By Carol Beuchat PhD

Some breeds seem to have much higher rates of cancer than others. Flat-coated Retrievers, Bernese Mountain Dogs, and Golden Retrievers are a few breeds that seem to be getting hit hard. There are some good breed-specific data on cancer in purebred dogs that I've summarized on the Cancer page in the "Health Data" section of the ICB website, and I've reproduced the relevant summary graph here.

The clear losers here are the Irish Water spaniel, Flat-coat, Wirehaired Vizsla, and Bernese Mountain Dog, but there are many breeds that suffer more than 30% mortality due to cancer. At the other end, the Shih Tzu, German Spitz, Dachshund, and Lhasa Apso fare better than most. These data are from a survey study, and of course cancer can occur in any breed and there are no doubt Flatcoats and Berners that are cancer free into ripe old age (I hope so). But we can take these data as a general indication of breed susceptability.

An interesting thing about cancer in dogs is that large dogs seem to have higher rates of cancer mortality than smaller dogs. I took the same data in the graph above and plotted it as a function of height at the withers, and you can see from the resulting graph (below) that mortality from cancer increases with size in dogs. (Note that the axes have been logged so the scale of x and y axes is not linear).

There are a few interesting things here to notice. First, bigger dogs have more cancer. This could perhaps be related to a hormone called IGF1 (insulin-like growth factor), which has higher levels in larger animals and is also linked to increased risk of cancer. In fact, there is a population of people in Ecuador in which the disorder known as Laron syndrome occurs, which is caused by a mutation in the growth hormone receptor (GHR) gene. These patients have very short stature, and they also very rarely get cancer or diabetes. This seems to be linked to abnormally low levels of IGF1 - lower levels of IGF1 reduce the risk of cancer and other age-related disorders.

So, are IGF1 levels higher in larger dogs (or lower in smaller dogs)? It seems so (Spichiger et al 2006; Kimberly et al 2011) -

Could selection for large size in dogs be incidentally increasing the rates of cancer? Who knows, but it seems like something worth looking into.

The other interesting thing about my graph of cancer mortality vs height is that there is a lot of scatter around the line. For breeds of a particular height, cancer mortality can vary two-fold or more. For instance, for their size, the Shih Tzu has an exceptionately low rate of cancer, while the cancer mortality of Welsh and Staffordshire Terriers is much higher than you would expect for their size. Among larger breeds, the Irish Water Spaniel, Bernese Mountain Dog, and Flat-coated Retriever are still stand-outs, while the cancer rates in the Pointer and Newfoundland are relatively low.

Tossing all of these numbers onto a graph as we did at the top, you can easily spot the breeds at the high and low ends of the scale.  But this doesn't reveal the underlying and very important relationship between cancer mortality and size. It raises two very interesting questions -

1) why do larger animals have higher mortality from cancer, and 

2) why do some breeds have much lower rates of cancer than other breeds of about the same size?

Boy, wouldn't we like to know.

You can learn more about the genetics of dogs in ICB's online courses.

Visit our Facebook Groups

ICB Institute of Canine Biology
...the latest canine news and research

ICB Breeding for the Future
...the science of animal breeding 

By Carol Beuchat PhD

A dog with one testicle can kiss a show career goodbye because it fails to meet the minimum threshold of quality for a male dog. Presumably this is because not having a matched pair reveals some deficiency of maleness, despite the fact that the dog can nevertheless be fertile. Most breeders send the cryptorchids on a trip to the vet that removes them from the gene pool before they are sent off to meet the family that will be their wonderful forever home.

If being cryptorchid was really all that bad, we would expect that there would be some detrimental effect to the dog. If not surgically removed, the undescended testicle can develop tumorous tissue. On the other hand, could the cryptorchid dog have some advantage over a normal one?

A group in the Netherlands decided to look into this, and they pulled together data for 11,230 litters in 12 purebred dog breeds. The Dutch studbook registers all pups by about 8 seeks of age, so the data they used are for puppies that reached the age of weaning. Cryptorchid animals were identified at 12 months or older. Animals that had produced a cryptorchid offspring were designated as "carriers"; all other dogs were designated as non-carriers (we will call them "normal").

They looked at the number and sex ratio of the offspring that were produced by three kinds of matings:  carrier-to-carrier (CxC), normal-to-normal (NxN), and C females to N males and vice versa (CxN). The data from the CxN crosses were not significantly different from the NxN matings, so we'll just summarize the CxC and NxN data here.

The first surprise was that, in most of the breeds studied, carriers bred to carriers produced significantly larger litters than NxN. 

Furthermore, the CxC litters had a biased sex ratio, producing significantly more males than females, and slightly fewer females than NxN litters. In CxC, an average of 24% of males were cryptorchid, in a pattern that is compatible with a simple autosomal recessive inheritance, with effects of modifier genes, gene-gene interactions, and environmental effects affecting the actual phenotypic expression in the cryptorchid animals. They did not observe any sort of "intermediate" condition in the NxC crosses.

