Effects of runs of homozygosity-based versus genomic relationship matrix-based future inbreeding penalties on genetic gain, variance components, and inbreeding depression in dairy cattle

Managing inbreeding while maintaining genetic gain is a critical challenge in dairy cattle breeding programs. Current North American genetic evaluations adjust breeding values for expected future inbreeding using genomic relationship matrix (GRM) approaches, but runs of homozygosity (ROH)-based methods may offer advantages for detecting recent autozygosity. This study compared ROH-based versus GRM-based inbreeding penalties across 20 generations of simulated genomic selection. Using AlphaSimR (version 1.5.3 in R v 4.3.1), we simulated dairy cattle populations under 3 genetic scenarios: additive only, low nonadditive effects σD2σD2σA2σA2=0.10, where σD2 is the dominance variance and σA2 is the additive variance, inbreeding depression = 0.30 σG, where σG is the genetic SD, per 10% inbreeding delta [ΔF]), and high nonadditive effects σD2σD2σA2σA2=0.50, inbreeding depression = 1.20 σG per 10% ΔF). Each penalty method was tested at 3 intensities (0.5 ×, 1 ×, 2 × the inbreeding depression parameter) with 10 replicates per scenario-treatment combination. The GRM-based penalties achieved lower inbreeding accumulation than ROH-based penalties, with differences of 0.13 to 0.22 in genomic F at matched penalty intensities. However, this enhanced inbreeding control was associated with reduced genetic gain, with GRM-based selection achieving 0.6 to 1.7 genetic SD less gain than ROH-based selection. The GRM-based selection retained more additive variance (66%–107%) compared with ROH-based selection (63%–74%), with values exceeding 100% indicating partial recovery of variance depleted during burn-in. Dominance variance retention was also greater for GRM-based selection (105%–114% vs. 93%–98%), reflecting maintained heterozygosity. Interestingly, GRM-based selection accumulated more inbreeding depression than ROH-based approaches despite lower inbreeding levels, potentially reflecting more effective purging under the stronger directional selection maintained by ROH-based methods. Results illustrate trade-offs between inbreeding control and genetic gain, with optimal strategy depending on breeding program objectives and trait genetic architecture.