Analytical validation and field testing of a specific qPCR assay for environmental DNA detection of invasive European green crab (Carcinus maenas)

Abstract

Environmental DNA (eDNA) methods are providing tools for detecting invasive species in aquatic environments. Targeted qPCR assays applied to eDNA samples promise to overcome limitations of traditional methods, especially for early detection. The European green crab (Carcinus maenas) is considered one of the most successful invasive species globally due to the large range it has invaded and negative impacts on native species, marine habitats, and shellfish industries. We developed, laboratory-validated, and field-tested a specific qPCR assay for the detection of green crab from eDNA samples. We also show that the assay can detect green crab in bulk DNA extracted from plankton samples. Assay design, optimization, sensitivity, and specificity testing generally followed the validation pathway recommended by the World Organization for Animal Health for assays used to manage global aquatic animal health and infectious disease. Assay specificity was verified in silico and in vitro by laboratory testing 26 nontarget species, none of which showed potential for amplification. Assay sensitivity was appropriately high, with the limit of detection approaching two gene copies/μl. The assay was field-tested on eDNA samples collected from filtered seawater at five sites on the Pacific coast of Canada known to harbor green crab based on historical monitoring data; green crab DNA was amplified from all sites. We also present early pilot field testing of the assay done on bulk DNA extracted from plankton samples from four sites from Australia, two sites with and two sites without reported records of green crab presence. Green crab was detected at both sites with known green crab records. Significant inhibition was recorded for some plankton samples but not for eDNA samples. This is the first qPCR assay for detection of European green crab, providing researchers and managers with a valuable new tool to aid early detection and ongoing monitoring.

INTRODUCTION

The European green crab Carcinus maenas (Linnaeus, 1758) is one of the International Union for Conservation of Nature's (IUCN) world's 100 worst alien invasive species and is one of the most widely distributed aquatic invasive species (AIS) globally (Darling & Tepolt, ). Its native range spans from northwestern Africa through Atlantic Europe including northern Norway and Iceland (Gillespie, Phillips, Paltzat, & Therriault, ). Through ballast water exchange, commercial shipping, and other human mediated activities, the species has been introduced and established abundant invasive populations in South Africa, Australia, Japan and both the Pacific and Atlantic coasts of North America (Carlton & Cohen, ; Darling & Tepolt, ; Gillespie et al., ; Therriault, Herborg, Locke & Mckindsey, ; Thresher et al., ). On the Pacific coast of North America, green crab arrived in a single invasion to the San Francisco Bay area in the late 1980s and have since expanded via larval drift northward more than 1,000 km to the central coast of British Columbia (BC), Canada (Gillespie et al., ; Kuris, Lafferty, & Torchin, ; Tepolt et al., ).

Invasive green crab populations create ecological and economical challenges for the regions they invade (Gillespie et al., ; Therriault et al., ). They negatively impact marine biodiversity through competition, predation, and physical habitat alteration as well as through modifications to plant, invertebrate, and fish community structure (Kuris et al., ; Therriault et al., ). Green crab are also of significant concern for commercial fisheries and aquaculture industries because they are major predators of clams, mussels, and juvenile fish and compete with lobster and native crabs (Darling & Tepolt, ; Gillespie et al., ).

Recent increases in global trade and travel provide an avenue for further spread and new introductions of AIS. There has been a shift by some governments toward developing regulations, monitoring programs, and action plans to address this issue. The Canadian Action Plan to Address the Threat of Aquatic Invasive Species and Australia's National System for the Prevention and Management of Marine Pest Incursions both aim to address new introductions of aquatic invasive species and evaluate costs and impacts associated with eradication and management of established populations (Australian government national control plan for the European green shore crab, Carcinus maenas, ).

At the forefront of managing, AIS is the ability to detect, identify, and monitor species presence; development and implementation of validated surveillance tools has therefore become a priority for decision-makers at local, regional, and national levels (Bax et al., ; Darling & Blum, ). Environmental DNA (eDNA) based methods, in which DNA is extracted directly from environmental samples and used to detect organisms, provide an exciting opportunity to complement traditional techniques for surveying AIS (Bott et al., ; Darling & Blum, ; Goldberg et al., ; Westfall, Therriault, & Abbott, ). Environmental DNA was first used for the detection of American bullfrog Rana catesbeiana (Shaw, 1802) and has successfully been used for detecting a variety of amphibian, fish, and aquatic invertebrate species (Balasingham et al., ; Evans & Lamberti, ; Ficetola, Miaud, Pompanon, & Taberlet, ; Geerts et al., ; Pont et al., ; Rees, Maddison, Middleditch, Patmore, & Gough, ; Sassoubre, Yamahara, Gardner, Block, & Boehm, ; Westfall et al. ). Despite the research success of eDNA methods to date, published studies are highly variable with regard to methods around field sampling, assay development, DNA extraction, technical replication, and data interpretation (Goldberg et al., ; Rees et al., ; Tsuji, Takahara, Doi, Shibata, & Yamanaka, ) and little-to-none have published validation data on diagnostic sensitivity and specificity. As such, there is a need for standardized and validated methods given the potential of eDNA to augment AIS surveillance globally.

