Detection of non-native freshwater fishes using environmental DNA.

Abstract

Invasive non-native species are a serious conservation threat, and management to mitigate their impact depends on accurate survey methods. Early detection enables a rapid response while an invader is still localised. Recently, environmental DNA (eDNA) techniques, based on detection from DNA present in water samples, have been applied to search for a range of organisms, from invertebrates in small ponds to cetaceans at sea. The aim of this thesis was to investigate the use of eDNA methods to detect non-native freshwater fish. Sampling methods and species-specific primers were developed for pumpkinseed Lepomis gibbosus, topmouth gudgeon Pseudorasbora parva, sunbleak Leucaspius delineatus, and fathead minnow Pimephales promelas. All four species were detected within 24 hours of introduction into aquarium tanks, and DNA of L. gibbosus was detected within 6-12 hours of their release into experimental 5m x 5m ponds. Further validation of the techniques was obtained in a survey of ponds for P. parva on a site where the species was known to occur. The methods were then applied to a real-world scenario, mapping the distribution of P. parva on a complex of fishing ponds where an eradication attempt was in progress. The eDNA survey proved the persistence of a small population in one lake, and provided evidence for the patchy distribution of eDNA in a still water body. A laboratory protocol based on nested PCR was developed, and determined in laboratory trials to be more sensitive than conventional or quantitative PCRs. This more sensitive technique was then applied to a repeat sampling at the fishing pond complex, confirming the previous results in only providing positive detections at the one pond. The nested PCR protocol was used to assess a second eradication attempt at this site. Moving the method development from still to flowing water, eDNA surveys were conducted over three days to map the distributions of Lepomis gibbosus and Leucaspius delineatus in two river catchments in southern England (Sussex Ouse and Hampshire Test). This demonstrated the utility of the method for rapidly assessing distribution, with the results broadly matching previous knowledge gained through conventional survey methods, but with the addition of the discovery of L. delineatus detections upstream of the ponds containing a known source population in the River Ouse. Successful management of non-native species often depends on accurate detection of small populations (or reliable interpretation of negative results), and environmental managers need to know the sampling effort (and therefore allocation of financial resources) required to attain the necessary level of sensitivity. To this end, a field trial was conducted in which a known low density (100 fish per hectare) of two species (rainbow trout Oncorhynchus mykiss and barbel Barbus barbus) was stocked into six fishing ponds, and electrofishing compared directly with trapping and electrofishing. Different detectability was observed in the two species: the eDNA sampling proved the only effective method for detecting O. mykiss (detected on all six ponds), but B. barbus was not detected in all ponds, by any method. This trial, and the project as a whole, demonstrate that eDNA surveying is an effective method for detection of non-native fishes to inform management decisions, but that surveys need to be tailored to the ecology of the target species. The DNA, particularly of rare species, has been shown to be patchily distributed within the water body (at detectable levels), and a high level of sampling effort may be required to detect low density populations.