How well can we really identify breeds of dogs?

Our friends over at the National Canine Research Council have been having some fun with DNA testing of late.

Last week I noted some new research from the AVMA is indicating that DNA testing may just be debunking any studies ever done on dog bite studies by breed. Basically, every dog bite by breed study out there is based on people's visual identification of the dog breed -- and we're finding out that people really aren't that good at identifying breeds of dogs by looks -- and in fact, it is more or less impossible.

But when you put images with it, it is really telling.

So last week, the NCRC published a "Find the Labrador Retriever mix" visual ID game. Similar to other version of the Find a "Pit bull" game, it turns out that a lot of mixed breed dogs that we assume to be mixed breeds that are predominently common breeds don't often contain any of those breeds at all.

Most people, including many who work in animal rescue, are not all that great at knowing the looks of all the different dog breeds. The AKC/UKC combined recognize over 150 different breeds of dogs....but at this point, there are over 400 recognized dog breeds. And often, when these breed intermingle, they don't carry with them looks of any of the breeds they represent.

However, because we are familiar with a few dog breeds, we have become pretty good at grouping dogs together in groups with other similar looking dogs.  Is the dog about 80 lbs and black?  "Lab Mix." Boxy head and muscly? "Pit bull".  Long hair and snout?  "Shepherd mix".  Never mind that there are several breeds of "Shepherd" that don't look all that much alike -- and many of the "shepherd mixes" don't really look like any of those breeds.

We group them, because in our mind, it's easier to classify them. Besides,  what does a Chesapeake Bay Retriever/Chihuahua mix look like?  And what does that mean to a potential adopter?  Mostly nothing. But it's a cute black dog that kind of looks like a Lab, so let's call it that.

So what do our classifications mean?  It means we group a lot of mixed breed dogs of uncommon breeds into a grouping of more popular breeds. So go to PetFinder, your local shelter, or look at any bite study. Almost anywhere you look for classifications by breed, you see "labs", "shepherds" and "pit bulls" as the most common breeds.  Most often it's not because these dogs share the same DNA -- but because the LOOK similar to each other and it they are more easily categorized that way.

I think this is a really important distinction. When communities say they are putting restrictions based on "breed" because they say they are "inherantly dangerous based on their DNA" -- we need to be honest about what that means. These communities are not making decisions based on DNA, or breed at all -- but on LOOKS. They are determining that a LOOK is aggressive - not genetics. Which is an even crazier notion than the genetics argument which doesn't hold up to science either.

It's turning out that the more we learn about dogs and breeds, we find out the less we ever knew in the first place.  And that pretty much every study we have ever done about dogs and breeds is completely irrelevant now because we didn't identify the breeds right in the first place.

Comments

[Lisa]

I'm all for visual breed IDs when they're being used as shorthand to convey the general appearance and size of a dog. But I'm careful about making visual breed IDs without lots of disclaimers, because I know that people misconstrue and misuse sloppy breed IDs to support stupid ideas.

So, while I'd still probably call most of those dogs 'some kind of Lab mix,' I love to see things like this that prove that I'm wrong a lot of the time.

[Rinalia]

I have mixed feelings on this issue.

While I'm curious about the genetic markers that lurk amidst my dog's DNA, I would probably fail to see how knowing my dog has an allele from a certain breed would garner me any sort of awesome knowledge.

Even stranger, I've read about people who want to get testing done to "see if their dog would fit the apartment lifestyle" even though the dog already fits the apartment lifestyle due to his individual personality. If he had a random allele belonging to a Border Collie, how is that useful if that allele codes for, say, eye color and has little to do with the mystical "does well in apartments" gene?

Yep, I'm curious. It's fun to do this sort of thing, because that is how we humans work - we want to quantify and define and know very specific things even if they have nothing to do with the bigger picture (or the individual dog's actual personality).

And the ultimate in silliness is this idea that if a dog has an allele coding for an effing phenotype that that, BY NECESSITY, has anything to do with behavior. While physical traits may certainly be linked to behavioral ones, it would be insane to argue that b/c my dog has the allele coding for Lab coat color that she will be the best duck-retrieving dog on earth. Or that because my older dog has an allele coding for, I dunno, rose-pricked ears in pit bulls that she is going to be hanging off the noses' of bulls anytime soon.