The primary aims of this study were to develop a TaqMan qPCR assay to detect European green crab from eDNA samples; laboratory validate its fitness-for-purpose prior to field implementation; and undertake basic field validation by testing filtered water at known invaded sites. We also include here pilot data generated in early testing of the assay on bulk DNA extracted from plankton, as these also support the utility of the assay for green crab detection. We followed the analytical validation pathway recommended by the World Organization for Animal Health (OIE) for qPCR assays used in global aquatic animal health infectious disease diagnosis, disease surveillance, and trade (International Office of Epizootics Animal Health Standards Commission, ) which are mirrored in Australia's guidelines for development and validation of assays for marine pests (Department of Agriculture and Water Resources, ). We report a detailed description of assay development, evaluation of necessary performance characteristics, and initial testing as an eDNA biomonitoring tool. This study not only provides the community with a useful early detection and monitoring tool for green crab, but will also facilitate improved consistency among studies for the development and validation of eDNA-based qPCR assays.

METHODS Assay design

The mitochondrial cytochrome c oxidase subunit 1 (COI) gene was targeted for assay design. Main factors in selecting this gene region were availability of reference sequences and sufficient variability over a short fragment for species-level detection of green crab. COI sequences from green crab and closely related organisms were downloaded from GenBank, grouped by family, and aligned in BioEdit (Hall, ) using the Clustal W alignment. This was done to infer useful diagnostic sequence regions conserved within green crab that were also able to delineate heterologous species. Green crab specific primers and TaqMan^® Minor Groove Binding (MGB) probe sequences were generated using Primer Express v2.0 software (Applied BioSystems^®) and then tested in silico using PrimerBlast.

Assay optimization

The 2-step qPCR assay developed here used forward primer 5′-ATGAACAGTCTATCCTCCTTTAG-3′, reverse primer 5′-GAAAGAACGCATATTGATAATAGTTG-3′, and TaqMan™ MGB probe (Applied Biosystems^®) 6FAM-AGTTGATTTAGGGATTTTC-MGB. The recommended optimal annealing temperature of 60°C was used; cycling parameters were 15 min at 95°C followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. Optimization experiments led to the following final assay concentrations: 12.5 μl 2× TaqMan^® Environmental master mix, 0.5 μM forward and reverse primer, 0.1 μM TaqMan^® probe, 2 μl template, and 7.75 μl of nuclease-free water to yield a 25 μl final reaction.

Optimal primer and probe concentrations were defined as the conditions generating the lowest cycle threshold (Ct) and were assessed using various matrices. Primers were optimized asymmetrically, testing forward and reverse primer final reaction concentration combinations of 50, 100, 200, 400, and 600 nM with a fixed probe concentration of 250. Probe concentrations of 100, 150, 200, and 250 nM with symmetric primer concentrations of 200, 300, 400, 450, 500, and 600 nM were tested in a second matrix. A final matrix was performed to assess the impact of low template DNA concentrations, characteristic of eDNA. A total of 76 possible primer, probe, and template combinations were assessed comprising of three green crab DNA concentrations: 8, 1, and 0.1 ng/μl with symmetric primer concentrations of 450, 500, 550, and 600 nM and probe concentrations of 90, 100, and 110 nM (Table S1). Statistical testing of optimization results and other results throughout this study were conducted using Graph Pad Prism 8 software.

Three qPCR replicates were used in assay development with one exception (details in specificity section). Assay optimization reactions were performed using green crab DNA extracted from two specimens collected from Barkley Sound, British Columbia (BC), Canada, and quantified by Nanodrop™ (Thermo Scientific).

