Another AGR Question

[Peregrine Farm]

Long time listener, first time caller.

I'm trying to get some information (hopefully from the geneticists on board) to assist with a research project regarding thoroughbred influence in top eventing horses.

Back in October, vineyridge asked about AGR, and was provided links that explain the theory behind it. She was actually asking on my behalf, to some extent.

Getting past what it is, is it viewed as being a more valid indicator of influence than blood %? i.e., is the science behind it accepted as valid? Is AGR controverisal? Can the geneticists out there comment?

On a related question that has probably been dealt with here before (I have seen differing views on this), is an analysis of blood % or AGR in a group of horses helpful, and if so, where does an analysis of data like blood % or AGR become unhelpful?

For example, let's say a horse has the following data line for Hyperion:

Name Linebreeding Lines Blood % Influence AGR
HYPERION 7s x 6S 2 2 (1) (1) 2.34% 6x7 3.32%

(The horse is Courageous Comet, by the way -- a top eventer).

So we've got 2 lines, and Hyperion shows up in the 6th and 7th. The blood % is below 3%, and the AGR is only a little better. Is this a non-event, genetically? Is there a blood % or AGR cut-off after which one can confidently say that there is little-to-no genetic impact for an ancestor?

On its face, 3% of a pedigree seems like no big deal. I have seen commentary that it is all a crapshoot, or that really only the first 2 generations matter, or that one should look at generations 4-6 (the "engine room" of a pedigree) as the primary source of information. The theory behind AGR suggests that older ancestors can push through to greater levels of influence than their generational position in the pedigree or blood % might suggest.

Is there a consensus on this? Even if there is no consensus, what are the leading theories? If I've got a modern horse with 20 crosses to St. Simon, a blood % of 4%, and an AGR of 2%, does that say anything about the horse?

Thanks for any insights.

PF

[Peregrine Farm]

Bumping this up. Does anyone have any answers or comments to the original post?

Thanks in advance.

[Shammy Davis]

Sounds like an interesting discussion, but my grasp of basic genetics is lacking. From the original AGR link provided by PZ, it appears that their is a mathematical formula but I'm not sure exactly what significance to the layman can be derived from the results. I'm still at a loss to what the coefficent of inbreeding really tells us. From what little I've read, it appears that no scientist, except for the deepest thinkers, seem interested in AGR.

There are some "deep thinkers" among the members. I'm surprised none of them have taken the leap and jumped into this thread with both feet.

[xfactor fan]

Are you using the Sporthorse data base?

[Pan Zareta]

For example, let's say a horse has the following data line for Hyperion:

Name Linebreeding Lines Blood % Influence AGR
HYPERION 7s x 6S 2 2 (1) (1) 2.34% 6x7 3.32%

(The horse is Courageous Comet, by the way -- a top eventer).

So we've got 2 lines, and Hyperion shows up in the 6th and 7th. The blood % is below 3%, and the AGR is only a little better. Is this a non-event, genetically? Is there a blood % or AGR cut-off after which one can confidently say that there is little-to-no genetic impact for an ancestor?

Iirc, in an open population the cut-off for statistical significance in regard to genetic 'relatedness' is about 0.333...%. But the TB is a closed population so the cut-off is probably higher. Personally, I'd consider Hyperion if not a genetic non-event then an exceedingly minor and relatively insignificant event in Comet's ped.

If I've got a modern horse with 20 crosses to St. Simon, a blood % of 4%, and an AGR of 2%, does that say anything about the horse?

If the horse is a TB I'd say he has about the average amount of St. Simon.

[vineyridge]

PZ, there are two major factors in eventing, in particular, that are needed for success, other than the usual trainability and a good gallop. One is scope and power in jumping; the other is movement that is dressagey because of the way dressage is judged now. What I believe Peregine Farms is attempting to do is identify which TB lines have historically produced top quality (4*, Olympic and WEG) top finishers. As far as I know, the data that PF is using comes from the PQ and Allbreed subscriber databases.

I've been investigating this for a few years, although not in the same depth, and it does truly appear that there are certain TB lines that do excel over time. For instance, I'm almost convinced that the movement needed for the dressage phase is 99% correlated with Bay Ronald in all his manifestations, and the closer and more ancestors the better. I also know that probably 99% of all TBs do have some or much Bay Ronald. In my research, Hyperion has showed up at least once and often more on the first page of the pedigree in ALL the pure TBs who have done GP dressage worldwide in the past thirty years. There are very few of them, and I will admit to developing hypotheses on the basis of history rather than numbers. The Germans developed their dressage horses from the Dark Ronald line of Bay Ronald, and they still look to that line when they want to add TB.

