Genetic evaluation of udder health traits in Spanish Holstein cows

The aims of this study were to estimate the genetic parameters of clinical mastitis (CM) and SCS traits, and to compare the performance of genetic evaluations of CM traits using univariate and bivariate analyses (CM-SCS). Data were edited according to the Udder Health Golden Standard harmonization, and then 6 CM traits and 6 SCS traits were considered, as the result of combining 3 lactation classifications (1, 2, ≥3) and 2 milking periods (early, late). The linear mixed animal models included the ratio of period at risk as a covariate, herd-year of calving, month of calving, and lactation-age as fixed effects, and the permanent environmental effect for traits of ≥3 lactations. Prevalence of CM in early lactation was similar regardless the lactation number (5%-6%), and the estimated heritabilities were 0.01. Prevalences in late lactation ranged from 10% to 24% and heritabilities ranged from 0.03 to 0.05. Estimated heritabilities of SCS ranged from 0.06 to 0.16 with univariate analyses. Somatic cell count (therefore its log-transformation SCS) showed a higher probability of correctly identifying healthy cows than infected cows but there was still up to 36% of healthy cows for which CM was not detected by SCS. Genetic correlations between CM-SCS traits ranged from 0.36 to 0.95, and SCS in lactation 3 and later did not add extra information to SCS in the second lactation for predicting CM. Regarding reliabilities of EBVs for CM traits, bivariate CM-SCS analyses led to substantial increases with respect to the single-trait model for sires (7%-12% more in first lactation and 16%-28% more for second lactation). Sire's rank correlations for CM between univariate and bivariate analyses (0.47-0.92) suggest that discarding sires could be more accurate than selecting candidates for sires of dams. We can conclude that SCS in first lactation could be useful to supplement CM data in first and second lactations to improve udder health genetic evaluation.