“Heritability” is another word that requires some explanation. You do not even hear it or read it all that often outside the field of genetics. It is mentioned only twice in Racehorse Breeding Theories, for example. Considering that “heritability” affects ALL breeding theories, you would think that the word would pop up more often than that in the course of a fair-sized book on the subject of breeding theories, but it does not.

The dictionary lists “heritability” only as the noun form of the adjective “heritable,” which is defined thusly: “1. Capable of being inherited : HEREDITARY. 2. Capable of inheriting or of taking by inheritance.” Which is not very useful.

My last post generated lots of comments, more than any other post I have ever written, which is gratifying. It also generated some back and forth among readers, which is also good. I had to step in only once to clarify a point.

At this point I will quote a particular exchange between two readers. Bill wrote:

“So If only the sire and dam are important. Then breeding a champion to champion should = a champion.”

Ned Williams replied:

“No. Still very difficult to breed a champion. The randomness of genetics makes it very difficult. However, your chances are improved. By this I mean you have a greater chance than if the champion + champion is 2,3,4 generations back in the pedigree.”

The latter’s response to the former was absolutely correct. It captures the essence of “heritability” without using the word. You might say that “heritability” is the “randomness of genetics” that makes transmission of racing class so difficult and unpredictable other than in general terms of probability. It is more frequently referred to as regression toward the mean, the average, the norm, take your pick of the word for “middle.”

Now I will quote two paragraphs from Racehorse Breeding Theories, Chapter 2, “Genetics and Gene Mapping,” page 23. Dr. Gus Cothran is an animal geneticist with the University of Kentucky.

“A sensible approach and a reasoned practicality are necessary for any successful breeder, and Cothran threw in a comment that ought to make anyone think long about his operation. He said, ‘Most studies of racing ability find that its heritability is about 30 percent, and that’s not surprising, when you consider what it is.’

“Racing ability is not a genetic trait like coat color. So it does not transmit from parent to offspring in the same way that a purely genetic characteristic would do. Athletic ability in the Thoroughbred is the result of many traits, many factors, and many relationships among the horse and his caretakers, health, environment, and so forth.”

The statement that heritability in the Thoroughbred is only about 30% means that genes account for only about 30% of the final outcome in terms of racing ability or class. All other factors account for the remaining 70%. Some studies say that heritability is more like 40%. It is somewhere in that range of 30%-40% anyway. And it is definitely less than 50%.

Now I am going to try to explain this matter in my own words, using an analogy (as usual). Imagine that it were possible to measure racing class in the Thoroughbred on a scale from 0 to 150 or so, with 100 being average, 150 being a “superhorse,” and 50 being the lowest-class claimer (albeit a winner).

What happens when you mate a sire who is a 140 (the equivalent of a G1 winner) with a mare who is also a 140????? As Ned Williams explained to Bill, you should definitely NOT expect that the average progeny of such a mating to be 140 as well. If heritability is 30%, you should expect the average progeny of such a mating to be a 112. If heritability is 40%, you should expect the average progeny of such a mating to be a 116.

How do I arrive at those numbers???? If you breed a 140 to a 140, you should expect to get a foal who is also a 140 only if heritability is 100% (which it is not). If heritability is 30%, then you would expect to get a 112 on average because 12 is 30% of 40, and 40 is how much both parents were better than average (100). So you should expect an average result of 112 (100 plus 30% of 40, or 12).

If you mated a 140 sire to a 130 dam in this same scenario, use the average of the sire and dam and multiply by .30. The average of the sire and dam would be 135. Three-tenths of 35 is 10.5. So you would expect a 110.5 (100 plus 10.5) on average from such a mating.

All this math makes a certain amount of sense if you think about it. Say that 140 is a G1 winner, 135 is a G2 winner, 130 is a G3 winner, 125 is a decent stakes winner, and 120 is a cheap stakes winner. Therefore, 112-116 would be a decent allowance horse. And a decent allowance horse is about what you would expect on average from mating a 140 to a 140.

But as Ned Williams pointed out, you have a much better chance of getting a 140 from the mating of a 140 to a 140 than from the mating of a 130 to a 130 (or a 120 to a 120 or a 110 to a 110 or a 100 to a 100 or a 140 to a 100, etc.). And increasing your chances of getting that 140 (a G1 winner or better) is what breeding is all about.

I should probably stop at this point and give readers some time to assimilate these thoughts. My next post will provide some specific numbers on how “heritability” decreases “influence” using as an example Northern Dancer and some other sires.

This entry was posted in Uncategorized. Bookmark the permalink.

17 Responses to “Heritability”

  1. Rod says:

    Does 2×2 still equal 4?

  2. jim culpepper says:

    Good genes or bad, offspring cannot inherit what is not there, only the ” genetic variance” of what is present, and just like the average hand of cards, most often the offspring is average, even if you double the number of Aces and other high cards. Even though mutations are thought by laymen to serve as wild cards, odds are 100,000 to 1 that they are a harmful wild card. Breeders can increase the count of high cards by different means and still will hit the middle of the bell curve most of the time.