Specificity testing

The Pacific Northwest of North America is reported as having only one green crab COI haplotype, whereas multiple COI haplotypes are known to occur in Australia (Darling & Tepolt, ; Hatzenbuhler, Kelly, Martinson, Okum, & Pilgrim, ); as such, 25 green crab samples from Australia were tested here for assay inclusivity (Table ) comprising seven samples from Gulf St Vincent, SA; six from Port Phillip Bay, VIC; four from Lakes Entrance, VIC, three from Hobart, TAS; and five from Batemans Bay, NSW, Australia. Nonspecific amplification and primer dimer formation were assessed by generating a melt curve of qPCR products by using Brilliant II SYBR^® Green qPCR master mix (Agilent Technologies) and omitting the TaqMan^® probe.

Species used, specimen sampling locations, and results of specificity testing for the green crab qPCR assay developed here

Phylum	Order	Family	Genus	Species	n	Sampling Location	Average Ct	qPCR Reps
Arthropoda	Decapoda	Cancridae	Cancer	productus	2	Port Hardy, BC, Canada	No Ct	3
		Cancridae	Glebocarcinus	oregonensis	2	Victoria, BC/Port Renfrew, BC, Canada	No Ct	3
		Cancridae	Metacarcinus	magister	2	Port Hardy, BC, Canada	No Ct	3
		Cancridae	Metacarcinus	gracilis	2	Port Hardy, BC, Canada	No Ct	3
		Cheiragonidae	Telmessus	cheiragonus	2	Port Hardy, BC/Croften, BC, Canada	No Ct	3
		Epialtidae	Pugettia	gracilis	2	Port Renfrew, BC/Tofino, BC, Canada	No Ct	3
		Oregoniidae	Oregonia	gracilis	2	Victoria, BC/Nanaimo, BC, Canada	No Ct	3
		Panopeidae	Lophopanopeus	bellus	2	Nanaimo, BC, Canada	No Ct	3
		Penaeidae	Penaeus	latisulcatus	5	Gulf St Vincent, SA, Australia	No Ct	2
		Portunidae	Portunus	pelagicus	5	Gulf St Vincent, SA, Australia	No Ct	2
		Portunidae	Carcinus	maenas	25	Gulf St Vincent, SA/Port Phillip Bay, VIC/Lakes Entrance, VIC/Hobart, TAS/Batemans Bay, NSW, Australia; Barkley Sound, BC, Canada	20.3	3
	Siphonostomatoida	Caligidae	Caligus	sp. 1	5	Spencer Gulf, SA, Australia	No Ct	2
		Caligidae	Caligus	sp. 2	5	Spencer Gulf, SA, Australia	No Ct	2
		Caligidae	Caligus	sp. 3	5	Spencer Gulf, SA, Australia	No Ct	2
		Caligidae	Caligus	sp. 4	5	Spencer Gulf, SA, Australia	No Ct	2
	Stomatopoda	Squillidae	Squilla	mantis	5	Ionian Sea, Italy	No Ct	2
Chordata	Phlebobracnchia	Ascidiidae	Ascidiella	sp.	5	Gulf St Vincent, SA, Australia	No Ct	2
		Cionidae	Ciona	intestinalis	10	Gulf St Vincent, SA, Australia	No Ct	2
Echinodermata	Forcipulatida	Asteriidae	Asterias	amurensis	10	Port Phillip Bay, VIC, Australia	No Ct	2
Mollusca	Mytilida	Mytilidae	Xenostrobus	securis	5	Port Phillip Bay, Australia	No Ct	2
		Mytilidae	Modiolus	modulaides	5	Jeju Island, Japan	No Ct	2
		Mytilidae	Arcuatula	senhousia	10	Gulf St Vincent, SA/ Port Phillip Bay, VIC, Australia	No Ct	2
		Mytilidae	Musculus	viridulus	5	Jeju Island, Japan	No Ct	2
		Mytilidae	Musculus	cumingianus	5	Wellington, New Zealand	No Ct	2
		Mytilidae	Perna	viridis	5	Singapore	No Ct	2
		Mytilidae	Perna	canaliculus	5	Nelson, New Zealand	No Ct	2
		Mytilidae	Trichomya	hirsuta	5	Gulf St Vincent, SA, Australia	No Ct	2

Specificity testing on related and co-occurring species was performed in Canada and in Australia (Table ) on 121 specimens from 26 species (Table ). In Canada, species for specificity testing were chosen by examining minimum evolution phylogenies of the COI barcode region of species known to occur in or near coastal waters where green crab occur (Castelin et al., ). Two specimens of each of the most genetically similar species (Cancer productus, Metacarcinus magister, Glebocarcinus oregonensis, and Metacarcinus magister) and select representatives of more dissimilar species (Pugettia gracillis, Lophopanopeus bellus, Telmessus cheiragonus, and Oregonia gracilis) were tested. In Australia, specificity was tested on either five or ten individual specimens (based on availability) of geographically co-occurring species from Gulf St. Vincent, South Australia, and other taxa available from the South Australian Research and Development Institute's (SARDI) collection (Table ).

