Investigation of Regional Bias in Dairy Cattle Genomic Evaluations

The objective of this study was to determine if important amounts of bias exist in genomic evaluations of dairy cattle in various regions of the United States. Genomic and traditional data was collected from three regions, Maddox Dairy in Riverdale, CA (West Coast), Iowa State University (Midwest) and Pennsylvania State University (East Coast). Genomic samples of 936 registered 1st lactation fresh Holstein heifers from Maddox Dairy in Riverdale, CA were collected using the BoveSNP50 Bead Chip. Iowa State University (ISU) supplied genomic, parental average and PTA results from 334 Holstein cows whose genomic samples were collected using the BoveSNP50 Bead Chip. Pennsylvania State University (PSU) supplied 2 data sets, one of 68 Holstein cows using the BovLD genomic test and another set of 210 Holsteins collected using BoveSNP50 Bead Chip. The traits evaluated in this study were net merit, productive life, somatic cell score, daughter pregnancy rate, milk, fat, fat%, protein, protein %, daughter calving ease, daughter still birth, predicted transmitting ability type and total performance index for all dairy farms. Genomic and traditional reliabilities for each farm were also given. A simple regression of genomic PTAs on PA and/or traditional PTA were calculated for each trait. Genomic evaluations were regressed against their PTA’s computed from PA or traditional PTA. The expected value for that regression was 1 and any deviation from 1 concluded that bias existed. In the end, Maddox Dairy expressed the most extreme deviations from 1 and ISU expressed the smallest deviations from 1. Current genomic evaluations may have been biased for cattle raised in western environment. Bias may be a result of epigenetic defects, G X E interaction or non-stable LD