Bolero, the great German dressage stallion was 3/4 TB. His sire was Black Sky (1966), and his damsire was Bleep (1956). Those stallions brought almost all of the great TB lines for sport together, often in multiples (Blandford, for example).

I know the jumps racing people in the UK and Ireland have a bloodlines "recipe" that they use in breeding; what I, and I believe PF also, would like to develop is a bloodlines recipe for eventing.

[xfactor fan]

There's a huge difference between looking at the pedigree, and looking at the horses in person.

For example what is it about Hyperion that translated into eventing? Was it conformation? Muscle type, flexibility? Then look for Hyperion lines that have these same qualities.

Suppose you were looking to breed palominos, just looking at the pedigree. And go to Milkie. Breeding to two chestnut Milkie lines will never get you a palomino. But if you look at the pedigree, select for the quality you want, then do the breeding there's a much higher chance of striking gold.

I suspect a lot of the information that PF is looking for can be found in the Sporthorse data base. Wonderful resource. And they respond and fix mistakes. One of the things that is nice about that site is the ability to post multiple photos and videos.

[brogers]

There was a paper just released recently that looked at the inbreeding coefficients in thoroughbreds and stated that the 'paper inbreeding' coefficient did not have a strong association with the actual genetic inbreeding of the horse in question. Some were more and some were less inbred than their 'paper inbreeding' coefficient suggested. This sort of discounts any evaluation of horses based on this in my mind.

[vineyridge]

While I'm sure this is true with genome level study, how is a breeder supposed to evaluate whether a stallion and mare would be a good fit? All that is normally available now is "paper" and the actual horse phenotype and accomplishments. But if the latter, then doesn't breeding just come down to 'breed the best to the best and hope for the best"? That's just pure gambling on the genetic lottery even if the breeder tries to use phenotype for improvement.

Are you suggesting that every horse prior to breeding should have a genome study done and a matching service based on the genome of the particular horses? That could get VERY expensive when you break breeding down to that level.

There was a paper just released recently that looked at the inbreeding coefficients in thoroughbreds and stated that the 'paper inbreeding' coefficient did not have a strong association with the actual genetic inbreeding of the horse in question. Some were more and some were less inbred than their 'paper inbreeding' coefficient suggested. This sort of discounts any evaluation of horses based on this in my mind.

[Peregrine Farm]

Thanks for the responses. What vineyridge reports about my research is correct, and I also use the sporthorse-breed-database as well when needed. It has a little less functionality when it comes to generating linebreeding reports.

Perhaps the historical approach is all we can really take into account unless we are getting genome level studies, given the variability in what genes and associated traits are being transmitted for any given breeding. I would hope, however, that over a sufficiently broad enough sample set, you'd start to see some averaging and perhaps trends. Viewed from that perspective (i.e., not from the perspective of one individual horse) does using AGR or blood % make sense?

The way I am starting to look at it is whether the AGR% of a distant ancestor is the equivalent of a single horse that is closer up in the pedigree. For example, a single horse in the fourth generation is (presumably on average over a large sample set) contributing 6.25% of the genes. I think most people assume that there is still some influence from horses in generations 3-5. When distant ancestors are linebred sufficiently to get to that high of an assumed AGR%, I will use them in evaluating the pedigree as a whole. We then have to use the historical perspective to evaluate whether the distant ancestor would contribute qualities we are looking for in eventers (although the same thing could be said about racing -- they are just somewhat different qualities - speed, stamina, temperament, etc.).

Is that approach valid (acknowledging all of the inherent variability) in trying to draw breeding conclusions? Even if the analysis is imperfect, isn't it better than nothing?

And if it is of little or no value, anybody know why PQ provides linebreeding reports and AGR information?

[Pan Zareta]

And if it is of little or no value, anybody know why PQ provides linebreeding reports and AGR information?

Finding individuals that are more or less inbred then suggested by the inbreeding coefficient derived from pedigree alone is to be expected. Such reports are based upon averages. They don't account for the fact that very few individuals receive exactly 25% of their autosomal recombinant DNA from each grandparent, exactly 12.5% from each great-grandparent, etc. Once you reach the 5th generation (or equivalent) the average contribution is 3.125%. It could be more, but one must also acknowledge the possibility that it could be less, or none at all.