Artificial (gBlock) positive control

A double-stranded 100 bp gBlock fragment was designed and synthesized (Integrated DNA Technologies) for use as a qPCR positive control to avoid potential laboratory contamination caused by using a tissue-derived positive control. The gBlock fragment contained green crab specific primer and probe sequences and was otherwise comprised of random nucleotides corresponding to no known sequences, making it readily differentiable from the target COI amplicon via melt curve analysis. The gBlock (starting concentration: 2.17 × 10¹⁰ gene copies/μl) was serially diluted 10-fold from 1.0 × 10⁻² to 1.0 × 10⁻⁸ and compared to a quantified 10-fold serial dilution of green crab DNA (starting concentration: 540 ng/μl) to ensure exhibition of similar properties under optimized reaction conditions. Resulting Ct values from each dilution of green crab DNA and gBlock were compared.

Limit of detection

The limit of detection (LOD), defined as the absolute minimum number of copies detectable by the assay (Purcell, Getchell, McClure, & Garver, ), was determined via a 10-fold serial dilution of the gBlock fragment from 1 × 10⁻⁴ to 1 × 10⁻¹¹. Each dilution was run in ten qPCR replicates. The lowest concentration of gBlock producing a Ct value in at least five of ten replicates was considered the LOD with 50% chance of detection (Biggs et al., ; Polinski et al., ). The limit of quantification (LOQ) was defined as the lowest number of target copies reliably detected (i.e., with coefficient of variance <20%). LOQ was calculated using a 10-fold serial dilution of the gBlock from 1 × 10⁻⁴ to 1 × 10⁻¹¹; this was tested with ten qPCR replicates per dilution in three different runs for repeatability. The lowest concentration of gBlock producing a Ct value in all ten replicates in all three runs was considered the limit of quantification (Nevers et al., ; Raymaekers, Smets, Maes, & Cartuyvels, ). LOQ results were statistically tested using a one-way analysis of variance. Amplification efficiency was determined for each run by plotting Ct value versus gBlock dilution and calculating the linear slope. The coefficient of determination (R²) value was also calculated for each run. The 10-fold dilution series of gBlock was subsequently used as a qPCR standard and was run in triplicate on every qPCR plate of field samples.

Pilot field testing: plankton samples (Australia)

Initial field testing of this assay was performed on plankton samples in Australia prior to the extensive laboratory optimization of this assay for eDNA-based detection in Canada, which was pursued at a later date. Reaction conditions used for the pilot testing were based on initial optimization and not the ones used here to determine LOD/LOQ. Regardless, since plankton is another sample type that holds promise for enabling early detection of green crab, we have included these pilot results here.

Between 60 and 288 plankton tows were done at four Australian ports: Brisbane; Gladstone; Hobart; and Melbourne (Figure ; see Table for sample sizes per location). Field surveys were designed using the Monitoring Design Excel Template (MDeT). Plankton samples were collected using a conical plankton net (Sea-gear^©) fitted with a Sea-gear^© flow meter and towed behind a vessel at ~1 kt. The length of tow was calculated based on GPS coordinates, with 100 m tows used as standard. Effective tow length was calculated based on flow meter readings and then compared to GPS distance as a measure of sampling efficiency.

Map of plankton sampling sites in southeast Australia, as follows: (F) Melbourne; (G) Hobart; (H) Gladstone; (I) Brisbane

Plankton was coarse-filtered using 2 mm mesh and concentrated to ~40 ml by tilting water through the cod end mesh. Samples were transferred to 120 ml sample tubes containing 80 ml RNAlater (Qiagen^®) and placed on ice. Between field sites, plankton nets, and other sampling equipment were cleaned in 60°C freshwater containing 200 mg/L active hypochlorite. An Artemia sample quality control (Giblot-Ducray & Bott, ) was used in all preserved samples and was added to tubes prior to sampling. In the laboratory, plankton samples were individually vacuum filtered on paper filters (Filtech, grade 1803). Paper filters were transferred to 50 ml screw cap tubes and freeze-dried.

