Detecting Fecal Contaminants on Produce, a New Occupation for Sniffer Dogs

My wife and I recently flew from Paris to Atlanta.   While we were waiting to pick up our luggage to go through customs, a handler with a beagle began to walk beside the carousel, sniffing bags that had been removed from it by passengers.  I assumed at first the beagle was a drug dog, then noticed the screen suspended above the carousel, which actually showed a short film about the beagle. There was no sound, but the film was subtitled, like a silent movie.  “What is the beagle doing?” the screen asked.  Then there were pictures of various food items—a head of lettuce, a bunch of radishes, a package of meat.  Then X’s crossed out each item to let anyone watching know that these items were forbidden.  The dog was shown sniffing a suitcase and sitting down in front of it and looking at the handler.  The suitcase was then opened by the overly willing passenger (probably an ICE officer in civilian clothes), revealing a bunch of radishes. (Think twice before you try to smuggle radishes from France.) 

Another agricultural use of dogs was recently described in the Journal of Food Protection, and quite likely foreshadows a new industry for dog trainers and handlers. 

Fecal Contamination of Produce

Microbes, such as Escherichia coli, Salmonella, and Listeria, can get onto fresh produce through feces of rodents, birds, and other animals.  The feces can be put directly on the produce by the animals or can be brought into a field or factory through irrigation or processing water.  A group of researchers from the School of Veterinary Medicine at UC Davis and from various facilities of the U.S. Food and Drug Administration have published a paper analyzing how effective dogs might be in detecting fecal contamination of produce.  

The team had three dogs trained for the experiments, all mixed-breed females.  Dog 1 and Dog 2 were given what was called “indirect detection training,” under which the target scent was a sterile gauze pad saturated with a mixture of feces and water.  The feces were collected from vertebrate species commonly found in or near agricultural fields, specifically dog, cow, horse, black-tailed deer, feral pig, coyote, Canada goose, sheep, and human. Dog 1 and Dog 3 were given “direct detection training,” under which dogs were trained to recognize fecal contamination of romaine lettuce, spinach, cilantro, and whole Roma tomatoes placed in specimen storage containers.  Thus, Dog 1 received both types of training, but Dogs 2 and 3 only received one or the other. 

Indirect Detection Trial Procedures

In the indirect detection trials, lettuce, spinach, cilantro, or tomatoes were put into 30 bags, and 8 of those bags were filed with either 0.025, 0.25, 2.5, or 25 grams of feces (two bags at each amount).  Gauze pads were suspended were suspended by strings were suspended inside the bags but not in contact with the produce for 24 hours.  The gauze pads, which were 4-ply cotton pads, were removed after a day and separated to create separate samples. The samples included gauze pads from bags in which there was no fecal contamination.  Samples were placed in special holders in three rows of 10, with contaminated samples distributed randomly. Trials were double blind as the dogs, the handler, and the data recorder did not know which samples had pads with fecal contamination.  The handler made sure the dog examined each holder.  For each alert, the dog was rewarded with some time with a chew toy.  Care was taken to remove secretions—at least obvious secretions—by dogs sniffing close to holders. 

Direct Detection Trial Procedures

For direct detection trials, 8 of 30 bags of produce had feces added, but in amounts of 0.0025, 0.025, 0.25, or 2.5 grams.  After the first trial, the largest amount was dropped and an even smaller amount, 0.00025 grams was added.  Instead of three rows of ten, as in the indirect detection trials, there was one row of 12 containers, which were different than those used in the indirect detection trials and had holes drilled in the top shortly before a trial began.  To avoid rewarding the dogs for an incorrect response (which had been possible in the indirect detection trials), the dog was praised verbally for an alert during the direct detection trials.  

Indirect Detection Results

In the indirect detection trials, Dogs 1 and 2, the ones used in these trials, missed detecting fecal contamination in most samples where the gauze had been exposed to fecal contamination.  Dog 1 was nearly twice as effective as Dog 2 in alerting to pads from samples that had fecal contamination, but also alerted more often to pads that had not been in bags with fecal contamination.  Both dogs were more likely to detect fecal contamination on Roma tomatoes than on cilantro or spinach.  When the produce in a bag was contaminated with 2.5 grams of feces, the highest possible amount, both were considerably more accurate.  However, the results indicated that this procedure, as conducted in the experiment, was not effective.  

