The family tree of dogs

We've had both our dogs DNA tested -- purely for our own entertainment, not because we have any concern about "pure breeding" -- and both of them gave us results that were quite a shock.

First, there's Guinness, whom the rescue agency told us was a black lab/akita mix.  You can see why:

Turns out he is neither -- he's American Staffordshire terrier, husky, chow, and Dalmatian.

Then there's Rosie, who we thought sure would turn out to be fox terrier/beagle:

Once again, not even close.  She came out to be a mix of about ten different breeds in which Australian cattle dog predominates.  Not a trace of hound, which is surprising not only because of her facial features, but her temperament.  We've had hounds several times before, and they are sweet and loving... and stubborn, headstrong, and selectively deaf, all of which describe Rosie perfectly.

I'm not sure that it's reasonable to expect a fifty-dollar mail-order dog DNA test to be all that reliable, mind you.  In Guinness's case, though, there are features that do make sense -- the ebullient disposition and square face of the AmStaff, and the curly tail and thick, silky undercoat of his husky/chow ancestry.  Whatever its accuracy, though, it's fascinating that any signal of ancestry at all shows up in a simple saliva test.

Especially given that just about every dog breed in existence traces back to wild dog populations in only a few thousand years.  That's an extremely short time to have any evolutionary divergence take place.  But genetic testing has become sophisticated enough that we can now retrace the steps in dog evolution -- creating a family tree of dog relationships encompassing 321 different dog breeds (including several sorts of wild dogs).

A team of geneticists led by Jeff Kidd of the University of Michigan, Jennifer R. S. Meadows of Uppsala University, and Elaine A. Ostrander of the NIH National Human Genome Research Institute did a detailed study of two thousand different DNA samples containing over forty-eight million analyzable sequences.  They identified three million SNPs -- single nucleotide polymorphisms, or "snips" -- that were characteristic of certain breeds. 

"We did an analysis to see how similar the dogs were to each other," Kidd said.  "It ended up that we could divide them into around twenty-five major groups that pretty much match up with what people would have expected based on breed origin, the dogs' type, size and coloration."

Interestingly, wild dogs and "village dogs" -- dogs that are somewhere between domesticated and feral, something you find in a lot of towns in developing countries -- have significantly more genetic diversity than domestic breeds do.  This, of course, contributes to their vigor (and, conversely, is why many "pure" dog breeds are susceptible to particular health problems).  It's also why it's so easy to identify behavioral characteristics of particular breeds, like the cheerfulness of golden retrievers, the intelligence and independent nature of huskies, and the nervousness of chihuahuas.

And the fact that if you want to partake in an exercise in frustration, try to housebreak a cocker spaniel.

If you take the time to read the original paper -- highly recommended, because it's amazingly cool -- you'll get to see the final "family tree" of dog breeds and see who's related to whom.

Now y'all'll have to excuse me, because Guinness wants to go outside and play.  I wonder what gene controls the trait of Wanting To Retrieve Tennis Balls For Hours.  Because whatever it is, I think Guinness has like fifty copies of it.

****************************************