Comparison of models for genetic evaluation of survival traits in dairy cattle: a simulation study

Abstract

Three models for the analysis of functional survival data in dairy cattle were compared using stochastic simulation. The simulated phenotype for survival was defined as a month after the first calving (from 1 to 100) in which a cow was involuntarily removed from the herd. Parameters for simulation were based on survival data of the Canadian Jersey population. Three different levels of heritability of survival (0.100, 0.050 and 0.025) and two levels of numbers of females per generation (2000 or 4000) were considered in the simulation. Twenty generations of random mating and selection (on a second trait, uncorrelated with survival) with 20 replicates were simulated for each scenario. Sires were evaluated for survival of their daughters by three models: proportional hazard (PH), linear multiple-trait (MT), and random regression (RR) animal models. Different models gave different ranking of sires with respect to survival of their daughters. Correlations between true and estimated breeding values for survival to five different points in a cow's lifetime after the first calving (120 and 240 days in milk after first, second, third and fourth calving) favoured the PH model, followed by the RR model evaluations. Rankings of models were independent of the heritability level, female population size and sire progeny group size (20 or 100). The RR model, however, showed a slight superiority over MT and PH models in predicting the proportion of sire's daughters that survived to the five different end-points after the first calving.

Genetic Prediction Models and Heritability Estimates for Functional Longevity in Dairy Cattle

“…Jamrozik & Schaeffer, 1997;Mrode & Swanson, 2004) and other traits such as susceptibility to mastitis and fertility (e.g. Carlen et al, 2005;Tsurunta et al, 2009), conformation, body condition scores, feed intake, heart girth measures (Schaeffer, 2004) and survival (Jamrozik et al, 2008).…”

“…In RR modelling, binary observations (0 = culled, 1 = survived) are assigned to each discrete unit in the cow's lifetime, such as per lactation or per month after first calving, after which a linear model with RRs for an animal genetic effect can be fitted to this data for genetic evaluation. Breeding values for survival can then be generated for both cows and sires and for each point on the trajectory (Jamrozik et al, 2008). Linear regression of observations of the trait under consideration on indicator variables is performed with the animals' additive genetic effects fitted as random effects.…”

“…Linear modelling of binary survival has been carried out in Canadian Holsteins (Boettcher et al, 1999), Czech Fleckvieh (Zavadilová et al, 2009), Norwegian Red (Holtsmark et al, 2009), South African Jersey cattle (Du Toit, 2011), Italian Brown Swiss (Samoré et al, 2010) and others (e.g. Jamrozik et al, 2008;Zavadilová & Štípková, 2012). Linear models have also been used to analyse survival based on actual and projected records, using currently available information (VanRaden & Klaaskate, 1993).…”

“…The criteria used for such model comparisons include predictive abilities, estimated variance components, heritability estimates (e.g. Holtsmark et al, 2009), correlations of estimated and true breeding values, proportion of sires' daughters that survived to specified end points, and ranking of sires between models (Jamrozik et al, 2008). Holtsmark et al (2009) compared TRMs, TCMs, linear multi-trait, linear repeatability, linear crosssectional and Weibull frailty models in terms of estimated variance components, heritability estimates, sire EBVs and predictive abilities.…”

“…Longevity has assumed various trait definitions (Vollema & Groen, 1997;Brotherstone et al, 1998), all of which are based on age at culling or death (uncensored) or censoring and survival to a given age or predetermined period within or across lactations (Jamrozik et al, 2008;Forabosco et al 2009). It has been scored as a binary trait (e.g.…”