Direct Detection Results

When dogs were able to sniff the produce itself, as opposed to a gauze pad that had been kept near the produce for 24 hours, they were much more accurate.  This is not surprising as there was presumably a higher odor concentration using this approach.  Here, each dog sniffed 720 containers, 156 of which contained some amount of feces.  The researchers found that “Dogs 1 and 3 had 11.1 and 23.6 higher odds of alerting, respectively, when encountering treatment samples compared with control samples.”  Dog 1 was significantly more likely to incorrectly alert in the presence of a control sample than was Dog 3. 

The amount of fecal contamination proved to be crucial.  When the amount of contamination was greater than 0.025 grams, the probability of detection achieved 75%, and reached almost 100% at 2.5 grams: 

“In other words, for samples with ≥0.025 g fecal contamination, the probability of collecting samples of produce with fecal contamination is 5- to 30-fold higher (500 to 3,000%) when using a dog than when randomly selecting produce samples across a field, as is sometimes done during investigations.”

Implications for Agricultural Inspections

The advantage of the indirect detection approach was that vegetables were not exposed to the dog, which prevented cross-contamination between the dog and the sample.  The researchers note that unfortunately this approach “did not result in acceptable levels of sensitivity for any but the highest levels of fecal contamination.”  The direct approach was more successful, in that the dogs exhibited 76% and 86% sensitivity, respectively, in detecting more than 0.25 grams of fecal contamination.  (For an explanation of the terms “sensitivity” and “specificity,” see a prior blog on the use of dogs to detect lung cancer.)

California Ground Squirrel

Ground squirrels, common in produce fields, defecate between 2% and 8% of their body weight per day.  The researchers state that, based on their results, “scent detection dogs might be able to detect as little as 1% of a ground squirrel’s daily fecal load if deposited on, for example, spinach or cilantro or after foliar irrigation when feces or scat are deposited in furrows and along beds of leafy greens.” 

The broader implication of this is stated as follows:   “A protocol that uses a fecal scent detection dog to first screen all produce samples and then test only those to which the dog alerted can increase the probability of detecting contaminated produce by up to 3,000%, depending on the background prevalence of fecal contamination in the field.”

The researchers conclude that “the use of scent detection dogs will allow us to prioritize produce samples for analytical testing and thereby optimize the detection of both feces and the associated microbial pathogens that so often accompany fecal contamination.” 

Conclusion

The results achieved with the dogs might be improved by different training regimens, or the use of different breeds, possibilities that the researchers concede may be true. Using dogs over longer periods than was the case with these experiments might also improve accuracy. 

The dogs were not as accurate as some prior research found in other contexts, but context is important.  Researchers looking to use dogs in cancer detection have insisted on consistently high success rates before this application can move into clinical environments, but the same levels of success should not be required for using dogs in detecting fecal contamination of produce.  If it can be verified that a protocol involving canine screening of samples, in order to prioritize which truckloads or other units of vegetables should receive further testing, would increase the probability of detecting contaminated produce by up to 30 times over the present approach, this makes a strong case for implementation despite the fact that some contaminated lots would still be missed.   

It is to be hoped, therefore, that if these results are verified, detection dogs may soon be put into service in real-world commercial agricultural operations.  Another example of such a use of dogs might be in detecting oil in fish hauls after a spill.  A new canine industry may be in the offing.

Partyka, Melissa L., Bond, Ronald F., Farrar, Jeff, Falco, Andy, Cassens, Barbara, Cruse, Alonza, and Atwill, Edward R.  (2014).  Quantifying the Sensitivity of Scent Detection Dogs to Identify Fecal Contamination on Raw Produce.  Journal of Food Protection, 77(1), 6014.  

This blog was written by John Ensminger and L.E. Papet.