Livestock Update, October 2000
Scott Greiner, Extension Animal Scientist, Beef, Virginia Tech
Understanding EPDs: Part 2
Note: This article is Part 2 in a three part series dealing with EPDs.
In last month's article, the basic definitions and applications of Expected Progeny Differences (EPDs) were discussed. This month's article will focus understanding specifics related to effective utilization of EPDs.
In addition to directly comparing the EPDs of bulls, it is useful to understand where a particular bull ranks in the breed for traits of interest. This will give a general idea as to the genetic merit of the bull compared to all others within the breed. It is important to understand that the average EPD for any trait within a breed is not 0. One reason for this is genetic trend. Genetic trend refers to the improvement in genetics that has taken place over time within a breed due to selection. Over the years breeders have selected for increased growth, milk production, etc. As this selection has occurred, the average EPDs for bulls within a breed for these traits has also increased. The following table depicts average EPD values for non-parent bulls of several breeds:
Consider a bull that has a yearling weight EPD of +25. If this bull happens to be a Charolais, he is around breed average for his genetic potential to transmit yearling weight (Charolais breed average yearling weight EPD = +22.0). However, if this bull would happen to be an Angus, his yearling weight EPD would be about 40 pounds below breed average (Angus breed average yearling weight EPD = +56). In this case, a yearling weight EPD of +25 would be interpreted quite differently in a Charolais bull vs. Angus bull. This demonstrates one reason why EPDs cannot be directly compared on bulls of different breeds. Also, it important to note that the EPDs in the above table do not reflect genetic differences for the traits between breeds, as the EPDs cannot be directly compared across breeds. These average breed values are not directly comparable due to the fact that each breed calculates its EPDs from its own data set. Since the data sets (performance records, pedigrees, etc.) are independent, there are few animals that would be found in more than one breed's records used to calculate EPDs. Without these ties, and without merging data into one data set, the calculated EPDs are not comparable across breeds.
Breed Percentile Rankings:
An understanding of where an animal ranks in the breed for a particular trait EPD is extremely valuable as a selection tool. Breed associations also publish percentile ranking tables in their sire summaries so that bulls can be specifically evaluated as to where their EPDs rank in the breed (top 10% vs. bottom 20%, etc.). Percentile rankings can be misleading if not used in the proper context. For example, a Simmental or Gelbvieh bull that ranks very high in the breed for milk EPD (top 10% for example) may not necessarily be ideal in a commercial crossbreeding program. A bull with breed average (or below breed average) genetic merit for milk may produce daughters that are more optimum in their milk production, resulting in females that potentially are a more efficiently match for feed resources and maintain more optimum reproductive potential. Similar examples could be given for other traits. Perhaps most importantly, the general merit of the breed for each trait needs to be considered along with where an individual bull ranks within that breed.
Along with EPDs for the various traits, sire summaries also contain accuracy values associated with each EPD. These accuracy values range from 0 to 1.0. The accuracy value is a measure of the certainty we have in the EPD. High accuracy values (>.80) indicate that the EPD is less likely to change when more performance records are included in the EPD determination for the animal. High accuracies are indicative of animals that have a large number of progeny records. The accuracy level also gives us an indication of how much the EPD may change. EPDs with low accuracies are more likely to change with additional information, as compared to EPDs with high accuracy values. Consider the following two Angus bulls:
|Birth Weight EPD (Accuracy)|
|Bull A||+2.0 (.90)|
|Bull B||+2.0 (.25)|
Bull A and Bull B have equivalent birth weight EPDs. However, Bull A has an accuracy of .90 vs. .25 for Bull B. Therefore, we expect the "true" birth weight EPD for Bull A to be between +0.8 and +3.2 (there is a 67% probability that the "true" EPD will fall in this range). For Bull B, his "true" EPD may be from -0.4 to +4.4. If these are considered calving-ease bulls to be mated to heifers, we would prefer Bull A due to his reliability and decreased risk relative to Bull B. This is one reason that artificial insemination of heifers to high-accuracy, calving-ease bulls is widely practiced. We are more confident in the published EPD of Bull A, and there is less risk involved with mating him to heifers. An accuracy of .90 (or greater) would be typical of a well-proven bull AI bull, whereas an accuracy of .25 would be common for a yearling bull. Does this mean that young, low accuracy bulls should be avoided? No. Research has clearly demonstrated that EPDs are much more effective than individual performance measures (individual birth weights, test gain, adjusted weights, ratios, etc.) for predicting future offspring performance. Even on young yearling bulls, EPDs are the most effective selection tool available. Keep in mind that with low accuracy values, there is an equal chance that the "true" EPD is more desirable vs. less desirable than the EPD reported. With traits such as birth weight EPD, specifications for specific EPD values may need to change to allow for this increased risk associated with low accuracy.
Visit Virginia Cooperative Extension