[Peregrine Farm]

It could be more, but one must also acknowledge the possibility that it could be less, or none at all.

Thanks for keeping this discussion alive PZ. I recognize that it is not perfect information; it will be some time before we get anything close to that, if ever. For predictive purposes, however, is there a consensus that it is better than nothing (and what most of us realistically have to work with)?

Thanks, PF

[Shammy Davis]

I recall reading a published paper by the late Ann T. Bowling at UCAL-Davis where she discussed the use of the term "blood related." As I recall, she suggested that the use of the term is very confusing as "blood" really isn't an aspect of genetics and can't be quantified.

I know from other sources, particularly when speaking to "inbreeding" that if a breeder is to pursue inbreeding a particular line, it is recommended that breeding twins (realizing in horses that this is rare) or sire/dam to their foals should be avoided and, for obvious reasons, that siblings and half siblings are the better choice. I'm not the brightest candle on the cake when it comes to genetics but I know from my experience breeding Chesapeake Bay Retrievers and horses that in the case of line breeding, a litter when all the pups have a consistent look and temperaments that is the optimal result. It is so important, IMO, that breeders visually recognize the qualities they are seeking and then visually examine the results to insure breeding success. I for one am lost when it comes to this gene mapping, but as long as I have my glasses on in a kennel or paddock I can visually recognize and compare traits and characteristics.

Dr. Bowling in another article about genetic myths advises that horse breeding is inituitive supported by a great deal of luck. Neither of which is teachable to other people. I guess she was saying that with breeding there are "no teachable moments" when it comes to horse breeding.

[cewright]

This is an interesting thread and thanks to all for the information. I am trying to understand how specific traits are stamped onto a line of horses. My assumption is that the expression of such fixed set of traits is what defines a breed. I am familiar with the history of the King ranch and their work to establish a model quarter horse using the tool of line breeding to fix desired traits into their population. This work has been continued in the QH world by other large breeders. For example, several years ago at the QH Congress, Carol McWhirter gave a delightful history of the breeding program at their ranch http://danmcwhirter.com/home/ which is well known in the western pleasure world. Seemingly, these breeders identified physical traits which correlate to desired performance. ( Perhaps an example of breeding the best to the best repeatedly?) One of the nuggets passed on by Carol was the observation that they monitored the success in their program by watching the cull rate. As the desired trait became more established in their breeding population, the number of culls dropped and finally flattened out.

It is my understanding from that presentation that genetically, the horses at McWhirter ranch are as diverse as any other population of QHs. Perhaps there is a genetic index that could be used to identify a unique population but unless one knew exactly what the uniqueness was and how it correlated to performance it would be useless. Further, the TB breeders do not have the large herds of similarly bred mares to establish type as seen in the QH world. I believe one is much better off looking at the individual in front of you and making your own evaluation based on your goals and experience.

Thoughts? Insights? Corrections?

Thanks

Chuck

[Pan Zareta]

For predictive purposes, however, is there a consensus that it is better than nothing (and what most of us realistically have to work with)?

The linebreeding analysis definitely has its uses but imho among 'paper evidence' is less predictive than, for example, the performance records (in any & all disciplines) for the sire, first three dams, and their progeny. Fwiw, I breed and purchase working ranch and sport (H/J) horses. After looking at the horse itself (and/or the sire & dam), that information is what I want to see. It can tell you a lot about soundness, trainability, and aptitudes. It can also tell you what's worked particularly well, or not, before in terms of matings, which is when I start taking a look at the linebreeding analyses.

she suggested that the use of the term is very confusing as "blood" really isn't an aspect of genetics and can't be quantified

Literally, "Blood %" as shown in the linebreeding reports here equals the average amount of recombinant DNA contributed by a particular individual.

My assumption is that the expression of such fixed set of traits is what defines a breed. I am familiar with the history of the King ranch and their work to establish a model quarter horse using the tool of line breeding to fix desired traits into their population.

Expression of a fixed set of traits is what defines a type. "Breed" in its modern usage implies more arbitrary qualifications, usually based on pedigree. The KR fixed a particular type without regard to breed. They did lean heavily on a "breed" (TB) in doing so, but only those that exhibited and could reproduce the traits they wanted. Somewhat ironically, the type fixed by the KR was not the "bulldog" type QH the AQHA was originally determined to preserve and perpetuate.