|MOTA, RODRIGO - Council On Dairy Cattle Breeding|
|CESARANI, ALBERTO - University Of Georgia|
Submitted to: World Congress of Genetics Applied in Livestock Production
Publication Type: Proceedings
Publication Acceptance Date: 2/18/2022
Publication Date: 7/7/2022
Citation: Mota, R.R., Cesarani, A., Van Raden, P.M. 2022. Comparison of single-step and multi-step evaluations for U.S. milk, fat, and protein. World Congress of Genetics Applied in Livestock Production. Front Matter, Commun. 323, pp. 1356-1359. https://doi.org/10.3920/978-90-8686-940-4_323.
Technical Abstract: Predictive abilities of single-step (SS) and multi-step (MS) evaluations were compared using current and truncated domestic phenotypes for U.S. yield traits. Accuracy from SS was several percentage points higher than MS to predict either MS or SS current genomic estimated breeding values (GEBV) of 2,619 Holstein and 345 Jersey bulls from 4-year earlier data. Regressions for GEBV were close to 1.0 and similar for MS and SS for all traits and both breeds. Accuracy advantages of SS have increased since a previous study, perhaps indicating more bias from genomic preselection in MS. The MS methods put too much weight on pedigree evaluation, whereas the SS model had no polygenic effect and was more stable because new phenotypes have little influence on marker effects. Further testing will include more traits, foreign data from Interbull, and crossbreds to determine the potential to use SS methods.