Overview and study site
The laboratory and field investigations took place in three stages: (1) development and laboratory testing of PCR protocols and their eDNA detection sensitivity; (2) a coordinated, pre-eradication survey of the seven water bodies at a commercial fishery in south-east England using conventional and eDNA approaches; and (3) a post-eradication survey of two of these water bodies, the angling pond subjected to eradication measures (henceforth the ‘infested pond’) and the adjacent ‘holding’ (i.e. quarantine) pond where the rescued (i.e. large, commercially-valuable) fishes were held during and after the eradication work (details given here below).
The commercial fishery (latitude 51°N, longitude 0°E) is the same angling venue where a previous eDNA survey, using cPCR (Davison et al. 2017), demonstrated P. parva to be present in only one of the venue’s seven human-made angling ponds (areas of 0.5 to 2.4 ha). These ponds are surface-water fed only, i.e. not connected with each other nor with an adjacent stream that flows along the eastern side of the venue (for a map, see Fig. 1 in Davison et al. 2017), and any outflows from the ponds discharge into their own gravel and reed bed filters that do not retain surface water. An invasive population of P. parva was discovered in the infested pond at least as early as April 2004. An eradication attempt was conducted by the fishery owners (intensive trapping combined with introduction of a piscivorous fish species), but the persistence of a low-density population of P. parva was confirmed by cPCR of water samples and focused intensive trapping at the locations where DNA of P. parva was found (Davison et al. 2017).
Protocol sensitivity testing
Sensitivity tests were conducted using DNA extracted from P. parva dorsal muscle tissue (DNeasy Blood and Tissue Kit, Qiagen, Hilden, Germany) collected from a population in southern England; the sequence has been deposited in the open-source database Genbank (www.ncbi.nlm.nih.gov/genbank) with the accession number KR092385 (Davison et al. 2016). Several different approaches to defining limit of detection (LOD) and limit of quantification (LOQ) have been suggested, as reviewed by Hunter et al. (2017). In the present study, LOD is defined as the minimum amount of target DNA at which positive detections were recorded in one or more replicates (following the definition used in other eDNA studies, e.g. Takahara et al. 2013; Tréguier et al. 2014; Biggs et al. 2015). LOQ is generally defined as the lowest amount of target DNA that yields an acceptable level of precision and accuracy (IUPAC 1995; Tréguier et al. 2014). In the present study, LOQ has been specifically defined as detection in 100% of replicates as per Agersnap et al. (2017). Tests to determine LOD and LOQ were applied to two sources of DNA, referred to hereafter as “total DNA” and “plasmid DNA”. “Total DNA” refers to DNA extracted directly from muscle tissue, and therefore comprises both genomic and mitochondrial DNA. “Plasmid DNA” refers to targeted mitochondrial DNA obtained using cloning to create a plasmid solution for use as a standard, enabling calculation of DNA copy numbers. Concentrations of both total and plasmid DNA were measured using a Nanodrop® ND1000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA) and calculated with the software ND-1000 v3.8.1 (Thermo Scientific).
To obtain the plasmid DNA, a preliminary cPCR using total DNA from P. parva was performed to amplify the 350 base-pair target region (Table 1). Cloning was performed using a TOPO® TA Cloning® Kit for Sequencing (Invitrogen, Carlsbad, CA, U.S.A.) with PCR 4-TOPO® vector including competent cells (Escherichia coli), following the manufacturer’s recommended protocol. Bacterial colonies were grown on agar plates with ImMedia™ Amp Blue (Invitrogen). Colonies not displaying blue colouration were selected and inoculated in a liquid medium containing 40 mL of LB-Medium (MP Biomedicals, Santa Ana, CA) and 50 μg/mL of Ampicillin. The plasmids were isolated using QIAprep® Spin Miniprep Kit (Qiagen) and tested with a cPCR to verify the success of the incorporation of the mitochondrial cytochrome c oxidase subunit I (mtCOI) target gene sequence into the plasmid. Copy numbers for plasmid DNA standards were calculated from DNA concentrations and base-pair lengths using the equation of Godornes et al. (2007).
