Combined single-step evaluation of functional longevity of dairy cows including correlated traits

Abstract

Background: For years, multiple trait genetic evaluations have been used to increase the accuracy of estimated breeding values (EBV) using information from correlated traits. In France, accurate approximations of multiple trait evaluations were implemented for traits that are described by different models by combining the results of univariate best linear unbiased prediction (BLUP) evaluations. Functional longevity (FL) is the trait that has most benefited from this approach. Currently, with many single-step (SS) evaluations, only univariate FL evaluations can be run. The aim of this study was to implement a "combined" SS (CSS) evaluation that extends the "combined" BLUP evaluation to obtain more accurate genomic (G) EBV for FL when information from five correlated traits (somatic cell score, clinical mastitis, conception rate for heifers and cows, and udder depth) is added.

Results: GEBV obtained from univariate SS (USS) evaluations and from a CSS evaluation were compared. The correlations between these GEBV showed the benefits of including information from correlated traits. Indeed, a CSS evaluation run without any performances on FL showed that the indirect information from correlated traits to evaluate FL was substantial. USS and CSS evaluations that mimic SS evaluations with data available in 2016 were compared. For each evaluation separately, the GEBV were sorted and then split into 10 consecutive groups (deciles). Survival curves were calculated for each group, based on the observed productive life of these cows as known in 2021. Regardless of their genotyping status, the worst group of heifers based on their GEBV in 2016 was well identified in the CSS evaluation and they had a substantially shorter herd life, while those in the best heifer group had a longer herd life. The gaps between groups were more important for the genotyped than the ungenotyped heifers, which indicates better prediction of future survival.

Conclusions: A CSS evaluation is an efficient tool to improve FL. It allows a proper combination of information on functional traits that influence culling. In contrast, because of the strong selection intensity on young bulls for functional traits, the benefit of such a "combined" evaluation of functional traits is more modest for these males.

Fig. 1

Correlations between GEBV for FL obtained with the CSS evaluation in 2021 without any performances for FL and those with a USS evaluation in 2021 for each population per year of birth

Fig. 2

Correlations between GEBV for FL obtained with the complete CSS evaluation in 2021 and those with a CSS evaluation in 2021 without any performances for FL for each population per year of birth

Fig. 3

Survival curves of females born in 2014 and 2015 ranked according to their GEBV based on the USS evaluation in 2016

Fig. 4

Survival curves of females born in 2014 and 2015 ranked according to their GEBV based on the CSS evaluation in 2016

Fig. 5

Observed mean difference in days of productive life between deciles of heifers based on USS or CSS evaluations

Fig. 6

Average standardized milk production per GEBV deciles for functional longevity obtained with the USS evaluation in 2016 (decile 1 = the best animals for functional longevity)

Fig. 7

Average standardized milk production per GEBV deciles for functional longevity obtained with the CSS evaluation in 2016 (decile 1 = best animals for functional longevity)

Fig. 8

Survival curves of daughters of sires born in 2011 according to their GEBV based on the USS evaluation in 2016

Fig. 9

Survival curves of daughters of sires born in 2011 according to their GEBV based on the CSS evaluation in 2016