Can DNA Decipher the Mix?

Unraveling the genetic tapestry provides clues to breeds and their mixes
By Jane Brackman PhD, November 2008

A mongrel dog is like a box of chocolates: You never know what you’re going to get. And therein lies the appeal. What’s more fun than serendipitous unpredictability all bundled up in puppy fur? But when that puppy grows up, we inevitably make assumptions about her ancestry based on how she looks and behaves. Our logic goes like this: “If my pooch is long and low to the ground, and she never barks, she must be a Corgi/Basenji mix.”

But it’s much more complicated. The genes—and there may be hundreds—that work together to make a Corgi look like a short-legged Shepherd may be completely different than those responsible for a Basset Hound’s low-slung carriage. With some exceptions, scientists cannot yet connect genetic dots to specific traits. But they have discovered something tangible that measures some of the differences between breeds: genetic patterns of organization displayed on a scatter graph that answer the question, “What’s the same and what’s different?”

A scatter graph provides a symbolic visualization of DNA, wherein each individual dog contributes one point. The resulting pattern indicates the type and strength of the relationship between individuals. The more the points cluster around each other, the more alike they are.


Until only a few years ago, scientists couldn’t identify the differences in genetic material that might explain profound variations in the Canidae clan. From wolf to West Highland White Terrier—they all looked the same under the microscope. Then, in 2004, Elaine Ostrander and her colleagues at the Washington-based Fred Hutchinson Cancer Research Center published data indicating that as much as 30 percent of the dog’s genetic material accounts for breed variation (Science, May 2004).

In addition to simplifying methods used to find markers for breed-related disease, the researchers identified patterns of “breedness” and tracked the history of breed DNA. At the same time, by following mitochondrial DNA, genetic material passed down from mother to offspring without changing, they traced the breed’s journey.

Depending on how much time is attributed to a generation and how many generations are involved, scientists can estimate how much time has passed. Based on this tracking, it has been suggested that it took 5,000 years to develop and refine a handful of the world’s 350-plus breeds, and about 400 years to create the rest.

Research indicates that four distinct breed groups are ancient: (1) Middle Eastern Saluki and Afghan, (2) Tibetan Terrier and Lhasa Apso, (3) Chinese Chow Chow, Pekingese, Shar-Pei and Shih Tzu; Japanese Akita and Shiba Inu, (4) Arctic Alaskan Malamute, Siberian Husky and Samoyed. Although the 13 breeds look different, they are so closely related that they are represented by a single genetic cluster. It’s likely they all originated from the same stem-parent—proto-breed, if you will—who roamed the Asian continent.

As humans migrated from one place to another, this ubiquitous proto-breed trotted along, bringing with her the ingredients needed to cook up all the breeds we’re familiar with today. Her offspring performed work unique to each geographical region, such as hunting, hauling or guarding. Isolated and mating only with each other, “accidental” breed types exhibiting consistent shape, color and behavior emerged.

No matter what historians might claim—scent hound to sight hound, bird dog to bad dog—evidence produced through genetic research indicates that all remaining breeds have been concocted in the last 400 years. Although closely related to one another, they can be identified as distinct based on the way their DNA separates.

How They Do It
Sue DeNise, vice president of genomic research at MMI Genomics Inc., which developed the Canine Heritage Breed Test specifically for mixed-breed analysis, talked to us about how her company analyzes canine DNA. “We’ve been doing testing for AKC parentage verification for a long time,” she notes. “We initially started working in the cattle business, looking for genetic markers in order to trace what was important to cattle breeders. Out of that whole-genome association study, we had purebred and crossbred cattle, so we asked, ‘What can the markers tell us about underlying traits in breeds of cattle?’” Their discovery paved the way for the companion animal program, which was modeled on what they learned with cattle.

“We look at ’breedness‘ among dogs. Our canine database is built with 10,000 samples of 108 breeds. We ran 400 markers to identify the best markers for a ’breedness‘ test against 38 breeds. We created a panel of 96 pieces of DNA that split dogs into their identified pure breed. In our preliminary test, we found that individual purebred dogs cluster with other purebreds.” Initially, MMI chose 38 AKC registered breeds from their database, selected for their popularity based on number of registrations. Recently, as DeNise notes, they increased the number to 108 breeds. This jump in breed recognition required testing thousands of markers to identify the just over 300 markers that characterize these 108 breeds.

Constructing Breeds
Like all species, domestic dogs are on an evolutionary journey, starting at wolf and going somewhere yet to be determined. We tinker with evolution, but might be surprised to find out we don’t control it. Our concept of a breed—that individuals within the breed look alike—is nothing more than a snapshot of the DNA time line, taken while we’re doing the tinkering.