Conventional PCR (350 bp)
The cPCR used in all field surveys, and in sensitivity testing (referred to hereafter as cPCR 350) used specific primers to amplify P. parva DNA, designed to amplify a 350 base-pair sequence of the mtCOI gene (Table 1). Specificity of these primers was tested in silico against all sequences in the NCBI Genbank database using the NCBI Primer Blast software (www.ncbi.nlm.nih.gov/tools/primer-blast/; Ye et al. 2012). The primers were also tested experimentally in cPCRs against 0.1 ng genomic DNA extracts from Cyprinidae species which are likely to occur at the study site: common carp Cyprinus carpio, common bream Abramis brama, roach Rutilus rutilus and rudd Scardinius erythrophthalmus, with none of the triplicate cPCRs showing amplification for any of these species (Davison et al. 2016).
A further pair of cPCR primers, referred to hereafter as cPCR 101, were designed to amplify a target region of 101 base pairs. The purpose of this primer pair was to enable a direct comparison in sensitivity tests with the pair targeting a longer region (cPCR 350) to assess whether length of target region affected sensitivity. This primer pair was used only for comparative sensitivity testing in the laboratory, and it was not used in the field surveys.
Conventional PCRs were performed with a total reaction mixture of 20 μL containing 2 μL of DNA samples (total DNA, plasmid or eDNA), 0.5 μM of each specific primer, 10 μL of HotStar Taq® Plus DNA polymerase 2× (Qiagen Fast Cycling PCR Kit) and 2 μL of Coral Load Fast Cycling Dye 10× (Qiagen). De-ionised water was added to obtain the total mixture volume. The cycling conditions were 95 °C for 5 min, followed by 35 cycles of 96 °C for 5 s, 62 °C for 5 s and 68 °C for 12 s, with a final extension at 72 °C for 1 min. PCR products were visualised after 60 min of electrophoresis migration on 2% agarose gel, stained with SYBR™ Gold Nucleic Acid Gel Stain (Invitrogen). For both laboratory validation trials and eDNA field samples, five cPCR replicates were analysed in each machine run, on three discrete machine runs (i.e. 15 replicates in total).
Quantitative and nested PCR
Specific P. parva primers and probes were designed for qPCR to amplify a 101 base-pair sequence of the mitochondrial cytochrome c oxidase subunit I (COI) gene, occurring within the sequence amplified by the cPCR 350 primers (Table 1). The primers were successfully tested for specificity in silico against all sequences in the Genbank database using NCBI Primer-BLAST software, in which no species likely to be present at UK freshwater sites corresponded to the primer pair to within four base-pair mismatches (a level of mismatch within that used for assessing specificity by recent fish eDNA studies, e.g. Harper et al. 2019). The primers were also tested experimentally against genomic DNA of C. carpio, A. brama, R. rutilus and S. erythrophthalmus, with no amplification observed. Real-time qPCRs were performed using an Applied Biosystems™ Step One™ system (Applied Biosystems, Foster City, CA, USA) using the following thermocycling profile: 2 min at 50 °C, 10 min at 95 °C, followed by 35 cycles of 15 s denaturation at 95 °C, and 60 s annealing-extension at 60 °C.
PCRs were performed in a 20 μL reaction mixture containing 2 μL of total DNA, plasmid or eDNA, 1 μL of assay mix (18 μM forward and reverse primers and 5 μM probe) for the targeted species (Applied Biosystems™), 10 μL of TaqMan® Genotyping Master Mix (Applied Biosystems™) and 7 μL of de-ionised water. Samples and standards were analysed in triplicate. The standard curve comprised a range of five or six dilutions of a selected standard (plasmid or total DNA), acting as positive samples to confirm reaction efficacy. The dilution series was constructed from the standard on the day of analysis. Finally, the lengths of the qPCR products were checked using 2% agarose gel electrophoresis after addition of DNA Gel Loading Dye (6×) (Invitrogen).
The nPCR protocol consisted of two steps: (1) a cPCR, using the cPCR 350 primer pair and the protocol described above; and if the initial cPCR produced a negative result, then (2) a qPCR was performed on 2 μL from each completed cPCR. Five cPCR replicates were performed on each sample. Each cPCR replicate that produced a negative result was then subjected to qPCR in triplicate.
