Environmental and training factors affect canine detection probabilities for terrestrial newt surveys

Abstract

Amphibians are of great interest to scientific research, but many populations are highly threatened and declining worldwide. Although varieties of aquatic survey methods exist, traditional methods for terrestrial surveys are very time-consuming but often not very effective. A novel method to detect terrestrial amphibians is the use of wildlife detection dogs. While their use for and factors affecting detection rates of mammals and reptiles are well documented, scientific literature on amphibian detection dogs is just emerging. The purpose of this study was to investigate the effects of environmental (habitat, weather) and training factors on detection probabilities for a newt detection dog. An experienced wildlife detection dog was trained and tested on smooth (Lissotriton vulgaris) and great crested newts (Triturus cristatus). Environmental and training parameters were recorded for 101 test trials and used as explanatory parameters in a binomial GLM. I found that detection probability strongly depended on temperature, whereby optimal temperatures varied by habitat. Detection probabilities were lowest in short grass, but there was no strong difference among forest habitats. They were higher for males than females and for great crested than smooth newts. For this dog, detection probabilities were also higher if the dog was working off the leash than on it, and when the dog was cooperative than fatigued. Dog performance increased over time with a strong increase at the beginning and a plateau at 92% detection probability. However, detection rates of this specific dog slightly decreased when the dog was working for more than two hours. The findings of this study provide a valuable basis for future deployments of this and other amphibian detection dogs. Dogs can certainly work in a variety of different habitats, although directed off-leash searching with enough time in complex habitats and specific training for small species with low detection distances may enhance their performance. The study design might also consider the temperature and humidity at which the dog will be deployed. A regular assessment of the detection dog using blind tests will give an indication of its reliability. Assessments similar to this study may further be used to estimate detection probability for a particular dog under given field conditions. Regular blind tests will show when the detection rates reached a plateau, which may then give an indication of on its reliability. Altogether, results suggest that newt detection dogs may provide a highly promising survey method, which can certainly be transferred for detecting other amphibians in terrestrial habitats.