The authors ruled out the possible effect of reporting bias, but could only speculate on what might cause the larger, male-biased litters in the CxC dogs. They concluded that there can be a selective advantage to cryptorchidism through an increase reproduce success. Cryptorchidism occurs in both wild and domestic animals, and results similar to these for dogs have been reported in pigs. In fact, cryptorchidism might persist in animal populations because CxC matings produce larger litters and therefore more carriers. Selection for larger litter sizes in dogs might complicate efforts to eliminate cryptorchidism.

By Carol Beuchat PhD

More than 450 years ago, on the nineteenth of July in 1545, one of King Henry VIII's warships had an unfortunate accident off the cost of the Isle if Wight while trying to repel a French invasion force with their sights on England.  The Mary Rose, together with her crew and the ship's dog, took a hit and went down in less than 50 feet of water. Silt and sediment deposited by the tides covered her, and she rested there until 1971, when she was rediscovered.

As it happened, the sediment created anoxic conditions that preserved some contents of the ship in near perfect conditions for four and a half centuries. Among the material excavated was the skeleton of a small male dog, maybe 18 months or two years old, and preserved well enough for the extraction of DNA from a tooth for analysis. The skeletal structure of the Mary Rose dog suggested a mongrel type, and DNA analysis of microsatellites suggested similarity with the Jack Russell or perhaps English Springer Spaniel. But since many of today's breeds didn't exist at the time this dog was alive, it's hard to say what he was except perhaps a "terrier-type" along for the trip to keep the rats at bay.

The reconstructed skeleton of the Mary Rose dog, nicknamed "Hatch", was on display at Crufts in 2010. You can read more about it here.

By Carol Beuchat PhD

I have two bucket lists. One for the things I want to experience and accomplish in life, which despite regular culling continues to exceed what I can reasonably expect to do in my time on this earth. The other bucket list is a list of topics related to genetics and breeding dogs that I want to write about. This list gets longer by the day, and I can only hope that at some point it will be so long that I can start tossing off the ones that are so far down the list that I have no hope of ever getting to. This might make the list look shorter, but it won't reduce the importance of addressing those topics.

The myth of hybrid vigor in dogs is a topic I've been putting off, hoping it would fall off the bottom of the list. But it keeps coming up, and the urgency of addressing it keeps smacking me in the face. So it's time.

Let me say first that this will be about the biology of genes. Genetics is complicated, and it's more important that I write something dog breeders will understand than mention all the little nuances that I wouldn't gloss over if I was writing for a scientific audience. The thing for you to understand is that there is a truth, supported by decades of data as well as a scientific explanation that is firmly rooted in our fundamental understandings about the way genetics works. So I'm sure some people will respond with "But..." and bring up something I didn't talk about that they hope will dismiss all of this as nonsense. There is much more we could talk about, but there are no "but"s that will change the simple facts, and don't let anybody try to convince you otherwise who is not prepared to haul out their documentation.

Over and over, I run into discussions about "hybrid vigor" and why (supposedly) it doesn't occur in dogs. The arguments usually center on some discussion about purebred and mixed breed dogs, and the position supported is that purebred dogs are just as healthy as mixed breed dogs. These ideas get very wide and enthusiastic support from dog breeders and even get extended as a reason why Doodles and other deliberate crosses are a disaster. Published studies are pointed to as supporting this, and even people with relevant professional expertise weigh in with support. But as I said before, there is a truth here that we really need to talk about.

Before we can address the hybrid vigor myth, we need to talk about inbreeding. This is going to on a bit, but inbreeding and hybrid vigor (which we're going to refer to with the scientific term "heterosis") are inextricably linked, and we can't adequatly explain heterosis without understanding some basic consequences of inbreeding. So bear with me. 

Inbreeding is the mating of related animals. They can be closely related or distantly related, and we might want to distinguish between close inbreeding and "less close" linebreeding, but for genetics it's all inbreeding, and that's what we will call it here.

Related animals are likely to share alleles as a consequence of common ancestors, so mating related animals makes it more likely that offspring will inherit two copies of the same gene. We would refer to the alleles at that locus as homozygous (i.e., the same). If there are two different alleles at a locus, it is heterozygous. As a consequence of simple inheritance then, inbreeding increases homozygosity and reduces heterozygosity.

Animal breeders noticed long ago that although inbreeding had the advantages of improving the predictably and uniformity of offspring, there was also an effect that could be generally described as "loss of vigor" (Wright 1922; Lush 1943; and I discuss this here). These effects could be very subtle and even be overlooked as reflecting the normal variation in the quality of a group of animals, but research over the last 100 years has confirmed that this phenomenon is real (Charlesworth & Charlesworth 1987; Charles & Charlesworth 1999; Charlesworth & Willis 2009). We call it "inbreeding depression".

