Use of Multitrait Evaluation Procedures to Improve Reliability of Early Prediction of Survival

Abstract

Multitrait, across-country evaluation procedures were adapted to improve reliability of survival breeding values by combining direct information with yield and functional traits: milk persistency, somatic cell count, fertility (male and female), and calving ease (direct and maternal). A set of bulls was selected from the Austrian Simmental population based on mean original reliabilities of at least 0.50 for yield traits or of at least 0.20 for functional traits. Only breeding values above these limits were retained. The breeding values were deregressed, assuming that they were obtained by single-trait, sire-maternal grandsire models. An expectation-maximization restricted maximum likelihood algorithm based on the multitrait, across-country evaluation equations was used to compute genetic correlations among all of these traits. These equations were solved, and the reliabilities of the solutions were also estimated. Mean rank correlation between direct and combined breeding values was 0.85 with values as low as 0.67 for the group of youngest bulls. Direct (original), indirect, and combined reliabilities were compared to appreciate the impact of our procedures on improvement of reliability of survival breeding values. This improvement, dependent on the level of reliability of direct and indirect information, could be up to 0.24 for animals with low direct reliability and high indirect reliability. For young bulls born in 1992 without reliable direct survival information but with already reliable information on yield and functional traits, mean reliability improvement was 0.13. For all bulls this value was still 0.06.