Dogs Flagged 8 of 11 Buried Jars of CWD Positive Whitetail Feces

ICanine Results

Three Dogs Flagged 8 of 11 Positive Samples in Field Tests

At the Penn Vet Working Dog Center Mallikarjun and colleagues trained two Labrador retrievers and a Finnish Spitz that were all 8 years old, then published the work in Prion in 2023. The USDA provided feces gathered in four US regions. The National Veterinary Services Laboratory used obex or retropharyngeal lymph node tissue to confirm each deer's status.

IILab Wheel

Each Dog Flagged 2 of 5 Positive Samples in Blind Tests

Lab testing used an 8-port steel scent wheel. Training included 79 negative and 54 positive samples that were all in jars under lids, so the dogs never touched feces. Positive samples came from deer with earlier-stage lymph-node-only infection and from deer with later-stage involvement in both brain and lymph nodes. Each dog next completed three double-blind sessions with three trials apiece. Each trial included three novel negatives and either one positive or none, drawn from 5 novel positives and 24 novel negatives.

Every dog flagged 2 of the 5 positives for a sensitivity of 40%, while specificity ranged from 90.4% to 95.2%. The dogs spent significantly longer at positive samples than at negatives, which the authors took as discrimination even when a dog did not give its trained alert. The authors counted full alerts on only 3 of 9 encounters with lymph-node-only samples, which may have been tougher to learn.

The authors also note that the blind test rewarded correct alerts only some of the time and the dogs had never practiced that in training, so the change may have cost them hits.

IIIOutdoor Tests

False Alerts More Than Doubled Between Lab and Field Tests

Field training started with jars lined up on open grass and progressed to offset lines. Training then moved to jars in tall grass and weeds before progressing to buried jars with only the lid visible. Each dog searched six areas in the final test. Three contained a positive sample, two contained only negatives and one contained empty lids. Jari and Charlie searched an unfamiliar site, while Kiwi searched familiar fields near the center because of COVID-related trouble.

Lymph-node-only samples did better in the field test at 5 of 6 compared with 3 of 5 for later-stage samples, which the authors put down to more experience with the lymph-node-only odor. False alerts climbed from 5 of 65 negatives in the lab test to 14 of 78 in the field test, a rise the authors link to the unfamiliar search area. Our read is that Kiwi was the one dog on familiar ground and barely rose from 2 of 21 false alerts in the lab test to 3 of 26 in the field test, while Jari and Charlie more than doubled their false-alert rates. Kiwi also detected the fewest field positives, so one dog supports the authors' explanation without proving it.

Handlers correctly called 5 of 6 blank areas and backed every true alert, but they also judged 8 of the 14 false alerts to be correct.

IVScreening Role

Dogs Could Check Leading Edges Without Anyone Handling a Deer

The authors write that dogs can survey large areas without anyone finding or handling a live or dead deer. The authors say dogs could help most at the leading edges of managed deer areas, around known positive captive facilities and along boundaries of herds still testing negative. RT-QuIC is a lab assay that would confirm whatever the dogs flag, leaving fewer samples for the lab to test. The authors also describe dogs as a significant investment both up front and over time.

Feces matter because deer can be infected and still look healthy. Signs can take up to 2 years to appear after infection and a deer without signs may still shed infectious prions that bind to soil and remain infectious for at least 2 years, the authors note.

The authors point to modeling that says the best management response requires catching CWD while fewer than 1% of deer carry it. FDR's clean-record paper shows that hunter samples can be too small to do that job. The median county among Wisconsin's 21 counties without a detection tested 60 deer in the 2025 CWD year, and none reached the 299 negative tests needed to rule out 1% prevalence.

Our read is that the best fit for dogs is around infected captive facilities in counties such as those on FDR's watch list and along the edges of known zones, where one confirmed hit would shift testing.

V2026 Study

Dogs Needed Training to Move From Small Fecal Amounts to Large

A 2026 study by Mallikarjun and colleagues in the Journal of Wildlife Diseases examined a practical gap. Newer CWD dog work aims to cut contamination risk by training on cotton aids that pick up the odor of 1 to 2 grams of whitetail feces over 24 hours without touching the feces, while a working dog often meets far more feces in the field. Four detection dogs trained with those small-amount aids were then tested with aids exposed to 22 grams.

The dogs did not significantly distinguish positive from negative large-amount aids in the first test. After three training sessions that rewarded alerts to the larger aids the dogs improved significantly, leading the authors to conclude that the dogs did not generalize on their own. We read only the abstract and it does not give hit rates.

Our read is that a dog's training odor has to match the odor it will meet, and the false alerts in the 2023 field test point in the same direction.

VIJarred Samples

Three Pet Dogs Searched Jarred Samples Only and Never Wild Droppings

A team of three dogs is small, and the field result comes from 11 positive encounters. The dogs were pets with handlers who were new to field searches.

Every sample was held in a jar beneath a lid. The authors list the varying freeze-thaw history, temperature, origin and water content of wild droppings as conditions the dogs will need to handle. Test positives had never been thawed while field negatives had different freeze-thaw histories, and only 4 of the 26 field-test negatives were novel because new samples ran low.

Postmortem tests may miss very early infections and some false alerts could therefore have been early cases, though the authors consider the unfamiliar area the likelier cause. The 2023 study made no comparison between dog alerts and RT-QuIC results.