BW question

[JHH]

I would encourage EVERYONE who is even slightly interested in the subject to go on Google or Dogpile or any other Search engine and type in "EPD's" - or -"Extra Progeny Differences.

DOC HARRIS

I THINK you mean Expected Progeny Differences. Some breeds use a different term but that seems to be the most common one that I see.

As for the birth weight question. Alot of it depends on acurate reporting of weight and how far back the reporting may go.

I asked a question similar to this one and I still think that you should have an average calf out of an average BW bull. But it would depend on the feed and the dam also.

EPDs are just tool or guide they are not perfect. JHH

 

[DOC HARRIS]

I THINK you mean Expected Progeny Differences. Some breeds use a different term but that seems to be the most common one that I see.

JHH - Yeah, you are right! I had a slip of the brain! If you have been reading any of my posts you probably are aware of the fact that I know what EPD's mean!! :lol2: :nod: But - I must be getting old!

DOC HARRIS

 

[JHH]

I THINK you mean Expected Progeny Differences. Some breeds use a different term but that seems to be the most common one that I see.

JHH - Yeah, you are right! I had a slip of the brain! If you have been reading any of my posts you probably are aware of the fact that I know what EPD's mean!! :lol2: :nod: But - I must be getting old!

DOC HARRIS

I know that, Just couldn't resist.

 

[Tom Underwood]

I read Dun's post with interest, since I have noticed that some highly accurate pulls produce highly variable calves in terms of BW,WW,MM such as Feltons 517. Other bulls produce more consistent offspring, all other things being equal (dam, environment, forage, etc).

Some bulls with highly variable offspring are "proven," because hundreds or more progeny have been recorded among many herds. Some very consistent bulls are less proven, albeit the bull's standard deviation for BW, WW, etc within given contemporary herds is less than proven bulls.

Any thoughts in terms of variability versus accuracy for EPD's among the experts?

 

[DOC HARRIS]

Tom Underwood - Your question is indicative of a lot of thought and well-phrased. In Biological terms, "variable" means 'tending to deviate somewhat from the 'type'.

Accuracies signify a reliability which can be placed on the particular trait or characteristic, and is established by the number of progeny and ancestral records determining the indicated mathematical symbolic numbers. Highly accurate EPD numbers are very reliable. Conversely, the performance of an individual with low accuracy values may be different from what his EPD's might indicate and what you might expect. Accuracy figures will NOT tell you what an individual's progeny will be. They tell you whether the EPD is reasonably reliable - or - just a good guess! AND - one vital factor which you must remember is this: there are TWO sets of genes, TWO sets of EPD's, and TWO determinations of accuracies to consider in every individual. Some breeder's seem to have a tendency to concentrate their attention on "Just the Bull" - or - "Just the Cow" - and overlook the fact that it takes "Two to Tango!" to produce that perfect seedstock individual.

My thoughts regarding "variability" versus "accuracy" encompasses the notable fact that the accuracies are influenced by TWO parents, whose EPD's will flucuate. It is difficult - yea- impossible to establish "- - -all other things being equal", particularly with the 'other' gender!

One more critical factor to consider which will have a bearing on the determination of 'accuracy' numbers is the phenotype of the opposite gender from the one being considered. Phenotype and Genotype must be considered in all judging and determinations.

DOC HARRIS

 

[Tom Underwood]

DOC,

I am fairly new to the cattle business, but I am an old hand when it comes to statistics. I have repeatedly asked the Hereford Association for the equations used for EPD's, and they have always refused to divulge the information. I think it would be useful for breeders to study the equations and offer ways to improve it, especially breeders with advanced mathematics degrees and some practical experience.

I have observed that some bulls with hundreds of progeny and high EPD accuracies for a particular trait (say, BW) produce a wider range of results than other bulls that are similarly proven with similar accuracies.