It's just frustrating that we cling desperately to this notion that how a dog looks determines how that dog will act. And now I have to be frustrated that we might now define a dog by a few alleles floating around, alleles that may have nothing to do with behavior.

[Sue Cosby]

Having raised bunches of litters with moms (foster dogs with fathers of unknown parentage) I can say that the overwhelming majority of puppies looked NOTHING like their mothers. Visual identification is at best unreliable guesswork and at worst - a joke. Behavior attributes can be a better indicator for some but for most, I prefer the label MUTT.

[JAL]

Breed, IF the breeding has been done WITH breed temperament in mind, can predict behavior.

Working, sporting and even just companion dog breeders who are serious about preserving a breed's intrinsic traits have to wash dogs out of breeding for not having those traits of personality and behavior to the extent called for, even if the parents were exemplary. It's work to preserve those things, not automatic based on a piece of paper.

Cosmetic exterior appearance is useless.

And in the case of mass produced/commercially bred dogs, especially after a number of generations, BREED becomes pretty useless for predicting behavior.

Dogs need to be judged as individuals.

[KC KS Kills Dogs]

Brent thanks for this most informative post. This is one to share with those silly city council know it alls.

[Sarah]

Fun! I was right on with the "Significant" one - not so much on the "Some" dog.

[MichelleD]

If pit bulls are dangerous based on DNA then that DNA should be easily seqenced and easily related to dangerous behavior traits.

This hasn't been done for humans despite actual scientific study and efforts to do so.

[Karen]

I find this subject fascinating. Thanks for posting this, Brent. I volunteer for a rescue group and I always tell potential adopters that I have no idea what breed the foster dog is and then I explain why I will not put a breed label on the dog. I adopted 2 of my foster dogs and would love to know what they are. They are sisters and look almost exactly the same. I can tell them apart but most people can not. They are beautiful dogs and I sure would like to know what they have in them but I can not afford the DNA test.

[Julie]

I must mention animal behaviorist Robert Sapolsky's book Monkeyluv again... seriously, the man is brilliant and his explanations of the ridiculousness of saying genes cause behavior are not to be missed.

[Charlotte]

" Most people, including many who work in animal rescue, are not all that great at knowing the looks of all the different dog breeds."

Yes, most people are not good at identifying dog breeds (purebred), but why do you mention those in "animal rescue" as people that WOULD know their breeds if ANYONE did?? "Rescues" have become the go-to people for anything to do with dogs anymore, but they are some of the most unknowledgeable people around when it comes to identifying breeds!

There ARE many people who know their dog breeds in and out -- and it's NOT "rescue" and it's not veterinarians and it's not animal control. Those that know their breeds are people like AKC dog show judges and people that train dogs for a living and many evil BREEDERS! But you never hear of THEM being consulted to determine breed. Determining breed by LOOKING AT THEM is not difficult when you know your stuff! AND they will give you a better idea of what's in a MIXED breed than any DNA test!

[Brent Toellner]

Charlotte,

I don't necessarily disagree with you that breeders tend to be more knowledgable about dog breeds than most folks in rescue or shelters. They at least know what dogs are SUPPOSED to look like. However, most spend little to no time around mixed breed dogs, and generally consider anything that isn't on the breed standard a mixed breed (whether it is or not) and because they deal almost entirely in purebreds, don't really know what mixes are in a lot of cases unless they visually look like the purebreds (which they often do not). But true, if they look like their purebred counsins, they will likely be better at this, because are generally much better versed in less common dog breeds and can recognize those similarities.

But the idea that someone can be better at breed ID of mixed breed dogs by eyesight as being more accurate than a DNA test (as long as the breed trying to be determined is actually in the database) seems like a bit of a stretch.

[small dog clothes]

Great information to share at my blog. Determining the breed of certain mix breed dogs can be very difficult unless there is close similarity in visual appearance against the pure breeds.

[Hannah]

Yeah knowing the right breed of dogs really help in many manners like predicting the possible proneness of dogs to such disease by breed, knowing their temperament and behavior as they differ in every breed..