A single-step evaluation of functional longevity of cows including data from correlated traits
Résumé
A routine genetic evaluation of dairy bulls based on the length of productive live (LPL) of their daughters corrected
for milk production was developed in France in 1997. Initially, only the functional longevity (FL) breeding values of
males were available. The FL genetic evaluation was then progressively improved by accounting for time-dependent
changes in environmental factors and by computing approximate FL evaluations for cows. It was also found during the
following decade that there were routinely collected traits related to fertility, conformation or udder health which could
serve as predictors of FL, using an approximate multiple trait approach. Combining these direct and indirect sources
of information led to more robust FL genetic evaluations of bulls and cows. With strong selection on production, the
economic importance of functional traits such as FL and its predictors gradually increased. At the same time, genomic
evaluations and selection became central. However, genomic evaluations of low heritability traits such as FL were
usually characterized by poor reliabilities. We propose a ‘combined’ Single-Step (CSS) approach to obtain genomic
evaluations (GEBV) of FL mimicking multiple-trait evaluations. This is illustrated in the context of the Montbéliarde
breed, using or not information from predictor traits. Survival curves of a complete cohort of cows born in 2014-2015
were calculated, showing a large superiority of the CSS evaluation: among genotyped cows, 50% of the best 10%
cows with the univariate SS reach a productive life of about 1,400 days, i.e. 402 days more than the worst 10% cows.
This difference reaches 501 days with the CSS evaluation. The corresponding figures for ungenotyped animals are
substantially smaller (187 vs 258 days). In other terms, CSS evaluations allow farmers to detect at birth the genotyped
heifers that are more likely to have a long productive life in their herd. In contrast, CSS evaluations of males are
much less pertinent to select young bulls, because selection intensity of males is much higher in selection programs.