17 The Genetics of Pedigree Breeds

Once the dog genome and the vast range of SNP markers were made available in 2005, scientists were quick to capitalise on the bonanza by re-drawing the evolutionary tree of dog breeds constructed with mitochondrial DNA over twenty years previously. In 2010 Nature published the results of a comprehensive study of 912 dogs from 64 breeds.¹ The authors were a group of thirty-six scientists led by veteran dog biologist Robert Wayne and Bridgett M. vonHoldt. Reassuringly, all branches of the resulting tree led back to the wolf, ruling out any major contribution to the dog genome from other species that, theoretically at least, might have entered the gene pool through the male line. See a visual representation of this tree, or phylogram, on the next page.

The relationships defined in this tree are based on the overall similarity between the sequences of the autosomes, the thirty-eight pairs of dog chromosomes not involved in sex determination. Breeds that cluster together share more identical DNA sequences than breeds that are far apart. Unlike the mitochondrial tree (Wayne/Vilà, page 17), the autosomal tree is drawn in a circular pattern. Breeds are displayed around the perimeter and the deduced links between them are towards the centre. The branching order is a rough approximation of the time that has elapsed since the breeds were established. From there, branches radiate out from the wolf, first to the so-called ‘ancient’ breeds, the Basenji, Australian Dingo and Chow Chow, then the Asian and Arctic spitz breeds. The next branch leads to the Afghan and Saluki, and at the end of the next branch are the Samoyed and American Eskimo dog, grouped closely together as one might expect with the Arctic breeds. After these, the next major limb of the tree carries all other breeds.

This is a good point at which to remind ourselves that, unlike thoroughly researched human family trees for example, where all the links are known, these trees represent the most likely rather than absolute scenarios. We are, after all, drawing them as best we can to fit the genetic data. The absolute accuracy of the tree can never be guaranteed. Neither is it strictly speaking an evolutionary tree, though there are evolutionary inferences to be drawn from it. Although it might look like a real family tree that starts at the centre and grows outward, it is really a diagrammatic representation of the genetic similarities between different breeds of dog. It is reasonable to assume that breeds on two close branches might have originated from a common founder, but it may not be quite so simple. So, for example, although the Samoyed and American Eskimo dog are placed at the tip of a long branch it doesn’t mean that all the dogs of both breeds are descended from a single ‘common ancestor’. As we shall see in a moment, there has been a great deal of mixing between breeds that we know about but which doesn’t show if we treat the diagram like a genealogy. This kind of diagram is formally known as a phylogram rather than a tree in an attempt, not always successful, to avoid the confusion with true genealogies.

There are ways, which I won’t trouble you with, for getting the tree that ‘best fits’ the data, and that is what is reproduced on page 117. The same algorithms also produce alternative trees that don’t fit quite so well, perhaps differing in the detail of some of the branches. These might be closer to reality, but not by much.

The most revealing aspect of the autosomal tree is the way that most breeds group with others of the same type as defined by the UK and US kennel clubs. Scent-hounds group with scent-hounds, mastiffs group with mastiffs, herding dogs with other herding breeds and so on. This is quite different, if you recall, from the tree derived from mitochondrial DNA, where individual dogs of the same breed are often descended from different ancestral females. This isn’t quite as surprising as it sounds at first. Although mitochondrial DNA contains the genes that enable cells to convert food to energy using oxygen, these are only a small fraction of the thousands of genes that do everything else. The fact is that the individual dogs of the same breed have different mitochondrial ancestors, and also that dogs of different pedigree breeds can share the same mDNA ancestry. It is a reflection of the fact that most of the features that distinguish breeds, like appearance, temperament and so forth, are controlled by many other genes. It is the similarity in all these other genes that explains the remarkable clustering not only of breeds but also of breed type in the genome phylogram. Thinking of the phylogram as a summary of similarities in all genes, it becomes less of a surprise to find similar breeds and breed types clustering together.

