A study of 18,385 dogs finds that breed explains just 9% of behavior

You might want to sit down for this one.

A new study is putting serious pressure on conventional wisdom about dog breeds. According to an analysis that included 18,385 dogs, very few behaviors are actually associated with particular breeds.

"For some of the behaviors, we just couldn't find much of a difference between the breeds," genomics researcher Elinor Karlsson, a co-author of the paper, told reporters at a press briefing." Some behaviors, such as howling or certain types of play, did seem to differ from one breed to another. But those population-level averages don't always translate to individual dogs.  "Even if the average is different, you've still got a really good chance of getting a dog that doesn't match what people say that breed is supposed to be," she says.

"You can get this if you talk to anybody who's owned like eight dogs of the same breed. They'll tell you all the reasons why all of those dogs are different from one another," Karlsson says.

The study was published Thursday in the peer-reviewed journal Science.

A large study produced surprising results

Karlsson began studying the connections between dog genes and behavior to better understand the genetic factors of obsessive-compulsive disorder in humans. "It [is] a complex disease that involve[s] many different genes. We had maybe 200 dogs and we just couldn't find any more," she says. Then she had a realization. Every time she told someone about her research, "they immediately pulled out their cell phone showed me a picture of their dog, and started to tell me everything about their dog's behavior."

"It suddenly occurred to me that this was in fact the data that I was looking for," she says. Karlsson and several colleagues developed a citizen science project where members of the public are invited to contribute information about their pet dogs to help researchers answer questions about dogs, cats, and ticks. 

For the study on dog breed behavior, the researchers used data about the behavior and physical appearance of 18,385 dogs, 49 percent of which were purebred. The researchers sequenced the DNA of 2,155 of the dogs included in the study. 

That data contained fewer patterns than most people would expect. "Breed offers little predictive value for individuals, explaining just 9% of the variation in behavior," the authors write. Some traits, like responsiveness to direction and commands, appeared to be somewhat inheritable and are more common to members of certain breeds (e.g. border collies and Australian cattle dogs) and less common in other breeds (e.g. beagles and chihuahuas). For many other traits, however, "breed is almost uninformative" in guessing how an individual dog will behave, the authors say.

Breeds are a relatively new concept in the history of dogs

The researchers say this surprising lack of consistent behavior from one breed to the next is a product of humans' and dogs' shared history. Before some wolves became dogs, they probably survived by scavenging garbage from human settlements, according to biologist Kathryn Lord, another co-author on the paper. (That's true for the majority of the roughly one billion dogs on Earth today, with pet dogs accounting for a fraction of the total population.) The ancestral dogs that were less fearful and better at getting along with humans probably received preferential treatment from humans, making them more likely to reproduce and pass their genes to future generations.

Humans have used dogs to do certain jobs, such as herding, for much of our shared history, but dogs as we know them are relatively new. It's only "in the last 120 years or so, starting in the Victorian era, that we've had what you call modern breeds," Lord says. Those dogs are based on a "standard... where somebody has taken a bunch of dogs, bred them together, and at some point [said] 'any dog that we're going to call a golden retriever has to [descend] from this set number of dogs... Those will be golden retrievers, and they will look a certain way,'" she explains.

It turns out that physical traits are easier to breed for than most behaviors.

"People are very good at finding patterns, and I think that they find patterns even when there aren't [any. That's] a lot of what people are seeing," Karlsson says.