Page 256 - Linear Models for the Prediction of Animal Breeding Values 3rd Edition

14 Survival Analysis

14.1 Introduction

Survival is one of the most important functional economic traits in livestock production,
affecting profitability through the rate of replacement and farm production levels.
In dairy cattle, the average herd life or survival of dairy cows has an economic value
approximately half that of protein yield on a genetic standard deviation basis (Visscher
et al., 1999). Consequently, most of the earlier research work on survival in terms of
genetic evaluation and inclusion in breeding programmes has been in dairy cattle.
Various traits have been defined as the basis of evaluating survival in the dairy
cow. These usually include some measure of survival for a period or length of life such
as stayability until certain months of life defined as a binary trait (Everett et al., 1976),
or in terms of the length of life or length of productive life (VanRaden and Klaaskate,
1993), or number of lactations (Brotherstone et al., 1997) or survival per lactation
as a binary trait. Linear models are generally used – either a repeatability model
(Madgwick and Goddard, 1989) or a multivariate model (Jairath et al., 1998).
Similar definitions of survival have been applied to other livestock species. The length
of productive life between first farrowing and culling has been analysed in pigs
(Tarrés et al., 2006; Mészáros et al., 2010). In rabbits, survival has been defined as
the length of productive life, referring to the days between date of the first positive
pregnancy diagnosis and date of culling or death (Piles et al., 2006).

14.2 Functional Survival

Another important element of evaluating survival is the concept of functional survival
or longevity. Functional longevity refers to survival that is independent of production
such as milk yield for dairy cattle or litter size in pigs. The reasoning is that voluntary
culling is based mostly on production, thus adjusting for production (usually at the
phenotypic level) in the analysis of survival produces EBVs for animals that defines
their ability to avoid involuntary culling.

14.3 Censoring

The traits used in survival analysis involve measuring the length of time between two
events, usually a start and end point (also called ‘failure’). However, at the time of
analysis, some animals might still be alive, not having had the opportunity to reach
the end point. Their measure of survival is based on their current status and does not

240 © R.A. Mrode 2014. Linear Models for the Prediction of Animal Breeding Values,
3rd Edition (R.A. Mrode)