  3. Byron Rogers says:


    You haven’t quite got the term “heritability” correct.

    Factors including genetics, environment and random chance all contribute to the variation between individuals in their observable characteristics. The heritability of a trait within a population is the proportion of observable differences in a trait between individuals within a population that is due to genetic differences. Because heritability is a proportion, its numerical value will range from 0.0 (genes do not contribute at all to phenotypic individual differences) to 1.0 (genes are the only reason for individual differences).

    For instance, some humans in a population are taller than others; heritability attempts to identify how much genetics are playing a role in part of the population being extra tall and how much of that is environment or other factors. Heritability is measured by estimating the relative contributions of genetic and non-genetic differences to the total phenotypic variation in a population. Heritability measures the fraction of phenotype variability that can be attributed to genetic variation.

    This is not the same as saying that this fraction of an individual phenotype is caused by genetics. That is an important distinction to understand. A heritability of 0.40 informs us that, on average, about 40% of the individual differences that we observe in, say, speed may in some way be attributable to genetic individual difference. It does NOT mean that 40% of any horses speed is due to his/her genes and the other 60% is due to his/her environment which is what you have incorrectly represented above.

    Heritability of a trait, in our case racing speed, can also change without any genetic change actually occurring. In terms of thoroughbreds, it is possible for the heritability of speed as a trait to increase if the environmental variation decreases, causing individuals to show less phenotypic variation. As an example if a racing environment starts to decrease the variability of races and homogenize towards sprinting (so all important races are run between 7-9 furlongs), the heritability of speed increases because non-genetic factors are contributing less variation and genetic factors become more predictable. Thus heritability is specific to a particular population in a particular environment.

    • ddink55 says:

      And you have failed to explain why this “is an important distinction to understand.”

      • Byron Rogers says:

        You said “The statement that heritability in the Thoroughbred is only about 30% means that genes account for only about 30% of the final outcome in terms of racing ability or class.”

        That is not right. The 30% represents the variability that we can attribute to genetic differences in a population. Heritability is a population concept. It tells us nothing about an individual. No matter what the numbers are, heritability estimates tell us nothing about the specific genes that contribute to a trait.

        Heritability measures the fraction of phenotype variability that can be attributed to genetic variation. This is not the same as saying that this fraction of an individual phenotype is caused by genetics.

      • ddink55 says:

        OK, that is clear as MUD.

  4. Ned Williams says:

    All I can say is: “Now I’m getting confused”!! I thought heritability was the percentage chance that a trait would be carried on in the next generation. While I had to read ddink55’s post twice, the math makes sense and we know that the genetics make sense. He clearly explains that it is a factor of 30% or 40% going forward. I am not sure I quite understand what Byron is driving at. As I said in an earlier post “I am a slow learner”. Hmm…

    • Byron Rogers says:

      Ned….David might be clear, but he is wrong.

      It is not the percentage chance of something being carried on, it is the variability of that trait across a population. I clearly can’t explain it well enough but here is a link that does a pretty decent job of it (better than I could do here at least).


      • ddink55 says:

        “Need to Know and All That Good . . . . ”


        BR might be right, but he is splitting hairs and obfuscating the important issue. The link is actually somewhat helpful if you want to try to follow the “logic” behind BR’s hair splitting.

        So here is what I am going to do. I am going to stop using the word “heritability.” The WORD itself is unimportant. The important issue I am trying to describe is how “influence” diminishes from the racehorse to the sire and from the sire in the first generation to that same sire in the second generation. Perhaps a better word for what I am trying to describe is “regression” (as in regression to the mean, the average, the norm, take your pick of a word for “middle”).

        (Call me any name you like, dear, I will never deny it/Farewell, Angelina, the sky is erupting, and I must go where it is quiet.”)

        BR is not arguing that regression does NOT exist. Only that “heritability” is the wrong word for it. OK, I can live with that.

        All you really need to know is that regression does exist and that it does operate on both individuals and populations. How can it be otherwise???? if something does NOT pertain to individuals, how can it pertain to populations???? Populations are just collections of individuals. If a thing does not affect individuals, it does not affect populations, and vice versa.

        Incidentally, BR still has ZERO CREDIBILITY as far as I am concerned. He is one of those persons who loves to split hairs and thus obfuscate the important issues. “The lady doth protest too much, methinks.” I could tell you all the reasons why BR has ZERO CREDIBILITY, but I don’t think I will do that. After all, “Diplomacy” is my middle name, as you may or may not have noticed. Need to know and all that good . . . .

