Assessment of genetic diversity in local breeds

The present study aimed at assessing genetic diversity in selected local livestock breeds in Northern Germany, and furthermore, examining some of the indices used in describing diversity in populations. The study begun with an assessment of genetic diversity in the German Angler (RVA) and Red-and-white dual-purpose (RDN) cattle breeds. Analyses of the RVA and RDN pedigrees, consisting of 93,078 and 184,358 records, respectively, revealed a number of Red Holstein (RH) sires that were influential in these breeds. Even when the pedigrees were combined and analysed, four RH individuals alone had a cumulative marginal genetic contribution of over 9% and were at the top of the list. The results also revealed low levels of classical inbreeding for RVA (1.39%) and RDN (0.41%) with moderately high effective population size (Ne) estimates of 156 and 170, respectively. However, using purging-based approaches such as the ancestral history coefficient, the levels of inbreeding rose slightly and the corresponding Ne values were reduced to 50 and 53 for RVA and RDN, respectively. Additionally, estimates of effective number of founders and ancestors indicated an unbalanced genetic contribution in the founder populations, where some individuals had many more offspring than others. Notably, the quality of pedigrees used for the analyses was low with mean number of complete generation equivalent of 5.59 and 2.7 for RVA and RDN, respectively.
A follow-up study was performed to assess genetic diversity at the genome level, investigate gene flow events, and to shed light on the usefulness of runs of homozygosity (ROH) as a means to estimate inbreeding. The Illumina BovineSNP50 BeadChip information on RVA, RDN, RH and 11 outgroup breeds were made available for the different analyses. For RVA, RDN and RH, observed heterozygosity (Ho) estimates were 0.374, 0.356 and 0.363, and the estimates of ROH-based genomic inbreeding (FROH) were 0.021, 0.045 and 0.053, respectively. Linkage disequilibrium-based Ne estimate for five generation ago was highest in RVA (113) and lowest in RDN (67). Population structure analyses found a high level of admixture between RVA and RH than between RDN and RH, and for the first time, connectedness between RVA and Norwegian Red cattle was uncovered. Furthermore, the investigation showed moderate to high positive correlations between FROH and pedigree-based inbreeding (FPED), and even when FPED was zero, traces of inbreeding were detected by the ROH-based measure. A third investigation was launched to analyse genetic diversity in the German White-headed mutton (GWM) sheep using both pedigree (N = 19,000) and genotype (N = 46) data. Additionally, different measures of ROH inbreeding were investigated. From the pedigree analysis, average generation interval was 3.24. Estimate of average FPED and Ne for the whole population were 0.01 and 132, respectively, and differed from the estimates computed for a subpopulation of animal born in the last four years of pedigree data, which were 0.04 and 99, respectively. Only 59 ancestors explained 50% of variation in the breed. Genome-based Ho and Ne were 0.389 and around 50, respectively. Based on 16,852 single nucleotide polymorphism (SNP) markers, three separate population structure analyses pointed to the flow of genes from other breeds, especially Berrichone du Cher to GWM. To investigate different ROH inbreeding measures, FROH was calculated based on either ROH length or on the number of SNPs in ROH. Inbreeding estimates derived from the two measures were 0.045 and 0.046 and highly correlated (rp = 0.99). Furthermore, total FROH was effectively partitioned into values for specific chromosomes. In a final investigation, the outcomes of two ROH detection software including PLINK and RZOOROH were compared to address the lack of consensus on the choice of parameter thresholds in calling ROH across studies. With array data on 52 Mangalitza pigs and whole genome sequence (WGS) data on six Schleswig Draught horses, the analyses showed comparable estimates of FROH across different models and detection software, although the number of ROH detected varied greatly. In the horse, an array-like dataset consisting of 26,932 SNPs was compared to the WGS data and the latter proved to be more efficient in detecting small size ROH.