Given the many questions on EPD's I thought that I would post for all to study. It should answer most questions about EPD's and help in using EPD's as a Selection Tool.
What is an EPD
An EPD, or expected progeny difference, is just what it sounds. It is a prediction of the difference between the average performance of future progeny of an individual and the performance of a theoretical reference animal, an animal with an EPD of zero. To make this comparison fair, we assume similar environments and mates of the same genetic value. An EPD is then a prediction of progeny performance relative to some standard. It is expressed in trait units. For growth traits, the units are pounds.
For example, a particular sire might have an EPD of +1.5 for birth weight. This means that he is expected to produce calves 1.5 pounds heavier on average than the theoretical bull with a zero EPD for birth weight. More meaningfully, this sire is expected to produce calves 4.5 pounds lighter than a sire with and EPD of +6.0 (6.0 - 1.5 = 4.5) or four pounds heavier than a sire with an EPD of -2.5 [1.5 - (-2.5) = 4.0]. As you can see, EPDs are designed to compare animals, nothing more.
How are EPDs calculated?
EPDs are calculated using statistical procedures that fall into a general category called Best Linear Unbiased Prediction or BLUP for short. BLUP pro
cedures properly weigh vast amounts of information on individuals and their relatives.
What advantages do EPDs have over ratios?
Ratios are measures of individual performance and are calculated on a within-herd basis. They contain no pedigree or progeny information and are really only appropriate for comparing animals within the same herd or contemporary group. EPDs, on the other hand, are calculated from all sources of informationpedigree, own performance and progeny. EPDs are comparable across herds.
How can we compare EPDs on animals in different herds if their calves have been raised in very different environments?
BLUP Procedures account for environmental differences through information on the genetic relationships between animals in different herds. This is similar to having "reference sires" in each herd. The more closely related the animals in two herds are, the better job BLUP procedures can do in adjusting for the differences in environment between those herds.
How much do EPDs depend on pedigree vs. own performance vs. progeny information?
The relative importance of these sources of information depends on the amount of information available from each source. The EPDs of young animals will be based entirely on pedigree and individual performance. EPDs for animals with large numbers of progeny will be based almost totally on progeny information.
If I mate a bull only to the best cows, won't that make his EPDs look better than they should?
No. The procedures used to calculate EPDs account for the merit of the mates to which an animal is bred.
What does an EPD of zero mean in terms of actual performance?
The actual performance of progeny of a sire with EPDs of zero will depend heavily on the environment in which those calves are raised and the genetic value of the dams. To take weaning weight as example, we can expect the zero EPD sire to produce heavier calves in better environments. And if a cow herd is genetically fast growing, we can expect heavier weaning weights from the zero EPD bull than if the herd were less growthy.
It is not possible, therefore, to generalize about the absolute performance of progeny of zero EPD animals, or for that matter, the absolute performance of progeny of animals with any EPD. The only good way to get a feel for what particular EPD levels mean in your own herd is to try well evaluated sires and establish your own baselines. Having done that, you should be able to predict how new sires will do by comparing their EPDs against those of the sires you have used and are familiar with.
Where does the zero point come from?
For starters, zero is not breed average (though it can sometimes be close). In the Red Angus analysis, zero is function of the performance of foundation animals, animals with no recorded parents. As such, the zero point or base, as it is often called, does not change over time as the breed changes genetically. This is handy because it means that a given EPD will always have the same biological interpretation; +10 for milk will always imply the same level of genetic ability for milk production.
Can EPDs be compared across breeds?
No. Or at least not yet. Different breeds have very different bases (zero points), and this makes it very difficult to compare EPDs across breeds. It is possible to define the base any way we like, and soon most breeds may define their base as the average genetic value of all calves born in a given year, say 1975. This would at least provide a common definition for what is meant by zero. EPDs still could not be compared directly across breeds, however, because all breeds were not performing at the same level in 1975. Before long, we will have conversion tables that will allow comparison of EPDs across breeds.
How do I determine breed average?
Probably the best way to get a feel for breed average is to study the statistics which are published at the front of the sire summary. These vary from sire summary to sire summary, but at least they will provide the average EPDs and the range of EPDs for the listed sire in each trait. These should give you a good idea where the breed currently stands in terms of EPDs.
