Sniffing out bee disease

American foulbrood (AFB) is a highly infectious disease that infects bee larvae. It costs the beekeeping industry millions of dollars a year and threatens the role of bees as pollinators. DownUnder Honey, in partnership with dog trainer Pete Gifford and Massey University researchers, is developing a scientific methodology to train dogs to reliably detect the disease.

American foulbrood

American foulbrood (AFB) is caused by spore-producing bacteria Paenibacillus larvae. The spores can survive for decades in soil. The spore has a tough outer layer that makes it difficult to kill – it can survive many disinfectants and boiling water. Beekeeping equipment from infected hives needs to be autoclaved at 160°C for 12 minutes to kill the spores.

Controlling AFB

In New Zealand, control of the disease depends on self-detection and reporting – a legal requirement for all beehive owners. Detection relies primarily on visual assessment. This means early infections and spores can be overlooked. The Management Agency National American Foulbrood Pest Management Plan audits hives where hotspots of disease are reported. An audit inspection is time consuming. It involves taking a hive apart to look at all the frames and components. This means fewer than 0.5% of hives are audited annually. The possibility of a dog trained to detect AFB would make a significant difference to the management of AFB. A reliable detection dog could easily inspect every apiary in an area like Wairarapa within a week.

AFB spores are also in honey boxes that are randomly put out in the field every year and may have come off an infected colony spreading the disease. A dog would also be able to detect spores in these boxes before they were deployed.

Reliability and repeatable results from dog detection are important to bring beekeepers on board and to pay for dog training and implementation.

Dog detection methods aren’t new, but to date, they have been unreliable with a lot of false positives. One of the reasons behind this is that these detector dogs were trained to detect stress pheromones or decomposing larvae. Honeybees emit stress hormones for many different stressors, including varroa, and decaying larvae could be attributed to a number of issues.

Why dogs?

Dogs are frequently used by people to ‘sniff out’ diseases and pests. They’re also employed to detect bombs, help in search and rescue or disaster situations or to detect drugs, plants and foods being brought into a country illegally.

There are many physiological reasons why a dog is such an excellent sniffer. A dog nose has a larger surface area of olfactory epithelium (tissue inside the nose cavity) than a human. This means that dogs have around 300 million smell receptor cells (depending on the dog breed) to only around 6 million in humans.

Lots of animals have a remarkable sense of smell – the elephant is said to have the best – but dogs have the added advantage of being a domesticated animal. Could you imagine training an elephant to detect disease?! As domesticated animals, dogs work well with people, and we can utilise their desire to please us and their natural play behaviour and prey drive.

The team

Highly experienced in breeding and training detection dogs, Pete Gifford has worked with search and rescue organisations, Corrections services and the military. He has trained dogs to detect everything from tobacco and drugs to humans with early-stage cancers and other diseases.

One day when picking up a foster dog, Pete was asked if he could train a dog to detect AFB. From here Pete was introduced to beekeeper Jason Prior at DownUnder Honey.

It all started with a conversation between a man who knew nothing about bees and beekeeping and a man who knew nothing about training detector dogs.

Pete Gifford, K9 Search Medical Detection

Together, the men received funding for training and trialling a new dog detection method. To assure a rigorous approach with reliable results that beekeepers and regulatory bodies could trust, they teamed up with Massey University veterinary science researcher Dr Neroli Thomson, who developed the scientific methodology, ran the trials and is the lead author on the end research report, and Plant and Food Research apicultural scientist Michelle Taylor, who cultured the AFB spores for the ‘target odour’ samples.

Pure odours

The trials used a sterile and pure AFB target odour to train and test the dogs. This pure odour is an important point of difference. It means the dogs were trained to only detect the actual AFB bacteria.

Pete explains that pure odours present interesting challenges. “When we smell a roast cooking, we smell Sunday dinner, whereas a dog breaks everything down and can smell the meat, the fat, the carrots, the potatoes, the roasting pan.” This meant an initial training phase had to involve desensitising the dogs to odours associated with the AFB target odour delivery that wouldn’t be encountered in beehives – for example, the flask and saline solution it came in.

Project phases

Pete started with four dogs he identified as having the drive and disposition for the work at hand. “It’s all about their play and prey drive,” he says. Alongside his standard training room, Pete set up a cleanroom with a multi-bucket carousel for the volatile organic compounds – the odour samples. The cleanroom enables testing without contamination odour issues and was used for the trials in the final phase.

Pete started with training the dogs to systematically check for scents in a series of boxes in his standard training room, and in the second phase, he selected the two best dogs to continue with the AFB samples in the cleanroom.

Final testing trials were conducted so that both Pete and the dogs did not know where the sample was within the carousel. Dog trainers can unconsciously signal to their dogs where a sample is, so blind testing is vital. All trials were recorded with a CCTV set-up.

Both dogs achieved a perfect score in the final trials, which suggests the dogs could indeed recognise an odour from P. larvae spore samples. Dr Neroli Thomson says, “For each dog, the probability of this result occurring by chance alone is very small – less than 0.1%.”

The next step will be trials involving real hives on location in the beekeeping industry.