  5. Ned Williams says:

    So I now understand the meaning of heritability, however let’s get back to the meat of ddink55’s post, which I contend is accurate. I believe that it is fundamentally correct to argue that breeding a “140” to a “140” (or champion to champion as Bill put it) gives one a better chance of getting a stakes class performer. However, the variability of genetics (and/or regression to the mean as ddinks55 calls it) does not make it a likely proposition, only more likely than if you breed a lesser performing first dam and sire. I would further contend that one would have a better chance at producing a stake caliber horse by breeding a “140” to a “140”, even if neither had a “deep or blue blood pedigree”, as opposed to crossing a “magnificently bred” first Dam and a “magnificently bred” sire who have poor racetrack performance records. The key for me, is that racing class is still a better predictor of a foals potential than a pedigree’s list of names and crosses that are several generations back. As I said earlier, I am a slow learner, however, Why am I wrong in my thinking???

    • Byron Rogers says:


      I should have said earlier that while I took umbrage at the use of Heritability, the proposition that David suggests is in agreement with a lot of what we see in terms of populations of horses.

      There is increased variability of results if you use contrasting phenotypes, i.e if you bred a great sprinting mare to say Tiznow, you have more chance of getting a slow one than if you bred the same mare to say Speightstown but in general, while there can be improvement made if selected for, the breed tends to go back to the norm, even with the best material used.

  6. Ned Williams says:

    Thanks. I am learning. By “best material” do you mean racing class or pedigree? (Or best case, if you are quite wealthy, both)

  7. Bill says:

    Carry back comes to mine. So if Saggy was a 130 at best. Joppy was 112-116 how did Carry back
    become a 150 with looking at just sire and dam. or was Carry Back a 120 running against 100.
    2. Law of Regression states that average parents tend to produce average offspring; Plus parents tend to produce plus offspring and extreme parents tend to produce offspring which are less endowed than their parents. I think this happend with Secretariat and Citation, I think you have to seperate Racing ability and Siring ability which proves one does no lead to the other.
    if two Olympic champions marry and have children, their average child will certainly not be an Olympic champion, although usually better than the general population, The children simply are not likely to inherit the extremely fortunate combination of genes that create greatness. And recent genetic studies are showing that many genes are not simply dominant and recessive as once assumed, but “co-dominant” in nature, so as to work in combination…Conversely, two very ordinary humans could have a child that becomes an Olympic champion—not because of a great environment, but by the luck of the genetic lottery their parents’ genes happened to combine in an unusually fortuitous manner.

  8. Ned Williams says:

    Bill when you state: ” I think you have to seperate Racing ability and Siring ability which proves one does no lead to the other.” You are missing a central point in all of this. Obviously one does not lead directly/exclusively to the other. However, the better the racing ability of one’s parents, the better chance one has of getting above average racing ability in the offspring. The problem lies in the genetic mix found in each individual. This is random and cannot be accounted for. This is what makes breeding so difficult and interesting (and why we are talking about it here). What ddink55 has stated so well and done the difficult work (math) to verify, in earlier posts, is to debunk the concept than names and crosses that are found back in a pedigree have meaning to predict racing potential. (ddink55 I am not trying to insult you by putting words in your mouth… I apologize and please clarify if I have). What I feel to be true, having tried to listen and learn (and I have learned quite a bit from ddink55’s postings) and through first hand experience, is that racing ability (of the first dam and Sire) is a better predictor of racing success. This is not to say that it is foolproof in any way. It is simply better than using pedigree. I would be thrilled if you or someone else could give me a better predictor for a foals’ future racing success. This is the Holy Grail of breeding thoroughbred racehorses. I am guessing that you are searching, like me and others, for the best answer to this vexing question. Good luck and please share your thoughts and answers as we try to tackle this question. Ddink55 has done some very heavy lifting to help us from following blindly down dead end roadways, for that I am very thankful.


    • ddink55 says:

      No need to apologize at all. You are dojng a fine job on your own. Saves me some work. Some effort anyway. Sometime I forget that all this is FUN (and not WORK).

      You did correctly glom on to the most dubious of Bill’s statements. Yes, racing ability and siring ability are two separate things. And one does not always lead to the other. But your BEST chance of finding a good sire is among the BEST racehorses. Try breeding only to unraced sires and see how long you last.

      My only other thought is that Carry Back was not a 150 (but probably right around 140). Joppy (his dam) had a record of 7-0-2-0 for an SSI of 0.19. She was well below a 100 in terms of racing ability. Such dams do produce 140s, but not very frequently. They appear to produce 140s more frequently than they really do only because they constitute such a LARGE portion of the broodmare population. Ditto for unraced mares.

      Keep up the good work. Boojum

      PS Call me Boojum if you like. It is much easier to deal with than ddink55.

  9. Ned Williams says:

    Thank you. I have read many blogs, books and articles and tried to learn from each of them. However, it was not until I found your writing that I began to feel like things were becoming clearer in an admittedly very murky endeavor. We will know we have had some success when foals turn into runners and win on Saturday afternoons.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s