Wildlife detection dog training: A case study on achieving generalization between target odor variations while retaining specificity

摘要

Wildlife detection dogs are required to correctly discriminate target wildlife species odor from nontarget species odors (specificity), while enabling some degree of target odor variation (generality). Because there is no standardized training protocol, and little knowledge on training efficiency, we conducted a case study to test a dog's training efficiency in detecting 2-week-old wild otter (Lutra lutra) feces (spraint) odor among feces odors from 6 other large mammal species that often share the otter's natural habitat, including fox (Vulpes vulpes), hare (Lepus europaeus), roe deer (Capreolus capreolus), and cattle (Bos taurus). The dog was trained using a standard multiple-choice carrousel in a stepwise protocol. We started with odor samples from fresh captive otter spraints and progressed toward 2-week-old spraints from wild otters among other mammalian dung odors and tested for specificity and generality after each training step. We show that training on only 2 variations of spraints from captive otters enabled the dog to detect all desired spraint odor variations in our protocol, indicating a rapid generalization to variations of spraint odor the dog was not trained on, while retaining specificity. Testing such concept formation of target odors should be included in detection dog training and certification and could serve as a quality control measure of detection dog performance. (C) 2016 Elsevier Inc. All rights reserved.