One company is working with Purdue University to train canines to detect diseases in hemp and cannabis plants.
Columbus, Ohio-based Bio Detection K9 (BDK9) offers canine and technology-based services to detect viruses and pathogens in human hosts and agricultural settings. The company recently partnered with Purdue University to develop processes for training canines to detect diseases in hemp and cannabis crops.
William Schneider, Ph.D., chief scientific officer at BDK9, says the company’s initial goal was to train canines to detect tree diseases.
“Those projects went particularly well, and then in the process of figuring out how that worked, we came across better methods that allowed us to train dogs to detect pathogens somewhat directly,” he says.
According to its website, since 2011, the company has trained canines to detect the following diseases in agricultural settings: citrus canker, huanglongbing (a bacterial disease of citrus), plum pox virus in peaches and plums, and four separate viral diseases in tomatoes. The company has also trained canines to screen humans for the COVID-19 virus.
Schneider says as BDK9 began to develop these processes, it started to think about what commodities could benefit from these services.
“Typically, because canines as a detection tool are somewhat pricey in terms of per plant diagnostics, we tended to focus on plant systems and cropping systems where individual plant value was quite high, [and] we knew that cannabis was a growing field,” he says.
Schneider says that he knew Janna Beckerman, Ph.D., professor of plant pathology in the Department of Botany at Purdue University, from her previous work studying diseases in cannabis plants, adding that the company then decided to partner with the university to advance its processes for training canines to detect diseases in cannabis crops.
“We wanted to expand into doing commodity, and we also wanted to expand into a new direction that would sort of overall decrease what the cost of doing canine-based diagnostics was and add distance-based imaging as a way to triage a plant and focus the efforts of the dogs exactly where we needed them,” Schneider says. “That’s where John [Couture, Ph.D., agricultural researcher at Purdue] comes in because he already has experience doing imaging on insect pests in plants, and so, this partnership was sort of a nice trifecta where all the skills came in, and everybody brought something to the table.”
“One of the biggest problems with managing plant diseases is by the time you diagnose the problem, that is, you see the symptoms and the signs of the pathogen, it’s too late and the problems there,” Beckerman says, adding that for many diseases, especially viral diseases, it’s important to detect them early on because they could spread to other crops.
“So, any process that would allow us to identify early stages of infection and invasion allows growers more options as far as how they are able to manage the problem,” she adds.
BDK9 and the university are specifically training the canines to detect botrytis, “a systemic mold infection in Cannabis,” according to Medicinal Genomics.
Beckerman says they have worked with the plants at all growth cycle stages. “We have done plants that are 6 to 8 inches tall, and we’ve done plants that are taller than that. And next month, we’ll be doing plants that are even larger and later in production,” she says.
The process includes flying programmed drones and satellite imagery over a field of crops to pinpoint if and where diseases are in the field. The canines are then sent to the specific areas that are picked up by the drone or satellite to identify the diseased plants.
“The [canine] runs along a row of plants at about one second per plant until it finds a positive, at which point it is trained to sit there, and the handler will usually check to make sure it’s not trying to mess around with it,” Schneider says. “And when we’re all convinced that the dog is not lying to us, the dog’s rewarded with a toy and a little bit of playtime. And then the dog gives the toy back … and eventually, we start up again, [and] we head down the row.”
Couture expresses that the canines get tired, especially for large-acreage grows. So, flying drones or satellite imagery over the field beforehand to highlight infected areas helps “increase the efficacy of the dog’s ability to hone in on spots and then identify those [specific infected plants].”
“So, what we did, and this was prior to anything flying in the air, is we had plants that were inoculated with botrytis, and then we collected their optical signatures, or the way that light reflects off of them,” Couture says. “And using those optical signatures in very early stages of the inoculation or infestation process, we could use the way that they look based on the way light reflects off of them to discriminate between inoculated and non-inoculated plants.
He adds, “One of the drawbacks [we found] is that … it’s not a very specific signature. So the same signature that we’ve seen comes up with many plant diseases, a lot of plant-insect infestations, or a lot of plant stress responses, but the benefit is that you have the dogs, which are very precise and trained to dial in on one specific stress. So, I think long term, if you can fly [over] a field and say, ‘Here are five or 10, or however many areas that don’t read like normal vegetation should read’, then you can send in the dogs [to] very specific areas to validate that ‘Yes, it is, or no, it is not; it’s something else.’”
Schneider adds that the canines can also sample an entire plant without destroying it and provide the growers with results in real time.
“Generally speaking, for every disease we’ve worked with, it is more accurate than any molecular test because there are no sampling issues,” he says. “All the typical tests that are used to detect diseases involve some level of sampling. … You can’t do an ELISA [enzyme-linked immunosorbent assay] test on an entire plant, but a dog can do it.”
Beckerman says that when using molecular tests to detect diseases in plants, you risk getting false negatives.
“When you try to do any sort of molecular test, you’re subsampling the plant, you’re not usually doing the entire plant. … So, you could have a diseased plant, but if you take a leaf that isn’t actually infected, you could end up with a false negative,” she says. “Conversely, if you end up doing many of these molecular tests, they are so powerful [that] you can get a lot of false positives in that you might detect the pathogen, but you might not actually have a disease problem.”
Schneider says the first set of dogs are almost ready to go, and they will be deployed to a company in Jamaica—which has problems with botrytis specifically—to work their first job. He says most projects typically require two dogs and one handler, but it depends on the density of the plants.
“So, [the dogs] are validated. … It’s just now a question of getting them used to being operationalized, and [that] usually takes a little bit of practice, but not too much,” Schneider says. “We feel we are at a point where we not only have deployable botrytis dogs, but we feel like it wouldn’t take long to make more if the need was there. I think the bigger picture, though, is we work with Purdue because of their expertise to guide us, to do different diseases than botrytis because …. different geographical growing regions for cannabis experience very different problems. … So, it’s specialized based on location.”
Beckerman says they are slowly expanding to look at pythium, a fungus that typically causes root rot.
“Pythium [is a] huge problem for field-grown cannabis here, but also a fairly common problem worldwide for cannabis in other field operations and greenhouse operations as well,” Beckerman says.
Schneider says they also envision eventually having a dog core that could be deployed to several locations to detect a wide variety of diseases. “For example, the clients we are looking at out in the west coast; they’re very interested in keeping mother plants healthy. In that particular case, we’d be screening for things like [viral and viroid diseases], and then things that will affect the propagation scale,” he says. “Once we get over this initial hurdle of how to teach a dog to look at a cannabis plant as a target, it becomes quicker [to train them] for each of the [other] diseases.”
He adds that the overall goal in working with Purdue is to eventually get to a place where they can remotely screen a field of hemp or cannabis to identify trouble spots while they are training the dogs. Then, they can send the dogs to the location to precisely identify symptomatic plants and initializing infections.
“We fully anticipate that this will be a commercial product that’s available by the end of the year,” he says. “I’m hoping that it won’t just be the botrytis dogs. I’m thinking that the pythium dogs are actually not that far behind. Then, we would probably move to some of the [diseases] that affect mother plants not too long after.”
Read the full article here