You've reached the Virginia Cooperative Extension Newsletter Archive. These files cover more than ten years of newsletters posted on our old website (through April/May 2009), and are provided for historical purposes only. As such, they may contain out-of-date references and broken links.

To see our latest newsletters and current information, visit our website at

Newsletter Archive index:

Virginia Cooperative Extension -
 Knowledge for the CommonWealth

Using reliability in breeding programs.

Dairy Pipeline: November 1996

by Bennet Cassell
Extension Dairy Scientist, Genetics and Management
Virginia Tech

Each of the traits evaluated on AI bulls has its own reliability, which measures the accuracy of the genetic evaluation for that trait. For most bulls, milk, fat, and protein will have the highest reliabilities because more daughters have performance records on those traits. Reliabilities for type traits tend to be lower since not all daughters are scored. Heritability affects reliability and, as a result, different type traits have different reliabilities. Low heritability will cause somatic cell score (SCS) reliabilities to be lower than for milk, even if all daughters have SCS data. Heritability of productive life (PL) is the lowest of all traits with PTA's calculated on a national basis. Heritability is less than 10% for PL expressed in mature daughters and less than 5% when predicted from type data on young cows. Reliability is a guide for how much to use a bull whose PTA is high enough to justify any use. I recommend that Reliability for milk be used for this purpose because it is the most important trait economically and many AI progeny testing programs were designed to evaluate production. There is no magic number which makes a bull "safe" for heavy use. Bulls with sample code "S" had semen distributed to 40 or more herds and are being marketed by the organization which sampled them. Reliabilities for milk for such bulls usually is in the high 70's to low 80's. Privately sampled bulls may have Reliabilities in the same range. However, there are many ways for a bull to have daughters in many herds. Daughters of some bulls are predominantly bred and reared on a single farm and trucked out to other herds just prior to freshening. This looks like, but is not equivalent to, a good AI progeny test. AI sampled bulls with a complete "first crop" in milk and some daughters in second lactation can be used to breed 40 or 50% of a herd if PTA's are high enough. Privately sampled bulls almost never deserve such heavy use, regardless of how high PTA is for milk or any other trait. It is worth the trouble to check out how a bull was sampled rather than buying small amounts of semen on many proven bulls. If too many proven bulls are used, the best ones never get a chance to have a major impact on a herd. Be wary of bulls with high risk progeny tests, but don't be afraid to make heavy use of bulls which were sampled correctly.

Visit Virginia Cooperative Extension