The amazing olfactory abilities of the dog are well-known to any dog owner. With a sniff of the door jamb, my dog knows who is on the other side of the front door before I open it, he can locate one of his toys hidden in a drawer, and he heads straight for the person with treats in their pocket when we go to the dog park. We have made good use of this portable detection system to find lost children, locate bombs, detect cancer, and many, many other things. The dog's nose is truly remarkable.

Just as amazing as its sensitivity to odor is its design, which is much more complex that it might first appear. Inside the muzzle is a complicated system of passages, the turbinates, that both humidify the air and also direct it to the olfactory organ for scent detection. This design must accomplish both respiration and scenting as breathing and sniffing, respectively.

To determine the sensitivity of scent detection, they placed a "smell" source, a chemical called 2,4-dinitrotoluene (DNT), in a small tin with holes in the side to allow vapor to escape. They used sophisticated chemical sensors (ambient ion mass spectroscopy) to document scent detection.

Finally, to visualize the flows of air, they used high speed video and "schlieren imaging", which uses optical techniques to photograph density gradients in air flow, as well as "theater fog" for some demonstrations.

What they found is fascinating. When the dog exhales, air is directed downwards and outwards. When the dog is sniffing a surface, the airflow from exhalation actually draws the odor from the object being scented towards the nose. This effectively extends the "aerodynamic reach" of the nose. The inspiration-expiration cycle during scent detection is repeated very quickly, about 5 times per second.

The air flow patterns that result from the simple design of the dog's nose increase the efficiency of scent detection by 8 times over that for the steady "inspiration" used in scent detection devices. This is achieved simply by directing the exhaled air away from the source of the odor in a way that draws the scent towards the nose. In fact, the farther away the scent, the greater the advantage of the dog's nose over a commercial device with continuous inspiration; in one set of experiments, the dog's nose was 4 times better than the detector at a distance of 10 cm from the source of the scent, and 18 times better at a distance of 20 cm.

Using what they learned from the dog, these researchers were able to construct a "bioinspired" scent detector that "sniffs" instead of continuously drawing in air and directs the exhaled air in a way that draws in scent from the source to the intake. When the air intake was directly above the source of the scent (distance = 0 in the graph below), continuous "inspiration" was as good or better than sniffing. But as the scent source moved away, the advantage of sniffing got better and better.
​

Needless, to say, the results of this study should be of great interest to the military and in situations where efficient odor detection is critical. A simple alteration in the design of odor detection devices, inspired by the anatomy of the canine nose, can result in spectacular improvements in efficiency.

Now why didn't we think of that?

REFERENCES

Staymates ME, WA MacCrehan, JL Staymates, and others. 2016. Biomimetic sniffing improves the detection performance of a 3D printed nose of a dog and a commercial trace vapor detector. Scientific Reports 6:36876; DOI: 10.1038/srep36876.

In more and more dog breeds, cancer is becoming a significant health issue. Irish Water Spaniels, Bernese Mountain Dogs, Golden Retrievers, Flatcoated Retrievers, Irish Wolfhounds, Scottish Deerhounds, and more have cancer rates of 40% or higher. Many research studies have been initiated to study the genetic basis of these cancers in the hopes of coming up with a genetic cause that can be managed by selective breeding. But it is becoming clear that the genetic backgrounds of most cancers are complex and poorly understood.

A major new study has been published that uses a systems-level approach to explore the tangle of genetic relationships and interactions associated with various types of cancers in humans (New Pathology Atlas maps genes in cancer to accelerate progress in personalized medicine) Researchers are finding that there are causal "themes" that account for many types of cancers. For example, many of the genes associated with cancer are involved in regulation of cell growth and proliferation, the processes that result in the growth of tumors. We are beginning to identify "prognostic" marker genes that are correlated with clinical outcome and duration of survival, some of which are shared among many types of cancers.

The paper is a heavy but fascinating read. But as part of the project, the authors have also created the Human Pathology Atlas as part of the Human Protein Atlas. The Pathology Atlas contains information on 17 major types of human cancers. Especially interesting is this chart (below) summarizing the number of genes associated with  favorable or unfavorable prognosis for each type. For most cancers, there are hundreds or even thousands of genes that affect both positive and negative prognosis.
​

We could be doing something like this for dogs. Accumulating the data that will be necessary for such an analysis will take significant effort and resources. But the rates of cancer in dogs are far higher than in any other animals, including humans (Do dogs have more cancer than other mammals?). We could collect data for the thousands of dogs each year that are affected by cancer and make it available to researchers for studies like this one. This will take infrastructure and organization, but from studies like this one, the model already exists. Breeders and dog lovers are usually enthusiastic supporters of research if they can see it benefiting the health and welfare of dogs.