DNA was extracted using the commercial DNA extraction service operated by SARDI (Ophel-Keller et al., ). The efficiency and consistency of SARDI's method to extract DNA from environmental samples has been confirmed in comparison with other commercial methods (Haling et al., ). Prior to DNA extraction, a standard amount of internal control was added to each sample to assess DNA extraction efficiency and PCR inhibition (Ophel-Keller et al., ). Carcinus maenas and internal control DNA were quantified in each sample by qPCR performed on a QuantStudio7 real-time PCR system (Applied BioSystems^®). qPCR reaction conditions were 5 μl 2× Qiagen Quantitect Probe Master Mix, 0.4 μM final concentration forward and reverse primers, and 0.2 μM final concentration of probe, with 4 μl template and nuclease-free water in a 10 μl final reaction volume. Each PCR plate included no-template controls, positive controls, and calibration standards. Three qPCR replicates were performed per plankton tow; a tow was considered positive if one or more replicate generated a Ct.

Field validation: eDNA samples (western Canada)

The green crab qPCR assay was field-tested on eDNA extracted from filtered water samples from five sites in Barkley Sound, BC, Canada, for which historical monitoring data confirmed the presence of green crab. Sites were as follows: Mayne Bay, Pipestem Inlet, San Mateo Bay, Ritherdon Bay and Hillier Island (Figure ). Six eDNA water samples were collected at the same time and location at each site.

Map of eDNA water sampling sites in Barkley Sound, Canada, as follows: (A) Hillier Island; (B) Pipestem Inlet; (C) Mayne Bay; (D) San Mateo Bay; (E) Ritherdon Bay

All eDNA field sampling equipment was decontaminated using 10% v/v bleach for 30 min, rinsed thoroughly with distilled water, and sterilized via autoclave. Water samples (1 L) were collected by submerging a 1 L Nalgene bottle approximately 1 m into the water using a pole held off the front of the boat. Samples were placed on ice and transported to the laboratory where they were vacuum filtered using a 47 mm cellulose 1 μm pore size filter (Whatman^®). Filters were placed in lysis buffer (Buffer ATL, Qiagen^®) and digested with proteinase K overnight at 56°C. DNA was extracted using a Qiagen^® DNeasy Blood & Tissue kit following manufacturer's protocol with three spin column transfers for each sample. A single field blank, comprised of filtering 1 L of distilled water for each site (N = 5) and DNA extraction blanks (N = 10) were included. DNA extracts were quantified on a spectrophotometer (Nanodrop™, Thermo Scientific).

Inhibition testing

Inhibition was tested for at each site. For eDNA samples, the gBlock artificial positive control was spiked into one aliquot of eDNA per site that had already tested negative by qPCR for green crab. The final concentration of gBlock after spiking was 2.17 × 10⁻⁵ gene copies/μl; a shift in expected Ct greater than three cycles in the spiked samples was considered evidence of inhibition (Hinlo et al. ). For the bulk DNA samples extracted from plankton, inhibition was estimated in each sample by a scale factor calculated using results from the internal control qPCR and comparing the results to the same control extracted from distilled water. Extraction efficiency was calculated using the Artemia sample control, the internal control, and results from equivalent samples spiked in distilled water.

RESULTS Assay optimization and specificity

We successfully developed, optimized, and analytically validated new qPCR primers and probe targeting a 100 bp fragment of green crab COI. Optimal final primer and probe concentrations were 500 and 100 nM, respectively (see Table S1 for Ct results of all combinations). Low template DNA concentration (0.1 ng/μL) was found to have no significant impact (p = .645) on optimal primer and probe concentrations as measured by a two-way ANOVA with Geisser-Greenhouse correction. Results of experimental tests for nontarget amplification were consistent with high specificity of the assay; no amplification occurred in all 26 nontarget species tested, including ten other crab species (Table ). In addition, all 25 green crab specimen from four geographically different locations in Australia known to host multiple COI haplotypes of green crab (Darling, Bagley, Roman, Tepolt, & Geller, ) were successfully amplified.

Artificial positive control

The gBlock gene fragment designed for use as an artificial positive control and quantification standard was readily discernable by melt curve analysis from the true 100bp green crab amplicon: The gBlock product melted at 78°C whereas the true green crab product melted at 82°C. Primer dimer formation was not visible on either of the melt curve plots and both curves exhibited a single peak at the respective melting temperatures. Comparison of the serial dilution of both gBlock and green crab DNA indicated no significant difference in Ct value between them (p = .94) based on an unpaired t test with Welch's correction (Figure ). This confirmed the gBlock to be valid for use as an artificial positive control.

