By Carol Beuchat PhD

The genetic status of a dog breed today was shaped by its history. The number of founder animals, population explosions and crashes, geographic separation, popular sires, and many other things have effects that shape the genetic trajectory of a breed.

A few days ago a breeder forwarded to me the population statistics for her breed in the UK, and to my surprise it contained not just the registration records her breed but for all the UK breeds back to 1908. Digging through data is like an archaeological investigation; you never know what you will find. I love discovery so I dove in, and I am truly astonished at the information in these tables and tables of numbers. Maybe this is all old hat to somebody out there, but it was a real eye-opener for me.

The data are just registrations by year with the UK Kennel Club. The database I was sent runs up to 2009, and I added the data to 2013 using the 10 year summaries available on the UK KC website. I graphed the registrations per year for each breed and also by groups. The numbers varied wildly by breed - Labradors are registering more than 40,000 dogs per year, while many breeds log fewer than a hundred. To save time I didn't want to draw separate graphs for every breed, so I tried to group them as I went, and as it happened this proved to be quite revealing. Some of the breeds should be graphed again on a different scale, but for now I hope that the information will be useful for most breeds. (The website location of the graphs is at the bottom of the page.)

The first thing that struck me as I started graphing was a dip in registrations between 1970 and 1980. It is very obvious in this graph of sporting dogs, looking a bit like somebody stuck their finger in the cake. As you will see, this dip appears in almost all of the breeds in every group. It is not a dip in popularity of a breed, but a very large, systematic drop in registrations. It doesn't seem to correspond to a war (you can see a dip around WW II in many breeds), and I'm at a loss to explain it. A plunge in the popularity of dogs? A postal strike? A rebellion against registration? I have no idea. It is not just the odd dip, but a steep decrease for several years followed by a rise over several years, usually to about the level of registrations before the crash. If anybody has an explanation for this, I would love to hear it. [A little homework turns up two possible explanations - a parvo outbreak in the 1970s and a severe economic downturn. More on the possible implications of those events later. (Sorry K, you were 11 hours and about 6 people too late to take credit for parvo. But keep playing - anyone can win!)]

Apart from just being an interesting anomaly, if this drop in registrations reflects a real drop in breeding, then it represents a bottleneck that lasts more than a generation. It is likely that there was a loss of genetic diversity from the breed as the population plunged, then in the rapid recovery the relatively small numbers of breeding dogs would produce a founder effect. It's not likely that breeders were being vigilant to avoid losing lines as the population crashed, and the "best" dogs might have done than their fair share of breeding on the up side. Together, these two events could substantially change the composition of the gene pool (see Hitting the bottle: the genetics of boom and bust). And the really puzzling thing is that it was like a global extinction event, exactly the same in breed after breed. This wasn't a bottleneck for a breed, it was a bottleneck for purebred dogs in the UK, and it was profound.

The next thing that is so striking in these data really shouldn't come as a surprise. Many breeds have gone through not just the bottleneck in the late 1970s, but others as well and some were even more dramatic. The English Cocker was enormously popular at the time of WW II (graph above), when registrations for most breeds dropped. And surging with the English Cocker was an odd mix of other breeds - the English Setter, Pointer, Whippet, Norwegian Elkhound, Smooth Dachshund, Scottie, Cairn Terrier, Bedlington, German Shepherd, Pekingese, Chow Chow, Bulldog, Keeshond, and an amazing boom and bust in the Pembroke Welsh Corgi. In many breeds there was also a dip in registrations around the time of WW I (1914-1918).

Apart from these bottlenecks explained by major historical events, there were many very severe population explosions, crashes, and bottlenecks. The Yorkshire Terrier went from 26,000 registrations in 1990 to only 5,000 a decade later. The Pug has gone from 1,000 in 2003 to about 8,000 in little more than a decade. In French Bulldogs, registrations have gone from 700 to 7,000 in only 7 years, and the Bulldog is gaining ground fast as well. You will find other really remarkable shifts in breed popularity if you explore the graphs.

Most of the Working breeds, Terriers, and Pastoral breeds are all headed down. Registrations of German Shepherds have been dropping since before 2000, and the one herding breed on the upswing is the Belgian Malinois although registrations are still less than 200 per year. It's fascinating to see how surges in popularity of one breed mesh with the drop in registrations of another. In the late 60s, everybody wanted a Beagle, Basset, or Afghan, then just as quickly the registrations of those breeds dropped off. Weimaraners have been steadily dropping for the last 15 years, while Vizslas have been increasing steadily since 1990.

I don't know what to make of the Poodles. I have talked elsewhere about the explosive growth in the Poodle population beginning in about 1965, but the data for Poodles in the UK have scarcely a bump at that time. There was, however, a huge surge in popularity of Miniature Poodles in about 1950, and a smaller one in Toy Poodles about 5 years later; both were way down by 1975. [I'll be taking a closer look at the poodle data.]

