Improving the accuracy of predictions for cow survival by multivariate evaluation model
journal contributionposted on 2023-03-09, 03:55 authored by Majid KhansefidMajid Khansefid, Mekonnen Haile-MariamMekonnen Haile-Mariam, Jennie PryceJennie Pryce
Context: Cow survival measures the ability of cows to survive from the current to subsequent lactation. In addition to economic gain, genetic selection for survival could improve animal welfare by increasing the adaptability and resilience of the cows to both environmental and health challenges. However, survival is a complex trait because it results from a diverse range of reasons for culling of cows from the herd. Consequently, the accuracy of genetic predictions of direct survival are often low. Aims: Our aim was to increase the accuracy of predictions of survival in Holstein and Jersey sires by including important predictor traits in multi-trait evaluation models. Methods: Phenotypic and genetic correlations between survival trait deviations (TDs) and 35 routinely measured traits (including milk yield, fertility and type traits) were estimated using bivariate sire models. Survival TDs for 538 394 Holstein and 63 839 Jersey cows were used in our study; these cows or their close relatives also had milk, fertility and type traits records between 2002 and 2019. These genetic parameters were required to assess the potential usefulness of predictor traits for the prediction of survival. Key results: Survival was genetically correlated with milk, fat and protein yields, overall type, composite mammary system and fertility TDs in both Holstein and Jersey. Further, most of the type traits related to feet and legs, and rump, were also correlated with survival TDs in Jersey. For sires, the accuracy of predictions for survival increased by 0.05 for Holsteins (from 0.54 to 0.59) and for Jerseys (from 0.48 to 0.53) through the use of multivariate models compared with univariate models. Conclusions: Survival was genetically associated with traits affecting voluntary and involuntary culling and when included in multi-trait genetic evaluation models, they moderately improved the accuracy of genetic prediction of survival. Implications: Predictor traits can be used to increase the accuracy of predictions of survival through the use of multi-trait models. The inclusion of breed-specific predictor traits should be considered, especially for Jerseys in genetic evaluations of survival.