I think we can do this, and if we build it, they will come.

REFERENCES

Uhlen et al 2017. A pathology atlas of the human cancer transcriptome. Science 357: 666 (eaan2507 (2017).
​

I've written before about the plight of the Doberman. This breed that was once hugely popular now sees declining registrations. (I can show you data for the UK Kennel Club, but unfortunately we lack the information to assess its status world-wide.) We do know that the gene pool is small and the dogs are highly inbred, with COIs in the neighborhood of 40%. The breed has high rates of cancer and a tragic incidence of mortality from dilated cardiomyopathy (DCM), which is sudden heart failure.

I have compiled a substantial amount of data about the genetic status of the breed (see references below), and there should be no question that the Doberman is in desperate need of rescue. But I am unaware of any effort (at least in the US) that is focused on increasing the size of the gene pool of the breed, which would require a cross-breeding program unless there are some populations hidden away that are dramatically different genetically. Further, there is no program dealing with the issue of population genetic management to minimize the loss of genetic diversity at the level of the entire breed.

If the information already available wasn't sufficient to convince breeders of the urgency of undertaking a genetic rescue, I am not optimistic that even more data will push anybody towards taking action. However, I stumbled on some very interesting data that are worth looking at, and for the sake of completeness I am posting them here. Note that these are not my data; they are (apparently) data for dogs in Russia (http://doberbase.ru), with information supplied by owners, so they can't be assumed to be either complete or unbiased. All I have access to are tables and charts on the website that present single numbers (e.g., a mean or maximum), not the values for individual animals. The data are for dogs born between 1982 and 2007. Because you can't have mortality data on dogs that haven't died yet, the statistics don't run to the current date.

Since the mid-1980s, the number of dogs born per year has hovered between 1000 and 1200 per year.

For dogs born in the early 1980s, the average life expectancy was 14 years or better, but it has been declining steadily since. Dogs born in the early 2000s have an expected average lifespan of about 10 years - 50% less than 30 years ago. 
​

The decline in lifespan is also reflected in the graph below, which shows that the percentage of dogs that live more than 10 years has been declining over the last 3 decades, and the fraction that don't survive past 8 years is increasing.
​

The website does not present data for all causes of death, but there are statistics for both cancer and DCM. These graphs tell a stark story. Since the 1980s, more and more dogs have been dying of cancer and DCM (top graphs), and they die at younger and younger ages (bottom graphs). Note that these increases in mortality cover a period when the number of dogs born per year has remained relatively constant (see the first graph above).

If you have a ruler and pencil, you can print off a few of these graphs and extrapolate to the hypothetical date of extinction. When I do that for the life expectancy data, extinction is predicted in about 2030, only 13 years from now. In fact, dogs born today would have an expected lifespan of less than 5 years if this extrapolation is correct. Of course, there is no way to know if this line continues on the same trajectory for the next few decades, but there is no compelling reason to think that it won't; there have been no breakthroughs in eliminating  DCM, and research on cancer is unlikely to reduce the number of animals afflicted. In fact, extrapolation of the line for the incidence of DCM from another body of data indicates that 100% of the breed will be afflicted by 2040. If  cancer doesn't drive them to extinction, DCM will.

These data simply add to the substantial body of very convincing evidence that this breed is rushing headlong towards extinction. 

A lay person viewing this situation from outside the fancy would no doubt expect to see custodians of the Doberman focused like a laser on the goal of turning this ship around. At the very least, there should be alarm, a sense that this is a desperate time for this breed. But there is no alarm, no anxiety about impending doom. In fact, very little is happening that will make a difference to this breed.

A new non-profit called the Doberman Diversity Project aims to genotype 1,000 Dobermans "to help Breeders improve the health and longevity of the breed by creating affordable access to cutting-edge, high-resolution DNA testing and online breeding tools...for one important goal: improved longevity & genetic health in the Doberman worldwide." While I applaud the efforts of this group, I don't see how this will improve the situation of the breed. With a small gene pool and 40% inbreeding (so everybody has the same genes and they are highly homozygous), there simply isn't much to work with - selection requires genetic variation, and there isn't any. We can do little more than push the same peas around on the plate. There isn't much time either - the life expectancy for dogs born 5 years from now in 2022 is only about 3 years. The window of time for doing something to save this breed is very narrow.