Inbreeding depression is not an increase in the incidence of genetic disorders of relatively high heritability like PRA or cardiomyopathy or hemolytic anemia. Rather it refers to loss of what biologists call "fitness", which encompasses the breadth of traits that affect an animal's ability to successfully pass its genes on to the next generation. Animals that die before reproducing have a fitness of zero. Animals that successfully reproduce but don't properly care for their offspring, which die as a consequence, also have a fitness of zero. Animals that have a high level of fitness produce offspring that go on to reproduce themselves and thereby perpetuate their genes in the population, and animals that do that less effectively or fail entirely have low or zero fitness.

In the context of breeding animals and plants, when we're talking about inbreeding depression we're usually referring to the collection of traits that affect reproduction and lifespan such as fertility, offspring size, pre- and post-natal mortality, maternal care, resistance to disease, and general "vigor and vitality". These effects have been documented in many thousands of studies and in all manner of organisms, and although there is much yet to be learned about it, there is no debate about the fact that it is a real phenomenon in both wild and domestic animals (Nicholas 1995).

Why do inbred animals have lower fitness? Because inbreeding results in an increase in genomic homozygosity, and homozygosity reduces fitness, resulting in inbreeding depression (Charlesworth & Charlesworth 2009).  Homozygosity can reduce fitness because it increases the expression of deleterious recessive alleles, some of which are lethal and can result in death early in embryonic development. Likewise, there can be high levels of homozygosity in many alleles of small effect that control reproduction and other developmentally and physiologically complex processes.  Homozygosity also reduces the beneficial effects of "overdominance", in which heterozygosity at a locus is advantageous over homozygosity of either allele ("heterozygote advantage") (Charlesworth & Willis 2009). (I have discussed some interesting examples of heterozygote advantage here.) We could say much more about the genetics of inbreeding depression, and there are some sources in the references at the end that address the details, but these are the basics of what we need to understand about inbreeding depression as it is going to relate to the issue of heterosis.

Of course, it is important to note that inbreeding depression occurs in dogs just as in any other mammal. As the level of inbreeding increases in dogs, conception rate declines, sperm count is reduced, litter size decreases, pre- and post-natal survival is lower, and lifespan is shorter. Dogs demonstrate inbreeding depression in the same ways as other mammals and vertebrates in general. I have summarized some data for dogs on the ICB website  here, here, and here.)

Okay, heads up if you've drifted off. Here's the punchline:

Increased homozygosity results in inbreeding depression in dogs, so the reduction of inbreeding depression that results from increasing heterozygosity is heterosis. It's as simple as that.

There are a few important points to emphasize. First, the sorts of genetic disorders that breeders commonly need to deal with that result from recessive mutations, such as copper toxicosis, centronuclear myopathy, multi-drug resistance, or exercise-induced collapse are not what we're talking about. Yes, these things can certainly affect a dog's ability to pass on its genes to the next generation, but inbreeding depression is not about a specific disease but rather the steady compromise in functions that can chip away slowly at the health and productivity of populations of animals over the generations. Reversing the effects of inbreeding depression is not about fixing just a gene or two. 

Notice that nowhere in here have we talked about hybrid vigor only applying to crosses between species, which is often argued as a way of rendering the topic irrelevant in discussions about dogs. In fact, there is some confusion here about the use(s) of several important terms. What defines a species is its inability to produce fertile offspring with other species, something biologists call "reproductive isolation", which can be a result of anatomy, physiology, geography, or even behavior. Two species don't usually tango. But there are a few animals that we consider separate species that can produce offspring that are healthy but usually sterile, such as the mules produced by crossing a horse with a donkey. Mules are "interspecific" (between species) hybrids. If these hybrids are generally sterile, we're not producing any hybrid vigor when we make these crosses. So the notion that hybrid vigor can't occur in dogs because it applies only to separate species is incorrect.

But we can also talk about the hybrids that are produced by crosses within a species, such as crossing different varieties or genetically distinct subpopulations or lines. These are intraspecific hybrids. When Mendel was sorting out the basic details of inheritance, he was making crosses between different cultivars or strains of peas, producing intraspecific hybrids. 

Like Mendel's peas, crosses between dog breeds would also produce intraspecific hybrids, usually referred to as F1 hybrids. If we're working with a breed of purebred dog in which there is inbreeding depression because of increased homozygosity, a cross with another breed will reduce the homozygosity of the offspring and result in hybrid vigor. But in fact, the dogs don't have to be different breeds, they can even be different lines within a breed. If these are populations are prevented from crossing for several generations they will drift apart genetically, and when they are crossed the offspring are likely to be more heterozygous than either parent. In fact, these crosses within breeds are widely used in animal breeding to manage levels of inbreeding and improve the vitality of the stock. (You can read more about this here.)

The notions of inbreeding depression and heterosis are what we call "settled science". The details might change in light of new research or ideas, but there is broad acceptance among scientists that the fundamentals are sound and likely to stand the test of time.