There are booms and busts aplenty, but there is also a large number of breeds that are hanging on my fingernails. The UK now has a list of native breeds threatened with extinction because of low registration numbers. There was a gasp of surprise when the English Setter showed up on that list, but a glance at the registration statistics show that it has been losing ground for 30 years. The alarm should have been raised long ago. There is more than just a handful of breeds that are registering only 20 to 50 dogs a year. A population this small cannot be genetically stable over the long term. Even a tiny push from artificial selection can head such a small population off in a different genetic direction, and of course the effect of genetic drift is greater in smaller populations. 

Unfortunately, wild swings in population size are usually not a good thing. If growth is uncontrolled, and especially if it favors popular animals, genetic diversity suffers as the less popular dogs are left at the curb. After narrowing the diversity of the gene pool during rapid growth, an uncontrolled crash narrows it further. The gene pools on either side of the boom and bust are likely to look very different, and almost certainly for the worse. A population can grow if reproduction is balanced so the the number of dogs increases but the allele frequencies in the gene pool are held stable. Likewise, a cull that includes the full breadth of the gene pool in the downsizing breeding population will help protect as much as diversity. But it's not likely that either of these happen during the dramatic swings in popularity of dog breeds. Everyone rushes to breed to the most popular dogs during the surge, and most of the smaller breeders probably fall out as the population crashes. Every breed will have a different story, but overall the histories of these pedigreed dogs in the UK provide example after example of inadequate genetic management that has surely been to the detriment of the gene pools of the breeds.

There is lots to be learned in these graphs and I've only just begun to explore. If you understand a bit about population genetics, you will probably find them unsettling. You will also see that there are changes happening now in many dog breeds (decreases as well as increases) that breeders should recognize have the potential for producing long-term genetic damage to the breed if not managed. It would be great if there was a comparable database for dogs in the US, but I don't have one and the AKC no longer publishes their registration statistics. If it's because of falling numbers, they can see here that this is happening in the UK as well. But surely publishing only a list of AKC's "most popular" dogs each year only serves to drive the popularity of those breeds, to the detriment of the breeds at the bottom of the list.  The UK is registering 25,000 German Shepherds a year, and 45,000 Labradors. A few thousand off each of those and distributed to the breeds at the bottom of the list would help many breeds that are really struggling with very low numbers. (At the same time, however, the numbers of both of those breeds are dropping, and it would be useful to know why.)

All of the graphs are under the "Breed Projects" tab (as UK KC Graphs) on the ICB website for lack of a better place.

My apologies in advance if the graph of your breed would be better graphed on a different scale. In time they will all be redone, but I hope that you will nevertheless find them useful. And I want to know whose finger it was in the cake.

By Carol Beuchat PhD

By Carol Beuchat PhD

Renal dysplasia is a serious and complex genetic disorder that occurs in many breeds of dogs. In 2011, a paper by Mary Whiteley and others was published in the respected journal PLoS ONE that claimed to have found a marker for the disorder (you can access that publication here). A DNA test was subsequently made available through Whiteley's company DOGenes.

After the publication of that study, serious concerns were raised about its validity. In 2012 the editors of the journal PLoS published an extraordinary "Expression of Concern" above the abstract of the online article, which I have reproduced below (and you can read here).

In short, the journal editors state that some issues of concern were not addressed during the peer review process, and that they undertook their own review of the study. From that, they concluded that the study was seriously flawed, and that the Cox-2 promotor suggested to be linked to the disorder was in fact a neutral DNA variant.

This is what they said:

"In light of the concerns outlined above, the PLoS ONE editors are issuing this Expression of Concern in order to make readers aware of the concerns about the reliability of the results and conclusions reported in the article."

Nevertheless, DOGenes continues to offer their DNA test based on that study and there is no mention on their site of actions of the journal editors.

The DOGenes test is NOT listed among available DNA tests on the website of the Orthopedic Foundation for Animals.

However, it continues to be listed in the University of Pennsylvania PennGen genetic testing database. In October 2014 I contacted PennGen and asked about this test. They told me that PennGen cannot be responsible for the validity of the DOGenes test, and that a more recent paper by the same author did not address the original concerns.

Note that in the absence of a valid DNA test for renal dysplasia, it can be managed effectively using Estimated Breeding Values like any other heritable trait.