What is a "significant" difference in EPDs?
A mistake that breeders often make is to interpret small differences in EPDs as being meaningful. A sire with a yearling weight EPD of +45 might be slightly genetically faster growing than a sire with a +40 yearling weight EPD (or he might not be, considering the error involved in prediction), but the truth is that a five-pound difference will be undetectable in most herds. My advice is to divide EPDs into five categories: extremely high, high, moderate, low and extremely low. Once you have places an animal in one of these categories for a given trait, forget about his or her precise EPD.
What EPDs are best?
Beef cattle breeding would be a lot easier if we could assume that the highest EPDs are the best. Unfortunately, it's not that simple. When we increase one trait, growth rate for example, we often pay a price somewhere else, perhaps in supplemental feed costs or reproductive performance. A delicate balance of levels of traits is required, and the balance differs with varying climatic, nutritional and economic environments. One of the most enduring challenges in beef cattle breeding is to determine the best balance of trait levels (and associated EPDs) for a given environment and clientele.
What is meant by accuracy?
The accuracy value reflects the amount and relevance of the information use to calculate an individual EPD. Accuracy values range from zero (very poor) to one (extremely accurate). Accuracy measures the reliability of an EPD or the degree of risk associated with using a particular animal on the basis of its EPD. Highly accurate EPDs are very reliable; there is little risk that the progeny performance of an individual with high accuracy values will, on average, be much different than the EPDs indicated. On the other hand, the average progeny performance of an individual with low accuracy values may be quite different from what his EPDs suggest.
A common misconception is that accuracy values tell us how variable an individual's offspring will be. They don't. Accuracies simply tell us whether the estimate is based on good hard data or whether it is little more than a guess.
What is good accuracy?
The answer varies according to each person's feeling about risk. Here is my breakdown:
Low Less than .40 Unreliable, but still a best guess
Low/Medium .40 to .60 Worth looking at, but risky
High/Medium .60 to .80 Quite trustworthy, make comparisons with some confidence
High Greater than .80 Good accuracy, compare with confidence
What are direct and maternal EPDs?
Traits like weaning weight have what is called a maternal component. This means they are affected not only by a calf's own genes, but by genes in the calf's dam that influence the environment she provides for the calf. For weaning weight, we assume the maternal component is primarily a function of milk production. So we call the EPD for the maternal component of weaning weight a milk EPD.
The calf's own genes for growth have a direct effect on its weaning weight. We call the EPD that predicts the effect of these growth genes the EPD for weaning weight direct or simply the weaning weight EPD.
There are, then two basic types of EPDs for maternally influences traits: 1) a direct EPD which estimates the value of genes passed on from an individual to its offsprings which directly affect offspring performance, and 2) a maternal EPD which estimates the value of genes passed from an individual to its daughters which affect the environment those daughters provide for their offspring.
What is a total maternal EPD?
Neither the direct nor maternal EPD predicts entirely the performance of offspring of an individual's daughters since that performance will be a function of both direct and maternal effects. A third EPDs, the total maternal EPD, combines direct and maternal predictions. For weaning weight, the total maternal EPD predicts the relative performance of progeny daughters of an individual, taking into account both the inherent growth of those calves and the milking ability of their dams.
Who calculates EPDs?
BLUP procedures require the solution of many thousands of simultaneous equations, and for that reason BLUP analyses are performed at universities where the necessary computer facilities and expertise are located. These analyses are normally done once a year, usually just prior to production of the sire summary.
Can I get EPDs for animals not listed in the sire summary?
Yes. EPDs for all animals, males and females, from newborns to those long since dead, are updated each time a BLUP analysis is performed. Many of these are routinely reported on weaning and yearling reports and dam summaries. EPDs on other animals are available from RAAA.
Some EPDs don't have accuracy values. Why is this?