There was an article by Dan Sayers in the July 2017 issue of Showsight Magazine about the notion of "the preservation breeder", an attribution that has recently become popular in the dog fancy. This term has a history in animal breeding, and it refers to the protection of the "genetic resources" of a group of animals - genetic diversity, the genes for ancestral traits, and the genetic health needed for successful reproduction and adaptation to changes in the environment.

Oddly, in his article about dog breeding, Sayers defined the term "preservation breeder" in the context of a cultural rather than genetic resource, as it would apply for example to things like historic buildings under the management of the National Park Service (NPS). By this definition, he says, "anyone who breeds purebred dogs is a preservationist". 

I'm not going to quibble about terminology. I am, however, going to point out that extinction eliminates the possibility of preservation, whether of genes or culture, and the data clearly have the Doberman headed for extinction. It simply isn't the case that producing a litter of purebred puppies is preservation, if that process pushes the breed one generation closer to disappearing entirely.  If we are not preserving the genes required to build a healthy dog, we will lose the breed, both genetically and culturally. The organization that understands this in terms of animal breeding and genetic resources is not the NPS, but the FAO (Food and Agriculture Organization of the United Nations). This organization, through its Domestic Animal Diversity Information System, is tasked with monitoring the status of the domesticated animals (and plants) that feed the people of the world. They are very much focused on genetics rather of culture.

"Preservation breeder" as used now by the dog fancy has little to do with preserving a breed, at least as commonly defined and as as used by Sayers. Preservation breeding requires sustainable breeding - the point being to produce a next generation of animals that is as healthy and fit as their parents. This requires more than simply producing a litter of puppies from purebred parents. That we are not maintaining health from one generation to the next is obvious from the trajectories of the graphs above for lifespan and for DCM and cancer. This is most certainly not preservation breeding.

After my previous essays about Dobermsns (see the references below), many people contacted me about "doing something" about Dobermans before it is too late. But the efforts of a few individuals that are peripheral to the main core of breeders in the fancy will make little difference. (We're not talking here about the puppy mills.) Furthermore, there is initial outrage and motivation to do something, but invariably it wears off, the voices grow quiet, and people go back to doing what they were doing before.

As Sayers notes, "when disease threatens a breed's future and the health and welfare of individual animals is at stake, it may become imperative to consider doing the unthinkable" (italics mine). 

The breeders at the core of the fancy are adverse even to the notion that the breed is on genetic last legs. If "the unthinkable" is unthinkable, then what plan do breeders have to save this breed? Is this at the top of the agenda of the national breed club? Is the health committee worried about specific genetic issues instead of extinction? Who will step up to the plate and save this breed before it is too late?

Indeed, my question to all breeders: What kind of preservation breeder are you?
​

Beuchat C. 2016. Are we watching the extinction of a breed? (or, Why are we focused on consequence instead of cause?)

Beuchat C. 2016. Are we watching the extinction of a breed? (part 2)

Beuchat C. 2017 An update on the genetic status of the Doberman Pinscher

For example, the graphs below show the probability of degenerative joint disease for dogs of different ages as a function of their distraction index (DI), which measures how far the ball of the hip joint can be displaced from the socket. 

In the Rottweiler (top graph), for a DI = 0.75 (on the horizontal axis), there is a 50% risk of DJD by the time the dog is 35 months old, and 50% of the Rottweilers with a DI = 0.95 will display DJD by the time they are 16 months old. The greater the DI, the earlier in life the dog is likely to display signs of degenerative joint disease. 

Compare these data with those for the German Shepherd Dog (GSD; below). For this breed, the curves are shifted to the left. By 35 months, a GSD with a DI of only 0.55 has a 50% risk of DJD (versus 0.75 for the Rottweiler above), and the DI at 50% risk for 35 month old dogs is about 0.7, compared 0.95 for Rottweilers.

In fact, at all ages, the probability of DJD in GSD is greater than in the other breeds that have been examined at > 24 months old (the line for GSD is shifted to the left relative to other breeds; below). 

The bottom line here is that breeds vary in their sensitivity to hip laxity in the development of degenerative joint disease. A Labrador Retriever with a distraction index of 0.5 has a low risk of DJD by 24 months (about 10%), while the same DI for a GSD is associated with a 40% risk of DJD by the same age. A DI of 0.5 presents a relatively low risk of DJD in a Rottweiler or Golden Retriever (about 10%), but a significant risk (40%) in a GSD.

