The success of using trained dogs to locate sparse rodents in pest-free sanctuaries

Context. Better techniques to detect small numbers of mammalian pests such as rodents are required both to complete large-scale eradications in restoration areas and to detect invaders before they become abundant or cause serious impacts on biodiversity. Aims. To evaluate the ability of certified rodent dogs (Canis familiaris) to locate Norway rats (Rattus norvegicus) and mice (Mus musculus) or their scent trails at very low densities in field conditions. Methods. We experimentally tested two rodent dogs by releasing small numbers of laboratory rats and mice in a 63 ha rodent-free forest sanctuary and then determining if the dogs and their handlers could find the rodents and their scent trails. We divided the enclosure into two halves, east and west of the midpoint, and alternated releases daily between the two areas to minimise residual scent between consecutive trials. Radio-tagged rats or mice were released a total of 96 times at random locations that were unknown to handlers, followed for 50-100 m, then caught and either placed in hidden cages at the end of the scent trail or removed from the forest. Handlers and their dogs had up to 6 h to search for rodents. Key Results. Despite the extremely low density of rodents in the effective research area of 32 ha, both dogs were highly successful at finding rodents, together locating 87% of rats and 80% of mice. Handlers reported few false positive detections. We found that well-trained dogs can effectively cover 30-40 ha of steep forested habitat in half a day (6 h). Conclusions. Despite the limitations of our study design, we conclude that well-trained rodent dogs may be able to locate wild rodents at low densities in forest situations. Implications. Our results support the ongoing use of certified dogs to detect rodent survivors and invaders in conservation areas in New Zealand and elsewhere. Additional research is required to trial dogs on experimentally released wild rodents and to compare the cost-effectiveness of dogs with other detection methods.