A Classic Case of Coincidence?

re: A Classic Case of Coincidence?

Byron Rogers — Tue, 11 Jun 2013 21:14:44 GMT

Sceptre,

Correct. The performance data related to their racetrack performance. The production related data for the sire and dams in that generation wasn't easily done as you had horses born where the sire data for the grandsire, let alone the sire, was still developing so that "point in time" data was impossible to do. It was also really hard to come up with a standard of production performance across the entire breed (working on that!). With that said, there is a strong relationship between racetrack performance and subsequent production, even with opportunity normalized for, so the racetrack performance values suffice.

If you take something as simplistic as Equinome's genetic test, Dawn Approach is a C:C variant in intron 1 of MSTN, his sire New Approach is a C:T variant and his sire Galileo is a T:T variant. With MSTN having an influence on muscle fiber type proportions (as per Petersen, et al), Dawn Approach is a completely different horse to Galileo in terms of muscle physiology. This is where I think that genetics (recombination), regression of statistical data and heritability studies agree in thoroughbreds. The third generation and beyond of a foal in terms of real influence on outcome is just a name on a page. 5 years ago when we developed TrueNicks any combination beyond the 3rd generation on the sire's side and the 4th on the dam's side (so the grandsire of the sire and dam), recessed to the norm making their calculation meaningless.

Breeders hold onto "observations", "suspicions" and "theories" like passages from the King James Bible. No better one in recent memory than the well constructed fable of the X-Factor - take a long passed ancestor, attribute mystical powers to her and weave a story for all to believe. That was something straight out of the Da Vinci code..."the lost sacred feminine".

I do however take on board your latter comment about scrutinizing the statistical studies though. Many don't consider opportunity or variables that influence outcomes.

re: A Classic Case of Coincidence?

sceptre — Tue, 11 Jun 2013 18:05:45 GMT

Obmar:

I haven't read your Dr. Boyd's paper, but can say with absolute certainty that your interpretation/application of its message is incorrect. While it is nearly impossible to inherit a 100% COMPLETE chromosome from a grandparent, etc.-owing to crossing over, etc., it is extremely likely that a goodly amount (on average, appox. 25%) of each grandparent's chromosomal material will be inherited by the grandchild. I suggest you do some more reading.

re: A Classic Case of Coincidence?

Obmar — Tue, 11 Jun 2013 08:43:23 GMT

Startling in the concept and astonishing in its aspects is the recent breeding premises of a renowned scholar, Dr. William C. Boyd, Author and member of the faculty at Boston University of Medicine. Dr. Boyd a recognized authority on genetics and Physical Anthropology has offered up a breeding principle which, If true, makes a virtual farce of millions of words of turf literature for a century; he make the breeding student who pores over the lineage charts a chump and a fool; and make those of us who believe in the blood of broomstick, Domino, and *Teddy the subject of a colossal guffaw. Dr. Boyd states alas, that recent studies have shown that an individual horse or human my not inherit even a single chromosome from its grandfather and that the chance of inheriting this single chromosome from his grandparent is 1 in 8,388,608. Now as we see it, the fact that scholar Boyd admits, by implication, that a chance does exist, even though one in eight million, that a Citation say, can get a chromosome from his grandad Bulldog.

1950-1952.

re: A Classic Case of Coincidence?

sceptre — Mon, 10 Jun 2013 18:51:09 GMT

Hi Byron,

Very nice of you to respond, and I hope you enjoyed your trip. You offer several interesting observations, and allow me to discuss a few:

When you say third and fourth generation "performance data", should I assume you're referring to the specific racing performance data of those influences, i.e. (ex.) Buckpasser appearing in the 4th and his earnings/start, etc.--rather than the data related to him as a sire, etc.? No doubt the data related to the latter would be a far more cumbersome, if not impossible task, but likely far more relevant. So, let's take a look at Tapit's pedigree, and his Foggy Note. I think you're suggesting that Foggy Note has little if any impact on Tapit's success as a stallion. I can't PROVE otherwise, but I stongly suspect it does-and not just by ASSOCIATION.