Inbreeding depression and hybrid vigor are predictable and understandable consequences of changes in genetic heterozygosity that can result from particular breeding strategies. Breeders of other domestic animals figured out long ago how to use heterosis to their advantage by deliberately crossing within or across breeds to produce higher levels of heterozygosity in offspring than in the parents. Many long-time dog breeders understand heterosis and use it to good effect in their breeding programs.

So no, hybrid vigor in dogs is not a myth. But why do dog breeders persistently insist that it is?

This topic is on my bucket list. Stay tuned.

Charlesworth D & B Charlesworth. 1987. Inbreeding depression and its evolutionary consequences. Ann Rev Ecol Syst 18: 237-268.

Charlesworth B & D Charlesworth. 1999. The genetic basis of inbreeding depression. Genet Res, Camb 74: 329-340.

Charlesworth D & JH Willis. 2009. The genetics of inbreeding depression. Nat Rev Genetics 10: 743-796.

Dickerson, GE. Inbreeding and heterosis in animals. J Anim Sci 1973: 54-77. (pdf)

Lush J. 1943. Animal breeding plans. The Iowa State College Press, Ames, Iowa. (pdf)

Nicholas FW. 1995. Veterinary Genetics. Oxford Science Publications.

Wright S. 1922. Coefficients of inbreeding and relationship. Am Nat 56: 330-338.

By Carol Beuchat PhD

Every breed has a history that has shaped the population of dogs we have today. For breeds recognized by kennel clubs, some number of animals were selected to be the founders of the "pure breed" and registered in a stud book. All subsequent members of the breed must be able to trace their lineage directly to these original dogs, and this is assured by the requirement that only dogs with registered parents can be registered themselves. The nice thing about this is that we should have very complete records of a breed's history that can be used to understand how a breed has changed genetically over time. (In fact, there are surprisingly few complete pedigree databases that are publicly available, much to the detriment of effective management of the  breed. A topic for another day...)

One of the first things we do at the Institute of Canine Biology as part of the genetic analysis of a breed is look at how the population has changed over time. (You can see several of these ongoing analyses under "Breed Projects" on the ICB website.) They are always very revealing and often there are some real surprises.

These are data (see graph below) for the number of Afghan Hounds born per year from the early 1920's to the present day from ICB's analysis of the breed's pedigree database. In the early years, a few hundred dogs were registered per year, with occasional increases and decreases in popularity through the 1950s. In 1957, an Afghan Hound, Ch Shirkhan of Grandeur, won best in show at Westminster, and - coincidence or otherwise - the popularity of the breed exploded. In 1960, there were about 1,000 registrations per year, in 1970 it had risen to 5,500/yr, and in 1975 it was close to 12,000/yr.  Registrations didn't double, or even triple, in those 15 years -they increased by twelve-fold.

Then, even more rapidly than they increased, Afghan registrations dropped, to about 5,000/yr by 1980 and half that by 1990.  Over a period of just 30 years, the breed experienced a boom and bust of truly epic proportions.

The domestic Afghan was never bred to be a working dog. It was a lovely bauble, prized for its beauty and noble demeanor. So when the population exploded, it was no doubt the "best" dogs that did most of the breeding, rather than a representative cross-section of the dogs in the gene pool at that time. And when the population crashed, it probably wasn't with the foresight of assuring that breadth of the gene pool be preserved.

You've probably heard about founder effects, popular sires, and bottlenecks. A few popular dogs become overrepresented on the increase, and less popular lines die out on the downslide.  These dramatic swings in the size of a population can wreak havoc on genetics.

When a population grows by balancing the contributions of all of the dogs to the next generation, the frequency of alleles in the gene pool stays about the same as it grows. But if there is selection on a particular feature, or a popular sire, the genetic balance of the gene pool will shift, with the frequencies of some alleles increasing while other alleles become rare or are lost entirely. The nature of the gene pool of the population at the peak of its growth can be dramatically different than it was before the surge in popularity.

From the height of the Afghan hound's popularity, breeding dropped dramatically. Only a fraction of the available dogs were bred, so the genes of some dogs were carried on but many were not. It's as if you reached into a large bowl of Jelly Bellies, selected 8 at random, and replicated just those to create a new bowl of candy. Chances are that new bowl is missing a lot of the great flavors in the original mix, so the population of beans in the new bowl will look very different from the one it was sampled from.

Lest you think that the history of the Afghan Hound is unusual, there are plenty of other examples of breeds that have very similar histories. The Standard Poodle also has an interesting story. From its early history in the late 1800's, it gradually increased in popularity until the start of WWI, when the population dropped by about 70% (from the ICB Poodle analysis). After the war there was a resurgence of popularity until WWII when once again there was a decrease in breeding. After that, however, things really picked up and the number of dogs produced per year quadrupled in the three decades between the mid-1940s to the 1970s. That's a pretty impressive increase in popularity as you can see in this graph, but it pales in comparison to the explosion in growth that followed.

