Differences in estimates of genomic and conventional inbreeding: Reasons and Implications

Genomic inbreeding coefficients, when calculated in a way to reflect the true level of homozygosity of an individual at genomic markers, result in considerably higher inbreeding trends per year at a population level when compared with the pedigree-based estimates. Using data from dual-purpose Fleckvieh cattle, we compare inbreeding trends and estimates of inbreeding-effective population sizes calculated from genomic data to those using a conventional pedigree-based approach. By repeating the calculation of genomic inbreeding trends using different subsets of SNPs, we illustrate how genomic inbreeding coefficients are shaped by the effects of trends in allele frequencies over time. Three SNP subsets were defined according to specific patterns of allele frequency changes over time: SNPs whose allele frequencies move toward intermediate frequencies, SNPs whose allele frequencies move away from intermediate frequencies and SNPs that show no directional allele frequency trends at all ('neutral'). As expected, genomic inbreeding trends based on subsets of 'neutral' SNPs only show the closest agreement with the pedigree-based estimates which highlights the fundamental theoretical difference between both approaches. This demonstrates that the validity of the overall estimate using all SNPs depends on their distribution across the genome. It therefore represents some kind of 'average' of effects caused by the balance of SNPs with differential allele frequency changes. Genomic estimates therefore can provide the basis to monitor both the distribution, amount and progress of neutral sites in the genome as well as the changes caused by drift and strong artificial selection at other sites. As a conclusion, we argue against the use of traditional probabilistic measures of inbreeding and in favor of the development and use of approaches capable of reflecting the complexity of various processes at the genome level, given that a sufficient and representative part of the breeding-population is genotyped for at least a decade.