To be more specific, let's assume that we're talking about two bulls with a 0.90+ accuracy for BW and an EPD value of 4.2. My experience is that one bull will produce 100 calves with an average birth weight of 85 in my region, with two thirds of the calves having "adjusted birth weights" between 81 and 89 pounds. The other bull will average birth weight of 85, with two thirds of the calves having adjusted birth weights between 75 and 95 pounds. Further, in statistical terms, "tails" get longer, meaning there are instances of 65 and 105 lb calves in the case of the second bull.

I am trying to make my question very focused. It's not about calving easy or birth weight, but about how EPD's account for the consistency versus the variability of a bull in terms of calculating EPD's. I believe most scientists and breeders believe that the greater the genetic variability of the animal, the greater the variability that should be expected in its progeny, whatever trait is being observed.

Finally, I am not attempting to imply whether variability is good or bad. If one's looking to change, variability is good. If one's looking for consistency, variability may not be so good.

Personally, I am so new to breeding and have so much room for improvement that I am attempting to collect all of the genetics I can from within the Hereford breed among hundreds of animals. I carefully maintain my "controls," and I am enjoying the experiments. Obviously, my goal is not to market cattle in the next few years to other registered breeders, nor is it to maximize the dollar return over the next several years from beef produced on my farm sodl to feeders or packers. I limit myself to Herefords because I do not have the resources or time to consider multiple breeds, and I picked Herefords because of their historic characteristics, abnormal genetic variation among the current breed, and personal preference in terms of appearance.

 

[MikeC]

Tom, from what I understand, it's all about averages. Too many variables in genetics to do otherwise.

You might e-mail Dr. Keith Bertrand at the Univ. of Georgia for the eqautions. I believe he formulated, or at least had a hand in doing so.

There are a few examples on the BIF website.


J. Keith Bertrand
Professor
B.S., University of Florida
M.S., Iowa State University
Ph.D., Iowa State University

Edgar L. Rhodes Center for ADS
University of Georgia
Athens, GA 30602-2771

Phone: 706/542-0964
Fax: 706/583-0274
Email: [email protected]

Description of Research and Teaching Interests
The main thrust of Dr. Bertrand's research program deals with the development, improvement, and implementation of genetic evaluation models for the analysis of large sets of livestock field data with the ultimate goal of predicting breeding values for economically important traits. Live animal research projects are also conducted to test the efficacy of predicted genetic values under real world conditions. Dr. Bertrand is the leader of the beef cattle genetics research group within the Animal and Dairy Science Department. Currently, this group provides genetic values in the form of expected progeny differences for 10 beef breed associations located in Canada and the U.S. Undergraduate curriculum responsibilities include teaching two courses: Introductory Animal and Dairy Science and Introduction to the Genetics of Livestock Improvement. Dr. Bertrand is active in undergraduate advising and in graduate advising at both the M.S. and Ph.D levels.

 

[Tom Underwood]

I do not believe there are too many variables today. I used to work with many millions+ row databases in huge data warehouses when I worked for a large Internet company. Even if we only work with averages, are we talking mean, median, or mode? Further, I believe it would be entirely possible and pragmatic based solely upon statistics to assign a relative value to the data collected per breeder.

Finally, why not put up a copy of each breed's database (all of the tables, unfiltered) and let whoever would like to query them in any way they'd like? The data wouldn't be harmed since it's only a copy, and perhaps some crazy breeder looking for some pattern that no one else is could develop a herd that would change the breed. That's the way things work in the Internet world. I'd personally pay the few thousand dollars in servers it'd take for the AHA to make its data available.

Thanks for the contacts, and I look forward to seeing what I can learn from them.

 

[rocket2222]

I do not believe there are too many variables today. I used to work with many millions+ row databases in huge data warehouses when I worked for a large Internet company. Even if we only work with averages, are we talking mean, median, or mode? Further, I believe it would be entirely possible and pragmatic based solely upon statistics to assign a relative value to the data collected per breeder.

Finally, why not put up a copy of each breed's database (all of the tables, unfiltered) and let whoever would like to query them in any way they'd like? The data wouldn't be harmed since it's only a copy, and perhaps some crazy breeder looking for some pattern that no one else is could develop a herd that would change the breed. That's the way things work in the Internet world. I'd personally pay the few thousand dollars in servers it'd take for the AHA to make its data available.

