Friday, August 13, 2010

Telling Tigers Apart: Scent Lineups Identify Individual Tigers from Scat

Dogs have been used to monitor the status of endangered species for some time, and a number of examples have been mentioned in prior blogs. (See Certification of Tortoise Detector dogs? May 20, 2009; Types of Detection Dogs—How Many Can You Name? March 5, 2010) In all cases I know of, however, the dogs were not trained to distinguish between individual bears, tortoises, ferrets, or whatever. Like drug dogs or cadaver dogs, they are trained to recognize any scent that belongs to the target category. An exception involves dogs that are being used to monitor the status of the Amur, or Siberian tiger (Panthera tigris altaica), found in the Russian far east. This involves training dogs under procedures similar to those used for scent lineups in criminal investigations where the dog is asked if there is a match to a scent from the crime in a lineup of scents from foils and a suspect in the crime.

Monitoring tigers has involved track size measurements, camera traps, and genetic analysis of hair and scat. Genetic analysis has been largely ineffective because of the low genetic variability of the remaining Amur tiger population. Camera traps often malfunction in cold temperatures, though winter may be the best time to survey these animals.

Two researchers from the Lazovsky State Nature Zapovednik (preserve) began training five dogs to develop the ability to distinguish tigers by scat. The training followed Dutch forensic practices, but the dogs were rewarded with food, not play. Dogs were selected at six to eight weeks based on observed play and food drive, and training began at four to six months of age. Dogs were trained to sniff the scat of one tiger, then alert to a match in a scent lineup of seven scat jars.

Dogs were first trained to smell jars by being presented with a row of 15 to 20 jars filled to the top with wadded paper on some of which was a small piece of hot dog. The dogs walked in front of the handler and were taught to smell each jar for the possible treat. After this was learned, treats were removed and jars were placed further apart and covered with scent boxes. A passive alert, a sit, was taught as the alert. The test jar, containing the scat the dog was supposed to match from the lineup, was placed on a 15 cm high stool. (See diagram.) Dogs were taught to smell the test jar first and received a reward from the handler on sitting at the matching jar. First this was done with a command but the command was soon removed so that the handler could remain blind to the correct target jar. Also, scat from different tigers was slowly introduced into every jar in the lineup but only one was a match.

When a dog chose the correct jar more than 70% of the time, more matching jars were introduced so that a dog might make one, two, or three matches in every trial (see diagram). Training was completed in two to four months, and performance usually continued to improve for four to six weeks after training was completed. Each trial took 10 to 15 minutes and dogs performed three to seven trials each day. Dogs that turned out to be unmotivated to work more than 10% of the time were retired.

Scat was obtained from tigers in zoos, circuses, and wild Amur tigers. Scats from known wild tigers (some of which were radio-collared) could be collected when there was a single set of tracks. The diet of wild tigers consists of red deer, wild boar, sika deer, and other animals, including dogs. Care had to be taken that collected scat did not grow mold, which could interfere with a dog’s ability to detect individual scent. Scats were aged about two days. Researchers found that dogs are attracted to some scats, as has been found by Adee Schoon and Tadeusz Jezierski with dogs used in forensic scent lineups. Of 58 scats, dogs were determined to be attracted to three for no apparent reason.

In 521 trails, dogs correctly chose one of seven scats with an average rate of 87% (individually varying from 79% correct up to 89% correct). When trials were repeated with different dogs, accuracy increased to 98%. Where two samples in the lineup matched the trial scent, dogs were accurate 84% of the time. Dogs were more accurate in identifying wild tigers than with captive tigers, perhaps reflecting the nearly identical diets of some of the captive tigers.

The researchers recommended that identifications be based on repeated tests of two or three dogs. They also recommended that scent identification be combined with other methods of following tigers.

Linda L. Kerley and Galina P. Salkina, Using Scent-Matching Dogs to Identify Individual Amur Tigers from Scats. Journal of Wildlife Management, 71(4), 1349-1356 (2007).  See also, Linda L. Kerley, Using Dogs for Tiger Conservation and Research.  Integrative Zoology, 5(4), 390-396.

No comments:

Post a Comment