Many of the traits in dogs that we would like to select for or against are influenced by more than one gene. In fact, there can be dozens or even hundreds of genes behind a trait, each having only a small effect, but collectively they can substantially influence phenotype. For instance, osteosarcoma is an aggressive cancer of the bone that is relatively common in larger breeds of dogs like the Greyhound, Irish Wolfhound, and Rottweiler. DNA studies have identified 33 loci associated with osteosarcoma that together explain 55-85% of the phenotypic variance in these three breeds, indicating that this cancer is highly heritable (Karlsson et al 2013). Many of these genes are found in regions of the chromosome associated with bone differentiation and growth, but they appear to be different regions in each breed. So far, at least, there is no "osteosarcoma gene" in dogs, and with dozens of potential risk loci it seems unlikely that there will be a simple genetic test coming anytime soon - or perhaps ever.

With no hope of a genetic test anytime soon for osteosarcoma in dogs, breeders should be able to significantly reduce the incidence of this cancer in their breed by using estimated breeding values. This requires only a pedigree database and the information about affected animals among a dog's relatives. A computer generates a number that reflects the "genetic value" of a dog for a particular trait, in this case osteosarcoma. By using this estimated breeding value (EBV) to identify dogs at lower risk for osteosarcoma, breeders can efficiently reduce the incidence of this cancer in the breed.

The real beauty of EBVs is that you don't need to know the genes involved in a trait or how it is inherited. You don't need to base selection decisions on a phenotype that might be only partially determined by genetics. The estimated breeding value is a numerical estimate of the what an animal will produce in its offspring.

EBVs can be used to improve selection against many genetic disorders. Heritability of other cancers are also relatively high: lymphoma in Golden Retrievers has a heritability of 44%, and histiocytosis in Bernese Mountain Dogs is 30%. Heritability of hypothyroidism in Beagles is 33%, mitral valve disease in Cavaliers is 33%, and epilepsy in Belgian Tervuren is 77%.

When a disorder pops up in a breed, breeders usually rush to identify the responsible gene so they can develop a test and select against. However, for many disorders a test is not likely to be forthcoming - not soon, and maybe not ever. Estimated breeding values offer a way to select against a problem much more efficiently than using phenotype alone, and all you need is a health database.

Several kennel clubs are already providing EBVs to breeders for hips (e.g., Finland and the UK), and these resources will continue to develop for other disorders. A website that will provide EBVs for hip and elbow dysplasia for dogs in the US based on the public data in the OFA database is in the works at Cornell, and ICB will be working with several groups of breeders in 2015 to get their data online and train them how to use it. For breeders that have been struggling to manage genetic disorders, sometimes for decades, we should be able to start seeing improvement soon if breeders will use it.The website is under development, but you can have a look here.

We will have an EBV course available online sometime in the next few months, but if there is a group in your breed that is interested in getting in on the initial stage of training and implementation of EBVs, please contact ICB and we can discuss what would be involved.

ICB's Population Genetics course lays the foundation for understanding how EBVs can be used as part of a strategy for managing genetic disorders in purebred dogs. The next class starts 5 January.

By Carol Beuchat PhD

The first step in planning a breeding is to evaluate the traits you want (and don't want) in potential sires. For the ones that are heritable - i.e., at least partially determined by genetics, what you would really like to know is whether the dog has the specific genes to produce the traits you want in the offspring. We have DNA tests for a few useful things like coat color and some diseases, but for the most part the genes that are actually in the dog are an unknown.

If you're an experienced breeder, you know that you can look at a dog's relatives - parents, aunts and uncles, litter mates, and especially offspring to get an idea of what a dog is likely to throw. What you are trying to do is predict the "breeding value" of a dog; that is, how likely it is to produce offspring with a particular trait. Sometimes the picture is clear, but sometimes it's complicated and you are left to trust your gut with fingers crossed behind your back. And the very best you can do is make a subjective, no-doubt biased, and very qualitative guess.

There are better ways to do this. If you know a trait is inherited, and you also know that a dog gets half of its genes from the sire and half from the dam, then you can make some predictions about how likely a dog is to be carrying the genes of a particular ancestor. You do this sort of thing when you use coefficient of inbreeding to estimate the probability of producing puppies that will be homozygous for a recessive allele inherited from both sides of the pedigree. We can do the same thing to come up with a quantitative (i.e., numerical) estimate of a dog's breeding value for a particular trait, called the "estimated breeding value" (EBV).

You will remember that phenotype (P) depends upon both genetics (G) and environment (E):

P = G + E

Here, "environment" includes any factor that could influence the trait that isn't genetic. If we can figure out what G is, we will have information about how likely a dog is to have the genes for a particular trait. For example, when you're looking at hip scores, or at one descended testicle, or an x-ray of dysplastic elbows, or the MRI of a Cavalier King Charles Spaniel suspected to have syringomyelia, you are assessing phenotype only. You have no idea how much of the variation you see from animal to animal for that trait is attributable to genetics and how much to things that are not genetic. Because selection can only operate on genes, knowing the true breeding value for a trait in a particular dog is extremely valuable.