My very strong guess is that the variability of (gene) inheritance among full-siblings is the central cause for the disparity, and not "foal rank of the sire". That said, you're alluding to chromosomal aberration as a cause-i.e. the older the stallion, the greater the liklihood for chromosomal aberration and its negative effect on outcome. I think the jury's still out on this, but I've suspected as much-for some stallions more than others (Raise A Native and Tom Rolfe come readily to mind).

Much of this debate centers around one's personal (life-long) observations vs some statistical studies. So, which better mirror reality? Before you embrace the latter (statistical studies) perhaps better first scrutinize those studies.

re: A Classic Case of Coincidence?

Byron Rogers — Sun, 09 Jun 2013 13:08:02 GMT

Sceptre,

I had meant to reply to this earlier but I have just got back from a trip to Australia and getting over the jet lag.

This type of pedigree analysis is primarily where Alan and I don’t so much disagree, but diverge in opinion.

While I would be the first to admit that Alan has significantly more experience and success with breeding racehorses, unlike him, I give very little stock to anything that happens outside the third generation of a pedigree. I used to, but it just didn't stack up to statistical scrutiny. I am more of the “Big Data” Joe Estes/Tony Morris/David Dink type with the immediate ancestors providing much of the genetic merit of a given horse.

Most of the modeling that I have done with data to date gives significant weighting to the racing class of the immediate parents. That is no surprise to you. In the case where the racing class is unknown, as it often is with mares that are unraced, the model has created performance based relationships between the grandparents with the racing class of the broodmare sire and granddam of a foal being somewhat important in that respect. I have tried throwing in third and fourth generation performance data to see what impact it has on the performance of its descendant but the regression model kicks these data points out as being non consequential even when I think that there is not a lot of performance data in the first two generations to make a decision on (hence my thoughts about them just being names on a page).

That is not to say that pedigree relationships are not important. Alan sort of implied something that we have modeled that looks promising in terms of prediction and that is coefficients of relatedness (to 6 gens) and coefficients of inbreeding (to 8 gens) with hypothetical matings between immediate ancestors. One of the more interesting one is the hypothetical mating of the sire and the broodmare sire for instance (which leads towards nicks in general).

In the case of “full siblings” as you mentioned above, there are actually quite a few data points that really discriminate between their potential as racehorses. One of the more interesting ones is the “foal rank of the sire”, that is, what foal number was the foal in the sires career, was it the first foal born by the sire or the 752nd? That has a decent impact on the outcome. Zabeel just sired his 43rd individual G1 winner last weekend but for me the fact that he had 10% SW/Fls first 10 crops but from his current 3yo, 4yo and 5yo crop he has 226 foals with just 7 Stakes winners (3.5% SW/Fls) tells me something and it is not related to the quality of mares that he covered. Ditto, but less influential is the foal rank of the mare.

Alan was talking about writing something further on all of this which would be interesting to further this discussion.

Byron

re: A Classic Case of Coincidence?

Alan Porter — Wed, 05 Jun 2013 12:00:02 GMT

MK Austin

Re: inbreeding and linebreeding. While the definition might not be the same as with most of the livestock world, the convention with thoroughbreds has become to call inbreeding a duplication or multiple instance of an ancestor within four generations, and linebreeding duplications or multiple instances at five or six generations. In fact if you look at any five cross pedigree generated by The Jockey Club through Equineline, you will see multiple ancestors listed as "Inbreeding."

re: A Classic Case of Coincidence?

MK Austin — Wed, 05 Jun 2013 02:50:22 GMT

The terms "inbreeding" and "line breeding" seem to get tossed around, almost interchangeably.

