Genetic Diversity

Photo by Susan Trow.

Some high-profile people say that the top priority for making breeding decisions for all dogs, including Scottish Deerhounds, should be to increase genetic diversity. I believe this recommendation is a mistake that could harm, rather than improve, Deerhound health.

by John Dillberger, DVM, PhD

Reprinted from the September/October 2016 issue of The Claymore.

Diversity is a popular buzzword in the United States right now, and it always has a positive connotation. We are taught that exposure to people with different skin colors, ethnic backgrounds, religious beliefs, world views, and sexual orientations is not just a reality we must accept, but good for us. More diversity is good, and less diversity is bad.

Maybe that is why some folks have decided that too little genetic diversity is making pure-bred dogs unhealthy and are urging dog breeders to aim for more genetic diversity. To get a sense of what is being advocated, and why, visit a website called “The Canine Diversity Project”. There you can read dozens of essays and blogs on genetic diversity. But be warned: these are not peer-reviewed scientific papers. Consequently, you should read critically, just as you would listen to a political speech.

I believe that many of the people who advocate breeding for more genetic diversity have good intentions and genuinely want to improve the health and life span of dogs. But I also believe that other advocates are not so well meaning, and would like to use genetic diversity as a means to attack the entire concept of pure-bred dogs. More importantly, I believe that all of them, well-meaning or not, are wrong.

Let me be clear. I think that breeding for greater genetic diversity is unlikely to produce healthier Scottish Deerhounds. In fact, I fear that doing so could harm overall breed health. In this month’s column, I want to explain why.

Selective Breeding and Genetics

Scottish Deerhounds have arisen through hundreds of generations of selective breeding. Breeders have intentionally selected for and against various traits like size, appearance, temperament, hunting ability, reproductive capacity, health, and life span. All of these traits are influenced by genes. Some are governed entirely by a single gene, while others are influenced by multiple genes.

Here I must stop and explain some concepts and terms. First and foremost, I have to define the term gene, because authors often use this word to mean different things.

When I use the term gene, I mean a section of DNA that has a function. The gene’s function might be to provide the blueprint for making a protein. A dog is estimated to have about 19,000 different protein-coding genes. Or the gene might provide the blueprint for making a piece of RNA, or be the site where some protein or RNA molecule attaches and influences the function of other nearby genes. I can find no estimate of how many of these sorts of genes each dog has. Let’s assume there are about 1,000 of them, and just say that each cell in a dog’s body contains 20,000 genes. But this isn’t the whole story, for two reasons.

First, every gene can come in different varieties, which are called alleles (pronounced uh-leels). You can think of alleles like flavors of ice cream. Some genes exist in only one flavor, but most exist in several flavors. If we use a protein-coding gene as an example, different alleles will code for different versions of the protein, some of which may work better, or worse, or differently than others, and some of which may not work at all.

Second, each cell in a dog’s body contains two complete sets of 20,000 alleles, one set inherited from its dam and the other from its sire. So, while it is correct to say that each cell in dog’s body contains 20,000 genes, it would more accurate to say that each cell contains 20,000 allele pairs. For each pair, one allele was inherited from the dog’s sire, and the other allele was inherited from its dam.

If the alleles in a pair are the same, then the dog is said to be “homozygous” for that gene, meaning that the gene is present in only one flavor. On the other hand, if two different alleles make up the pair, then the dog is said to be “heterozygous” for that gene.

Actually, not every cell in a dog’s body has 20,000 pairs of alleles. The exceptions are sperm and eggs, which contain only 20,000 single alleles. And the 20,000 alleles in each sperm or egg will not be exactly like the set inherited from either of the dog’s parents!

During the process of making sperm and eggs, the two sets of alleles that the dog possesses are “scrambled” to produce a single new set that contains some alleles from the dog’s sire and some from its dam. Therefore, the set of alleles passed on to the dog’s pups is different from either set of alleles in the dog itself. Moreover, the scrambling process takes place individually for each sperm and egg, so that no two sperm or eggs are likely to contain the exact same set of alleles.

To sum up, every dog has the same 20,000 genes, but only identical twins have the same pair of alleles for every gene. Every other dog has a unique set of 20,000 allele pairs that define it as an individual. The dog will be homozygous for some genes and heterozygous for others.

Genetic Diversity

Genetic diversity is not a term one can apply to an individual dog. Instead, it applies to a breeding population, which could be a specific dog breed or a subpopulation within a breed, such as North American Scottish Deerhounds or a particular line of Deerhounds.

