Inbreeding and using COI to analyze potential pairings (yet more on Pedigree Dogs Exposed)

One of the topics that Pedigree Dogs Exposed brought up was inbreeding, or the practice of breeding together close relatives in order to come up with a more predictable result than you’d get if you bred two unrelated dogs.

(people who know this already can check out here and resume close to the end, because I want to explore exactly what inbreeding is and why breeders do it)

In my family (my own background) there are closely related (first cousins or closer) women who are as short as 5’2″ and as tall as 5’11”, and from very light-boned and tiny-framed to big and broad. In Doug’s family, there are women from 5’0″ to 5’8″ and who span a similar range of frame sizes. That means that when we had female children, even though I am 5’6″ and he’s about 6’0″, we would never expect every child to grow up to be my height or even between my height and his height. It’s quite possible that I carry a bunch of genes that code for shorter height than I myself show, and it’s quite possible that I carry a bunch of genes for taller than I am. His situation is similar.

And, in fact, this is borne out in our family. We have an average to tall daughter with a strong frame, an average daughter with a narrow frame, a very short and tiny daughter who is proportional, and a baby who (so far) is tall but light.

We did not reproduce predictably when it comes to height and skeletal weight.

On the other hand, we have very little variation in our families when it comes to hair and eye color and skin color. We both come from families that are exclusively fair, with either blonde/brown or red hair, and eye color is very consistently blue. So we had absolutely no surprises when it came to the skin, eye, and hair color we produced–everybody’s blonde or red and has extremely light skin and blue eyes, and there’s no possibility for anything different. In this area, we will always reproduce very predictably.

This is, basically, just the way it works in dogs. If you breed together dogs with a wide variety of genetic expression behind them for a particular trait (this is also known as heterozygosity), you will produce dogs with a wide variety of genetic possibilities and therefore many different physical appearances.

If you breed together dogs with a narrow range of genetics behind them for a particular trait (homozygosity), you will produce dogs with a very narrow range of physical traits.

This can also work if you have just one parent who is very homozygous for traits. Most of us have a lot of different alleles that control which genes are expressed. I don’t want to get too much into junior high genetics, but if I’m Aa Bb Cc dd, and Doug is Aa Bb Cc Dd, our kids could be (AA, Aa, aa), (BB, Bb, bb), (CC, Cc, cc), (Dd or dd). Those lowercase pairs are recessive traits. If, on the other hand, I’m Aa Bb Cc dd, and Doug is AA BB CC DD, his capital letters will overwhelm all my lowercase letters and none of the possible recessive traits will be expressed. This will create a group of children who look much more alike than they would if he had a mixture of dominant and recessive alleles and so did I.

One of the most potent ways to reduce heterozygosity and reliably produce offspring that look very similar to the parents and to each other is to breed to relatives. You can immediately see how this would work–nobody’s contributing that black hair from across the ocean, or that olive skin.

And so this has become a major tool in the dog breeder’s repertroire.

Over a long period of time, you can develop an extremely reliable system for making predictable dogs. (This is not, by the way, how most breeds are formed–this is how some breeders operate within established breeds. It is important to keep this in mind.) Breeders who do this can end up with such a distinctive look to their dogs that it’s immediately apparent whose dog that is across the ring, even if you’ve never seen the dog before.

And it’s not only true that those breeders have very predictable or distinctive looks in their own dogs; because of the lack of heterozygosity their dogs are very likely to produce that look even when bred to unrelated dogs. So you can get that look by choosing one of their stud dogs to breed with your bitch, even though your bitch doesn’t look much like them at all.

An important variant on inbreeding is LINE-breeding, which is a form of inbreeding where the genes of one particular individual or a closely related individual are concentrated in the resulting puppies. For example, if I have a bitch and I really like her, but I think her great-grandfather on her mother’s side had a prettier head, I would not breed her to a relative on her father’s side, or to her own brother. I’d breed either to that actual great-grandfather or to his brother or to another dog related to him. I’d try to concentrate the genetic material of that handsome g-grandfather as much as I could, in an attempt to create puppies that look as much as possible like HIM, not like any other dogs.

Line-breeding is still inbreeding, but we give it a different term to show that we’re not just concentrating genetic material randomly (as you would if you bred brother to sister). You’re doing it to try to re-create the aspects you like of a single dog or group of dogs in the pedigree.