This is a bottleneck as the population crashes, followed by the founder effect as the population subsequently carries on from that subsample of its original gene pool. This is what happened in Afghans. There is clear evidence that the changes in population size of the breed were unbalanced genetically. Over the history of the breed, only about 25% of the dogs were bred (19% of males and 32% of females), which is typical of many breeds. This is like a little hole in the bottom of a bucket; you're slowly losing water, but not so fast that you can't ignore it if you're busy with other things. But for Afghans, the hole is getting bigger; for the period 2005-2013, 12,000 dogs were recorded in our pedigree database, but only 1,200 have produced offspring. That's barely 10% of the population. With every passing generation, the breed's genetic pantry - in fact, its heritage - is slipping away. A full 90% of the dogs being bred will pass nothing on to the next generation. (Note that we don't have a "complete" database for this and many other breeds. In the absence of cooperation from Kennel Clubs in making their registration records freely available, breeders have taken on the onerous task of constructing databases by hand, and complete or not these are what we have to work with.)

One thing that's important to remember is that the larger the swings in population size, and the faster they occur, the more dramatic the consequences can be for the gene pool. A gradual increase in population size is more likely to include a broader mix of animals, and likewise for a slower decline. Explosive growth and rapid crashes can change the composition of the gene pool dramatically.

The axes on the graph below are difficult to read so I've labeled some strategic points for you. First, the spot marked "1970", is the rightmost end of the axis on the graph above - that is, the graph above of the dramatic population explosion before 1970 is just a few little bumps in the same years of the graph below. Beginning in about 1970, the number of poodles produced went from about 1,000/yr to 8,000/yr only 35 years later in 2005. Just as in Afghans, however, the peak in Poodle popularity might now be waning, as registrations appear to be dropping dramatically in the last 10 years. This might reflect incomplete data in the most recent years, and time will tell. But if this drop in popularity is real, we know there will be repercussions for the gene pool. The danger of losing less common lines as the population contracts is very real.

Those familiar with Poodles know that tucked into the pages of the breed's history is a bottleneck caused by the sensational Wycliffe poodles of the late 1950s and 1960s. (Armstrong gives an excellent summary of the impact of this period on the genetics of the breed in a collection of papers here.) The impact of these dogs on the breed is so huge that nearly every poodle in the world has a pedigree that goes back to those dogs. The effect on today's dogs is obvious. One of the famous dogs of this era, Annsown Gay Knight of Arhill, is indicated by the arrow in the pedigree below of a contemporary dog, and all of his subsequent descendants are in red. Half of the breed - the dogs that should have been populating the left side of this chart - just slipped out of the gene pool and disappeared as breeders rushed to the Wycliffe lines. The unfortunate consequence is predictable; much genetic diversity was lost and the average level of inbreeding is relatively high, and the breed now suffers increasingly from genetic disorders and especially autoimmune diseases. Poodle breeders contemplating a potential breeding not only compute the coefficient of inbreeding, but also "% Wycliffe", as they try to reduce the predominance of these dogs and their descendants in their puppies. Ongoing DNA studies at UC Davis are are establishing the patterns of genetic structure in the breed, including the DLA haplotypes (the genes of the immune system), so that breeders can take advantage of the existing genetic diversity in the breed to reduce the level of inbreeding and improve the heterozygosity of genes in the immune system. The breed's large population size and global distribution make constructing its genetic landscape difficult, but they also make it more likely that pockets of genetic diversity do exist that can be used strategically by breeders to improve the genetic health of the breed. If the popularity of the breed is falling as suggested by the available data, these efforts could not be more timely.

If we dig into their histories, we would find that many dog breeds have surges in popularity like the ones we've seen in Afghans and Standard Poodles. Changes in hunting styles that resulted from improvements in guns fueled surging populations of retrievers who could fetch birds shot on the wing. Overland fox hunting on horseback as (mostly) sport required large packs of hounds of various shapes and sizes. Development of rocky farmlands in the British Isles produced an array of terrier breeds suited to digging out vermin in the cairns, and larger terriers were developed to dispatch the badger and fox. Military dogs to assist in the tasks of war popularized the Doberman and German Shepherd. Popular culture transformed 121 Dalmatians the movie into a lovefest that put Dalmatians in thousands of homes in suburban America, far from the nearest horse-drawn fire wagon. Similarly, Lassie transformed a working breed into every boy's companion. The popularity of Mexican food and Chihuahuas rose together in the southwest US, thanks to a certain fast-food chain, and a fascination with wrinkles made the Shar Pei the unlikely cover shot on hundreds of gift cards and calendars. 

