Expected Progeny Differences and Accuracy

With AOA having released updated expected progeny differences (EPDs) several times since beginning the genetic evaluation system, many breeders will have witnessed firsthand how EPDs for some animals change as additional data is added to the database. This serves to reinforce the concept that EPDs for an individual animal may change as additional information on that animal, on its relatives, and on genetically correlated traits becomes available. The new data provides an opportunity for a more precise estimate of animals’ genetic merit. The potential for an EPD to change from one evaluation to the next and degree of change are reflected in accuracy value for each EPD, so breeders who have been initially uncomfortable with the idea that EPDs may change, are provided with an indicator of potential change each time EPDs are updated.

As review, remember EPDs are a prediction of true genetic merit for an individual for a specific trait. The EPDs represent expected differences in performance of progeny resulting from the genes passed from the parent to the offspring. For instance, let’s assume you need to choose between two potential sires to use in your breeding program. The first, sire A, has a fiber diameter EPD of -2.9 micron and the second, sire B, has a fiber diameter EPD of -.1 micron. Assuming we were to mate these to females of similar genetic merit, we would expect the average difference of fiber diameter in their offspring to be 2.8 micron (-2.9 – -.1) with the progeny of sire A having the smaller average fiber diameter. When the EPDs are calculated, this is the best estimate of those two individual’s genetic merit, given the data available at the time and is always better than selecting solely on the sire’s own fiber diameter.

Associated with every EPD is an accuracy value. Accuracy ranges from 0 to 1.0 with higher values associated with more data and therefore, a better estimate of the animal’s true genetic merit. An accuracy of 1.0 would indicate that we have a perfect prediction of that animal’s true genetic merit. In reality we never get accuracies of 1.0 but can only approach that level of accuracy with large numbers of progeny.

Going back to the original example, the accuracy of A’s EPD was .3 and the accuracy of B’s EPD was .5. So our “faith” in the second animal’s EPD would be greater than that of the first animal and both would be better than selecting on the sire’s fiber diameter alone. We can also take accuracy one step further and determine how much change we might expect as more information is submitted to the AOA database. The value that represents this future change is known as “possible change.” The possible change value associated with .3 accuracy is .84 microns and the corresponding value for .5 accuracy is .48 microns. Notice that the higher accuracy is associated with a smaller possible change—reflecting our greater confidence in the EPD at higher accuracies. Correspondingly there is a lower potential for change as new data is reported (.48 microns is less than .84 micron). For those that are more statistically oriented, we can construct a confidence interval around each EPD by simply adding and subtracting the possible change value to the EPD. For the low accuracy individual (sire A) with the -2.9 EPD and an accuracy of .3, the confidence interval would be -3.7 to -2.1 micron. This range represents the range in which we are 68% confident the animals’ true progeny difference for fiber diameter lies. If this sire had a greater accuracy, our confidence interval would be narrower, also reflecting our greater confidence resulting from more data and higher accuracy.

EPDs are a tool for selecting breeding animals that are based solely on pedigree and performance information contained in the AOA database. As additional data is submitted, EPDs will change and accuracy values will increase on many animals giving us more confidence in those EPDs. Even though EPDs are more accurate than any other means of selecting breeding animals, as more information is compiled some animal’s EPDs will improve and other’s will not. The accuracy value is a representation of the “risk” associated with that EPD and as with investments, some breeders will be willing to take on more risk than others.