Genetic diversity is a measure of how many different alleles exist within a breeding population. For example, if we assume that each of a dog’s 20,000 genes has only two possible alleles, then the maximum possible number of alleles in any dog breed would be 40,000. If we found 30,000 different alleles in North American Scottish Deerhounds but only 20,000 different alleles in Australian Deerhounds, then we would say that Deerhounds are more genetically diverse in North American than Australia. In reality, some genes have only one allele, while others have dozens, and so the total possible number of alleles in the dog population as a whole may be much greater than 40,000.

No dog breed has all possible alleles. There are some alleles that make a dog look and act like a Deerhound and other alleles that makes it look and act like a Corgi. The breeders who created Deerhounds selected for the former alleles and against the latter alleles. As a result, the range of alleles that exist in the Deerhound population differs from the range of alleles in the Corgi population, and there are fewer alleles in either breed than in the dog population as a whole.

You can think of selective breeding as the process of weeding out undesirable alleles within the breed, thereby increasing the relative frequency of desirable alleles. The ultimate goal is to eliminate undesirable alleles entirely from the breed.

Of course, breeders cannot select for or against specific alleles; instead, they can select only for or against traits that are influenced by those alleles. The only exception is when a genetic test is available for an allele. So far, such tests are rare and have been developed only for alleles that cause health problems. They are used to select against the allele.

Actually, breeders cannot even select for or against single traits, because matings occur between dogs that each have good, bad, and indifferent traits. Breeders end up choosing sires and dams with as many desirable traits as possible and as few undesirable traits as possible. And whenever possible, they try to double up on desirable traits and avoid doubling up on undesirable traits.

There is one more challenge to selective breeding. Because alleles get scrambled more or less randomly into each sperm and egg, there is no guarantee that an allele responsible for a desirable trait will be present in the few of the sire’s sperm or the dam’s eggs that actually combine to make pups—unless the dog is homozygous for that allele. In that case, every sperm or egg will contain the desirable allele.

I said earlier that the ultimate goal of selective breeding is to eliminate undesirable alleles entirely from the breed. Now I can add a second goal, which is to have the breeding population contain dogs that are homozygous for as many desirable alleles as possible. A homozygous dog will “breed true,” passing on its desirable allele to every pup.

So if a dog “breeds true” for a desirable trait, then can a breeder assume the dog is homozygous for the allele that influences the trait? Unfortunately, no—while dogs that are homozygous will always breed true, dogs that breed true are not always homozygous. This brings us to a concept called “expression.”

For some alleles, a single copy is enough to produce the trait. Such alleles are said to have a dominant mode of expression. But for other alleles, two copies are needed to produce the trait. These alleles are said to have a recessive mode of expression.

To illustrate this concept, consider a hypothetical “coursing” gene that has two alleles. One allele provides a blueprint for a normal oxygen-carrying protein in blood, while the other provides a blueprint for a “mutant” oxygen-carrying protein that does not work as well. A geneticist might represent the normal allele with an upper-case letter “C” and the abnormal allele with a lower-case letter “c.”

If a dog has even one copy of allele C, then it can make the normal protein, and its blood will deliver oxygen normally to its muscles during exercise. But if the dog has only two copies of allele c, then it can make only the abnormal protein, which will not deliver oxygen as well to its muscles. A geneticist would represent the genetic makeup (or genotype) of a dog with two normal alleles as “CC,” of a dog with one allele of each type as “Cc,” and of a dog with two abnormal alleles as “cc.” For the coursing gene, allele C would be called dominant, while allele c would be called recessive.

In this example, the recessive allele “c” is undesirable because it reduces a dog’s top speed and stamina during a chase, and the dominant allele “C” is desirable. Coursing ability is a trait that a breeder might take into account. By choosing a sire and dam that course well, the breeder could hope to increase the likelihood of having allele C in the pups. But of course, a dog that courses well could be genotype CC or Cc, and therefore the breeder cannot know if the dog will pass on its coursing ability to all of its pups or only to some of them.

One final note: a dog with the Cc genotype would be called a “carrier.” This term could apply to either the trait (subpar coursing ability) or to the allele (“c”).

Breeding for More Genetic Diversity—A Worthwhile Goal?

I have explained how selective breeding can reduce the number of alleles within a breed, partly by eliminating undesirable traits (and the alleles the produce them) and partly by increasing the number of dogs that “breed true” with respect to a given trait because they are homozygous for the alleles that govern them. And fewer alleles equals less genetic diversity.

