Global breeding programs

An integrated pedigree-genomic simulation approach for assessing the robustness of genetic evaluations in Alpine cattle breeds

---

Abstract

Genetic evaluations in local cattle populations can be sensitive to limitations in pedigree recording, herd connectedness, and uneven genomic data availability. The Valdostana breed consists of three closely related subpopulations, Aosta Black Pied (ABP), Aosta Chestnut-Hérens (ACH), and Aosta Red Pied (ARP), with ABP and ACH jointly analysed due to their comparable genetic background. This study integrates real data and controlled simulations to evaluate the performance of the LR (linear regression) method under different conditions of pedigree accuracy, genomic reliability, and population structure. A complete pedigree of approximately 150,000 real animals was used to reconstruct the genealogical structure of the breed in AlphaSimR. Based on this structure, we simulated a genome comprising 29 autosomes, ~100,000 SNPs, and 10,000 QTL, assuming a moderately heritable trait (h²≈0.40). True breeding values (TBV), simulated genotypes, and phenotypes were generated under additive and residual variances of 0.40 and 0.60. Three pedigree scenarios were defined: (i) a baseline with a fully correct pedigree (P0); (ii) a scenario in which 20% of animals with known fathers were randomly assigned an incorrect sire (P20); and (iii) a more severe scenario where 40% received an incorrect sire (P40), while dams remained correct in all cases. Genomic scenarios explored genotype quality: (i) real genotypes available for the subpopulations (Greal); (ii) simulated genotypes without errors (Gsim); and (iii) simulated genotypes including a controlled proportion of random errors or missingness (Gerr). Scenarios were evaluated individually and in combination to assess the sensitivity of the LR method to different sources of information loss. Standardized 305-day milk yields were derived from test-day data; after aggregation, no permanent environmental effect was required. Genetic evaluations were performed with a univariate animal model including herd-year and age-at-calving. Heritability estimates from real data ranged between 0.25 and 0.30. In the simulated population, P0 produced high correlations between EBV and TBV (≈0.75-0.82) with negligible bias. Accuracy decreased under P20 (≈0.65-0.72) and more markedly under P40 (≈0.45-0.55), accompanied by increasing positive bias and reduced additive variance. Genomic scenarios followed similar trends: Gsim maintained high accuracy, whereas Gerr showed reductions consistent with moderate levels of genotype errors. Comparisons with real data indicated that ARP behaved similarly to P0-P20/Gsim, whereas ABP-ACH showed patterns closer to P20/Gerr, suggesting higher sensitivity to incomplete pedigree or genomic information. Overall, this integrated pedigree-genomic simulation framework proved effective for assessing the robustness of genetic evaluations in local cattle populations. The LR method consistently detected decreases in accuracy and emerging biases, emphasizing the importance of considering both pedigree and genomic data together to improve reliability in genetic assessments.Research supported by DUALBREEDING - PSRN 2014/2020 and by the EU-Next Generation EU, PRIN 2022 "GENO-VAL" (CUP G53D23004040006).

Keywords: 2026

How to Cite:

Gomez Proto, G., Mancin, E., Bernini, F., Oian, A., Rulli, E., Strillacci, M., Bagnato, A., Sartori, C. & Mantovani, R., (2026) “An integrated pedigree-genomic simulation approach for assessing the robustness of genetic evaluations in Alpine cattle breeds”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286342. doi: https://doi.org/10.31274/wcgalp.23939

Rights: 1