Breeds are created a number of ways. In simple terms, when breeders interfere with natural reproduction and rigorously select for traits favored by humans, specialized breeds like Retrievers, Spaniels, Hounds and Terriers are the result. Saving spontaneous mutations in a litter of dogs, repeating the breeding to get more of the same mutation, and breeding those dogs back to one another has resulted in the English Bulldog, Chinese Crested and Inca Hairless. More recent breeds, such as the Airedale, Australian Cattle Dog and Doberman, are the result of crossing older breeds to make new ones.

When kennel clubs closed gene pools in the late 19th century to suspend change in registered dogs, breeds drifted toward a more uniform stereotype. Until the early 1800s, an assortment of dogs with similar talents who could produce somewhat similar offspring were awarded the right be called a breed. Breeds evolved, flourished and disappeared when jobs were eliminated. Tumblers, who mesmerized prey by “winding their bodies about circularly, and then fiercely and violently venturing on the beast,” disappeared when guns came into widespread use. Turnspit dogs, who made a living running on a wheel to turn meat so it would cook evenly, received their pink slips when technology improved cooking methods.

By and large, Victorian society was not so pragmatic; sentimentality and commercial opportunity were catalysts for saving unemployed breeds from their inevitable demise. As a result, many Terrier breeds went from killing varmints in the barnyard to killing time in the Victorian parlor in less than a decade.

Whereas previously, a breed was a regional product maintained and preserved by a small community of knowledgeable people, commercial interest in the well-bred pet dog initiated a shift in breeding practices during the Victorian era. The old-money kennels operated as a pastime by the wealthy gave way to a large number of small, commercially operated kennels run by entrepreneurs of modest means and experience.

Germane to this tale is that, according to the unwritten rules governing canine physiology, anatomy and behavior go hand in hand. One cannot be changed without affecting the other. Victorian enthusiasts who were busily adding aesthetic traits to utilitarian breeds were creating not only subtle variations in type, but in many cases, modifications in behavior as well. As utilitarian breeds went from working hard to hardly working, many exhibited new physical and behavioral characteristics that were compatible with their augmented duties as companion animals. Breeders claimed the “sub-breeds” as their own, made up new names and registered each one.

However, no matter how they’re sliced and diced, reducing and suppressing genes so they aren’t expressed doesn’t mean they’ve been eliminated. They’re still lurking and, depending on the method used to analyze the DNA, the lurkers often show up in the results.

Deconstructing Breeds
The problematic aspect of analyzing mongrel DNA is that breeds were not all created at the same time. As DeNise explains, “As new breeds are developed, they may not appear as uniform as older breeds. When older breeds are crossed to create a new breed, there is some period of time before the new breed develops a unique DNA pattern of its own. In these cases, the more ancient breed sometimes appears in the new breed. The number of generations required to have a uniquely identified breed created from crossing of older breeds depends on the number of breeding animals in the new line, the severity with which the breed owners apply the standard, and the amount of introgressing [inbreeding, or breeding immediate relatives; line breeding, breeding close relatives; and backcrossing, breeding sibling to parent] allowed by the registration agency.”

Most people assume all mixed-breed dogs had a purebred ancestor at some time in their recent heritage. But in fact, this is not necessarily the case. When you run a mongrel’s DNA through a computer program, the algorithms attempt to group breeds together on a scatter chart. If the heritage of the dog is such that it is not in MMI’s database of 108 breeds, the program tries to find varieties that are most alike. Because at least one or two of the handful of ancient breeds are in every modern dog, sometimes the program will identify an ancient breed in the mix. “In the report we send to the client, we use the terms ‘primary,’ meaning the majority of the DNA matches a breed; ‘secondary,’ meaning less than the majority of the DNA but a strong influence nonetheless; and ‘in the mix,’ meaning the least amount of influence,” DeNise notes. That’s how you might get an obscure breed in the report. For instance, a 35-pound mongrel with a tablespoon of Husky and a teaspoon of Border Collie may also have a dash of Borzoi, because before gene pools were closed a century ago, Huskies were crossed with coursing hounds to add speed.

Don’t Judge a Pup by Her Cover
As MMI Genomics states on their certificates, “Your dog’s visual appearance may vary from the listed breed(s) due to the inherent randomness of phenotypic expression in every individual.” What this means is that you may look nothing like your parents, but you have Grandma’s great legs and Great Uncle Harry’s turned-up nose. All in all, though, no matter how genes are mixed and matched, your family members resemble one another. However, if Grandma was an Afghan Hound and Great Uncle Harry was a Pug, “random phenotypic expression” can be pretty extreme.

Researchers are intrigued by data that suggest expressed traits are somehow turned “on” and “off” by other genetic components, thus causing the wide variations in canine form and behavior. For instance, it’s possible that many breeds have the genetic potential for a black tongue, but only a few breeds have the molecular mechanism to switch that color on. So that black-tongued mutt may not have any Chow in the mix after all.

On the other hand, the results may show that a quintessential Heinz 57 has the genetic makeup of a single breed and it could be one she looks nothing like.

