Differing genetic trend estimates from traditional and genomic evaluations for genotyped animals as evidence of pre-selection bias in US Holsteins

Y. Masuda*1, P.M.VanRaden, I. Misztal*, and T.J. Lawlor

*Department of Animal and Dairy Science, University of Georgia, Athens, GA 30602
Animal Genomics and Improvement Laboratory, USDA-ARS, Beltsville, MD 20705
Holstein Association Inc., Brattleboro, VT 05301

2017 J. Dairy Sci. (?)
© American Dairy Science Association, 2017. All rights reserved.
Individuals may download, store, or print single copies solely for personal use.
Do not share personal accounts or passwords for the purposes of disseminating this article.


The objective of this study was to compare genetic trends from a single-step genomic BLUP (ssGBLUP) and the traditional BLUP models for milk production traits in US Holstein. Phenotypes were 305-day milk, fat, and protein yield from 21,527,040 cows recorded between January, 1990 and August, 2015. The pedigree file included 29,651,623 animals limited to 3 generations back from recorded or genotyped animals. Genotypes for 764,029 animals were utilized, and analyzes were by a three-trait repeatability model as used in the US official genetic evaluation. Unknown parent groups were incorporated into the inverse of a relationship matrix (H-1 in ssGBLUP and A-1 in BLUP) with the QP-transformation. In ssGBLUP, 18,359 genotyped animals were randomly chosen as core animals to calculate the inverse of genomic relationship matrix with the APY algorithm. Computations with BLUP took 6.5 hours and 1.4 GB of memory, and computations with ssGBLUP took 13 hours and 115 GB of memory. For genotyped sires with at least 10 daughters, the annual trends for PTA and GPTA were similar up to 2008, with a higher trend for ssGBLUP later (approximately by 36 kg milk, 2.1 kg fat and 1.1 kg protein for bulls born in 2010). For genotyped cows, the trends were similar up to 2006, with a higher trend for ssGBLUP later (approximately by 91 kg milk, 3.6 kg fat and 2.7 kg protein for cows born in 2012). For all cows, the trends were slightly higher for ssGBLUP, with much smaller differences than for genotyped cows. Trends for BLUP indicate bias due to genomic pre-selection for genotyped sires and cows. The single-step genomic BLUP seems to at least partially account for that bias and is computationally suitable for national evaluations.

Keywords: predicted transmitting ability, single-step approach, underestimation, unknown parent group