Detecting low densities of koi carp and other pest fish species in lakes using environmental DNA

Abstract

Environmental DNA (eDNA) is a quick, effective and sensitive biomonitoring tool that can be used to detect invasive species at low densities in aquatic habitats. The aim of my study was to develop a standard operating protocol (SOP) for the detection of invasive fish in New Zealand’s lowland lakes, specifically koi carp (Cyprinus rubrofuscus). Four separate eDNA surveys between October 2022 and April 2024 sampled 18 lakes in total across the Auckland and Waikato Regions in the North Island of New Zealand. I evaluated different aspects of eDNA sampling including sample collection methods, filter sizes, seasonality, laboratory analysis methods, along with the optimal number of samples and spatial locations to filter water from. My study found that eDNA was a simple, reliable, and effective tool that can be easily used to detect koi and other pest species at low densities across a wide variety of lakes. The efficacy of eDNA sampling was proven during a real-world koi incursion at Lake Waiwhakareke, where I successfully detected koi for the first time in this lake at very low densities using eDNA. Overall, I found that a coarser (5 μm) filter size outperformed the finer (1.2 μm) filter in every comparison and that the qPCR laboratory analysis method was slightly superior to the metabarcoding method, although it has the disadvantage of only being able to detect a single species. My study showed that the multisite composite sample collection method performed marginally better than the single collection overall, across 13 lakes in the Auckland Region. Targeting the littoral margins of the lakes was found to be significantly better in terms of metabarcoding read counts and species detections than sampling the pelagic/mid-lake regions at both Lake Puketirini and Lake Waiwhakareke. In the spatio-temporal study in the Waikato, I found that there was a higher degree of variation between the two seasons sampled (Spring and Autumn) than there was between the three lakes. The spring season performed best in terms of fish detections and metabarcoding read counts. In terms of the sampling number and density, I found that eight samples per kilometre of lake perimeter was found to be optimal to detect koi carp at very low densities at Lake Waiwhakareke and Lake Puketirini. However, further research is needed to better understand the uncertainties involved with eDNA sampling, including the amount of sampling required to provide high levels of confidence in the results. Using negative and positive controls at the time of sampling and during laboratories analyses, in addition to species distribution modelling may help ascertain the sensitivities of eDNA detection with a greater level of certainty.