Is this a bad thing? Does less genetic diversity result in poor health, diminished reproductive ability, or shorter life? Some influential folks think so. For example, the Veterinary Genetics Laboratory (VGL) at the University of California-Davis maintains a website, which says this about Standard Poodles as a breed:

“…a majority of Standard Poodles are relatively inbred and contain a minority of the existing genetic diversity. This has resulted in an increased incidence of heritable traits, including… recessive disorders such as PRA, Von Willebrand’s disease, and neonatal encephalopathy; possible recessive disorders such as juvenile renal disease, juvenile cataracts, and enamel dysplasia; and more complex genetic disorders such as autoimmune disease (e.g., SA, AD, IMHA, ITP, thyroiditis, chronic active hepatitis, masticatory myositis), allergies, hip dysplasia, elbow dysplasia, atrial septal defect, patent ductus arteriosus, degenerative myelopathy, and bloat. These various disorders appear to have resulted from… [alleles] that have been concentrated in certain lines as a result of inbreeding.”

Notice that the VGL claims that this whole laundry list of health problems has “resulted from” low genetic diversity; in other words, they believe that low genetic diversity is the reason that all of these health problems are more common in Standard Poodles.

Based on this belief, the VGL proposes to improve breed health by increasing genetic diversity. Not only does the VGL believe that this solution will work, they believe that implementing it is so important that breeding for greater genetic diversity should trump all other considerations when one plans a breeding. As they put it:

“Potential sires and dams should be first screened for genetic differences in the genome… Considerations of mate choices for genetic diversity should be balanced with other breeding goals, but improving genetic diversity in puppies should be paramount.” [italics mine]

To help breeders increase genetic diversity, the VGL would like to create a “mate-selection” business for dogs, which they describe like this:

“[The] mate selection service… will allow a breeder to identify, among all of the dogs tested, potential mates that would be most ideal for increasing genetic diversity in their litters. This service could be on a subscription basis and hopefully key information on potential mates would also be included. This might include pictures of the animal, age, phenotypic traits such as height, coat color or pattern, behaviors, show and field characteristics, health status, physical location, breeding fees, etc.”

I find this paragraph shocking. The VGL’s mate-selection service would suggest appropriate mates based solely on increasing genetic diversity. Other factors—including appearance, behavior, conformation, performance, and health status—might be taken into account, but are secondary considerations at best!

To reiterate—the folks at the VGL believe that selective breeding based on traits has increased health problems in dog breeds because it has decreased genetic diversity. And they also believe that the only remedy is to change course completely and breed selectively based on genetic relatedness instead of on traits. They understand that this may mean sacrificing breed traits in order to increase genetic diversity. But they think the situation is so dire, and the threat so immediate, that this must be done. This is akin to saying that, in order to save a dog breed, we may have to unmake it.

I strongly disagree. For one thing, reduced genetic diversity is not the cause of health problems. It is true that inbreeding can both reduce genetic diversity and increase the frequency of an allele that causes a health problem. But that does not mean that the one is cause and the other effect.

The point of inbreeding is to increase the frequency of desirable alleles, while decreasing the frequency of undesirable alleles. But as already discussed, breeders must mate dogs, not alleles or single traits, and so sometimes they end up increasing the frequency of an undesirable allele as well as the desirable ones. When that happens, the VGL would fix the problem by breeding to the most genetically dissimilar dog possible, so as to “dilute” the frequency of the undesirable allele. But this will also dilute the frequency of every desirable allele, basically undoing the entire breeding program. It is throwing out the baby with the bathwater. For Scrabble players, it is like having the letters SEPARAX and deciding to discard them all and draw again rather than simply discarding the X to see if you can draw the T you need to play the word “separate.”

But beyond this objection, I worry that the approach being pushed by the VGL could actually worsen breed health. Let me use two examples to illustrate what I mean.

For the first example, let’s assume a gene that comes in only two flavors: a dominant allele that results in good health (“B”) and a recessive allele that increases the likelihood of a bloat (“b”). And let’s assume that the possible genotypes (BB, Bb, and bb) are equally distributed in the population. Table 1 shows the six possible matings and the results they would produce, on average.

Table 1.    Possible Matings and Outcomes for a Hypothetical Recessive Allele “b” that Increases Bloat Risk

  Chances that a pup will be:
Possible Matings Normal (BB) Normal, but carrier (Bb) Prone to Bloat (bb)
Genetically similar
BB x BB 100% 0% 0%
Bb x Bb 25% 50% 25%
bb x bb 0% 0% 100%
Overall average* 41.7% 16.6% 41.7%
Genetically diverse
BB x Bb 50% 50% 0%
BB x bb 0% 100% 0%
Bb x bb 0% 50% 50%
Overall average* 16.6% 66.7% 16.6%
*Because we assumed the parental genotypes are equally distributed in the population, there is an equal chance of any of the six matings.

