A workflow for the relative quantification of multiple fish species from oceanic water samples using environmental DNA (eDNA) to support large-scale fishery surveys

While the number of published marine studies using environmental DNA (eDNA) has increased substantially in recent years, marine fish surveys are still scarce. To examine the potential for eDNA to support marine fisheries monitoring surveys, we optimized an eDNA isolation method, developed a multispecies assay and tested it on eDNA samples collected along the Pacific coast of the United States. Four commercial DNA extraction kits that exploit the capability of the nucleic acids binding a solid phase (two using a silica matrix and two magnetic beads) as well an organic separation method were tested. A species-specific multiplex qPCR assay was developed and tested to simultaneously target Pacific hake (Merluccius productus), Pacific lamprey (Entosphenus tridentatus) and eulachon (Thaleichthys pacificus). The specificity of the assay was tested in silico, in vitro and in natura. Environmental DNA isolation using phenol:chloroform:isoamyl purification with a phase lock was optimized and yielded the highest amount of total and target DNA and was used to extract 46 marine water samples for the detection of the three species of interest. The multiplex qPCR assay used in the quantification process was also optimized to provide convenience and accuracy. Pacific hake was present in 44% of the eDNA samples while the other two species were absent. Here, we present a complete workflow for the simultaneous detection and quantification of multiple marine fish species using eDNA. This workflow supports large-scale at-sea sampling efforts with preservation at ambient temperatures and has demonstrated DNA extraction efficiency and reliability. The multiplex qPCR assay is shown to be sensitive and specific for the purposes of simultaneously monitoring the relative abundance of multiple targeted fish species.


Introduction
The use of environmental DNA (eDNA) from water samples either to detect specific targets or characterize community composition is becoming a regular monitoring tool used in aquatic Environmental DNA can be used to infer the relative abundance of fish by developing species-specific primers and hydrolysis probes using quantitative PCR or qPCR [6,19,20]. The ultimate goal of this study was to develop an optimized qPCR protocol to be used to quantify commercially important species during marine surveys. The main target species chosen to test the methods was Merluccius productus. In order to maximize the information obtained from the samples collected with minimal additional effort, a multi-species qPCR assay can be used. To test the ease with which a multiplex PCR could be developed, two endangered, anadromous fish species often found in conjunction with M. productus along the Pacific Coast of the United States (U.S.) were also targeted: Entosphenus tridentatus (Pacific lamprey; Richardson, 1836) and Thaleichthys pacificus (eulachon; Richardson, 1836). Entosphenus tridentatus is a species of concern and in decline or extirpated in many areas of Northwest of the U.S. [21]. Little is known about the migration patterns of the oceanic life stage of Thaleichthys pacificus and the southern distinct population segment is classified as threatened under the Endangered Species Act (U.S.) [22].
Here, an optimized eDNA sampling and laboratory assay protocol for use in marine fisheries assessment is characterized. Five DNA isolation methods were tested to maximize the recovery of target species DNA and a sensitive, specific, and accurate multispecies quantitative assay was developed.

Materials and methods
The National Oceanographic and Atmospheric Administration (NOAA) Fisheries and Fisheries and Oceans Canada (DFO) undertake biennial coastwide Integrated ecosystem and Pacific hake acoustic and trawl surveys from Point Conception in California (United States) to Dixon Entrance in the north of British Columbia (Canada). Typically, east-west transects (ranging from 50-150 m isobath) are performed every 10 nautical miles over a three-month period of time (map in S1 File). Sample collection methods were approved by NOAA, NMFS.

