Estimates of genetic parameters for stayability to consecutive calvings of Canadian Simmentals by random regression models1

• Published in: Journal of animal science 2013-08, Vol.91 (8), p.3634-3643
• Main Authors: Jamrozik, J., McGrath, S., Kemp, R. A., Miller, S. P.
• Format: Article
• Language: English
• Subjects: Bayesian analysis; Cattle; Genetics
• ISSN: 0021-8812; 1525-3163
• DOI: 10.2527/jas.2012-6126

Stayability to consecutive calvings was selected as a measure of cow longevity in the Canadian Simmental population. Calving performance data on 188,579 cows and culling information from the Total Herd Reporting System were used to determine whether a cow stayed in a herd for her second and later (up to the eighth) calvings, given that she had calved as 2 yr old. Binary records (n = 1,164,319) were analyzed with animal linear and threshold models including fixed effects of year of birth by season of birth by parity number and age of cow at first calving by parity number and random effects of contemporary group (CG) defined as herd of birth within year by season, animal additive genetic effect, and a cow permanent environmental (PE) effect. All random effects were Legendre polynomial regressions of the same order, defined on the scale from second to the eighth calving. Bayesian methods with Gibbs sampling were used to estimate covariance components and genetic parameters for random effects of models and selected variables on the longitudinal scale. Bayes factors and analyses of mean squared error and correlation between observed and predicted observations indicated that the linear model with regressions of order 3 was most plausible for generating the current data compared with a fixed regression and other random regression (both linear and threshold) models of order up to 4. Estimates of variances for all random effects from the best fitting model changed with the calving number. Estimates of heritability decreased in time: from 0.35 (SD = 0.006) for stayability to second calving to 0.13 (SD = 0.004) for stayability to the eighth calving. Variance due to PE effect constituted the largest part of the total variance of stayability for all longitudinal points followed by genetic and CG components. Genetic effects of stayability to different calvings were relatively highly correlated, from 0.62 (SD = 0.011) to 0.99 (SD = 0.001), and correlation decreased with the time span between calvings. Correlations for PE and CG effects showed similar trends. Animal genetic effect seemed to be less variable on the longitudinal scale compared with other random effects of the model. The first 2 principal components explained from 95% (PE effects) to 99% (genetic effect) of the total variance. The overall level of genetic stayability curve correlated well (from 0.87 to 0.99, with SD < 0.006) with genetic stayability to different calvings and therefore could be used as a sing