, , ,

A dog can detect aromas as insignificant as one part per trillion. So why aren’t we using them more to help out in the winery? Camellia Aebischer spoke to Sonja Needs from the University of Melbourne, about her work with dogs on phylloxera and brettanomyces.

Sonja Needs - dogs in wineries

Sonja and her two doogs, Keely and Bertie who were a part of the trial

BAND-AID, SWEAT AND CHEESY are not the descriptors a winemaker wants to use when tasting a barrel sample. These are some of the characteristics of brettanomyces, which can be detrimental to a wine.

Sonja Needs – who is a winemaker, professional dog trainer and an academic at The University of Melbourne – has found a way to combining her oenology and canine skills. Needs has spent time testing the ability of dogs to detect phylloxera in the vineyard, and has now shifted focus to the unfavourable yeast, brettanomyces (also known as brett).

Besides an annual portrait for the national book Winery Dogs, a four-legged canine friend is more useful to winemaking than you’d think.


“Dogs can pick odours up eight or nine meters under the ground, they can pick up a person five or six stories up in a building.”


A dog’s sense of smell is far superior to a human. “Dogs can pick odours up eight or nine meters under the ground, they can pick up a person five or six stories up in a building,” Needs said.

“Cadaver dogs have picked up bodies buried eight meters under landslide.”



For the un-funded trial with phylloxera, Needs took a couple of dogs out to a vineyard in the King Valley, which the winemaker knew had a phylloxera infestation. They spend two hours searching and collected three bugs to use.

A test was set up, then repeated, and it was determined that the dogs could successfully and consistently identify the phylloxera scent. But conducting the trial was difficult and required more support and funding to go further.

“If you’re going to train the dogs properly you need to have bugs. If you want them to find live bugs you need to have the live bug. When the phylloxera are dead they don’t smell the same. So to source live bugs and do research without the help of the Victorian primary industries department was too difficult.”

Because of the hurdles and lack of funding, Needs set the venture aside and changed focus.

Keely on a scent board indicating Brett in Wine

Keely on a scent pad, sniffing out the bandaid aromas



Over the past six months, she has worked with a student on a research project compiling data on dogs and their ability to detect the brett yeast.

Using samples of 4-ethyl phenol, 4-ethyl guaiacol and 4-ethyl catecol, diluted with 10% ethanol and a control sample of just pure ethanol, the dogs were able to detect down to 0.5mg/l of the brett scent. Human threshold for detection is around 300mg/l.

Needs explained that the dogs are able to detect much smaller amounts, and can identify smells up to one part per trillion. However, for the purpose of training dogs to be useful in a winery, it’s important to not let them get too specific.

“If we actually want to train dogs and use them as a brett-detection tool, we don’t want them trained down too low. If you walk in to a winery and there’s a bretty barrel somewhere, there will also be a small concentration of the odour in the air.

“So, if they’re detecting such minute levels all around they’ll misguide us. We want it to be lower than at human level but not too low, and 0.5mg/l will be enough to detect really early on.”

The dogs were trained using food rewards, but one young pup named Bertie got a bit too excited and began giving false readings to earn treats.

“They had a really god hit rate with over 80% accuracy. That would have been better but one of the dogs was mucking around. He’s a young Labrador and because we started off using bits of hotdog or steak, he got too excited and kept giving an alert at any container.”

Needs and her student had to re-adjust the exercise and swap out rewards for slightly less exciting treats like pieces of kibble and chopped up carrot which delivered more consistent results.

“It was a nice little project to start but next semester and next year we’ll be following up with a Masters student,” said Needs.



The focus moving forward will be on detection of the yeast cells, then eventually detecting cells in the winery and picking up infected barrels or pieces of wood.

“The problem is that once we can smell [the brett infection], it might have been there for weeks, and every time you’ve used a wine sniff in the barrel you risk spreading it. So it could have already infected everything.”

Training the dogs on yeast cell detection can create a prevention plan much fluffier than a typical chemical test.

“Early on you’d run dogs through the winery, and have them indicate any barrel that they’re picking up taint in, so that you can pull it out before it spreads.

“The other way would be at the end of the season, after the barrels have been washed, we can run dogs through where the empty barrels are and indicate which barrels are infected.”

That way, winemakers can discard infected wood, or take a more rigorous approach to cleaning.

Needs will be supporting the project and conducting research until the end of 2018 when she hopes to have results published.