This means that interpretations of hip scores need to be breed-specific. This is highlighted as well in a study of breed-specific differences in hip laxity in smaller breeds. This graph shows the distraction index (vertical axis) for 15 breeds of dogs weighing from 4-16 kg (8-35 lb). This study focused on dogs with normal hips that were free of any evidence of DJD, so in these dogs hip laxity did not result in the development of hip dyspasia. The differences among these breeds are striking, with the highest DI found in the Dachshund, Pekingese, and Miniature Poodle. Note as well that these measurements varied little among individuals, as evidenced by the small standard deviation (SD) relative to the magnitude of the distraction index.

 
The degree of displacement in the breeds with the highest DI is really quite remarkable. This is a series of radiographs of a Dachshund, first in resting position (top left), then with distraction resulting from 12 kg (top right) and 20 kg (center, below) of tension.

In the third photo above, you can even see darker areas between the the femur head and acetabulum where a vacuum has formed as a result of the large displacement.

​

Why do some breeds have looser hips than others? We don't know. Interestingly, the breed with the tightest hips based on DI is the GSD, yet as we've seen this breed has the earliest and most significant development of degenerative joint disease. This seems to fly in the face of the notion among some breeders that loose hips in herding breeds provide greater flexibility that is useful in these working dogs. Even with relatively tight hips, the GSD is most significantly afflicted with joint degeneration that would compromise its function as a working dog. 

Most importantly, these data show that hip laxity alone is not a very good indication of whether a dog will go on to develop joint disease. Dogs could be removed from breeding based on DI that never develop DJD, and others could be allowed to breed based on a score that presents low risk in one breed but high risk in another. 

What about genetic tests for hip dysplasia? There are several published studies that claim to have identified genetic markers associated with the development of hip dysplasia, and some of these have been developed into commercially available tests. However, they are breed specific, and at least one has just been shown to be of very poor predictive value (Manz et al 2017).

Unfortunately, the situation for breeders wishing to reduce the incidence of hip dysplasia remains problematic. The best options remain consideration of the phenotype of related dogs (as in Estimated Breeding Values), and a genomic rather than marker-specific approach to identify genetic risk using thousands of markers (Guo et al 2011).

REFERENCES

​Arnbjerg J. 2017. Hip joint laxity in small dog breeds: a radiological study. SOJ Vet Sci 3(1): 1-5.

Guo G, Z Zhou, Y Wang, K Zhao, L Zhu, G Lust, and others. 2011. Canine hip dysplasia is predictable by genotyping. Osteoarthritis Cartilage 19(4): 420-429. . 

Manz E, B Tellhelm, & M Krawczk. 2017. Prospective evaluation of a patented DNA test for canine hip dysplasia (CHD). PLOS ONE 12(8): e0182093. https://doi.org/10.1371/journal.pone.0182093

Smith, GK, PD Mayhew, AS Kapatkin, PJ McKelvie, FS Shofer, & TP Gregor. 2001. Evaluation of risk factors for degenerative joint disease associated with hip dysplasia in German Shepherd Dogs, Golden Retrievers, Labrador Retrievers, and Rottweilers. JAVMA 219 (12): 1719-1724.

Should you breed a dog that doesn't have excellent hips or elbows? How much less than excellent is still "good enough"? Most dogs, especially of larger breeds, can't boast about having excellent hip scores. So how does a breeder decide what to do? You might see a a hip x-ray posted in a Facebook group with a plea for advice from other breeders. "Are these hips really that bad? Are they bad enough to remove a dog from the breeding pool completely?  

If all you have is the evaluation or x-rays for a single dog, you really don't have enough information to make a good decision. The reason is something called "heritability".

Many traits are affected by both genetics and environment. Heritability tells us about the relative contributions of genes and environment to the variation in a trait in a population. The key word here is variation. Some of the variation in a trait among individuals is because they have different genes, and some will be due to non-genetic (i.e., environmental) factors. The fraction of the variation accounted for by genes is the heritability of the trait in that population.

The heritability of a trait is specific to a particular population measured at a particular time. A different population could have a different heritability for a particular trait, and even the same population measured at different times can have different heritability. A trait can have a high heritability in a population of animals raised and measured under the same identical conditions (i.e., genes account for most of the variation in the trait), while the same trait can have low heritability in animals raised under different conditions (i.e., differences in environment account for a lot of the variation in the trait).

