A important in the development of the skeleton and mammary glands, ANTXR2 associated with adaptation to variation in climatic conditions and BMP2 which has been reported as strongly selected in both fat-tailed and thin-tailed sheep. Using the Rsb analysis for selection sweeps, the Dohne Merino vs SAMM shared all six sweeps regions on chromosomes 1, 10 and 11 with the comparison for Afrino vs SAMM. Genes such as FGF12 on OAR 1:191,3-194,7Mb and MAP2K4 on OAR11:28,6-31,3Mb were observed. The selection sweep on chromosome 10 region 28,6-30,3 Mb, harbouring the RXFP2 for polledness, was shared between Dohne Merino vs Merino, Meatmaster vs Merino and Meatmaster vs Nguni. The Dohne Merino vs Merino and the Meatmaster vs Merino also shared an Rsb-based selection sweep on chromosome 1 region 268,5 - 269,9 Mb associated with the Calpain gene, CAPN7. The study demonstrated some genetic similarities between the Merino and Merino-derived breeds emanating from common founding populations as well as some divergence driven by breed-specific selection goals.

Chapter 5 tested the hypothesis that production systems geared towards specific traits of importance or natural or artificial selection pressures influenced the occurrence and distribution of runs of homozygosity (ROH) in the South African sheep population. The ROH were screened and their distribution within chromosomes and between breeds were analysed to assess breed history and associated selected pressures. ROH were computed at cut-offs of 1-6 Mb, 6-12 Mb, 12-24 Mb, 24-48 Mb and >48 Mb. Analysis of the distribution of ROH according to their size showed that, for all breeds, the majority of the detected ROH were in the short (1- 6 Mb) category (88 %). Most animals had no ROH >48 Mb. Of the South African breeds, the Nguni and the Blackhead Persian displayed high ROH based inbreeding (FROH) of 0.31 ± 0.05 and 0.31 ± 0.04, respectively. Highest incidence of common ROH per SNP across breeds was observed on chromosome 10 with over 250 incidences of common ROHs. Mean proportion of SNPs per breed per ROH islands ranged from 0.02 ± 0.15 (island ROH224 on chromosome 23) to 0.13 ± 0.29 (island ROH175 on chromosome 15). Seventeen of the islands had SNPs observed in single populations (unique ROH islands). The MacArthur Merino population had five unique ROH islands followed by Blackhead Persian and Nguni with three each whilst the South African Mutton Merino, SA Merino, White Vital Swakara, Karakul, Dorset Horn and Chinese Merino each had one unique ROH island. Genes within ROH islands were predominantly associated with metabolic and immune response traits and predomestic selection for traits such as presence or absence of horns. In line with observations in Chapter 3, the frequency and patterns of distribution of ROH observed in this study corresponded to the breed history and implied selection pressures exposed to the sheep populations under study. Chapter 6 investigated (i) LD between adjacent SNPs, (ii) LD decay with increased marker distance, (iii) trends in effective population size over time and (iv) consistency of gametic phase in 13 South African sheep breeds South African Merino (n = 56), Merino (n =10); Mutton Merino (n = 10), Dohne Merino (n = 50), Meatmaster (n = 48), Blackhead Persian (n =14) and Namaqua Afrikaner (n = 12), the four pelt-colour based Swakara subpopulations of Grey (n = 22); Black (n = 16); White-vital (n = 41) and White-subvital (n =17) Dorper (n = 23); Afrino (n = 51) and unimproved Nguni sheep (n = 30). Linkage disequilibrium (r2) averaged 0.16 ± 0.021and ranged from 0.09 ± 0.14 and 0.09 ± 0.13 observed in the SA Merino and Dohne Merino respectively to 0.28 ± 0.29 observed in the Blackhead Persian sheep. Chromosome 10 had the highest LD with r2 values ranging from 0.10 ± 0.15 (SA Merino) and 0.12 ± 0.18 (Dohne Merino) to 0.28 ± 0.30 in Blackhead Persian and 0.29 ± 0.30 (SA Mutton Merino). Across the 14 breeds, LD decayed from 0.27 ± 0.30 at 0-10Kb window to 0.02 ± 0.03 at 1000-

2000 Kb window. A progressive decrease in Ne across generations across all populations was observed with effective population size of <500 for all the populations 66 generations ago decreasing to <250, 23 generations ago and well below 100, 13 generations ago. Highest correlations in gametic phase were observed within the 0-10kb window between pairs of Merino and Merino-derived breeds. The highest correlation observed with Nguni sheep was with Dorper sheep (0.33) within the 0-10kb window, which was similar to that observed with Blackhead Persian sheep and Dorper (0.32) again within the same window. The study reported considerable LD persistent over short distance in the South African sheep breeds. The implications of the observed LD, LD decay and consistency in gamete phase on applications such as GWAS, QTL mapping and GS were discussed. It was concluded that the South African sheep population is highly diverse with that diversity found both within and between populations. Genetic differences between fat tailed sheep population, Merino type breeds and the English Dorset were demonstrated as well as low levels of genetic diversity in small and indigenous breeds such as the Nguni, Namaqua Afrikaner and Blackhead Persian. The frequency and patterns of distribution of ROH observed in this study corresponded to the breed history and implied selection pressures exposed to the sheep populations under study. The utility of the OvineSNP50 Beadchip as a genomic tool for the South African Sheep population was also demonstrated. Keywords: Ovis aries; SNP data; genomic structure; production system; selection signatures; ROH