For these and many other breeds, explosive surges in population size are usually followed sooner or later by a fall that can be equally dramatic, and the gene pool we're left with isn't the same as the one we started with. The dogs might look the same, because we have been carefully selecting for the genes for type, but the reservoir of genes that run the rest of the dog - physiology, immune system, organ function, temperament - can be only a tiny subset of what we had before. Today's breeders, trying to use the husbandry methods handed down from generations of dog breeders, find themselves in genetic cul-de-sacs from which there seems to be no escape, or engaged in a game of genetic whack-a-mole, defeating one nasty recessive mutation only to have another pop up in what was thought to be a "healthy" line.

For most breeds, It's not your grandpa's gene pool any more. Like the cheetah and Przewalski's horse, the populations might appear robust and healthy, but because of their history, the genetics that support them are precarious. Protecting these breeds will require more vigilance now, and more cooperation among the breeders who have a collective desire and responsibility to protect the genetic heritage of our breeds. At the very least, we should be monitoring breed populations in the same way that we keep an eye on the gray whale and Mountain Gorilla. We should be able to identify in real time impending bottlenecks developing from popular sires, genetic lines in danger of extinction, dangerous levels of inbreeding, the emergence of new genetic disorders long before they are epidemic, and the booms and busts that can drive the genetics of a breed in an unwanted directions.

Think about the history of your own breed, and how your genetic landscape has changed over the generations. Assess what you have today, the problems that need to be tackled, and the path breeders need to take in the future to protect and improve on what we have. Collective efforts can be very powerful, and the payoff will come in securing the genetic resources of the breed for the future.

By Carol Beuchat PhD

You've all heard the breeder's maxim - "Breed the best to the best".  Breeders carefully evaluate the good and bad qualities of each puppy in a litter, saving out for breeding only the best quality animals. The rest are spayed and neutered and go off to enjoy their days in a loving pet home, or if not S/N they often go with a contract that restricts breeding or registration. The thought is that this protects the breeder's line and the number and quality of the puppies that are produced.

Let's do a little exercise that explores the genetic consequences of this breeding strategy. For this, you'll need a piece of paper and pencil, and maybe a calculator for some simple math. Go get those now...

Here's the scenario. You go out and find yourself a really nice bitch that you're going to use as the foundation for your line. Then you do months of homework evaluating dogs at shows and scrutinizing pedigrees and health data in the search for the perfect stud dog. When you make that decision you breed these two stunning dogs and produce the litter of your dreams. You watch them carefully as they grow, evaluating structure and temperament, how well they meet perfection in your mind's eye, and how you see each as the possible dog to produce the next generation. Finally, you pick out a really stunning puppy, and send the rest off to their carefully-chosen homes.

When the chosen pup reaches maturity, you repeat the process - evaluate the available mates, scrutinize the pedigrees, review the health test info, and so on until you settle on the perfect match. Again you breed, produce a nice litter, keep the best pup, and repeat, striving for improvement with each generation.

Okay, let's look at the genetics. Get your pencil and paper and make four columns labeled like this:

In the first column is the number of puppies in your dream litter (we'll work with 10). Now we're going to estimate how much of a parent's genes will get transferred to the collective genomes of the puppies. 

Think about it like this: If you have no puppies (0 in the table above), NONE of the genes in the parent dog get carried to the next generation, right? So under % of parent's genes preserved we have zero.

What if there's only one puppy in the litter? It will get HALF the genes of a parent, so under % of Genes Preserved we put 50%. If half of the parent's genes are preserved in that puppy, then the other half of the parent's genes are NOT preserved. So the second column is for the fraction of the parent's genes that have NOT been carried on in a puppy. If 50% of the genes of the parent are in one puppy, then 50% of the parent's genes are not in a puppy.

Now here it gets a little tricky. Let's say you have two puppies. You know each will get half of each parent's genes, but it's very unlikely that puppy #2 gets the other half of the genes that puppy #1 didn't get. Instead, you would expect puppy #2 to get on average just half of those. This is just probability - and what puppy #2 actually gets could be all the same genes as puppy #1 (unlikely), or none of the genes of puppy #1 (unlikely), or something in between (most likely). So half of the genes that puppy #1 didn't get is 25% of the parent's genome- write that in Column B (half of the remaining genes). So this column is the fraction of a parent's genes that have NOT been copied to the offspring.

Now, in the third column where we're keeping track of the amount of a parent's genome we have captured in puppies, we can add 25% to 50% (what we already had in puppy #1), for 75%. This means that a litter of two puppies will have about 75% of all the genes found in a parent, and that 25% of the parent's genes are not present in the puppies.

Okay, so we can continue this for litters of 3 pups, 4 pups, and so on, with each additional puppy adding to the fraction of parental genes that get passed to the litter. You're going to repeat the calculations we just did for each additional puppy. Don't get a brain knot over this. Take the number in "half of genes", divide by two, and write that in the blank cell below it. Now add that number to the last number in the % of parent's genes column, and write that number in the blank. Do this for each of the 10 puppies in the litter.