Sample collection
Sea water samples were collected aboard the NOAA ship Bell M. Shimada during the course of an Integrated ecosystem and Pacific hake acoustic and trawl survey in the summer of 2019. Among the 1873 samples collected, 20 were collected off the coast of Cape Flattery, Washington (US) (48 23.88 N, 125 28.39 W) specifically for methods evaluation, hereafter 'extraction test samples'. A CTD rosette with ten Niskin bottles (10 L) was deployed at a depth of 120 m at a site where a mixed aggregation of fish had been observed with the acoustic echo-sounders and caught by trawling at a depth of 103 m over a 175 m bottom. All ten Niskin bottles were triggered simultaneously and two replicate samples of 2.5 L of seawater were collected on deck from each of the bottles (n = 20). The rest of the samples from the survey ('survey samples', n = 1853) were collected along the voyage transects at varying depths, triggering two Niskin bottles on the rosette simultaneously at each sampling depth. One sample of 2.5 L was collected from each of these Niskin bottles. Negative sampling controls were collected routinely by filtering 2.5 L of distilled water from either the onboard evaporator (in the engine room) or from jugs of distilled water brought from the laboratory for this purpose. Three of the survey samples were used to test an organic extraction phase lock method -hereafter 'phase lock subset'and 43 survey samples plus 10 sampling negative controls -hereafter 'validation subset'-were used for final assay assessment (Table 1 and 'samples details & qPCR results' in S2 File). The rest of the survey samples were beyond the scope of this analysis and will be analyzed in a future publication.
All water samples (extraction and both subsets in survey samples) were immediately filtered using reusable 500 ml filter cups (KP-47W, Advantec1, Japan) to reduce storage volume of supplies and subsequent waste (normally burned at sea or packed back to shore). The cups were fit with 47 mm diameter mixed cellulose ester (MCE) sterile filters (composed of a combination of cellulose nitrate and cellulose acetate) with a 1 μm pore size (Advantec1, Japan) and supported on a triple-port manifold attached to a 1/8 HP portable diaphragm vacuum pump at -8 to -12 Hg (Gast, Benton Harbor, MI, U.S). Four filters from the extraction test subset were stored dry and frozen (-80˚C) in individual, single use Whirl-Pak (Whirl-Pak, Madison, WI, U.S.) bags and 16 samples in the extraction test sample set and all survey samples were submerged in 2 ml of lysis buffer [9] in 5 mL LoBind Eppendorf (Eppendorf, Hamburg, Germany) tubes and stored at room temperature. The filter cups and forceps used to handle filters were washed with 0.5% sodium hypochlorite and rinsed in distilled water obtained from the onboard evaporator before assemblage. The benches and other laboratory surfaces, as well as the spigots of the Niskin bottles were also wiped with sodium hyplochlorite prior to sample collection. Gloves were changed frequently and a fresh pair was always used when assembling filter cups (S3 File).

PLOS ONE
Pasadena, CA, US), hereafter 'AGF' (details and modifications of each method in supporting S1 File). The four frozen filters were extracted using the DNeasy PowerWater kit (Qiagen, Hilden, Germany), hereafter 'PW'. The corresponding sampling negative control was extracted using the AGF method. Regardless of extraction protocol, DNA was eluted in a final volume 100 μl of Tris-lowEDTA (TlowE). Low retention tubes and plates were used at all steps. DNA quality for all 20 samples in the extraction test was evaluated on a TapeStation 2200 system (Agilent, Santa Clara, CA, US). The total amount of DNA from all 20 samples and the sampling negative control was quantified using a Qubit1 3.0 fluorometer (ThermoFisher Scientific, Waltham, MA, U.S.).
To increase the throughput of the PCI DNA extraction and decrease repetitive pipetting for the anticipated 1800+ samples collected, the use of silicone vacuum grease (Dow Corning, Midland, MI, US) as a phase lock [24] between the organic and aqueous phases was evaluated (see S1 File) with the phase-lock subset of samples. The phase lock creates a stable interphase layer allowing easy decanting of the aqueous layer into the next extraction step.
The validation subset of samples was extracted using the PCI method with the phase lock modification (Fig 1). In order to avoid any co-precipitation of salts, the final PCI method was slightly modified by conducting the precipitation with isopropanol at room temperature and adding a second 70% ethanol wash (final DNA extraction method in S1 and S3 Files; additional experiments in S1 and S2 Files).
DNA from reference samples of M. productus, T. pacificus and E. tridentatus and related species used for the specificity tests was isolated in a separate laboratory using a DNeasy Blood and Tissue Kit extraction kit (Qiagen, Hilden, Germany) and eluted in 200 μl of TlowE.