Plot of cycle threshold (Ct) versus relative dilution of gBlock and green crab DNA. Each data point represents the average Ct value of three technical replicates. No significant difference between the two data sets was determined via unpaired t test with Welch's correction

Assay limit of detection

The limit of detection (LOD) and limit of quantification (LOQ) were assessed using the gBlock fragment in three separate qPCR runs (Figure ). A gBlock dilution of 10^–11, equivalent to the detection of 0.2 copies/μl, was determined to be the limit of detection with a 50% chance of detection. This corresponded to an average Ct value of 38 ± 0.61; no Ct value was produced in any replicates at dilution levels below this. The LOQ was observed at a gBlock dilution of 10^–10 corresponding to 2.01 copies/μl. The LOQ in all ten replicates in all three runs had an average Ct value of 35 ± 0.97, resulting in a coefficient of variance of 2.81. As such, a 10-fold serial dilution of the gBlock ranging from 10^–4 to 10^–10 was run as a quantifying standard on each plate. Reaction efficiencies and R² values were calculated for every run and ranged from 88% to 100% and >.98, respectively.

The linear operating range of the green crab qPCR assay developed here; mean Ct of 10 replicates is plotted against gBlock dilution factor for three independent qPCR runs

Field testing

Pilot testing of this assay (prior to full optimization and characterization, as described in the methods) successfully detected green crab from plankton samples from Hobart and Melbourne, two ports recorded as positive for green crab based on historical records (Table ). In contrast, no evidence of green crab detection was observed in plankton samples from Brisbane or Gladstone, both of which have had no historical reports of green crab presence (Table ). All extraction blanks and no-template qPCR controls showed no amplification.

Green crab qPCR results from plankton samples from four sites in Australia

Site	n plankton tows	n qPCR replicates	n tows with ≥1 positive qPCR replicate	Historical green crab?
Brisbane	148	444	0	No
Gladstone	260	780	0	No
Hobart	60	180	20	Yes (Wells, McDonald, & Darragh, )
Melbourne	288	864	23	Yes (Wells et al., )

Note:

Each plankton tow was assayed using three qPCR replicates. The number of plankton tows that returned a Ct value at one or more qPCR replicates is given and whether there are historical green crab records for each location.

For the eDNA water samples collected from Barkley Sound, BC, Canada, all sites were positive for green crab (Figure ). As is common with eDNA testing (e.g., Harper et al. ), not all six field replicates within each site were positive, and within each field replicate not all qPCR triplicates were positive (Table ). To summarize: 11/30 field samples generated a Ct in all three qPCR replicates; 12/30 field samples generated a Ct in 2/3 replicates; 5/30 field samples generated a Ct in 1/3 qPCR replicates; and 2/30 field samples generated no Ct in all three replicates. For calculations of average Ct, undetermined technical or field replicates (i.e., no Ct) were omitted from analysis. Average Ct values (of three technical replicates) from all 28 positive field samples across sites ranged from 34 to 38 (Table ; Figure ), which were below or equal to the assay's LOD (Ct ≤ 38). Average Ct values were usually higher than the assay's LOQ (Ct > 35), the exception being Pipestem Inlet, which had the lowest Ct values observed and is known from historical monitoring data to have a high abundance of green crab. All field blanks, extraction blanks, and no-template qPCR controls showed no amplification.

Average Ct values from green crab qPCR assay for each of six field sample replicates at five sampling sites in British Columbia, Canada, where green crab are known to occur (from trapping data). Each data point represents the average of three qPCR technical replicates; replicates with no Ct were omitted. Error bars represent standard deviation

qPCR results for six eDNA sample field replicates from five sites in British Columbia, Canada

[Table omitted. See PDF]

Note:

Each field replicate was tested using three qPCR replicates; the average of the three replicates is given (with "No Ct" replicates omitted).

Inhibition testing

In the plankton samples collected from Australia, PCR inhibition occurred in some samples from each port. Almost half (n = 128) of the samples from Gladstone had moderate to very high inhibition (scale factor: 5⁻¹ × 10⁴), and 14 samples from Hobart displayed inhibition (scale factor: 5⁻¹ × 10⁵). In contrast, the eDNA water samples collected from British Columbia, Canada, showed no evidence of inhibition at any site. The gBlock spike-in control test resulted in no significant shift in Ct values relative to expected for the selected samples from each of the sites (p = .99) measured by one-way ANOVA with Brown–Forsythe test.

DISCUSSION

We developed a novel qPCR assay for detection of European green crab and conducted initial pilot testing on plankton samples from Australia. Given promising results of this testing, whereby green crab detections were obtained using the assay at sites known to contain green crab and not from sites without, the assay was then tested in Canada for eDNA detection. At this point, we conducted a more thorough assay optimization, assessed its analytical performance characteristics, and successfully trialed it on eDNA collected from filtered water at 6 sites in western Canada known to be invaded by green crab.