Thanks for the contacts, and I look forward to seeing what I can learn from them.

Tom, years ago they used to print the formula for epd's in the Hereford World. It was a fairly simple thing. Now they incorparate so much data into each epd, hundreds of the animals offspring are taken into acount, where they are raised, contempry groups, "age slicing" is also used in the calculation. They will not give you the formula to calculate epd's because they don't know it. All data is sent to a lab, I think is in New Zeland, a company {computer} does the calculations there.

 

[Tom Underwood]

I was told Australia, but I was likely misinformed. I know it may seem complex and that it would take thousands of rows, but does that mean that the association should assume that breeders cannot comprehend it and therefore not offer it to them, even if the association itself cannot understand it?

My study of EPD's indicates that there is a large assumption of "mean reversion." In other words, EPD's place enormous weight on both the average EPD's for an entire contemporary group and the EPD's (however unproven they may be) of the dam and sire. If an extreme outlier appears to a highly unproven cow, the numbers do not change much. Even if the outlier repeats with regularity, the numbers do not change much. Because of this, dam EPD's are much more concentrated that sire EPD's. My basic understanding of genetics is that a dam contributes half of the genetics, and one would expect dam EPD's to follow the same curve as sire EPD's. After all, EPD's themselves deal with expected differences, not accuracies.

In pure statistical practice, you would expect a greater variance in EPD's among the progeny of unproven animals than among the progeny of proven animals.

I am not posting to disparage EPD's. I am a TPR breeder, and I collect and use data and find it valuable. I can also think of reasons that it would make sense to tilt EPD's towards proven animals. While not statistically sound, the results could be pragmatic.

 

[rocket2222]

On Feltons 517, he came along at the very end of the big frame era. There were alot of breeders out there that had alot of big frame cows with high BW's, good growth and no milk. 517 had it all, low BW, growth and MILK. Needless to say he was used alot. Some of his swings in BW WW etc numbers surely came from being bred to very large cows with huge bw epd's. He was also used alot to bring up the WW and YW and keep the milk on low BW cows with no growth numbers to speak of. This would account his low actual bw numbers. There were a couple of breeders that IMO that really came to the top of the breeders list during this period in time, one was Feltons the other was Witherspoons/MSU with Z03. They ignored the trends of the day and bred well balanced cattle with balanced epd's, and it payed off for them.

 

[Herefords.US]

Tom, here is a Powerpoint presentation that I found that contains some simplified explanation about how EPDs are derived.

http://www.ansi.okstate.edu/course/5303/lectures/epd/epdnotes.ppt

EPDs from the AHA appear to come from a complex matrix that's solved (twice a year) based on all the information gathered up to that point.

I'm not surprised that anyone at the AHA couldn't give you an actual formula. The basics of it could be easily explained to someone of your background, but I'd really be surprised if anyone at the AHA has the expertise and the full understanding of it themselves to be able to explain it. You'd probably have to talk to a professor that not only has a background in animal science and genetics, but considerable experience in working with statistics and complex matrices.

One of the big variables is the percent heritability of a given trait.

George

 

[rocket2222]

Yes, it could be Australia. I think epd's should be more biased towards the animals actual performance than a bunch of relatives.

 

[TEXANGRANCH]

I have only owned angus since the whole bw epd has become popular, but my collected data shows that a 70 lb calf is average for an angus under moderate diet/conditions/fleshyness, and that for every full count higher than 0, they add 5lbs to actual weight. For instance a bw +1 bull on average throws 75lb calves, a bw 2+ throws 80 lb, and so on. Therefore a bw +5 epd should throw a 95lb calf. A mature cow can handle that, but not always. We now use a maximum of bw 2.5 and we very rarely have a calf over 84lbs. I have only studied the angus, and other breeds have different scales. Also remember that actual bw depends alot on their intake at the time of the 3rd trimester.