Unknown parent and contemporary groups for genetic evaluation of Brazilian Holstein using autoregressive test-day models

Abstract

Test-day milk, fat and protein yields and somatic cell score (SCS) of Brazilian Holstein cattle were used to study the effect of fixed (HTDF) vs random (HTDR) contemporary groups (herd-test-date) with or without unknown parent groups (UPG) using an autoregressive test-day animal model for genetic evaluations. Therefore, four models were used in this study to evaluate these effects: HTDF with or without UPG and HTDR with or without UPG. A total of 4,142,740 test-day records from the first three lactations were used in this study. The data was provided by the Brazilian Holstein Cattle Breeders Association and obtained from 1994 to 2016. UPG were defined by geographic region, sex and birth year of the animals with missing parents resulting in a total of 133 unknown sire groups and 153 unknown dam groups. The additive genetic variances estimated with HTDR decreased in all traits (except for SCS), but also did the estimated phenotypic variances, leaving the heritabilities almost the same in both models and ranging from 0.13 for fat (3rd lactation) fitted with HTDR to 0.24 for milk (1st lactation) fitted with HTDF. The recovery of information performed on the HTDR models increased the lower reliabilities from 0.50 to 0.75, 0.54 to 0.66, 0.64 to 0.71 and 0.25 to 0.67 for milk, fat and protein yields, and for SCS, respectively (considering only bulls with 10 or more daughters). The difference in annual genetic gains between models (HTDR vs HTDF) was for sires (cows) of 30.66 (38.59) kg, 1.18 (1.35) kg, 1.26 (1.22) kg and -0.001 (-0.03) scores for milk, fat and protein yields and SCS, respectively. The contemporary groups as a random effect in the AR model is more relevant than considering just the UPG effect, but the combination of both may provide higher annual genetic gains. The rank correlations between HTDF without UPG and the HTDR with UPG models were higher for cows (from 0.77 to 0.85) than for bulls (from 0.73 to 0.82). The HTDR with or without UPG, were the models with the best results indicating that there was no significant bias in the genetic trends using Interbull's methods 1 and 2. In general, the HTDR model with UPG was the procedure that best fitted these traits and should be the model of choice for genetic evaluations and genetic trend analysis of longitudinal traits in Brazilian Holstein cattle.