A statement was made regarding Flower Alley: "inbred X3 Mr. Prospector". Upon review of the page, it was noted that Mr. Prospector was in the top and bottom of the pedigree in the 3rd gen., while Northern Dancer was present in the top and bottom in the 4th gen. This represents line breeding, NOT inbreeding. Inbreeding refers to extremely close crosses, such as colt/dam, something I've never seen in selectively bred Thoroughbreds. "Inbreeding" is often misused.

re: A Classic Case of Coincidence?

Alan Porter — Tue, 04 Jun 2013 16:34:55 GMT

This is not exactly the easiest stuff to concisely define, and it's a question of trying harmonize what we know about genetics and what we observe.

I would also stress the importance of "starting at the front" when considering a mating - that the sire and dam need to be a good match in terms of aptitude (depending on what you are hoping to breed), physique (phenotype), and temperament.

I also want to look at the TrueNicks rating, to confirm at minimum, that the cross has not been negative, and now the Key Ancestor report, which indicates what strains have and have't been positive influences for similarly-bred mares.

I'm also not advocating any kind of breeding theory as a substitute for class. I want to start with the best stock available.

Having said all that, I believe than inbreeding or linebreeding through similarly-bred channels often has an influence for upgrading (it's telling that a disproportionate number of generally disappointing stallions have done their best work when sent mares that give this kind of "clever" pedigree reinforcement).

I would also say that I would regard multiples of related ancestors further back in pedigrees than many are inclined to look, can exert a positive influence.

I guess the "tension" between genetics and observation comes when say that "all genes come from somewhere" - like the single source of Native Dancer's gray coat color - but note that any one ancestor in the sixth generation is likely to have contributed only around 1.5% to genetic make up of the foal.

So if we are going to seize on "Nearco in the sixth generation" and say that is why a horse is good - to credit one ancestor with as the source of a positive genetic variant, we're on dodgy ground.

However, it's not quite as simple as that. It's highly likely the difference between the elite and non-elite thoroughbred comes down to the additive effect of relatively small number of gene variants. In the human, there are about 200 different genes that have been associated with athletic performance, but given that the thoroughbred has been bred for performance for over 300 years (or 30+ generations), genetic variation in this regard is probably a lot less than it once was, a lot of positive traits have been fixed over the generations (remember, even a pretty slow thoroughbred would be an elite athlete compared to most non-thoroughbred when it comes to running over distances that a the norm for thoroughbreds).

What we call a "nick" would frequently be down to the sire and broodmare sire having genotype that gives a high probability of inheritance and expression of the maximum number of positive variants (and with those in harmony with the potential phenotype).

So if we have two stallions bred on a successful "nick" it's very possible that they have inherited similar variants, and might cross well with mares from similar backgrounds. Thus, when we inbreeding and linebreed to successful "nicks" or other positive combinations, the chances of the positive variants that worked well together being transmitted are far higher than in a more outcrossed pedigree (at least as far as we can go from a paper pedigree). It's why, as we discovered with our original TrueNicks study, that the accuracy of the ratings doesn't drop off over the generations in the way one might expect. Not specific inheritance, but the tendency of the same things to go on working with the same things

It's likely that there might be optimal degrees of inbreeding and linebreeding - that not only maximize the chance of expression of the positive variants that a parent has, but also brings in positive variants that the parent lacks (where outcrossing will diminish the likelihood of reinforcement, and close inbreeding is likely to fix existing positives, but less likely to contribute "missing" ones).

So using our Blenheim II/Gainsborough example, can we say for sure that Paynter, Oxbow, Awesome Patriot and Ghostzapper - the four stakes winners bred on an Awesome Again/Relaunch cross - inherited any specific positive genetic variants from those specific instances. No, absolutely not. Do I think that the similarities - of which Blenheim II/Gainsborough are an integral part - have something to do with why those four horses - three out of the top drawer - are the only stakes winners for the Awesome Again/Relaunch cross. You betcha!

