Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Animal Genomics and Improvement Laboratory » Research » Publications at this Location » Publication #412816
Research Project: Increasing Accuracy of Genomic Prediction, Developing Algorithms, Selecting Markers, and Evaluating New Traits to Improve Dairy Cattle

Location: Animal Genomics and Improvement Laboratory

Title: Validation of genomic prediction for economic traits in heifers across five U.S. dairy breeds

Author
item Toghiani, Sajjad
item Vanraden, Paul
item Null, Daniel
item Miles, Asha
item Van Tassell, Curtis - Curt

Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: 3/10/2024
Publication Date: 6/16/2024
Citation: Toghiani, S., Van Raden, P.M., Null, D.J., Miles, A.M., Van Tassell, C.P. 2024. Validation of genomic prediction for economic traits in heifers across five U.S. dairy breeds. Journal of Dairy Science. 107(Suppl. 1):129(abstr. 1513).

Interpretive Summary:

Technical Abstract: While previous studies validating genomic predictions have primarily focused on bulls, most genotypes in the National Cooperator Database now originate from cows. This study paired official within-breed genomic predictions (GPTA) and parent averages (PA) for genotyped heifer calves born between 2019 and 2021 using August 2021 data with their corresponding yield deviations (YD) for 18 traits. YD data became available after the heifers completed their first lactation and was extracted from August 2023 data. Using YD diminishes herd and other nuisance effects, providing a more unbiased assessment of prediction accuracy than validation using raw lactation performance records. The analysis included milk production YD for 340,578 Holsteins (HO), 56,405 Jerseys (JE), 1,436 Brown Swiss (BS), 437 Guernseys (GU), and 105 Ayrshires (AY). However, fewer YD records were available for other traits due to missingness. Reliability or squared correlations (R2) were divided by corresponding trait heritability because only the heritable portion of cow records can be predicted and varied across different traits and breeds. In HO and JE, the predictive ability of GPTA outperformed PA in predicting cow YD for yield, productive life, somatic cell score, fertility, and health traits. The improvement ranged from 33% to 142% compared to PA's predictive ability. However, the results for AY, BS, and GU breeds were less consistent due to the smaller number of genotyped heifers. The gains of R2 in those breeds were smaller and aligned with the published reliabilities of GPTA. The regression results for YD on GPTA and PA traits slightly exceeded the expected value of 2.00 when predicting the future trait YD using GPTA or PA. The larger number of observations and lower standard error for regression coefficient prediction in HO and JE contributed to more stable and consistent regression coefficients for all traits except Milk Fever and Heifer Livability. Our study suggests that herd owners may experience greater benefits from genomics than originally promised.