So why are you talking about goldfish in the title, then, you may ask.

COI is actually short for coefficient of inbreeding. The dog’s COI is a numerical expression of how many shared ancestors he or she has. For example, if you look at just the dog’s parents, the COI will be zero, because the parents are (obviously) not the same dog. But if you go back another generation, you see that the parents were half-siblings. This throws the COI up to around 15%. If the parents were full siblings, the COI is 25%. If the parents were full siblings AND the grandparents were related AND the great-grandparents were related (or the grandfather one one side was also the great-grandfather on the other side, etc.), you can quickly get up to a COI of 30-40%. At this level the puppies are quite inbred.

What often happens in dogs is that when you’re looking at two paper pedigrees, which usually list four generations, and imaginging a cross between these dogs, you see a few shared relatives but not a huge number. So you can conclude that you are not breeding closely related dogs. But when you expand the search to ten generations (from 64 dogs to 1024 dogs) you very frequently find that in fact you are breeding the equivalent of cousins. The ancestors in common were found in those prior generations. So the breeding you thought was between dissimilar dogs is not. There are several breeds that have high COIs across the board, like Australian Shepherds and Standard Poodles. The average COI in those breeds makes the typical breeding, even when very few or none of the dogs in the five-generation pedigrees match, closer than first cousins. You don’t see that until you get back to ten generations or so.

You can keep pushing COI further and further back, but at some point it becomes less useful because you start hitting the founding dogs of the breed and that can artificially inflate the COI (because those are behind every single dog in that breed). So ten generations is considered pretty standard.

OK, here’s where we come to the controversial part of my little tale.

Many dog breeders use the COI to help choose breeding partners, but they actually push toward a HIGHER COI rather than a lower one. They will seek out those individuals with high COIs because those are the ones that are going to make a more predictable puppy.

The generally accepted “formula” in show dog breeding is to breed closely (make high COIs) for the majority of breedings, bringing in an “outcross” (a lower COI) only occasionally. This is, invariably, related to me as “what my mentor told me to do” or “a respected older breeder told me to do.”

As I said, this is an incredibly potent method for making predictable looking puppies. You can get yourself a lot of champions that way.


I’ve been involved in many more species than dogs, and I can tell you that the only breeders who have the philosophy that closer breeding is preferable to outcrossing are the dog fancy and the import-bred Arabian horse fancy. The rest of every other group I’ve been involved with (cow, goat, sheep, rabbit, other horse) thinks that this is CRAZY and INSANE. They don’t even use the words like we do–it’s not an “outcross” if you breed to an unrelated sire; it’s just a normal breeding.

They breed relatives only for specific purposes and they don’t keep doing it, or they may deliberately inbreed for a couple of generations to make one inbred sire that can be used on a lot of unrelated females to create a specific result, but then they don’t keep inbreeding. The vast majority of their breeding decisions put together completely unrelated animals.

If you have no genetic heterozygosity, you have vastly reduced resistance to disease across the population (because every dog has the exact same genetic resources to use to fight disease; there won’t be some that do better than others), and you have a very real danger of ending up in a genetic corner with nowhere to go. There are about a hundred other reasons, some more or less important, but the bottom line is that there’s a reason that animals evolved behaviorally to seek out the least-related genes to pair with.

I understand the drive toward predictability in type, but we’re facing a situation in purebred dogs where so many breeders have done this for so many decades that we have a super tiny gene pool even in the common breeds. This is a situation that is instinctively understood to be unhealthy by just about everyone who is not breeding show dogs; as I said, in every production- or longevity- or health-based species (in other words, where “success” in that species means the ability for it to consistently do a job) they do not follow this strategy. The elevation of predictability–not even type, it has to be predictability in type–above other considerations is something that needs to be done with extreme caution.

It’s also very pertinent that we are under fire from welfare groups and the general public for doing this kind of inbreeding–and honestly there’s no difference in terms of genetics between linebreeding and inbreeding; those are labels we dog folk put on it to say we are inbreeding to a specific dog or set of dogs, but it doesn’t mean a lot objectively–and this is going to become a battleground.

