Expected Progeny Differences for Stayability in Angus Cattle Using a Random Regression Model

Stayability (STAY) is a measure of whether or not a female remains productive in the herd until a specified point in time, with that point traditionally being 6 years of age. Waiting 6 years before an observation is generated represents a limitation to obtaining accurate Expected Progeny Difference (EPD) values for STAY at the age of 6 (STAY6). Random Regression Models (RRM) have been suggested for use in national cattle evaluations to improve the prediction of STAY6 by including observations from earlier ages into the analysis. Therefore, the objective of this study was to compare EPD for STAY at consecutive ages using traditional methods with EPD for STAY obtained with RRM in Angus cattle. Calving performance data consisting of 1,233 females (progeny of 215 sires and 791 dams) collected from 1993 to 2012 at the Colorado State University Beef Improvement Center (CSU-BIC) was used for the study. Four STAY endpoints defined as whether a cow calved at age 3, 4, 5, and 6 given she calved as 2-yr-old were assigned observations (0, unsuccessful; 1, successful). These observations were used to calculate EPD for each of the STAY endpoints. Traditional STAY was evaluated for each endpoint (STAY3 through STAY6) using a univariate BLUP threshold animal model along with a probit link function to convert binary observations to an underlying normal distribution. Effects in the models included contemporary group (dam birth year and calf birth year) as a fixed effect and animal as random effect. Additionally, all STAY endpoints were evaluated using a linear RRM with Legendre polynomials as the base function. For the RRM, contemporary group and a linear fixed regression were included as fixed effects. The RRM predicted the genetic merit of the presence of a weaned calf at each particular age endpoint, therefore EPD were summed to obtain the individual’s genetic merit for the presence of a calf at 3, 4, 5 and 6 years of age. Pearson correlation coefficients between EPD at each endpoint and the corresponding RRM EPD were 0.59, 0.83, 0.82 and 0.77 for STAY3, STAY4, STAY5 and STAY6, respectively. Regression of predictions obtained from the RRM on traditional EPD were 1.76, 3.65, 4.76 and 5.05 for each of the consecutive endpoints. These results suggest that while both models are predicting similar genetic merit for individuals, the traditional method is under-predicting the genetic merit of individual animals when compared to the obtained with the RRM.