Location: Animal Genomics and Improvement LaboratoryTitle: Single-step genomic evaluation of crossbreed dairy cattle in the US
|CESARANI, ALBERTO - University Of Georgia|
|LOURENCO, DANIELA - University Of Georgia|
|TSURUTA, SHOGO - University Of Georgia|
|LEGARRA, ANDRES - Institut National De La Recherche Agronomique (INRA)|
|NICOLAZZI, EZEQUIEL - Council On Dairy Cattle Breeding|
|MISZTAL, IGNANCY - University Of Georgia|
Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: 3/10/2022
Publication Date: 6/19/2022
Citation: Cesarani, A., Lourenco, D., Tsuruta, S., Legarra, A., Nicolazzi, E., Van Raden, P.M., Misztal, I. 2022. Single-step genomic evaluation of crossbreed dairy cattle in the US [abstract]. Journal of Dairy Science. 105(Suppl. 1):45(abstr. 1117).
Technical Abstract: Dairy cattle evaluations are progressively moving to single-step GBLUP (ssGBLUP). We recently investigated the performance of ssGBLUP for US Holstein and multibreed genomic evaluations in the US. However, the latter comprised data from Ayrshire, Brown Swiss, Guernsey, Holstein (HO), and Jersey (JE) without crossbred data. Because many crossbred animals have been genotyped, the focus is now on including this information in genomic evaluations. Thus, this work aimed to explore a ssGBLUP model considering purebred and crossbreed data from the Council of Dairy Cattle Breeding (CDCB). Only phenotypes and genotypes of purebred Holstein, Jersey, and their crosses were considered. A total of 49M records (of which 580k were from crosses) of complete lactations with milk, fat, and protein yields were obtained from January 2000 to August 2021; incomplete lactations by August 2021 were projected. The pedigree contained 89M animals; genotypes at 79,294 selected SNPs, which were imputed within each breed, were available for about 5M animals, of which 4.48M, 589k, and 39k were pure HO, pure JE, and crosses, respectively. Predictive abilities of BLUP and ssGBLUP were evaluated based on two different runs: i) FULL, with all data in the model; ii) TRUNCATED, in which the last 4 years of data were removed from the model. Validation for cows was based on correlations between adjusted phenotypes (FULL) and (G)EBV (TRUNCATED), whereas for bulls on the regression of daughter yield deviations (DYD, in FULL) on (G)EBV (TRUNCATED). Validation animals were divided into purebreds and crossbreds, and the latter were split into groups depending on the breed proportion. Predictivities for purebreds were compared with those from single- and multibreed models. Predictivity from ssGBLUP was similar between purebred and multi-breed evaluations in earlier studies; therefore, including genotypes for crossbred animals should not undermine genomic predictions for purebred animals and should provide more accurate GEBV for crossbreds than the calculations based on breed proportion.