Genetic gain & Inbreeding

Use of Genomic Relationship Matrices to manage Genetic Diversity

---

Abstract

Genomic selection has increased the rate of inbreeding in livestock populations, raising concerns about long-term genetic diversity and adaptability. To address this, we evaluated the performance of different genomic relationship matrices in optimal contribution selection (OCS) schemes. Using stochastic simulation, we generated 50 replicates of a population with 10 generations of OCS. Inbreeding was constrained using genomic relationship matrices based on either genomic covariance (GVR), runs-of-homozygosity (GROH), linkage analysis (GFGLA), or true identity-by-descent (GIBD). The availability of true IBD allowed us to assess both the accuracy of coancestry estimates and the efficacy of inbreeding management. Changes to allele frequency were assessed by comparing drift- and homozygosity-based inbreeding. Under Hardy-Weinberg expectations, these measures are expected to be equal at neutral loci, and a divergence indicates that the inbreeding management itself alters allele frequencies, thus violating the assumption of neutrality. We randomly sampled 50 000 SNPs, which are used for management, along with 10 000 QTLs and 10 000 neutral loci, both masked. At this SNP density, some LD will occur between markers and neutral loci, meaning neutrality refers to a marker not being directly part of selection.. GFGLA showed the highest concordance with true IBD for both inbreeding and coancestry estimates, with a concordance correlation coefficient (CCC) of 0.9949 for inbreeding and 0.9988 for coancestry. It also provided the most accurate estimate of the rate of inbreeding, matching the IBD-based rate. GVR had high concordance for coancestry (CCC = 0.9833) but showed greater variance in inbreeding estimates (CCC = 0.8838). GROH performance was highly sensitive to the minimum ROH length: GROH with ROH >1cM overestimated inbreeding (CCC = 0.1327), while GROH with ROH >8cM (CCC = 0.9670) provided more accurate estimates.Simulation results showed that GFGLA was the only method to stay within the inbreeding target while minimizing frequency changes at neutral loci. GFGLA achieved the highest genetic gain per unit of inbreeding, with a slope of 98.83, compared to 95.95 for GVR, 94.72 for GIBD, and 92.86 for GROH >5cM as the best GROH scheme. GFGLA also preserved the most genic variance and maintained neutrality of allele frequency changes, with no significant difference between drift- and homozygosity-based inbreeding. In contrast, management using GROH and GVR caused directional allele frequency changes at neutral loci. In conclusion, GFGLA, provided a sufficient number of genotyped generations, is a robust and accurate approach for estimating IBD and managing genetic diversity. GVR offers high genetic gain but does not effectively manage the increase of homozygosity. GROH matrices require careful calibration to truly estimate IBD, as their performance depends heavily on ROH length thresholds. These findings further highlight the importance of selecting relationship matrices that reflect IBD to maintain genetic uniqueness in livestock populations.

Keywords: 2026

How to Cite:

Wæge, O., Yu, X., Berg, P. & Meuwissen, T., (2026) “Use of Genomic Relationship Matrices to manage Genetic Diversity”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2284561. doi: https://doi.org/10.31274/wcgalp.23591

Rights: 1