Coordinated pre- and post-eradication surveys
Water samples were collected on 6 and 7 September 2016 from 12 littoral zone locations spread at approximately equal distances from each other around the shores of all ponds, using the same statistically-designed sampling protocol developed specifically for these ponds (Davison et al. 2017). Water samples were collected at about 1.5 m distance from the bank using a 183-cm-long sampling pole fitted with a 500 mL polypropylene sampling cup (Camlab Ltd, Cambridge, UK), which, between samples, was disinfected thoroughly with Microsol 3+ sterilising solution (Anachem Ltd, Luton, UK) and washed with de-ionised water. New sampling poles and cups were used for each pond to ensure no contamination risk. The sampling cup was moved in a standardised manner from the bank, to the greatest extent reached by the pole, ensuring no contact with the bottom sediment. At each sampling location, three replicates of 300 mL water, obtained using the sampling cup, were injected through a Sterivex-GP 0.22 μm sterile filter cartridge (EMD Millipore, Billerica, MA, USA) using a 50 mL sterile syringe (Thermo Scientific) that is designed to attach directly onto the cartridge’s input opening. Cartridges from each location were sealed in individual plastic bags and immediately frozen (− 20 °C) for transportation back to the laboratory. On each sampling day, water from a sterilized bottle of de-ionised water from the laboratory was also filtered, handled and transported in the same manner as the pond samples, and analysed in the laboratory to test for contamination.
Conventional trapping in each of the angling ponds consisted of ten, previously-unused, rectangular minnow traps of 45 cm length and 25 cm width and height with 3 mm mesh, which were deployed on 7 September 2016 (i.e. same date as eDNA surveying, the use of new traps to avoid potential DNA contamination). Traps were baited using fishmeal pellets (21 mm diameter) and exposed for 12 h, with the numbers of fish captured recorded for P. parva only. Only five traps were used in the pond known to contain P. parva due to a periodical check of the traps revealing high numbers of P. parva captured. Once P. parva presence was confirmed, the traps were retrieved.
Post-eradication surveys (eDNA, trapping) were completed approximately six months after the fishery undertook procedures to eradicate P. parva. This consisted of complete drain-down of the infested (i.e. treatment) pond during which the larger and more commercially valuable fish were collected, passed through a salt bath (≈ 30 ppt) and placed into one of the adjacent ponds, henceforth the ‘quarantine’ pond. On 8 June 2017, three replicate water samples of 300 mL were collected (as described above) at 12 littoral zone locations from the treatment and the quarantine ponds. These samples were collected and analysed in the same manner as described above.
Laboratory processing of the pond-water samples
In the laboratory, DNA was extracted from the cartridges using a PowerWater Sterivex™ DNA Isolation Kit (MoBio, Carlsbad, CA, USA), with a final elution volume of 100 mL. The extracted sample was then diluted 1:5 in deionised water to dilute potential inhibitors (McKee et al. 2015), and a nPCR then performed using the conditions described above on 2 µL of diluted sample. To confirm that negative results in the qPCR were not detection errors (‘false negatives’) caused by PCR inhibition, five replicate samples from four locations per pond were spiked with 0.01 ng of P. parva total DNA and compared against controls of deionised water spiked with the same DNA quantity.
Sample extraction, PCR preparation and post-PCR analyses were each undertaken in separate rooms of a laboratory dedicated to molecular biology, observing strict anti-contamination procedures (no transfer of equipment between rooms; changing of lab coats when moving between rooms; thorough cleaning of all equipment and surfaces before and after use, and treating of equipment under UV light; use of sterile filter tips for pipettes). Increased risk of contamination is an important consideration with nested PCR protocols, due to the increased handling of amplified DNA. This risk was minimised by placing prepared reagents into well plates in a fume cabinet in a separate room from where the completed cPCR template was added, using different pipettes and gloves.
Differences between treatments in the sensitivity testing (plasmid DNA cPCR 350 vs. plasmid DNA cPCR 101; plasmid DNA cPCR 350 vs. plasmid DNA qPCR; total DNA cPCR 350 vs. total DNA qPCR) were tested by Permutational (Univariate) Analysis of Variance (PERANOVA). This was based on a one fixed-factor design consisting of Detection rate at two levels. PERANOVA was carried out in PERMANOVA + v1.0.8 for PRIMER v6.1.18 (Anderson et al. 2008), using a Euclidean distance, 9999 permutations of the residuals under a full model (Anderson and Robinson 2001), and with statistical effects evaluated at α = 0.05. Notably, the advantage of PERANOVA compared to ‘traditional’ (fully parametric) ANOVA is that the stringent assumptions of normality and homoscedasticity, which often prove unrealistic when dealing with biological data sets, are ‘relaxed’ considerably.