Abstract
First-service conception rate (FSCR) can be defined as the probability of a heifer conceiving in response to her first artificial insemination (AI). Given the binary nature of its phenotypes, FSCR has been typically evaluated using animal threshold models (ATM). However, susceptibility of these models to the extreme-category problem (ECP) limits their ability to use all available information to calculate Expected Progeny Differences (EPD). Random regression models (RRM) represent an alternative method to evaluate binary traits, and they are not affected by ECP. Nevertheless, RRM were originally developed to analyze longitudinal traits, so their usefulness to evaluate traits with singly observed phenotypes remains unclear. Therefore, objectives herein were to evaluate the feasibility of a RRM genetic prediction for heifer FSCR by comparing its resulting EPD and genetic parameters to those obtained with a traditional ATM. Breeding and ultrasound records of 4,334 Angus heifers (progeny of 354 sires and 1,626 dams) collected between 1992 to 2019 at the Colorado State University Beef Improvement Center were utilized. Observations for FSCR (1, successful; 0, unsuccessful) were defined by fetal age at pregnancy inspections performed approximately 130 d post-AI. Traditional FSCR evaluation was performed using a univariate BLUP threshold animal model, whereas an alternative evaluation was performed by regressing FSCR on age at AI using a linear RRM with Legendre Polynomials as the base function. Heritability estimates were 0.03 ± 0.02 for the ATM and 0.005 ± 0.001 for the average age at AI with the RRM, respectively. Pearson and rank correlations between EPD obtained with each method were 0.63 and 0.60, respectively. The regression coefficient of RRM predictions on those obtained with the ATM was 0.095. In conclusion, these results suggested that although a RRM genetic prediction for FSCR was feasible, a considerable degree of re-ranking occurred between the two methodologies.
Margret M. and Paul B. Baltes Award