Species-specific assay
Species-specific primers and hydrolysis probes for the cytochrome oxidase subunit 1 (COI) and 12S rRNA (12S) genes that targeted the three species of interest-Merluccius productus, Entosphenus tridentatus and Thaleichthys pacificus were developed (details in Table 2 and S1 File). Specificity was accounted for in silico using primerTree [25], an R package that uses the Primer BLAST search [26] against all sequences available in GenBank1 [27]. Results were filtered to narrow down to one or zero mismatches in each primer (S1 File). Non-target species found to match each primer set using the primerTree in silico specificity test were verified as either extinct or not co-occurring with the species of interest (S1 File), the latter being the case of the Merluccius gayi (Peruvian hake) and the Entosphenus similis (Klamath river lamprey). The probes themselves conferred additional specificity (S2 File). Reporter dyes on the hydrolysis probes were selected for optimal multiplex performance on an Applied Biosystems Quant-Studio 6 Flex Real-Time system (ThermoFisher Scientific, Waltham, MA, US).
Target DNA from the extraction test subset and the sampling negative control (distilled water) was quantified using a multiplex TaqMan assay. All amplifications were carried out in triplicate in 10 μl reactions containing 1 × TaqMan 1 Environmental Master Mix 2.0 (Thermo-Fisher Scientific, Waltham, MA US), forward and reverse primers (0.9 μM), hydrolysis probes (0.2 μM), and 2 μl of genomic DNA standard or eDNA. Cycling conditions were: initial denaturation step at 95˚C for 10 min, followed by 45 cycles of 95˚C for 15 s and 60˚C for 1 min. Seven 10-fold serial dilutions starting at ca. 1 ng μl -1 were used to build the standard curve. The standard curves starting points were quantified on each run using a Qubit dsDNA HS kit on the Qubit1 3.0 fluorometer and ranged from 98-103 pg μl -1 . The amplicons were sequenced on a 3500xL Genetic Analyzer using a BigDye terminator v3.1 cycle sequencing kit (ThermoFisher Scientific, Waltham, MA, US) to confirm that the signal obtained in the multiplex TaqMan assay was from the target species.

Inhibition testing
To expand the multiplexing capabilities and minimize PCR inhibition, all qPCR tests following the initial development of the assays were performed as above using the TaqPath ProAmp Multiplex Master Mix (ThermoFisher Scientific, Waltham, MA, U.S.) (see S1 File). A synthetic, linear, double-stranded oligonucleotide gBlocks1 Gene Fragments, herein 'gBlocks' (Integrated DNA Technologies, Coralville, IA, US), of 130 bp was synthesized from the D-loop sequence (GenBank Accession GU139002) of the extinct Euryapteryx curtus (broad-billed moa) as an exogenous internal positive control (IPC) of the amplification efficiency. A set of primers and a probe ( Table 2) were designed and synthesized for the IPC to amplify a fragment slightly longer than those of the target species to avoid a greater efficiency than the three fish species-specific assays [28]. The purpose of the IPC was to assess amplification inhibition rates without affecting the amplification of the target, thus avoiding false negative results (absence) of the species of interest. The inhibition cut-off value was set arbitrarily as the difference between the sample IPC and the non-template control average IPC values by rounding up the deviation found among the non-template control IPC replicates (i.e., ΔCq = mean C q eDNA sample-mean C q ntc ; inhibition = ΔC q > @C q ntc ). Samples for which the IPC failed to amplify were given an arbitrary value of 45 as if they had amplified in the last cycle.

Standards multiplexing validation
To enhance accuracy in the quantification and simplify the qPCR set up, gBlocks were also manufactured to provide standards for the target species [29]. The gBlocks comprise specific 130 base pairs (bp) fragments of the genes 12S for M. productus and COI for T. pacificus and E. tridentatus (S1 File). The performance of the standards from genomic DNA and the gBlocks were compared in the multiplex reaction and adding 1 μl of the IPC at a medium concentration range (1000 copies μl -1 ) in the master mix [30]. To build a standard curve, six 10-fold dilutions of the gBlocks were included starting from 10 5 copies down to 1 copy. In addition, M. productus-specific gBlocks were run in a specific singleplex to validate performance in the multiplex reaction individually and combined with the gBlocks of the two other species. Hereafter, target DNA concentration is expressed in number of copies per microliter.

