Coefficient of inbreeding, explained.

0:00 Same dog, three different COI results

Picture three officers in the same club. They each look up the same dog, and they each run its coefficient of inbreeding.

One gets three point one three percent. One gets four point six nine. One gets five point four seven.

Nobody made a mistake. All three did the math correctly.

They counted a different number of generations.

I'm Dusty. I'm a breeder, and I built Breed Ledger, which is software for breed clubs and registries. In this video I'll show you how that number is calculated, by hand, and why it moves. Then we'll go through the five decisions your registry can write down, so your members know what a COI on your records means.

0:49 What coefficient of inbreeding measures

COI stands for coefficient of inbreeding. It's the probability that the two copies at a given place in a dog's DNA, one inherited from each parent, trace back to the same ancestral copy. Geneticists call that identical by descent.

If the sire and dam share no known ancestors, the pedigree COI is zero. The more closely they're related, the higher it goes.

Two things it is not. It isn't a health result. The U.K. Kennel Club describes it as a measure of risk, and it doesn't tell you whether a given puppy will or won't have an inherited condition. And it isn't a verdict on the dog. Temperament and health testing still matter when two dogs are being paired.

1:31 Calculate COI by hand: Wright's path method

The standard way to calculate this from a pedigree is Wright's path method, published by Sewall Wright in the nineteen twenties. You can do it with a pencil.

Find an ancestor that appears on both sides of the pedigree. Trace a path from the sire, up to that ancestor, and back down to the dam. Count the dogs on that path, including the sire and the dam. Then multiply one half by itself that many times.

Here's the simplest case. The sire and dam are half siblings. They share one parent. For these examples, assume the shared ancestors aren't inbred, and nothing else connects the parents.

The path is the sire, the shared parent, and the dam. That's three dogs. One half, times one half, times one half, is one eighth. Twelve point five percent.

If they're full siblings, they share two parents, so there are two paths. Each one is worth twelve point five. Add them, and you get twenty-five percent.

That's nearly the whole method. Add the contribution from every valid path connecting the sire and dam through a shared ancestor. A path can't pass through the same dog twice. And the same ancestor can contribute through more than one valid path. There's one more rule, and we'll get to it in a minute.

2:47 Why more generations change the number

Now let's take an example dog. We'll call her Wren, and open her pedigree one generation at a time.

With two generations, parents and grandparents, there are no shared ancestors. The COI is zero.

Open a third generation, and a dog named Cedar appears on both sides. The path runs from the sire, up through Cedar, and back down to the dam. Five dogs in all. One half, five times, is three point one three percent.

That's the first officer's number.

Open a fourth generation, and two more shared ancestors appear. Each path has seven dogs on it, and each one adds about point seven eight percentage points.

That's the second officer's number.

Open a fifth generation, and something different happens. We find that Cedar's parents were full siblings. Cedar was inbred. In this example, the extra information changes Cedar's own COI, rather than adding another valid path between Wren's parents.

That's the rule I saved. When the shared ancestor is inbred, its path counts for more. You multiply it by one, plus that ancestor's own COI. Cedar's COI is twenty-five percent, so Cedar's path grows by a quarter. That adds another point seven eight percentage points.

Same dog. Same records. Three correct answers.

And this isn't only a tidy example. The Institute of Canine Biology recalculated real dogs at different depths. One went from about six percent at five generations to about ten percent at ten. Their advice on how many generations to use is, as many as you can.

4:21 How missing ancestors affect COI

Here's the part that catches people out, and it matters for any registry with older records.

Standard pedigree calculations usually treat founders, the animals whose parents aren't recorded, as unrelated to one another and not inbred. They have to assume something. An unknown ancestor gives you no path to count, so it adds nothing.

That means a thin pedigree can look better than a complete one, without the dogs being any less related.

Think about what that does in a club. A dog with twelve generations on record, carefully researched, can show a higher COI than a dog whose papers stop at three. The first dog isn't necessarily more inbred. You can see more of its family.

So read a pedigree COI as an estimate based on the ancestry recorded. Missing generations can hide shared ancestry and make that estimate too low. A low number from a short pedigree is not strong evidence of low inbreeding.

5:15 Pedigree COI vs. DNA results

Sooner or later, a member will send you a DNA test with a different number on it, and ask which one is right.

They estimate inbreeding in different ways. A pedigree COI predicts identity by descent from recorded ancestry. A genomic estimate uses the dog's own DNA.

It's commonly based on runs of homozygosity: stretches where the tested DNA markers match on both chromosome copies. The proportion of the assessed genome in those runs is used to estimate inbreeding.

Inheritance involves chance, so two full siblings have the same pedigree COI, but they don't carry exactly the same DNA. A pedigree can't see that difference. A DNA test can.

But a genomic result has its own conditions. Testing companies can use different markers and different cutoffs for what they count. So a number from one lab may not line up with a number from another, or with the pedigree figure.

A good habit for a registry is to label each number with where it came from, and to keep them separate. Don't average them, and don't compare one dog's pedigree COI with another dog's DNA result.

6:19 Five decisions for your registry's COI policy

None of this is a problem if your registry has decided how it handles COI, and written it down. Then a new registrar doesn't have to work it out again, and members know what they're looking at.

Here are the five decisions I'd put in front of a board.

One. Depth. How many generations do you count? Whatever you choose, state it next to the number. Five point four seven percent, five generations, tells a member far more than the percentage alone.

Two. Missing ancestors. How do you handle pedigrees with gaps? I'd show how complete each pedigree is alongside its COI, so nobody mistakes a short pedigree for a low-risk one.

Three. Lab results. Do you record genomic COI at all? If you do, record which lab it came from and keep it apart from the pedigree figure.

Four. Visibility. Who can see a COI: the public, members or officers? Your members may feel strongly about this one, so decide it on purpose.

Five. Use. Is COI information only, or is it part of a rule? And if it's part of a rule, who decides that, and how do they change it?

You'll notice I haven't given you a number to aim for. That's deliberate. Breeds start from very different places, so one universal cutoff doesn't travel well. The U.K. Kennel Club's guidance is to breed puppies at or below the breed average where possible, which is a benchmark built from that breed's own records. That guidance is about the puppies. For a proposed pairing, calculate the expected COI of the offspring. Don't average the parents' own COIs. And if you use a breed average, calculate it on a comparable basis, and record the reference population and the date it was calculated. A five-generation result isn't directly comparable to an average built from all the pedigree data available. Whether your registry uses something like that is a decision for your board, with your breed's data in front of it.

8:12 Free resources and Breed Ledger

I've put all five decisions into a one-page COI policy template. It's free, and the link is in the description. Bring it to your next board meeting. You'll find a link to the COI calculator there too, so you can run the same dog at different depths and see the number move.

I built Breed Ledger for clubs and registries. It calculates Wright's coefficient from up to eleven generations of recorded ancestry, using the same pedigree assumptions and path rules we've covered here. If you'd like to see it with your own registry in mind, you can book a demo with me.

Thanks for watching.