Now, let's see what we have here. (If you didn't bother to do the math and write out your table, I'm not going to do it for you and you wan't have a clue what we're talking about below. So go get your pencil and just do it.)

I want you to draw a little graph like this one. I've already added the two data points we figured out above, and you can add the additional points up to 10 puppies. 

Is this what you expected to see? What does this plot say to you? 

---------------

After you've thought about that a bit (go back and think about it if you skipped that question), let's also express this another way. In the 4th column of your table, calculate 100 minus the value in column 3. So for the 0 puppy, that's 100%; for puppy #1, it's 50%; for puppy #2 it's 25%, and so on.

Now draw another graph, with the same x axis but this time the y axis will be "% of genes lost".

These two graphs tell a very important story. From your first litter, with your special bitch and the spectacular sire, you saved one puppy. What fraction of your foundation bitch's genes will get passed to the next generation? What fraction of her genes are not saved and are lost to your future breeding efforts forever?

Each time you pick only the best puppy from a litter to breed, you only capture half of the genes of each parent. The other half get tossed out. But you originally picked that sire and dam because they had qualities you really liked. You can't expect to get the exact same qualities in the offspring because each gets only half the genes in each parent.  So what you're hoping for is a lucky blend of the genes from sire and dam that get passed to each puppy. It's that particular mix of genes, including the regulatory genes, the epistatic interactions, the epigenetic factors, and (don't forget) the environment that produced those parents, and likewise these things will determine the traits of the offspring. The number of different ways you could sample the genes of each parent and mix them together is what gives you the variation in each litter that you evaluate when choosing the best puppy. Without that genetic variation, there is less phenotypic variation, and you don't have much variety - good or bad - to choose from.

Do you really want to discard half the genes of each parent because they didn't end up in just the right mix to produce the dog of your dreams in that litter? Once tossed, you can't get those genes back except by crossing into your line a dog that has some of them, but no dog will have the full complement of the genes you want to add back. You're much better off keeping more than one puppy from your litter, and you can look at your graph now and figure out how much you're likely to gain with each additional puppy so you can balance that with how many additional puppies you want to stay in the gene pool

In the last graph we drew (from the data in the fourth column), what you see is the relentless attrition of the gene pool of your line based on how many puppies are kept in the breeding pool. You think you're saving the "best" genes in the "best" puppy, but you are only saving the best COMBINATIONS you got in that particular pup. If you had 100 puppies from that pair of parents, each one would be different and some might be even better than the best pup in that first litter. Just by chance.

I can hear you all grumbling that I'm completely clueless, because you can't breed a bunch of dogs from each litter; it would be a glut of puppies that you would need to find homes for, and who could afford it with all the DNA test costs, vaccinations, food, not to mention time and space. Of course, these are very real limitations. But if you think about this a bit, I think you might see that it might be prudent to consider the longer-term consequences to your line when deciding which and how many dogs to breed. You can't breed everybody, but breeding three or even two substantially increases the fraction of the genes you preserve. After 5 or 6 pups there are diminishing returns, so you can weigh cost and benefit. But breeding only one pup per litter is throwing out a lot of genetic raw material that would be better to keep in the gene pool.

What about an entire breed?
We've been thinking about this in terms of your personal breeding program and the line you're hoping to develop (or preserve). But exactly the same math applies to the gene pool of the breed as to individual breeding programs. Study after study has found that after a breed enters the "purebred" studbook, genetic diversity is lost very rapidly as a consequence of selective breeding.

For example, analysis of the pedigree data for 10 breeds of dogs in the UK KC registry found that 7 of those lost more than 90% (!!!) of unique genetic variability in only 6 generations (Calboli et al 2008). This is because only a fraction of the dogs in the population are bred, and the genes in the rest are lost to the breed. And if there is a popular sire among that small breeding population, the damage done is even greater.

If we go back to our genetic pantry analogy, you can understand how losing genetic variation will impact your breeding program in the future. You can't make dessert if you tossed out the last of the sugar. If spices represent the various alleles that occur for a particular gene, tossing out a random handful of those may wipe curry, chili, and cinnamon toast off the menu.

The way we have been breeding dogs for the last 100 years has been throwing out potentially useful (even essential) genes each and every generation. We might be getting better and better at making our favorite dish, but the menu has been getting shorter and shorter. It IS possible to lose enough genes that we can no longer make a dog that works, and if you consider that dogs that don't live past their prime are "broken", then we have a lot of breeds that are on the genetic precipice. We need to fix this or we'll lose them completely. And we need to stop breeding in a way that tosses out the genetic ingredients that we need to build healthy, beautiful, and functional dogs.

By Carol Beuchat PhD

Let's say we have a population of animals that are randomly breeding. Over generations, the average level of inbreeding in the group will increase, just as you can imagine it would if you and a random assortment of unrelated people were shipwrecked on a remote island - eventually, everybody ends up being related to everybody else. This is indeed what happens to populations of animals on islands, and it is also what happens in purebred dogs that have a closed stud book. And the smaller the population, the faster the degree of inbreeding increases.

