Genomic prediction for commercial body weight using information from fish measured under production conditions - A simulation study

In aquaculture selective breeding programs, growth is typically the primary breeding objective, and selection is often based on weight recorded on candidates under controlled conditions. However, this strategy may be inefficient for improving commercial weight due to differences in rearing conditions that lead to different genetic control of growth-related traits in each environment. To address this, genetic evaluations should incorporate records from sibs of nucleus candidates reared under commercial conditions. The optimal number of sibs required for recording, however, remains unknown. This simulation study, designed to mimic an aquaculture selective breeding program, evaluates how many fish need to be sampled under commercial conditions to accurately estimate breeding values for commercial weight and obtain large selection responses. Nucleus and commercial weight heritabilities were set at 0.30, with genetic correlations of 0.2, 0.5 and 0.8. Both BLUP and genomic BLUP (GBLUP) were evaluated over five generations in a nucleus population of 100 full-sib families with 20 candidates per family. Scenarios with 0–60 sibs recorded (and genotyped for GBLUP) under commercial conditions were compared. Incorporating data from commercial-reared sibs substantially increased genetic gains, with accuracy and response improving as more sibs were added. However, these benefits diminished as the number of sibs continued to increase for both BLUP and GBLUP. Incorporating commercial sib data increased inbreeding under BLUP, though the levels were lower as sib numbers grew. In contrast, inbreeding under GBLUP remained similar to scenarios without sib data whatever the number of sibs measured. The results show that sampling more than 50 sibs per family appears unnecessary, as marginal improvements in genetic progress and inbreeding are unlikely to justify the additional cost within the simulated scenarios.