Using Scent-Matching Dogs to Identify Individual Amur Tigers from Scats

1. What they did

Amur tigers are hard to count. They are secretive, rare, live at low densities and range widely. In the Russian Far East they are usually monitored by snow-track surveys, and the usual alternatives have problems there: DNA does not separate individuals well because Amur tigers have low genetic variability, and camera traps need long sampling periods and often fail in the cold. Dogs in Russia had been trained to identify individual tigers by scent, but the method had not been tested rigorously for field studies. The authors asked whether trained dogs can tell individual tigers apart from their scats.

• The dogs. They raised five dogs of different breeds from pups, chosen for their play and food drives, and trained them over 2 to 4 months, rewarding them with food (pieces of hot dog) and praise.
• The scats. They collected 58 scats from 25 known tigers: wild tigers (some radio-collared), and tigers in circuses, zoos and a rehabilitation center. A wild scat was only used when it could be tied to one individual: by following a single set of tracks and collecting only the scat along those tracks, at capture sites of radio-collared tigers, or by snow-tracking in areas far enough apart (or following tracks of tigers of different sex, judged by track size) to avoid following the same tiger twice. Scats were frozen, aired for 48 hours, and placed in jars under upside-down buckets with a hole, so the dog had to sniff rather than look.
• The scent lineup. The dog smelled a test scat, then sniffed seven jars set in a semi-circle and sat by the jar it judged to be from the same tiger. One person handled the dog while another set out the jars, so neither the handler nor the dog knew where the match was.
• The tests (December 2002 to March 2005). Single-choice tests (one match among seven jars); repeated tests (one dog given the same tiger on different days); cross-tests (three dogs given the same tiger on the same day); double-choice tests (two matches in the lineup); and a scat-age test using 11 scats from one tiger collected over 4 years.

2. What they found

Single-choice tests

• Across 521 trials, the dogs picked the right jar 87% of the time (chance was 14.3%).
• Each dog did far better than chance, with accuracy from 79% to 89%, and there was no significant difference between dogs.

Repeated tests, cross-tests and double-choice tests

• When a tiger counted as identified if a dog got it right in 2 of 3 trials on different days, accuracy rose to 98% (86 sets of repeated trials, dogs 1 to 4). Dogs 1 and 2 were 100% accurate; dogs 3 and 4 each missed one tiger.
• In cross-tests on the same day, 20 of 22 tigers (91%) were correctly identified by 2 of 3 dogs. One scat from a wild radio-collared tiger was matched to the same wrong tiger by three dogs on one day, but that tiger was never misidentified in repeated tests.
• With two matches in the lineup, dogs found both in 84% of 49 trials. Every mistake was sitting at only one of the two correct jars.

Scat age

• Four dogs matched 11 scats from one radio-collared female, deposited over 4 years, with 100% accuracy (40 trials).

The dogs themselves

• Three of 58 scats (from circus tigers) were dropped because the dogs were drawn to them for no apparent reason.
• Working lives varied. Dog 1 worked well for about a year, then refused to smell the jars. Two more German shepherd mixes never finished training. Dog 2 was 95.8% accurate in her first year but dropped to 78.1% in her second, as she showed signs of boredom.
• Mistakes: sitting at the jar next to the correct one (50% of mistakes), sitting at the jar that had been correct in the previous trial (6%), sitting at a wrong jar for no apparent reason (44%), and, in double-choice tests, sitting at only one of the two matches.
• 56% of mistakes involved scats from captive tigers. The authors suggest captive scats smell stronger and are often collected wet, while wild scats air-dry and contain more hair and bone. Similar diets, living conditions or close relatedness of captive tigers may also play a part.

3. What they concluded

• All five dogs identified individual tigers from scent in scat with high accuracy, despite different breeds, ages and working styles. Most mistakes probably came from training or trial design (for example a dog rushing to sit), not from being unable to recognize an individual’s scent.
• Repeated tests on different days improved accuracy much more than cross-tests on the same day, perhaps because conditions on one day can throw dogs off. The authors recommend basing identifications on repeated tests (on different days) by 2 to 3 dogs.
• Careful choice and care of dogs can extend their working lives: choosing dogs with a strong food or toy obsession and search drive, perhaps choosing older dogs (over 1 year) rather than pups, plenty of play and enthusiastic rewards, never correcting harshly, varying the routine, and treating each dog as an individual.
• Where possible, dog identifications should be checked against genetic results. Where that is not possible, as for Amur tigers, dogs should be used with another field method, such as track measurements or locations of radio-collared animals, to confirm identifications and build a reference collection of scats.
• Only one tiger’s scats spanned 4 years, so scat age needs more study.

4. How tracking was or was not used

Tracking was used to collect scats that could be tied to one known wild tiger. Teams followed a single set of tracks and collected only the scat along those tracks, collected scat during snow-tracking in areas far apart, or followed tracks of tigers of different sex, judged by track size. The paper also describes snow-track surveys as the traditional way to monitor Amur tigers, and suggests that in winter animals could be tracked through the snow to a scat for dog identification. Track measurements are recommended as one way to confirm identifications made by dogs.

5. Tracker or observer certification

Not reported. The paper does not say who did the snow-tracking or how skilled they were; staff of the Wildlife Conservation Society Siberian Tiger Project are thanked for collecting scats. This matters because the “known individual” status of the unmarked wild tigers rests on the tracking.

6. Other technologies, skills or innovations

• Scent lineups adapted from law enforcement: seven jars in a semi-circle, each under an upside-down bucket with a 9 cm hole, with a blind setup so neither handler nor dog knew the match.
• Scat handling: wrapping fresh scats in aluminum foil inside two zip-lock bags, freezing, air-drying for 48 hours before use, using only a small piece of scat, and boiling jars between uses.
• Practice runs before each trial to check the dog’s motivation; dogs unmotivated more than 10% of the time were stopped.

7. Implications for conservation (from the paper)

• Scent-matching dogs are a noninvasive and fairly reliable way to identify individual tigers, and presumably other animals.
• With mark-recapture methods, they may help estimate the abundance of Amur tigers and other large carnivores, and offer an alternative to DNA analysis where genotyping is impractical, unaffordable or ineffective.
• Because some tigers can be identified from scats deposited 4 years apart, dog identifications could suit both short-term (closed population) and multi-season or multi-year (open population) studies.