Two types of EPDs are returned to the breed association from the university. Parental EPDs are simply EPDs for animals with recorded progeny. These EPDs will have accuracy values associated with them. Nonparent or back solution EPDs are EPDs for animals that have not yet produced progeny. They are calculated somewhat differently from parental EPDs and will not have accuracies associated with them. (Accuracies for nonparents can always be assumed to be low because these animals have no offspring.) Some non parent EPDs are simply pedigree estimates, averages of the sire's and dam's EPDs for a trait. A nonparent EPD for milk , for example, must necessarily be a pedigree estimate since no other information is available on an animal that has not yet produced a calf. RAAA prints a "P" in the accuracy column for pedigree estimates. Other nonparent EPDs, weaning weight for example, will include a calf's own performance information. RAAA print a "P+" in the accuracy column for these EPDs.
What is an interim EPD?
New performance data arrive at the breed association daily, and so EPDs, especially for young animals, rapidly become out of date. To mitigate this problem, interim EPDs are calculated by the breed association as new performance information is processed. Interim EPDs combine the EPDs of a young animals' parents with that animal's new performance record and an estimate of the average genetic value of his or her contemporaries. In cases where no new observation has been recorded, interim EPDs may simply be pedigree estimates. Like any EPD, interim EPDs can be compared against
EPDs of animals in other herds.
Typical examples of an interim EPD would be the EPDs for yearling weight that a breeder receives after reporting yearling data on his calves. These EPDs contain pedigree information from the most recent university analysis and combine that information with the new postweaning performance records. Like nonparent EPDs, interim EPDs do not have accuracy values associated with them. RAAA prints a "P" or "P+" in the accuracy column depending on whether the EPD is a pedigree estimate or contains individual performance information. Actual accuracies will be low.
What is stayability?
The probability that a cow will remain in the herd for a given number of parities or to a given age, with the condition that the cow was given the opportunity to become part of the breeding herd (i.e. calved at least once). Stayability data consists only of calving records. Stayability is affected by breeders' culling criteria. Given the primary emphasis in most herds on pregnancy, stayability is likely to be more a measure of sustained fertility than anything else. Stayability EPD for bulls are the predictions of the genetic differences between their daughters probability of staying in the herd to at least the age of six years.
Glossary of Terms
ACCURACY (ACC) - A measure of the reliability of an EPD. Accuracy values are reported as decimal numbers between zero and one. Values closer to one indicate larger amounts of available information and greater certainty that a bull's EPD will not change significantly as more progeny information becomes available.
BIRTH WEIGHT EPD - A sire's expected progeny difference (lb) for mature dam equivalent birth weight of calf. Progeny of sires with lower birth weight EPD values will generally be born with less calving difficulty.
DAUGHTERS (DTRS) - The number weaning weight records of the sire's daughter's calves (grand progeny) used in the analysis.
EXPECTED PROGENY DIFFERENCE (EPD) - The expected difference (in trait units, usually lb) in performance of a bull's progeny when those progeny are compared to progeny of a sire with an EPD of zero for the trait in question. EPD values are relative. They do not indicate absolute levels of performance. Rather, they can be used to compare expected progeny performance of different bulls.
GROUPS (GRPS) - The number of contemporary groups in which a sire's progeny appear.
MATERNAL MILK EPD - The expected difference (due to milk production alone) in weaning weight of calves out of a bull's daughters compared to calves from daughters of a bull with a zero EPD for maternal milk. Maternal milk EPD values are measured in pounds of calf weaned and indicate milking ability.
PROGENY (PROG) - The number of progeny weaning weight records used in the analysis.
PROGENY PER DAUGHTER (PROG/DTR) - The average number of progeny records per daughter.
STAYABILITY - The expected difference in probability of daughters staying in the herd to at least the age of six years. Since cows are usually only culled for being open before the age of six, the EPD is primarily a measure of sustained fertility in female offspring.
TOTAL MATERNAL EPD - The expected overall difference (lb) in weaning weight of calves out of a bull's daughters compared to calves from daughters of a bull with a zero EPD for total maternal. The total maternal EPD is a measure of both growth and milking ability and is an indication of the size of calf a bull's daughter will wean.
WEANING WEIGHT EPD - A sire's expected progeny difference (lb) for 205-day, mature dam equivalent weaning weight. A sire's weaning weight EPD is an indication of growth to weaning. It does not indicate milk production.
YEARLING WEIGHT EPD - A sire's expected progeny difference (lb) for 365-day, mature dam equivalent yearling weight. The yearling weight EPD is a measure of growth from birth to a year of age. r most questions about EPD's and their Value as Tools
