Estimation & Prediction

Residual effects of genotype call rate on genomic breeding values in a commercial Atlantic salmon population

---

Abstract

Genomic selection (GS) using high‑density SNP arrays is central to genetic improvement in aquaculture. Standard pipelines apply genotype quality control (QC), including a minimum per‑animal call rate, followed by imputation of missing markers. It is often assumed that samples passing a lenient QC threshold (e.g. >90% call rate) are, after imputation, fully informative for genomic evaluation. However, the residual impact of pre‑imputation sample quality on the distribution of genomic breeding values (GEBVs) is poorly characterized. This study quantified the relationship between pre‑imputation call rate and GEBV variance for harvest body weight (BW), pigment (astaxanthin, mg/kg fillet), and cardiomyopathy syndrome (CMS) resistance in a commercial Atlantic salmon (Salmo salar) population.Genotypes from 48,645 Atlantic salmon from the 2023 smolt year class of the Mowi Norway breeding program were analysed. All animals were genotyped on a 68,582‑marker custom SNP array and imputed to 100% using Beagle 5.4. We considered three scenarios: (1) the full observed dataset, retaining samples with call rate ≥90%; (2) a high‑quality subset (HQ; call rate ≥99.5% for candidates and ≥95% for informants); and (3) a masked sensitivity trial, created by randomly masking up to 10% additional markers in HQ genotypes and re‑imputing, to separate effects of missingness from allelic noise. Multi‑trait GBLUP was used to estimate GEBVs, and the relationship between original call rate, GEBV variance, and Mendelian error rate (MER) was evaluated.A strong, systematic relationship was observed between original call rate and the distribution of GEBVs. Animals with call rate >97% displayed the full range of GEBVs, whereas animals between 90% and 96% call rate showed substantially reduced GEBV variance, and this pattern was consistent for BW, pigment, and CMS. Poor call rate continued to affect these individuals after imputation, as indicated by both lower GEBV variance and a marked increase in MER. In contrast, the masked HQ trial retained stable GEBV variance and MER across the same range of simulated missingness, demonstrating that imputation per se did not cause variance shrinkage.We conclude that low call rate has a persistent, negative residual effect on GEBV variance, even after imputation, primarily via allelic noise. Lenient call‑rate QC introduces a non‑genetic source of bias where an animal's probability of selection depends on genotype quality rather than true genetic merit.

Keywords: 2026

How to Cite:

Brand, W., Boison, S., Sae-Lim, P. & Baranski, M., (2026) “Residual effects of genotype call rate on genomic breeding values in a commercial Atlantic salmon population”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286377. doi: https://doi.org/10.31274/wcgalp.23968

Rights: 1