A species will usually have many subpopulations that are separated by geography, but the occasional animal will leave its resident population and move to a new one, taking a slightly different mix of genes with it that get introduced into the new group. Many species drive the juveniles away, or the offspring of one sex (especially males), which forces a bit of genetic mixing every generation that mitigates the tendency for inbreeding to increase over time.

Animal breeders working with relatively small populations of animals (hundreds or perhaps thousands, but not many millions) also need to manage inbreeding or it will increase relentlessly over time. They can do this using breeding strategies that mimic what happens natural populations. How would this work?

Here are some pedigrees of individual dogs (all the same breed) in which each animal is represented only once (these charts are produced by Pedigree Explorer). Founders are at the base (on the right), and the subject is on the left. (We looked at some pedigrees like this in our discussion of The Problem with Poodles.)

The first two pedigrees are similar and are typical of most of the pedigrees I've looked at of purebred dogs. The most recent generations (on the left) have some tight breeding, and like most breeds there are just a handful of dogs at the base. The coefficient of inbreeding (COI) of the first one is 45.7%; the second is 49.6% (!). A sib-sib mating would have a COI of 25% by comparison. These dogs are definitely inbred.

Here (below) are two more pedigrees, from the same breed but revealing a very different breeding strategy. The base of the pedigrees is the same as for the dogs above, and both of these pedigrees also show a bottleneck at about the midpoint. But there is a pronounced mushroom shape to the tops of the pedigrees, and you can see that this is because the recent inbreed combined two lines that were completely unrelated for 4 generations or more. In essence, the breeder has crossed two inbred lines to produce animals that relatively speaking are outcrossed, with COIs that are far lower (21.4% and 18.8%) than the examples above. Still inbred, but much lower than the first two examples.

Of course all of these animals are related if you go farther back in the pedigree, so how can this make a difference in the level of inbreeding? How can we use this to increase heterozygosity?

There are two things to consider.

1) The probability of inheriting a particular allele from the pair at a locus of a parent is 1/2, or 50%. With each added generation, that probability is reduced by half; so the probability of inheriting a specific allele from a grandparent is 25%, from a great-grandparent it's 12.5%, and so on. The deeper in a pedigree a particular ancestor occurs, the lower the probability of inheriting a specific allele from that animal. In the bottom two pedigrees above, the parents of the dog share no common ancestors in the most recent generations. In generations, in which there are no shared ancestors, there is no additional inbreeding. So any inbreeding must be deeper in the pedigree, and its effect is reduced by half for each additional generation to the shared ancestor.

2) The other thing we can take advantage of are two forces that will tend to make two populations increasingly different genetically over time - selection, either natural or artificial, and something called genetic drift. You know about selection. If you started with two genetically identical populations, then bred each independently but selected for the same traits in each, you would no longer have identical populations several generations later. In the shuffling of genes with each generation, and the selection of offspring to breed in the next generation, some genes will increase in frequency over time and some will decrease, and some might become fixed and others lost. The offspring in generation 4 of the first population will not be genetically identical to those of the second population, and if you were to breed together an animal from each, their lower genetic similarity will be reflected in a lower coefficient of inbreeding in the offspring.

So think about this. You have a breed with thousands of animals in a closed gene pool. If you can divide that population of animals into sub-populations and breed them independently for several generations, you can create for yourself groups of animals that you can use every several generations to produce animals with a lower lower the level of inbreeding, and also reintroduce genes that might be present in one subpopulation but lost from another. In essence, you can use inbreeding in subgroups in a clever scheme to manage inbreeding in the population as a whole.

In the good old days, there were large kennels that could keep dozens of dogs, allowing the luxury of managing multiple groups of dogs that could be used to produce dogs with nice type without close inbreeding. Most of today's hobby breeders have to manage with far fewer dogs, but you can still take advantage of this strategy with a little planning. Certainly, there are probably subpopulations in your breed already, but you might not know where they are. But a genetic analysis of the pedigree database can reveal them. In each of these three figures, analysis of the pedigree of a breed has revealed groups of dogs that are more likely to be closely related to each other than to animals in other groups. The first one, of Icelandic Sheepdogs, was used by breeders to identify pockets of useful genetic diversity at risk of being lost that could be used to reduce the level of inbreeding in the breed as a whole. The second and third graphs were used to identify the animals in a breed that were likely to be affected by a particular genetic disorder for which the genes were unknown (centronuclear myopathy and elbow dysplasia, both in Labradors). (Click on the figure to read more about these studies.)

Even in highly inbred populations, there is likely to be some "genetic structure" like this that breeders can take advantage of to manage the genetic health of their breed. Analysis of a pedigree database can produce graphs like these that can be used to identify the genetic subgroups in a breed as in the examples above, and at the same time identify those groups of animals at risk for potential genetic disorders so they can be controlled.