Canine scent detection: A new approach to detect invasive freshwater fish

Koi carp (Cyprinus rubrofuscus) were first introduced to New Zealand in the early 1960s. Since their introduction, they have contributed to the decline of New Zealand’s freshwater ecosystems through resuspension of sediment from their feeding behaviour and through competition with native species. Methods such as netting, electrofishing, and environmental DNA (eDNA) are used to detect and monitor invasive fish species. However, these methods can be labour intensive, expensive, and can have low sensitivity when fish are rare or elusive. Although scent-detection dogs have been used to detect a wide range of substances, diseases, and animals, there is limited information on dogs’ ability to detect aquatic species. This thesis aimed to determine if laboratory-based dogs could be used as a viable survey technique for invasive carp. Three studies were performed to investigate this aim. The first study (Chapter 2) investigated dogs’ sensitivity and specificity to carp scent using a multiple probe design experiment to determine if dogs had the potential to detect a low biomass of carp. A single dog was trained to use an automated carousel independently of its handler and assessed water samples from aquaria containing either no fish scent (non-target, n = 3), goldfish (Carassius auratus) scent (non-target, n = 5) or carp (target, n = 9) scent. The goldfish samples and six of the target samples were presented to the dog at a standard fish biomass concentration of 15.5 mg/L (equivalent to 310 kg fish/ha). The remaining three target samples (probes) were diluted to determine the dog’s detection threshold. Results showed that the dog could detect carp down to a dilution equivalent to a biomass of 9.3 kg/ha (i.e., 0.5 mg carp/L), well below the biomass threshold of carp known to cause significant ecological impacts (i.e., >100 kg/ha). The dog’s performance was then compared with eDNA, an existing survey method. Quantitative PCR conducted on DNA extracted from laboratory aquaria revealed that the species-specific primer could detect carp at 9.3 kg/ha, but amplification rates at this dilution were low, as were all dilutions below the limit of quantification (≈160 kg/ha). Collectively, these results suggest that laboratory-based dogs could, with further research, provide a highly sensitive method of carp detection. Depending on the location of the field site it could take hours or even days for water samples to reach the laboratory and be assessed by dogs. Without effective preservation the olfactory profile of the scent could change, making it unrecognisable to the dogs. While preservation of scent samples is commonly practiced in the literature, very few studies have investigated the impacts preservation methods can have on dogs’ ability to detect the target scent. The second study of this thesis (Chapter 3) used an ABACADA reversal design to determine if refrigeration, freezing or potassium sorbate could be used as water sample preservation methods. To assess this, the dogs (n = 2-4) performed a baseline evaluation (A) of unpreserved water samples from aquaria containing carp (target scent; n = 7), goldfish (non-target scent; n = 5), or no fish (non-target scent; n = 5) before preservation treatments were applied to the samples. The treatment phases (B = refrigeration, C = freezing, and D = room temperature with potassium sorbate) involved applying a different preservation method to the water samples seven or eight days prior to assessment by the dogs. The results indicated that freezing and potassium sorbate have potential as scent sample preservatives. However, further research, testing longer storage durations with a larger sample size, is required to determine the full efficacy of these preservation techniques. The final study in this thesis (Chapter 4) evaluated dogs’ ability to detect carp in lake water. The first experiment in this study was a proof-of-concept experiment to determine if dogs could detect a common biomass of carp in lake water, which likely contains more background odours than aquaria water. Results from this experiment revealed that dogs could accurately detect a standard biomass of carp (i.e., 310 kg carp/ha) in samples from three lakes with varying water qualities; Taupō, Rotoroa and Rotoehu. All dogs achieved sensitivity values >83.6% and specificity values >77.3% on all lakes tested. Experiment Two tested dogs' ability to generalise carp scent across lake samples that varied in novelty, trophic state, and carp biomass. The results from this experiment revealed that dogs' mean detection performance was above chance on some, but not all, lakes. To compare the dogs' detection performance to an existing survey technique, eDNA analysis was performed on water samples assessed by the dogs. Interestingly, the multispecies assay only detected carp in three out of five lakes. Quantitative PCR improved detection of carp but still failed to detect carp in a lake known to contain naturalised populations. These results indicated that dogs have the ability to detect carp in naturally sourced water. However, further research is necessary to determine the scope of their generalisation abilities given the limitations imposed by COVID-19 and their inconsistent performances during Experiment Two. The findings from this thesis provide evidence that laboratory-based scent detection dogs may have potential to be used as an invasive carp detection method. In addition to demonstrating that dogs and eDNA have comparable levels of sensitivity to carp in aquaria water and that freezing, and potassium sorbate have potential as water sample preservatives, this thesis has also indicated that dogs may be able to detect and discriminate carp in water sourced from natural aquatic systems. The use of laboratory-based dogs could provide researchers with a timely and cost-efficient method of invasive fish detection, potentially enhancing their ability to regularly and extensively monitor freshwater ecosystems for new incursions of invasive species.
The University of Waikato
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.