EBVs have been used with great success for many decades by breeders of other domestic animals. "But", I hear you say, "dogs aren't livestock". Can EBVs be used successfully in dogs?

Yes. Here are a couple of examples. You will see that in both, the efforts of breeders to control a genetic problem by selecting against an undesirable phenotype had little success before the adoption of EBVs.

Cryptorchidism in Boxers
Cryptorchidism (the failure of a testicle to descend) is a problem in many breeds, and historically, Boxers have had a high incidence of cryptorchidism. If neither testicle descends, the dog will be sterile because the heat of the body interferes with sperm production. But a dog with one testicle is fertile (although prone to testicular tumors). 

Most living Boxers can trace their pedigrees to four German stud dogs - Sigurd von Dom and his three grandsons, Utz von Dom, Dorian von Marienhof, and Lustig von Dom. All four of these dogs produced cryptorchid offspring. The first efforts to reduce the frequency of cryptorchidism in Boxers began in 1942, with a total ban on breeding cryptorchids (see the graph below). Nevertheless, the incidence of cryptorchidism increased in East German dogs over the next 40 years from about 6% in 1941 to 10% in 1981. In West German dogs, it increased from 7% in 1959 to 14% in 1985. In 1985, concerns about genetic diversity in the breed prompted the kennel club to encourage breeders to use lesser known males and breed away from popular sires, but this still did not improve the incidence of cryptorchidism. 

The unification of Germany in 1990 increased access to a broader gene pool, but cryptorchidism continued to increase unabated. In 1996, once again strict regulations on breeding were imposed. Bitches were excluded from breeding if they produced cryptorchid offspring in 2 litters, and sires were culled that produced more than 15% cryptorchids in at least 20 offspring. This policy reduced the frequency of cryptorchids, but it also removed 84% of the reproductive dogs from the breeding pool (not a good thing), and in any case improvement tailed off after a few years.

Finally in 2000, Germany removed all restrictions related to cryptorchidism and instituted the use of estimated breeding values (EBVs) to improve selection against cryptorchidism. Within only a few years they saw marked improvement (red in the graph).

Hip dysplasia in Hovawarts
EBVs have been used to successfully reduce the frequency of canine hip dysplasia (CHD) in several breeds. The data in the graph below are from a breeding program against CHD in the Hovawart. After efforts to reduce hip dysplasia using selection against phenotype failed to produce consistent improvement, severe breeding restrictions were instituted in 1984 that banned all affected dogs from breeding. This actually made things worse, reducing the number of unaffected dogs and increasing the number of dogs classed as borderline.

In 1989, selection based on EBVs was instituted, and this produced immediate, significant improvement in hip scores. In only 5 years more than 80% of all dogs had normal hips, and severely afflicted animals were nearly eliminated.

These are dramatic, but not atypical, examples of the improvements that can be achieved in just a few years by using EBVs to guide selection instead of phenotype. EBVs allow breeders to distinguish between dogs with bad hips and dogs with the genes for bad hips - essentially separating the potential influence of environment from the underlying genotype that the breeder is really interested in. This means that fewer animals will get removed from the gene pool, because dogs with a bad phenotype but good genotype for the trait of interest can be kept in the breeding stock for potential use.

Using EBVs for selection can produce one problem that is common to phenotypic selection as well. If everybody rushes to breed to the dog with the best EBV score, this will create a bottleneck and increase inbreeding if care is not taken to balance the reproduction of animals across the breadth of the gene pool. This should be a basic part of sound genetic management of a breeding population of animals anyway, regardless of the scheme breeders are using to make breeding decisions.

EBVs will be new to many dog breeders, but in fact they have been used for decades to guide breeding decisions of service dogs by Guiding Eyes for the Blind, Seeing Eye, and other organizations. Using EBVs, a well-run organization can manage genetic disorders, limit inbreeding even in a closed gene pool, and produce dogs with the traits that are important in a service dog. This improves the efficiency of the breeding program because more of the dogs produced are suitable for service.

Dog breeders can use EBVs to improve selection for the traits they want and reduce the ones they don't (Lewis et al 2013). EBVs can be used on any trait that can be evaluated by the breeder - temperament, size, herding ability, coat quality, heart disease, "showiness", hip dysplasia - anything you can judge to be better or worse, desirable or not desirable. EBVs are becoming available in more and more countries through the kennel clubs, and they are the most powerful tool now available for improving the health and well being of dogs.

Beuing R, G Beuing, P Pracht & N Janssen.l 2003 Cryptorchidism in dogs: prevention by breeding (from a seminar in 2003) (pdf)

Beuing R. Strategies in modern dog breeding. (pdf)

Lewis TW, SC Blott, & JA Wooliams. 2013. Comparative analysis of genetic trends and prospects for selection against hip and elbow dysplasia in 15 UK dog breeds. BMC Genetics 14:16 (pdf)