How Behavior of Nontarget Species Affects Perceived Accuracy of Scat Detection Dog Surveys

1. What they did

Detection dogs are domestic dogs trained to find a target odor, here the scat of a target species. Scats they find are usually DNA tested, and some surveys find that 3.7% to 44.6% of the scats collected came from other (nontarget) species. This has often been blamed on poor handler or dog training. The authors tested three other explanations: a nontarget animal urinating on a target scat, eating it and passing it again (coprophagy), or mouthing and moving it. In each case the nontarget animal’s DNA could cover the target’s DNA, while the target’s smell remains for the dog to find.

Their target species was the cougar, and the main nontarget species was the coyote. They worked in three parts:

• Does urine change a scat’s DNA? They put about 2 to 3 milliliters of domestic dog urine (the size of a typical urine mark) on six dried domestic cat scats, then tested the DNA on the outside of each scat.
• How accurate is the detection dog? One experienced detection dog, a male Chesapeake Bay retriever trained on cougar scat, was tested in open areas with lines of odor stations: cougar scat, nontarget scat (bobcat, canid or domestic cat), nontarget scat with cougar urine on it or about 3 meters away, cougar scat with cougar urine, and cougar urine alone. The dog was also tested on scats from domestic dogs that had eaten deer meat with hair (common cougar prey), and from domestic dogs that had eaten cougar scat.
• Do nontarget animals interact with cougar scat? Five pairs of captive coyotes were given cougar scats marked with orange glitter, and their regular food with blue glitter four hours later; a control pair got only regular food (orange, then blue). Trail cameras watched the pens, and the glitter showed what was in each coyote scat, which was then tested with the detection dog. In the wild, single cougar scats were placed in front of three trail cameras at each of eight sites in Missouri for 14 days.

2. What they found

Urine on scat

• All six domestic cat scats marked with dog urine were identified genetically as domestic dog.

The detection dog

• In the urine trials the dog gave no trained response to nontarget scats, with cougar urine (24) or without (20), or to cougar urine alone (16). It gave a positive trained response to cougar scats with cougar urine (12) and without (20).
• The dog did not respond to any of seven scats from domestic dogs fed deer.
• The dog gave a strong positive response to all six domestic dog scats that clearly contained cougar scat, and none to the two that contained only dog food (it urine-marked one of them).
• The six dog scats containing cougar scat were all identified genetically as domestic dog, from swabs of both the outside and the inside, with no sign of mixed DNA.

Captive coyotes

• Many coyotes carried the cougar scats out of the cameras’ view, which limited the camera data. Cameras did record at least one coyote eating a cougar scat at once, and coyotes rubbing, rolling on, picking up and urine-marking the scats.
• The five test pairs produced 16 scats: 6 mainly orange (cougar scat), 8 mainly blue (regular food) and 2 mixed. All had some orange glitter.
• The detection dog gave a strong positive response to all 8 mainly orange and mixed scats. For the 8 mainly blue scats, it gave a weak positive response to 6 and none to 2. It did not respond to the four control-pair scats.
• The authors believe the four-hour gap between the two feedings was too short to separate the two diets cleanly.

Wild animals at cougar scats

• The cameras recorded 139 visits: 41.0% by white-tailed deer and 59.0% by furbearers (raccoon 36.7%, coyote 7.2%, fox 7.2%, bobcat 5.7% and opossum 2.2% of all visits).
• Deer never touched the scats. Most furbearer visits involved no contact either, but 8.5% did: a raccoon and an opossum moved and smelled scats; a fox and three coyotes smelled and urine-marked them; and one coyote smelled, moved and urine-marked a scat.

3. What they concluded

• There are three workable explanations, besides handler or dog training errors, for why detection dog surveys collect nontarget scats: urine-marking, coprophagy, and saliva from mouthing or moving scats.
• A well-trained detection dog does not confuse target urine, or nontarget scat containing the same prey, with target scat. It will find target scat even when other odors mask it, because the target odor is what earns the reward. So it may correctly find a cougar scat that now carries coyote DNA.
• To the handler such a scat looks like the target species’, so the problem only shows when the DNA result comes back, and the dog may then wrongly be judged inaccurate. This may explain the 33% nontarget scats in an earlier San Joaquin kit fox study where red foxes overlap.
• The share of nontarget scats probably depends on whether generalist mesocarnivores such as coyotes, or free-roaming domestic dogs, live alongside the target species: 3.7% in Misiones, Argentina, where the local fox eats mainly fruit and small rodents, against 50% or more in Nebraska cougar surveys where coyotes are widespread.
• Swabbing the inside of a scat does not get around coprophagy. Collecting only the freshest scats, first clearing an area of old scat and then collecting every few days, noting which way up the scat lay, and swabbing the bottom of the scat can reduce urine contamination.
• Coprophagy within and between felids and canids may also spread viruses and parasites, which matters for disease studies.
• Nontarget scats will not change research conclusions as long as every sample is DNA tested, but they add costs. Ways to reduce them include varied training samples, enough handler trials (with known and unknown sample locations), cleaning areas, good collection and storage, and lab checks for errors.

4. How tracking was or was not used

There was no tracking by people in the usual sense. The study is about scat as sign: how detection dogs find it by smell, how handlers judge whether it looks like the target species’, and how other animals’ behaviour (urine-marking, eating and moving scat) changes it. Scats were identified by DNA rather than by appearance, and the detection dog’s trained responses were read by the handler and an observer.

5. Tracker or observer certification

Not reported for people: the handler and observers are not described. The detection dog had several seasons of survey work in Argentina and Nebraska, and its accuracy was tested in this study. In the captive coyote test the handler did not know which samples had been placed where.

6. Other technologies, skills or innovations

• Detection dog trained to give a set response at target scat (stop without touching it, stay quiet and hold until the handler arrives), with strong and weak positive responses described from ear set, tail wag and sniffing.
• Coloured glitter mixed with food to show what each coyote scat contained.
• Motion-activated trail cameras, some set to video only, to record animals’ behaviour at scats.
• Genetic species identification from swabs of the scat’s mucus layer, with steps to avoid contamination in the lab.

7. Implications for conservation (from the paper)

• Detection dogs combined with DNA testing of scat are an effective way to gather samples for population estimates of uncommon, wide-ranging or elusive species.
• Nontarget scats should not lead managers to doubt a well-trained dog without first considering urine-marking, coprophagy and mouthing by other species, judged case by case.
• Coprophagy between wild and domestic animals is new information to consider in studies of zoonotic diseases and the health of wild populations.