Borrowed from Western Livestock Journal, June 26, 2018:
—
Due to objective genetic predictions such as EPDs (expected progeny differences) and indexes, the cattle industry has made tremendous progress in production and efficiency. However, as the models that produce the predictions become more sophisticated and producers understand less of the mathematics behind them, some people are turning off from the technology.
This is a problem because, although calculation of modern genetic predictions have become complicated, the precision and reliability of the EPDs has likewise improved.
An EPD is defined as the difference in expected performance of future progeny of an individual, compared with expected performance at some base point for the population. EPDs are estimated from phenotypic and genomic merit of an individual and all its relatives. They are generally reported in units of measurement for the trait (e.g., lb., cm., etc.). EPDs are best used for comparing the relative genetic transmission differences to progeny between individuals.
What it boils down to is EPDs let a producer sort out genetic differences between animals, eliminating the “noise” of the environment. Some producers think they can do this better with their eyes or just a simple set of scales. This has been soundly proven wrong. The most glaring example of this occurred in Red Angus.
The breed was founded based on performance principles in 1954 with performance reporting as a requirement for registration from the very beginning. Although all Red Angus breeders had weights and measures from the beginning, the breed made no genetic progress for over 20 years. That all changed when it began converting this data into information in the form of EPDs. Since the breed started calculating EPDs, the genetic trend for traits measured has improved linearly.
Red Angus also studied the phenotypes for various traits and how they compared to the genetic predictions of the population. An example is weaning weight EPDs, which have been increasing linearly. This lines up perfectly with the breed’s adjusted weaning weights, which have improved at the same rate as the EPDs. EPDs have also allowed the breed to beat genetic antagonisms like increasing weaning weights without increasing birth weight.
Indexes are an even more powerful tool for genetic improvement. Certified Angus Beef studied when cows were flushed to either low or high $B ($Beef terminal index) bulls and all progeny were fed out and harvested. The progeny out of the high $B bulls were significantly better for all input traits into the index including weight per day of age, age at harvest, carcass weight, quality grade, and yield grade. The progeny of the high $B sires had $48.65 lower feedlot production costs and produced carcasses with $166.82 more value for a total financial benefit of $215.47.
The prediction models have also been proven to be unbiased. Cornell University did a retrospective study of the American Simmental Association’s cattle by going back and adding two years of data at a time. They then observed the differences in how cattle’s genetic predictions changed as they went from pedigree estimates through being proven sires. Animals changed up and down as the possible change chart indicated they would, as more information was added to the genetic predictions. They equally moved either up or down demonstrating no bias in the model producing the genetic predictions. If the model was biased, the predictions would tend to move in only one direction.
The basic input into genetic predictions is contemporary group deviations, and the models assume there is no environment by genotype interaction. Cornell also studied this in the Simmental population, and the assumption was validated as true.
That the models have been improving over time only makes the genetic predictions and indexes even that much more valuable.
Genetic predictions using field data were first introduced to the industry with the 1971 Simmental Sire Summary, but those early models were fraught with problems. The early models were based on sires and all dams were assumed to have equal genetic merit, which of course is not correct.
Early models also didn’t account for mating bias. The most common case of mating bias occurs when high-priced artificial insemination sires are only mated to producers’ top cows, so accounting for this bias is important. Over time, these and many more problems have been eliminated. However, with these improvements, the models have become ever more complicated and more of a challenge for the layperson to understand how they work.
This brings us to today’s modern genomic models, which are light years better than the old models, but the complicated statistics that go into the genetic predictions are admittedly hard to understand. The goal of the genetic predictions has always been to sort out what is genetic—thus will be transmitted to progeny—from what is due to environment. Marker-assisted selection is the ultimate way to determine genetic value because, by definition, genomics are not influenced by environment.
Adding genomics to traditional information that goes into genetic predictions—like contemporary group deviations, heritability, and trait correlations—all adds up to predictions that are more precise and reliable. They do a much better job of establishing genetic relationship between animals, as well as identifying markers associated with causative genes, all to improve accuracy of genetic predictions.
The whole goal to animal breeding is to improve cattle genetically. This means different things to different people—some are looking to optimize genetics to their environments while others are looking to maximize the genetic potential for traits.
Whatever a producer’s goal, EPDs and indexes are the best way to achieve it. Today’s prediction models do an unprecedented job of removing all the noise from EPDs and indexes, allowing producers to make the most informed genetic selection decisions possible.
It has been demonstrated time and again that visual evaluation and simple weights and measures are inferior substitutes for modern genetic prediction. Those who ignore objective genetic predictions do so at the long-term peril of their business’ ability to compete.
Performance pioneer Don Vaniman summed it up nicely in 1978 when he wrote, “Is it those who feel cattle that look good must perform, or those who accept that animals that perform look good?” — Dr. Bob Hough, WLJ correspondent
