This focus on carcass selection not only elevates the economic viability of beef production, but also contributes to the competitiveness of the nation’s cattle industry on a global scale. 

At the American Angus Association, the carcass expected progeny differences (EPDs) are calculated weekly using pedigree information and genomic profiles, as well as actual carcass phenotypes and carcass ultrasound records. The robust Angus data set of ultrasound traits includes 2.8 million animals and makes a valuable contribution to carcass evaluation by significantly enhancing the accuracy of carcass EPDs. 

The carcass EPDs have been a powerful tool for breeders and their customers to improve marbling (MARB), carcass weight (CWT), ribeye area (REA) and fat thickness (FAT).

Motivation

In genetic selection, the genetic parameters (heritability estimates and genetic correlations) are crucial for calculating EPDs, and updating them is a routine procedure. This is necessary because through time, the population undergoes changes and evolves due to selection. As more phenotypic data is collected, the genetic parameters can change. 

By updating them regularly, we can accurately estimate the genetic variation in the population and the relationships among traits. In multiple traits models, such as the carcass evaluation, accurately estimating the genetic correlations among traits is essential to take advantage of the accuracy that correlated phenotypes, particularly ultrasound records, offer.

Additionally, the carcass model includes ultrasound measurements for ribeye area (uREA) and fat thickness (uFAT) as indicator traits. The number of traits makes this the largest evaluation model within the Angus genetic evaluation pipeline, resulting in a high computational demand for EPD calculations. 

For two of the ultrasound traits in this model (uREA and uFAT), records for animals of different sexes (bulls and cows) were historically modeled as separate traits. However, modeling the sex effect within the contemporary group instead of different traits is the standard practice within genetic evaluations. This modeling is a strategy to reduce model complexity and optimize its computing time.

With that in mind, the Angus Genetics Inc. (AGI) team set out to conduct research on the carcass model with two objectives: first, to reestimate the variance components and second, to reduce model ultrasound records of different sexes as same trait and reducing model complexity. This research did not involve the model for predicting marbling EPD. 

More accurate genetic parameters 

The data set used in this study was from the official carcass model for the World Angus Evaluation (WAE), which included more than 5 million animal records. In addition to phenotypes, the pedigree used in the analyses contained more than 7 million animals, of which 1.8 million were genotyped.

Table 1. Heritabilities and genetic correlations between carcass and ultrasound traits

Table 1 shows the old and current estimated heritabilities and genetic correlations for the carcass model. The results show some significant differences in heritabilities, and genetic correlations given the previous carcass and ultrasound records reported by Angus breeders. Notably, the CWT heritability decreased from 0.44 to 0.40, whereas the heritability increased from 0.32 to 0.42 for REA and from 0.33 to 0.39 for FAT. For the genetic correlations, which explain the relationship between traits, the most notable changes are the correlations between CWT, FAT and uFAT, which are now all positive.

With changes in the heritabilities and genetic correlations among traits, EPD changes were expected. Overall, the correlation between the old EPDs and those calculated with updated variance components was 0.99 for CWT and REA, and 0.95 for FAT. When looking at the group of current sires (sires with at least one progeny registered in the past two years), the EPD correlations were 0.98 for CWT and REA, and 0.96 for FAT.

The high correlation means no substantial reranking in EPDs occurred, although individual changes in EPDs did occur. These EPD changes were due to changes in heritability estimates and strength of relationships among traits included in the model. In addition to EPD changes, small accuracy changes can also occur and the variance components are used in the accuracy calculations. It is important to note the new genetic parameters better represent the variation in the current population, thereby improving the EPD predictions.

Reducing model complexity

For the second objective, the carcass model was compared with a model that accounts for sex in the contemporary group instead of modeling it as different traits. In this research, updated genetic parameters were used for both models to allow a fair comparison between predictions. EPD correlations, validation and computing time were evaluated to compare both models. 

The correlation between carcass EPDs was higher than 0.99 between the two models, showing that only minor changes in the EPD rankings were observed when combining ultrasound traits for different sexes.

Table 2. Prediction Accuracy for old and current model

Additionally, a validation analysis was done to compare the EPD predictions of the two models. The validation involves calculating the prediction accuracy (Table 2), which is an indicator of the model’s effectiveness in predicting the EPD. Both models yielded similar results, which indicates that the new model does not compromise the accuracy of EPD predictions.

In addition to maintaining prediction accuracy, the current model greatly reduces computing time, with a reduction of approximately 30% in the total time it takes to run the old carcass evaluation. This means that the model implemented in May 2024 provides predictions just as accurate while also optimizing computing time and resources.

Take-home message

The carcass model was updated by reestimating the variance components and consolidating ultrasound traits, relying on  modeling the sex effect within the contemporary group alone instead of different traits. 

The research shows EPD changes were primarily related to the estimation of new variance components, while consolidating ultrasound traits minimally affected changes in EPDs. The correlation between EPDs from the old and current models is high, indicating no substantial overall reranking of EPDs; although individual animals may experience changes. 

This carcass model update has significantly reduced computing time for the weekly evaluation without compromising prediction accuracy, which is important for an on-time delivery of weekly genetic evaluations. These updates were implemented May 24, 2024, with the American Angus Association Sire Evaluation Report updates.