DeNise explains it this way: “In a population of any breed, there are dogs that are carriers but don’t exhibit phenotype [observable characteristics]. If you reduce the size of breeding population—creating what we call a bottleneck—you start increasing the frequency of deleterious traits, like dwarfism or white coat. If we looked at the DNA of, for instance, a group of white mini-German Shepherds, they would probably cluster with German Shepherds. After they’ve intermated for five to six generations, we may not come up with that. They would cluster with each other. If breeders were changing allele frequencies quickly, you could do it very fast.

“There are always contradictions that make you say, ‘Huh, that’s really weird.’ One odd thing that happens is due to some sort of random assortment of genes in mixed-breed dogs. The algorithm may identify a breed that is not consistent with the physical appearance of the dog. We sometimes get an indication of this when the certificate is printed with the picture of the dog provided by the owner, and the certificate is reviewed by our customer service department prior to mailing it to the pet owner.”

A 90-pound, wiry-haired mongrel who swims, chases balls and makes goo-goo eyes like a Golden Retriever and whose only pedigreed relative is a very distant Chihuahua confounds the process, but says a whole lot about the complexity of canine genetics and why some scientists devote their careers to studying canine evolution. Extreme variation in anatomy and behavior is unique to the domestic dog. If humans were an equally anomalous species, we’d weigh between 20 and 650 pounds and range in height from three to 10 feet. In dogs, adaptations change with such speed that scientists suspect there may be a clue in the canine genome that could reveal how evolution works.

Sorting It Out
Before they launched the project, MMI tested DNA from a street dog rescued from a Thai village. You’d think there would be no clusters of any kind, but the computer identified Chow and Akita in the mix. This isn’t surprising, because the free-living common village cur who populates most of the developing world may be the closest living relative to the original proto-breed. Findings suggest that Thai pooch stores a sizable chunk of the original genetic blueprint of every single living dog in her DNA. The question is, how much?

MMI can’t as yet define the percentage of “breedness” in mixed-breed dogs. One reason is that some breeds cluster loose and others tight. Why this happens isn’t clearly understood. German Shepherds, Standard Poodles and Collies cluster tight. Miniature Poodles cluster tight, but Toy Poodles cluster loose. Within their breeds, Labrador Retrievers and Beagles often cluster as two different groups. According to DeNise, “Labs from the United Kennel Club that are bred specifically for hunting and AKC Labs do not necessarily cluster as one breed. And AKC Beagles and Beagles bred specifically for research don’t cluster together either.”

She adds, “In my opinion, it’s possible that a population that increases rapidly doesn’t cluster as well as those populations that have remained static. This is because, as you increase a population to accommodate breed popularity, people breed everything, including animals that may not exhibit all the physical characteristics that are desirable.”

And even when dogs look alike, they can display behavioral differences. As DeNise notes, “We understand so little about how behaviors are coded. Many behaviors are learned, but there are probably multiple genes that are responsible for herding, birding, heeling—these kind of hard-wired behaviors.”

Scientists are eager to tease out genetic connections to breed-associated motor patterns. When wolves hunt, they display these behaviors sequentially: orient > eye > stalk > chase > grab-bite > kill-bite > eat. Artificial selection, however, extracts and segregates these patterns in incomplete sequences. In certain breeds, individuals perform the abbreviated motor pattern repeatedly. A Pointer who stops dead in her tracks and stands stock still with her front leg held rigid in mid-stride to indicate the presence and position of game is the lofty goal of bird-dog breeders. To wolves, it’s just a good meal interrupted.

Combinations of canine anatomy and behavior push and pull one another along in a rhythm of interconnected patterns in relationships that may not be as random as they appear. Like principal components of an automobile in which the size of the engine and the weight of the body directly affect efficiency, it appears that dogs, too, have integral parts wherein one component is proportionate to the other.

Researchers don’t fully understand the relationship, but they are making headway. As reported in Genetics (June 2008), a team of scientists identified a few single genes that regulate systems controlling skull shape, weight, fur length, age span and behavior. Because mutts are combinations of DNA from different breeds, they may hold the answer to how the genes influence multiple traits.

Scientists suspect that many evolutionary secrets are hidden in the dog genome. For dog lovers, deconstructing Molly or Max’s mixed-breed heritage is an interesting intellectual mystery to be discussed at cocktail parties or the dog park. For scientists, their genetic material is nothing less than an instruction manual for species building. Whereas populations evolve over the course of millennia through the process of natural selection, dogs can change so rapidly and abruptly that they represent evolution at hyperspeed. How it happens remains a puzzle. Now scientists are looking to mutts to find the missing piece.



Jane Brackman, PhD, is an authority on the cultural history of canine domestication and the author of two books on pets in 19th-century America. See her new pup, Barkley, and watch him grow on her blog.