WORCESTER—A new UMass Chan Medical School study on canine genetics — and what DNA can actually say about a dog’s personality — is gaining national attention after being featured this week in The New York Times.
The research, published Nov. 24 in PNAS, examined 151 genetic variants long promoted as predictors of canine behavior. According to UMass Chan, the team found that none of those variants reliably forecasted how an individual dog actually behaves.
The study was led by Elinor Karlsson, PhD, the Dr. Eileen L. Berman and Stanley I. Berman Foundation chair in biomedical research and a professor of genomics and computational biology at UMass Chan, along with Kathryn Lord, PhD, a research specialist at the medical school. Lord discussed the findings with the New York Times.
Using data from Darwin’s Ark, a community-science project with genetic and behavioral information from more than 3,000 dogs, the researchers tested whether widely marketed behavior-related genetic markers held up when compared to real-world data.
They didn’t.
“We looked at 151 genetic variants that have been linked to canine behavior and found not a single one predicted the behavior of individual dogs,” UMass Chan said in its announcement.
While none of the supposed behavior-linked variants showed predictive power, many of them did line up with aesthetic traits that define dog breeds, such as size, leg length, or ear shape. The study suggests earlier research may have been skewed by the genetic structure of modern breeds — a product of the last two centuries of selective breeding.
According to the PNAS paper, behavioral traits in dogs are “polygenic and complex” and are also shaped heavily by environment, training, and early-life factors. Because of that, the authors conclude, tests that focus on a small number of genetic variants “are unlikely to provide accurate predictions” about a dog’s personality or future behavior.
The researchers note that dog behavior shows only moderate heritability and that meaningful genetic prediction would require much larger datasets — potentially tens or hundreds of thousands of dogs — paired with individual behavioral assessments, not breed-averaged scores.
The work has implications for the booming market of consumer dog-DNA products, many of which claim to predict traits like sociability, fearfulness, or aggression.
The UMass Chan team’s findings indicate those predictions are not supported by evidence.
Despite the shortcomings of current tests, the researchers did replicate numerous strong associations for physical traits, demonstrating that canine genetics can reliably predict appearance — just not personality.
The full paper, “Genetic testing predicts appearance but not behavior in dogs,” is available through PNAS.
Have news, tips, or a story worth telling? Reach Editor Charlene Arsenault at carsenault@theworcesterguardian.org—because good stories (and great scoops) deserve to be shared.
The Worcester Guardian is an independent nonprofit news organization. Support local journalism by making a DONATION today.