During the work leading up to this definitive summary, the question arose as to whether genetic analysis could be used to predict the breed of a dog from its DNA alone. This was first attempted back in 1999 in a case before a German county court. The essence of the case was that a car collided with a dog and suffered substantial damage to its front end. Completely disregarding its legal obligations, according to the police report, ‘After the collision, the dog left the scene of the accident without proving his identity …’

The driver and owner of the car filed a law suit including a claim for damages against a local shopkeeper who he suspected was the owner of the dog. One of his two German Shepherds had apparently been treated for minor injuries around the time of the accident. The judge ordered a test on the injured dog to compare its mDNA to three hair fragments recovered from the damaged car. The sequences from the hairs left on the car and the defendant’s dog were both definitely canine but they differed from each other in their detailed sequences. After this result was brought to court, the case was dismissed and the suspect walked, or at least limped, free. Although the mDNA analysis cleared the dog, it was only later that it was appreciated that mDNA could not have been used to differentiate the breeds. For that, it was essential to have nuclear DNA sequences.

Heidi Parker in Robert Wayne’s lab was the first to explore in detail the clustering of breeds using nuclear genes, and in 2004 she published her results in Science.² This was the year before the dog genome sequence appeared, with its galaxy of SNP markers, and she had to rely on a different but still effective genetic system based on what are known as microsatellites. These are small segments of DNA that occur as repeated blocks a few bases long. They are very useful as genetic markers because the number of blocks in a run can vary and is comparatively easy to measure. Genetic fingerprinting, invented by the British geneticist Alec Jeffreys in the late 1980s, is based on microsatellites.

Parker and her colleagues used a panel of 96 microsatellites in 414 dogs from 85 different breeds, and tried to assign the breed from the genetic data alone. It was a remarkably successful exercise, assigning more than 99 per cent of ‘test’ dogs to the correct breed. Only four dogs were classified incorrectly: a Beagle was identified as a Perro de Presa Canario; a Chihuahua, as a Cairn Terrier; and two German shorthaired pointers, one as a Kuvasz, the other as a Standard Poodle. The correspondence was quite remarkable, with 410 of 414 assignments being spot-on, but it was nonetheless surprising that the ‘errors’ assigned dogs to apparently un-related breeds rather than something rather similar.

By 2004 genetic genealogy and ancestry testing for humans was well under way and it was no surprise that companies saw an opportunity to put the technique to good use in dogs. Not only might it appeal to the owners of purebred dogs, but it could also offer a way of determining the mixed ancestry of mutts. As with any commercial operation there is a balance to be struck between accuracy and price. There are no rules governing the fidelity of such tests, as there are for human health diagnostics, which must gain approval by the US Food and Drug Administration (FDA), for example. Testing companies compete on reputation and price just as they do in clothing, cosmetics and a thousand other consumer products. The science behind the tests is based on the type of analysis which Heidi Parker and her colleagues pioneered with the pedigree breeds. However, it does not follow automatically that the remarkable accuracy in breed assignment achieved by Parker with pedigree animals will translate seamlessly into a comparable accuracy when it comes to mixed breeds. Let me explain.

Parker’s original 2004 assignments were based on genetic similarities of a number of microsatellite genetic markers, while the more advanced 2010 treatment used SNPs. None of the markers in either system is diagnostic of the breed on their own. It is only when they are combined that the ‘most likely’ breed assignment can be made for individual dogs.

I’ve been testing DNA, in humans, for more than twenty years and am struck by the reputation for invincibility that DNA has acquired and has maintained over all these years. I think it began after the remarkable precision of genetic fingerprinting and its rapid and well-publicised application to some very grisly cases of rape and murder. The stunning accuracy of individual identification claimed for the technique, often to the level of one in several billions, was thoroughly and publicly tested in the courts. DNA secured convictions of the guilty and overturned wrongful incarceration of the innocent. Faced with the evidence, rapists changed their pleas to guilty and victims were spared a courtroom cross-examination. This is genetics at its most impressive.

The years have only enhanced DNA’s reputation for invincibility, and it has now become part of everyday vocabulary. Not in the original sense of the word (it isn’t a word, incidentally, but an acronym – best forgotten – for Deoxyribo Nucleic Acid), but as a metaphor for a mysterious essence. This is great news for geneticists like me who find themselves wrapped in the cloak of invincibility and assuming the Delphic power of the Oracle. But, in life as in myth, it only goes so far. Although a DNA sequence is an ultimate truth of a sort, it needs an all too human oracle to interpret it. And like the classical example, the exaggeration of its power is a perennial temptation.

In my own work on human ancestry, expectations are often unrealistically high. One woman was genuinely astonished that I was unable to tell her, from a DNA test on her mother, why her second cousin from Yorkshire had freckles. Another complained when I told him he probably had Celtic ancestors. ‘I know that already,’ he said. ‘How come?’ I asked. ‘Because I have dark hair and blue eyes.’