So, hard to attribute specific genetic variants for performance to any specific ancestor that deep into the pedigree, but easy to say such patterns are very often associated with success (although TrueNick ratings and Key Ancestor reports might frequently save us the trouble of having to dig them out). Where it gets a little more blurry is if you can ask whether the patterns have caused the "right" variants to come forward, even though they are not attributal to any one individual!

re: A Classic Case of Coincidence?

Obmar — Sat, 01 Jun 2013 09:04:40 GMT

" Once you start talking about names beyond that point, their impact in terms of performance as a trait recedes towards zero which is why people correctly discount the influence of distant ancestors." Byron are you saying you beleive that Nearco in the 6th or 7th generration has 0 impact on term of performance. If you do than why key on a

ancestor in a report. Help me here. why did Alan mention the Blenheim II/Gainsborough cross. If it ment 0 in term of performance.

re: A Classic Case of Coincidence?

sceptre — Fri, 31 May 2013 16:26:31 GMT

Much admired Alan's short, but important piece. Its general message should give pause to those who debunk the value of pedigree research. I suppose it was necessary to include those anecdotes, but doing so often adds fuel to the naysayers-for it's then so easy to offer anecdotes to the contrary (just imagine how more "convincing" yours may have been had Tiz Roses and All Tiz proved better runners)...A more thorough explanation of Alan's points would require books-as a matter of fact Alan has written a few on this very subject of patterns...At first, I felt it better not to engage in this discussion-Alan's piece was rather pristine, so why, perhaps, muddy the waters. But then I noticed Byron's comment, and while no doubt genuine, and authored by another learned source, may have in fact muddied the waters. I think the concept of heritability can be rather tricky and often misinterpreted in application. For example, the fact that "paper" pedigrees" may offer a relatively low % of performance predictability should not imply that the ACTUAL genes inherited offer the same degree of impact on performance-ex.: two full siblings share the same "paper" pedigrees, but can have quite different genetic compositions. Yes, the vagaries of intra and extra-uterine environment can substantially impact, but as these are today likely far better "controlled" I doubt their combined impact is in the range of 60-70%...As to Byron's distiction between "cause" and "association", I think, within this context, none exists. As somewhat an aside, our present understanding of genetics makes it entirely possible that a relatively distant ancestor could exert a meaningful contribution (+ or -) to performance. While it may be difficult, if not impossible, to employ this concept in present mating practices, should not negate this reality.

re: A Classic Case of Coincidence?

Byron Rogers — Thu, 30 May 2013 22:22:13 GMT

Jean

Most population studies on the heritability of performance as a trait peg it at ~30-40%. This means that while the parents of a horse contribute 50% each (roughly), in terms of the trait of performance they are responsible for ~15-20% each of the variance of performance (half of the 30-40%). The grandparents are then ~7.5-10% and great grandparents ~3.25-5%. Once you start talking about names beyond that point, their impact in terms of performance as a trait recedes towards zero which is why people correctly discount the influence of distant ancestors. Coming at it a different way, statistician David Dink has come to a similar conclusion in his work. The names recess to the norm after three generations and are generally just noise.

That is looking at cause, it is not to say that distant ancestors aren't associated with performance. There is a difference between cause and association which gets lost in these types of discussions.

re: A Classic Case of Coincidence?

jean-in-chicago — Thu, 30 May 2013 01:07:18 GMT

First of all, thank you, Alan, for using the word 'impact' correctly.

I always become quite annoyed with people who say discount the possible influence of ancestors more than a few generations back. If you look at Native Dancer's pedigree, there is only a single gray horse per generation for 17 generations back. It isn't until the late 1700s that multiple grays appear and that 2 grays are actually bred to each other.

Coat color is an obvious trait that is easy to follow. Who knows what other genes have been passed down for many generations and when they will appear together in a horse. Personally, I spend a lot of time counting up how often Nearco shows up through out a pedigree.