The assay's LOQ was determined to be 1 pg/L (gBlock dilution of 10^–10), which is within the range reported for other qPCR assays designed for high-sensitivity species-specific detection, although caution is needed in interpreting comparisons between studies as there are no standardized definitions for LOQ (or LOD) currently in use. Harper et al. () report an LOQ of 0.1 pg/μl for a signal crayfish (Pacifastacus leniusculus; Dana, 1852) assay. Similarly, measures of sensitivity indicating an LOD of 0.2 copies/μl for the green crab assay developed here are consistent with other eDNA studies detecting AIS. Geerts et al. () reported a detection limit of 2.5 ng/μl for Procambarus clarkii (Girard, 1852), whereas Xia et al. () described an LOD of 1 × 10⁻⁷ ng/μl for the freshwater mussel Limnoperna fortune (Dunker 1857).

Most of the positive field samples returned average Ct values between the assay's limit of quantification and limit of detection. This is not problematic as the purpose of the assay is primarily for qualitative use, to determine presence or absence of the target rather than to quantify target DNA concentration. These presence/absence data can be used for estimating the proportion of positive sites using site occupancy models to account for imperfect detection probabilities (Schmidt, Kéry, Ursenbacher, Hyman, & Collins, ).

Over the last decade, eDNA has shifted from a new tool used predominantly to detect microbial taxa to one used to detect macro-organisms for purposes related to ecosystem preservation (e.g., invasive species detection) and conservation of species (Thomsen & Willerslev, ). Environmental DNA applications necessarily push the limits of qPCR-based detection, and this emphasizes the need for impeccably optimized procedures. Beyond stringent analytical validation of the assay itself, several other factors need to be optimized to maximize eDNA-based detection which are yet to be investigated for green crab. Currently, the average volume of water collected for eDNA isolation by filtration ranges from 0.5 to 2 L (Tsuji, Takahara, Doi, Shibata, & Yamanaka, ; Rees et al., ) but can be more depending on the study (Goldberg et al., ). Our results demonstrate success detecting green crab from just 1 L of water collected; however, it is possible that increasing the volume of water filtered per sample may generate lower Ct values (i.e., higher detectability). We used three qPCR replicates per sample. Much higher levels of replication at the qPCR step have been recommended (Veldhoen et al., ) and are expected to increase detection probability, although this inevitably increases the cost per sample. Improving detection likelihood by increasing replication presents a conundrum for end users of eDNA assays about cost-benefit trade-offs between detection probabilities and financial investment. Using alternative methodologies to increase sensitivity has been explored by a few studies including nested qPCR (Stoeckle, Mishu, & Charlop-Powers, ) and metabarcoding (Lacoursière-Roussel et al. ; Westfall et al. ).

The majority of eDNA research has been in freshwater (Cristescu & Hebert, ). While qPCR assays targeting eDNA of marine species are beginning to appear in the literature (Medlin & Orozco, ; Rees et al., ; Wood, Zaiko, Richter, Inglis, & Pochon, ), there is a need to better understand variation in eDNA signal in the marine environment. Although expectations are that eDNA could travel far from its origin in a highly dynamic fluid environment like the ocean, new research suggests a surprising level of site fidelity to the marine eDNA signal. eDNA persistence in the marine environment can be limited to approximately 48 hr (Collins et al., ), which may limit the extent to which signal originating from one site is detectable at another. The potential for eDNA sampling to generate site-specific signals in the marine environment is exemplified by O'Donnell et al. (), who detected differences among communities separated by less than 100 m. Another study showed the marine eDNA signal to be endogenous to the site and water mass sampled, rather than changing with each tidal cycle (Kelly, Gallego, & Jacobs-palmer, ). These studies support the potential of eDNA to be useful for obtaining site-specific detection information on the presence of green crab, but this requires further experimental work to confirm.

It is suggested that eDNA detection of invertebrates is hindered by low DNA shedding rates of shelled organisms. This may decrease detection probabilities when targeting trace eDNA for these taxa, including green crab, for any given level of assay sensitivity compared to taxa with higher shedding rates (e.g., finfish). Forsström and Vasemägi () found the DNA release rate for Harris crab Rhithropanopeus harrisii (Gould, 1941) in aquaria was highest when animals were most active and when activity and stress declined the amount of detectable target eDNA dropped drastically (an 89.9% decrease after 2 days). Dunn et al. () noted factors such as reproductive biology can influence eDNA detection of invertebrates. In addition, Mächler et al. () found mixed results with the detection of macroinvertebrates using eDNA and suggested that low amounts of DNA released by these animals is a considerable challenge. As such, for some species, using plankton as the sample matrix and aiming to capture larvae may be the best approach. Results from Australian samples suggest this is a suitable approach for green crab. However, it requires an understanding of spawn timing and, ideally, to take into account vertical migration patterns of larvae in the water column (Queiroga, ; Queiroga, Moksnes, & Meireles, ) to optimize the chance of detection. Overall, a better understanding of assay diagnostic sensitivity and specificity, sampling methods, sample numbers, and optimum sampling season is important for interpreting surveys.