If you're trying to select for or against a trait that is influenced by both genes and environment, this presents a problem. If heritability of the trait you're interested in is relatively low, this will mean that phenotype - what you can see or measure about the trait - is not necessarily a good reflection of the animal's genotype, and of course it's genes you're trying to select for. 

So when you're trying to make breeding decisions based on traits affected by both genes and environment, you must remember that what you see is not necessarily what you get. A dog with moderate hip dysplasia might have been very overweight as a puppy, or loved jumping off the side of the front porch every afternoon to greet the mailman. Those hips might reflect the adverse influence of non-genetic factors - the "environment".

What do we do about this dog with the less than perfect hips? Was it environment? Or was it genes?

We don't know what genes affect the risk of hip dysplasia. But a dog will share some of its genes with relatives - parents, siblings, and offspring. We can learn more about the genes in a dog by knowing about the hip evaluations of closely related dogs. Let's look at some examples.

Here's a simple pedigree with information about the relatives of a dog you are thinking about breeding that has a less-than-stellar hip score (he's the fellow with the red dot on his head in the third row). You find out that the dog's siblings have acceptable hip scores, the sire and dam are also good, and in fact even back another generation the dogs either have good hips or no information. You would probably be making a good bet to assume that your dog has inherited favorable genes for normal hips.

If the pedigree looked like the one below, it would be a bit harder to decide. There is a sibling with bad hips, and there are litter mates with no information. (Maybe bad scores weren't submitted???) There's no info for the dam, and missing information in earlier relatives. Did two pups in this litter get genes that predisposed them to hip dysplasia, or did these two pups love to push each other off the front porch when they were little? Should you breed this dog? What would you do?

​Unfortunately, this next pedigree is probably more like what the average breeder will have to deal with. Bad hips, missing data, no patterns, and it's anybody's guess if all you have is this pedigree information to work with. What would you do with this one???

What you need for this problem is a statistical technique called "estimated breeding values" (EBV). This is a way of mathematically predicting the "quality" of the genes in a dog for a particular trait. You can determine the EBV for a particular trait from pedigree information, and some kennel clubs are now providing EBVs for hips and elbows to breeders.

However, with high resolution genotyping we can now determine genomic EBVs directly from DNA (gEBVs). Using DNA is especially powerful because we don't even need to know the relationships among the dogs we have data for. The software evaluates the DNA for each dog, accounts for whether they are affected or normal for the trait, and spits out a number that reflects the breeding value for a particular dog. 

The singular beauty of gEBVs is that they can be used for any trait, and you don't need to know anything about the genes involved or mode of inheritance. In a nutshell, you can manage a genetic disorder without first launching a research project to hunt for the genes. You simply need the trait information for as many dogs as possible, and their high-resolution DNA genotypes. 

Why aren't we doing this? Until now, we didn't have the DNA data or the computing tools, but the ICB Breeder Tool will have this capability. The same DNA test you would do for health testing in your breed can also produce a research-quality genotype that can be used to determine genomic estimated breeding values. This information can be provided for breeders who can then choose the breeding pair that has the lowest risk of producing problems in the offspring - or the highest probability of producing some desirable trait. Furthermore, we can assess in young puppies the genetic risk of disorders that only appear late in life. 

Because all the math is done behind the scenes, there are no privacy issues with data. A dog simply has a gEBV based on the information available for related dogs, without the breeders needing to know the specifics of that information. There is no need for secrecy, no fears of getting hung out to dry on Facebook. 

Genomic breeding values should revolutionize dog breeding. Even better, they should eliminate the need to invest thousands of dollars, and months or years of waiting, in research studies that might never produce information that is useful to breeders. We have invested millions of dollars in cancer, epilepsy, allergy, heart, and kidney research, and for all this we have just a handful of genes that breeders can test for. There is a better way.

We will be rolling out genomic breeding values in the ICB Breeding Tool as soon as we have enough data for a breed to do it. Talk to your fellow breeders about it. For this to work well, the more data available the better, even for dogs that are not intended to be bred, because their DNA and trait status will improve the power of the estimate of breeding value.

If you think you can get enough participation to set this up for your breed, please let us know and we will work with you to get it set up.

You can read more about genomic selection here -
Boichard et al 2016. Genomic selection in domestic animals: principles, applications and perspectives. (Open access)

Check out
ICB's online courses
​
*******************

Join our Facebook Group
ICB Breeding for the Future
...the science of dog breeding
*******************

Visit our Facebook Page
ICB Institute of Canine Biology
...the latest canine news and research
​