Returning to dogs, it’s my impression that many owners who have their mixed breed dog’s DNA tested expect a similar ultimate truth. Companies are understandably wary of disabusing their customers of the power of their product, and that makes unrealistic claims all the more tempting, especially in an intensely competitive market, and I have found it quite difficult to understand the claims of competing companies. It’s fun to have your dog’s ancestry tested, but bear in mind that there are limits to the accuracy of these tests, for the reasons we have covered. And the more mixed the ancestry, the more inaccurate the assignments. In my very skimpy reading of available tests I can scarcely believe that a proper breed assignment test can be sold for as little as £10.

I thought it was time I tried one of these tests and asked Ulla to look out for a mixed breed dog whose owners would be prepared to volunteer their dog. This was easier said than done. As we will see later in the book, Ulla interviewed several owners who regularly walked their dogs in London’s Hyde Park. Because it’s an affluent area, all the dogs that she met in the park were pedigree specimens and I was beginning to think we would never find a mutt to test. At the very last minute, only three weeks before the book manuscript had to be with the publishers, Ulla came across Archie in a local pub, along with Chris, the landlord, his wife Helen and their 10-year-old son George, who took the saliva sample. Archie was not a complicated mongrel; he was what is called a ‘designer’ dog. He was a Labradoodle, a cross between a Labrador and a Standard Poodle. They specifically wanted a Labradoodle to guard against allergies.

The DNA sample was sent to what looked like a reliable testing lab to see what they were able to deduce about Archie’s genetic composition. The results were turned around with admirable speed, so there was time to ask the family what they made of the genetic results. Their first reac tion was one of surprise. The DNA test revealed that Archie was a mixture of 62.5 per cent Standard Poodle, 12.5 per cent Labrador and 25 per cent Golden Retriever. Archie is jet black, so to find out that he is one quarter Golden Retriever came as a bit of a shock. I am confident that the results were technically accurate insofar as they reflected the best fit of Archie’s DNA to the company’s extensive database on pedigree breeds, and I have no intention of challenging the report’s conclusion. The more interesting question for me is this. Is Archie a mixture of five-eighths Standard Poodle, one quarter Golden Retriever and one eighth Labrador, as the DNA results tell us, or is he a Labradoodle, which is what the breeder told Chris and Helen when they bought him?

‘I suppose I’ll have to go with the science,’ replied Chris, a little reluctantly. Ulla then enquired whether, had he known the results of the genetic test, he would still have bought the dog. ‘Had I known that, I wouldn’t even have driven all that way to see him.’ But once they saw Archie they were hooked.

Labradors and Golden Retrievers are genetically very close. We can see that from the phylogram on page 117 where the two breeds occupy adjacent positions. Also, remembering how the Golden Retriever breed was developed by Lord Tweedmouth, one of the ancestors of Queenie, the original Golden Retriever, was Sambo the Labrador. Even if the DNA conclusions were not quite right, they are very close to what was expected. Assuming that the breeder is being completely honest about Archie’s pedigree, and there is no reason to doubt that, is he a Labradoodle or is he not?

Pedigree breeds are defined by the breed standard and not by genetics, at least for the moment. Although Archie is a crossbred dog, his two parents, one assumes, were a pedigree Labrador and a pedigree Poodle, each with its own breed standard. From a purely genetic point of view, that cannot be true. The simplest genetic explanation for the one-quarter Golden Retriever in Archie’s make-up is that the equivalent of one of his grandparents was a Golden Retriever. My own feeling is that the explanation is neither of these two apparently conflicting options. Instead, our judgement is dulled by the drowsy syrup of Horus, the Egyptian god of numbers. Once complex issues are reduced to numbers, we seem to abandon our sense of judgement. Genetically, there is no such thing as a ‘pure’ dog breed any more than in human genetics there is such a thing as a ‘pure’ race or a ‘pure’ ethnic group. In humans, this fallacy has led to dangerous misunderstandings and harmful discrimination. I doubt the fallacy of ‘purity’ will infect the dog world, but it is best to be on guard against its malign influence. It certainly does not bother Archie’s owners. Once he had recovered from the surprise about Archie’s part Golden Retriever heritage ‘revealed’ by the DNA test, Chris reached down to stroke his neck. He was still the same dog. He was still Archie. Helen and George smiled in agreement.