Notice that crossing genetically diverse dogs, rather than genetically similar dogs—breeding to increase genetic diversity, as the VGL advocates—does reduce the likelihood of producing a dog likely to bloat. But it does so at the expense of also reducing the likelihood of producing a normal dog that does not carry the undesirable allele and increasing the chance of producing a normal dog that is a carrier. To put it another way, the likelihood that a pup will carry at least one copy of the disease-related genetic factor is only 58% if the parents are genetically similar but 83% if the parents are genetically diverse. I do not think that increasing the frequency of carriers is a good thing.

For the second example, let’s again assume a gene with only two flavors. But let’s reverse the situation so that the dominant allele (“C”) increases the likelihood of cancer and the recessive allele (“c”) results in good health. Would breeding for genetic diversity now be a good thing? No! If we again assume that the possible genotypes (CC, Cc, and cc) are equally distributed in the population, then Table 2 shows the six possible matings and the results they would produce, on average.

In this situation, crossing genetically diverse dogs actually increases, rather than reduces, the likelihood of producing a pup predisposed to get cancer, which is another way of saying that crossing genetically diverse dogs increases the likelihood that a pup will carry at least one copy of allele C. This makes sense if one just stops for a moment and considers that breeding genetically diverse dogs means that at least one parent must carry the allele C, while breeding genetically similar dogs means that 1/3 of the matings will be between dogs that both lack C entirely.

Table 2.    Possible Matings and Outcomes for a Hypothetical Dominant Allele “C” that Increases Cancer Risk
  Chances that a pup will be:
Possible Matings Normal (cc) Prone to Cancer (CC or Cc)
Genetically similar
CC x CC 0% 100%
Cc x Cc 25% 75%
cc x cc 100% 0%
Overall average 41.7% 58.3%
Genetically diverse
CC x Cc 0% 100%
CC x cc 0% 100%
Cc x cc 50% 50%
Overall average 16.6% 83.3%
*Because we assumed the parental genotypes are equally distributed in the population, there is an equal chance of any of the six matings.

Even if one wanted to make a breeding decision based partly on trying to maximize genetic diversity—and I have argued above that this may not be a good idea—how likely is it that genetic diversity would be a deciding factor in a breeder’s decision? For one thing, this assumes that a breeder has access to a range of possible sires from which to choose.  For a rare breed like ours, the list of sires is usually pretty small to begin with—small in theory, and even smaller in practice. In addition, genetic diversity would be only one of many factors to weigh in a breeding decision. One still would consider all of the desirable and undesirable traits in each sire, the sire’s family, and the sire’s previous offspring and their descendants (if any). Given this, I suspect that genetic diversity would be the deciding factor in very few (any?) breeding decisions, even if it looked like a worthwhile goal. And I think it is not.

Think of it this way. The ideal Deerhound population would be one that is free of alleles that increase the risk of ill health and enriched for alleles the promote good health. In my two hypothetical examples with the recessive bloat allele “b” and the dominant cancer allele “C,” every Deerhound in the ideal population would be genotype BB and cc. In other words, they would be genetically identical (homozygous) for both genes. This ideal will not be achieved by aiming for genetic diversity. On the contrary, breeding to increase genetic diversity moves us farther away.


Some high-profile people say that the top priority for making breeding decisions for all dogs, including Scottish Deerhounds, should be to increase genetic diversity. I believe this recommendation is a mistake that could harm, rather than improve, Deerhound health. Instead, I believe that the best way to work toward a better Deerhound population in every respect, including health, is to:

  • Continue selective breeding based on desirable and undesirable traits, as Deerhound breeders have always done. This process has served us well. To cite just one example, breast cancer is a major health problem in most breeds, including mixed-breed dogs. The American College of Veterinary Surgeons reports that one in four bitches that go through their second heat cycle will develop breast cancer. But breast cancer is extremely rare in Deerhounds, presumably because past breeders have weeded out the allele(s) that predispose to it. We owe them a debt of gratitude.
  • Develop genetic screening tests for specific alleles that we do not want (such as alleles that increase disease risk) and use those tests to breed away from such alleles.
  • Develop genetic screening tests for specific alleles that we do want (such as alleles that increase disease resistance) and use those tests to breed toward such alleles.