NEW YORK — A dog's breed may not be a good predictor of how it may behave, a new genetic analysis of thousands of dogs has found.
Certain dog breeds are often associated with particular personalities and behaviors. Labrador retrievers, for instance, are expected to be friendly and energetic, while pit bulls are expected to be aggressive.
But a new research study drawing on survey responses from the owners of more than 18,000 purebred and mixed breed dogs, as well as genotyping analyses of a subset of those dogs, found that many breed stereotypes do not hold up. As researchers led by the University of Massachusetts Chan Medical School's Elinor Karlsson reported in Science on Thursday, although many behavioral traits are heritable, breed explains only a small portion of variation in behavior.
"Dogs are individuals," co-author Marjie Alonso from the International Association of Animal Behavior Consultants said during a press briefing. "Certainly, I think that the paper shows that there are some general differences between breeds. … But I don't think that we should really be deciding that breeds are the things that will tell us whether or not we're going to be happy with a dog or a dog is going to be happy with us."
The researchers' study drew on data collected through the Darwin's Ark project, an open-data resource they developed for collecting owner-reported dog phenotypes and genetic data. Owners completed a dozen short surveys — a total of 117 questions — asking about their dogs' behavioral and physical traits. They combined the answers to the behavioral questions into eight different factors, such as their human sociability or comfort with unfamiliar people, how well biddable they are or how well they respond to human direction, and their agonistic threshold or how easily they are provoked.
By comparing dogs from a one breed to a sampling of random dogs from the cohort, the researchers developed a population peculiarity score (PPS) to gauge whether dogs from one breed had behavioral traits that differed from other dogs. Overall, they found that there were only subtle differences between dog breeds and that no behavioral trait was unique to a certain breed.
"We found things like German shorthaired pointers were slightly more likely to point, or golden retrievers were slightly more likely to retrieve or Siberian huskies to howl than the general dog population," co-author Kathryn Lord from UMass said. "But because these behaviors pre-date breeds, we also see them in other breeds and other dogs."
Modern dog breeds, the researchers noted, were established during the Victorian Age about 120 years ago, while dogs had been under selection for thousands of years before then.
About half the dogs in the Darwin's Ark cohort were mixed breeds, enabling the researchers to explore whether breed stereotypes have a genetic component. To do so, they developed a linear mixed-effects regression (LMER) model for the different factors and questions to find that genetic breed ancestry accounts for about 9 percent of the variance in individual behavior factor scores.
By further comparing PPS and the LMER results, the researchers noted that some behavior differences in mixed-breed dogs did stem from differences in breed ancestry — mixed-breed dogs with more border collie ancestry, for instance, were more biddable — but others contradicted the accepted wisdom. Owners of confirmed golden retrievers tended to disagree that their dog was fearful of unfamiliar people, which fits the breed stereotype. But among mixed-breed dogs, the amount of golden retriever breed ancestry had no effect on that question, suggesting the factor might actually not be genetically driven.
Meanwhile, through a genome-wide association study of the behavioral phenotypes, the researchers homed in on 11 genomic regions that are associated with various dog behaviors. Howling among dogs was linked to a region between SLC38A11 and SCN3A, a voltage-gated sodium channel involved in speech and language development, while human sociability was linked to a region downstream of HACD1, which is involved in long-term memory.
"It could be that understanding human sociability in dogs could help us understand how brains develop and learn," first author Kathleen Morrill from UMass added. "So we're just scratching the surface there, but that's one [trait] that seems incredibly heritable."