The breeders of production species have managed to create very high-quality breeding programs without using a high COI. Showing goats or showing sheep or showing rabbits is just as demanding as showing dogs– in fact, in goats (with which I am most familar) it’s actually MORE demanding because the judges have to give critiques and because you have a linear appraisal system where the animal is compared to a mythical perfect goat. So your goat is exposed as crappy no matter how many other goats it has beaten in the show ring.

That’s an important lesson, I think. They did it, and continue to do it, by pairing animals that move toward a desired look and a desired production level, without using close breedings.
I do want to be clear on this, now that I’ve made everybody mad: COI tells you nothing about whether the two dogs are going to produce high-quality offspring. I could get a really low COI by breeding to an entirely different breed, but that doesn’t mean it’s a good move. You still have to choose a stud dog or bitch based on his or her quality and health testing and so on. And a low COI shouldn’t trump the more immediate issues of temperament or disease. COI is one more tool you use to structure a breeding program, especially a long-term breeding program, and it helps you understand how your prospective matings will or will not support your efforts and what I hope are the efforts of your breed club.

Committing to a lower COI as much as you possibly can–not to the detriment of the dogs, but as a general rule–will create a breed that is substantially more sustainable over the long term (I mean decades or centuries here, which is–I hope–an important part of how breeders are planning their breedings).
If I can say one thing about this and have it be remembered, it’s this:

There’s a story I heard years ago of a young woman who was putting in the roast for Sunday dinner. Her guests watched as she carefully cut off a portion of one end of the roast before putting it in the pan to cook. One family friend asked “Wow, that’s really interesting. Why do you do that?” She said, “My mom, who was the best cook I know, taught me to always do this.”

Fascinated, the friend called the woman’s mother. “Why do you cut the end off the roast?” She replied, “My own mother, who was a fabulous cook, always did it, and I have always thought it was important too.”

And so the friend called the elderly grandmother, related the story, and said “So what is it that makes cutting the roast end so critical?” The grandmother laughed and laughed. “My goodness!” she said. “I didn’t have a big enough roasting pan. I had to cut the end off the roast to fit it in the pan!”

The time of just following what an older breeder told us is over. Own your own decisions! You must know what you are doing and be able to explain exactly why you are making the breeding decisions you are making, and “so and so told me it worked” is not enough. We are breeding in a totally different and often overtly hostile environment and, if we are not already, we’re going to be under an incredible amount of scrutiny.

Inbreeding is a huge part of why we’re perceived as borderline animal abusers. So if you choose to inbreed (and you can’t avoid this by looking at a short pedigree; the many-generation COI is important), you need to be able to justify it based on facts and studies and in-depth knowledge of the dogs and the pedigrees, not on what someone else always did.

Or, someday, you or somebody in your breed is going to be sitting in an interview room being made to look like at best a fool and at worst a pervert. If you think I’m overreacting, go watch Pedigree Dogs Exposed.


3 thoughts on “Inbreeding and using COI to analyze potential pairings (yet more on Pedigree Dogs Exposed)

  1. Pingback: Implications of the Kennel Club decision on the Pekingese: Pedigree Dogs Exposed « Ruffly Speaking: Railing against idiocy since 2004

  2. While I agree with you generally, your take on COI is off. Outcrossing decreases “genetic diversity” by melting all of the strains and family types of a breed into one pot. The problem is you gloss over the negative recessive traits and those traits come back to haunt your breed, you have no repository of “clean” genetics to turn to.

    I totally agree that breeders need to know what they are doing and why they are doing things. You can’t ethically breed dogs by following a recipe, and you should have a strong idea of where you are going not only in this generation, but for several generations to come.

    Please Google “Of Peas and Pups” as it is the best free source of information on the topic.

    Wonderful and beautiful blog. You have fantastic passion for dogs and an informative and warm style that really shines through.

  3. Thank goodness! There is someone in purebred dogs who agrees with me! I have been preaching against inbreeding (and yes, that does include linebreeding) for years and I think I’ve run into about three people in the entire sport who agree with me. The rationalizations for inbreeding are incredible and creative. They usually end in accusing me of not understanding genetics. As you point out, it is only dog breeders and Arabian horse breeders who think this is okay. Shouldn’t that tell us something?

Leave a Reply

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

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

Facebook photo

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

Connecting to %s