For characterizing assay performance, we scored all replicates that returned a Ct value “positive” and included them in downstream analyses accordingly. It is important to recognize this was to meet the analytical goals of this paper and not to comment on how detections in the field should be interpreted. For this, validation of eDNA detections against another method is needed, with care being taken to understand and account for test conditional dependence, which is being pursued by the authors in another study. It is also important to recognize that the assay validation presented here applies to the Pacific coast of North America and Southern and Eastern Australia regions only. COI haplotype diversity in C. maenas is known to be high globally, and some invaded regions are home to multiple, evolutionarily “distant” haplotypes (Darling & Tepolt, ). Although this is not the case for the Pacific Northwest of North America, multiple COI haplotypes are known to occur in Australia (Darling & Tepolt, ; Hatzenbuhler et al., ). While sequence data were not available for all haplotypes when developing the assay, it was shown to detect DNA from green crab from as many localities throughout the region considered. However, before this assay is implemented in another geographic region, additional validation is needed to ensure that it successfully detects all haplotypes of green crab in the region and does not provide positives in response to native species.

The need for standardized assay validation will become critical as eDNA approaches grow into becoming a fundamental management tool for invasive species, species at risk and indicator taxa for environmental monitoring (Goldberg et al., ). There is substantial disparity among validation studies with respect to how LOD and LOQ are calculated, number of replicates used, number of samples included in the specificity panel, number and type of negative controls, and what material is used as a positive control. Here, we provide a model of an eDNA-qPCR assay validation approach that is consistent with standards applied in aquatic animal health-related fields and is consistent with Australia's Guidelines for development and validation of assays for marine pests (Department of Agriculture and Water Resources, ). To summarize, we conducted in silico assessment of developed primers, rigorously optimized reaction conditions, tested specificity both in silico and experimentally, developed an artificial positive control, determined limits of detection and quantification using a stringent definition of these two measures, and conducted a field pilot to confirm potential of the assay to amplify DNA from filtered seawater and plankton. We hope that in addition to providing a useful tool for moving toward forensic detection of invasive green crab, this study will also contribute toward a standardized approach for the development and validation of qPCR assays for use on environmental samples.

ACKNOWLEDGMENTS

The Canadian component of this work was funded by Fisheries and Oceans Canada's Program for Aquaculture Regulatory Research (PARR). Canadian authors wish to thank: Kara Aschenbrenner for sampling and molecular laboratory work; Tom Therriault for field sampling and site selection; Scott Gilmore for specificity testing advice; Geoff Lowe and Laura Hawley for guidance on following the OIE analytical validation pathway and gBlock development; and Aidan Goodall for laboratory work assistance. The Australian component of this work was funded by the Adelaide and Mt Lofty Ranges NRM Board as well as the Australian Government Department of Agriculture and National Biosecurity Committee. Australian authors are grateful to Andrew Irving, Graham Hooper, Fred Gurgel, and John Lewis for the provision of samples and Herdina, Alan McKay, Kathy Ophel-Keller and Teresa Mammone for laboratory assistance and advice. We are grateful to Michael Sierp, John Gilliland, Vic Neverauskas, Mae Allan, and Tim Carew for assistance and support. Thank you to The Root Disease Testing Laboratory (SARDI) for extractions, testing and automated reporting. Mandee Theil prepared and processed samples with assistance from Ian Moody, Leonardo Mantilla, and Nhu Y. Lieu.

AUTHOR CONTRIBUTIONS

CLA and MRD conceptualized the idea and acquired funding for the project. NJB designed the qPCR assay. CLA, KMW, KHW, DGD, and NJB designed the detailed methodology. KMW and CLA conducted field sampling. L-MDR and DGD conducted the laboratory work and analyzed the data. L-MDR wrote the first draft of the manuscript and CLA and MRD edited and provided feedback. All authors gave final approval for publication.

DATA AVAILABILITY STATEMENT

The raw data underlying the main results of the study can be found in the dryad digital repository (Roux et al., ), as well as on servers at the Pacific Biological Station, Fisheries and Oceans Canada.