Validation of the triple species-specific assay
The multiplex qPCR assay was examined in vitro for robustness, accuracy and specificity by testing genomic DNA from multiple individuals for each species, including 32 M. productus collected along the US West Coast and seven from the Salish Sea (described as two discrete populations), 33 E. tridentatus from along the coast of Washington and Oregon (U.S), and 45 T. pacificus samples (five individuals from nine different populations) ranging from northern British Columbia (Canada) to northern California (U.S.). The specificity was also tested by testing 13 DNA samples of four species in the order Gadiformes and one Lampetra ayresii (Western River lamprey) ( Table 3). The DNA samples were run in duplicate at ca. 3 ng μl -1 and the combined gBlocks standards in triplicate. To corroborate the specificity of the assay the sequences of dubious PCR products were sequenced on a 3500xL Genetic Analyzer. The validation subset of samples was quantified using the multiplex assay. Of the 43 survey eDNA samples, 75% were selected from sites or depths with high levels of chlorophyll a fluorescence (CTD readings>1 μg L -1 ) to evaluate whether chlorophyll a (common in surface water samples) could directly or indirectly inhibit the amplification. Samples exhibiting signs of qPCR inhibition were purified using a OneStep™ PCR Inhibitor Removal Kit (Zymo Research, Irvine, CA, U.S.) removal column and the qPCR was repeated.

DNA isolation methods comparison
All negative controls from the DNA extraction methods comparison were negative for total and target DNA quantification. Total and target DNA was measured in each sample tested from the extraction test subset (Fig 2). The average of total DNA isolated ranged from 0.29 ng μl -1 using ADV to 16.13 ng μl -1 obtained from PCI extractions (Fig 2 and S2 File). Target DNA (M. productus) ranged from 1.03 × 10 −5 ng μl -1 in the case of MA-PW to 5.17 × 10 −4 ng μl -1 for PCI extraction. Two extraction methods (MA-PW and ADV) showed mean C q values for M. productus beyond the lowest detectable standard, set as the limit of detection (LOD) (graph and results in S1 and S2 Files, respectively). The other three methods yielded detectable PLOS ONE amounts of M. productus DNA. The C q standard deviation of the three PCR replicates varied from 0.28 for PCI method to 1.66 for ADV (Figure in S1 File). The integrity of the DNA was variable between methods when visualized on the TapeStation (S1 File), with MA-PW and ADV being undetectable. PW generated smaller DNA fragments, most likely due to the mechanical lysis process. Regardless of the DNA extraction method, the ratio of M. productus DNA present was ca. 0.001% of the total DNA. Resulting sequences from the amplicons matched M. productus. In the phase lock subset of samples, the addition of the phase lock to the PCI extraction method yielded both more total and target DNA (Fig 2); this result was also supported by supplemental experiments in S1 and S2 Files.

Specificity and reproducibility
None of the non-template controls in the qPCR reactions showed amplification. All reference samples (M. productus n = 40, T. pacificus n = 45 and E. tridentatus n = 34) from the in vitro validation showed amplification with their respective probes (Table 3 and Fig 3). However, there was M. productus signal at low levels on 13 of 34 E. tridentatus samples with only two being at more than 1 copy μl -1 . The M. productus primer-probe set does not amplify Boreogadus saida (Arctic cod), Microgradus proximus (Pacific tomcod), Gadus chalcogrammus (Alaska pollock), and Gadus macrocephalus (Pacific cod). Contamination of reference fish tissues was inferred, which most likely occurred at the time of collection for samples that were collected concurrently. However, the majority of these non-target observations were < 1 copy μl -1 (Table 3). More specifically, M. productus signal was obtained in one G. macrocephalus and three G. chalcogrammus samples. These were confirmed via Sanger sequencing of the amplicon as M. productus sequences (trimmed sequences of �25 bp were attained and matched M. productus with a 100% identity) as opposed to a false positive reaction of the assay to G. chalcogrammus or macrocephalus DNA. One M. proximus out of the four also showed

PLOS ONE
some amplification for T. pacificus that was identified as such after sequencing, attributable to insufficient sterility measures at fishing trawl collection (see Discussion). All T. pacificus samples from Klamath and Sandy rivers and two from Elwha River showed amplification for lamprey that was found to be E. tridentatus contamination, attributable to co-occurrence in high numbers in those specific rivers (Gustafson pers. comm.). Hence, false positive and true negative rates initially close to 0 and 1 are fixed when the amplicon originated in the qPCR was sequenced (Fig 3 and S2 File).

Validation of the triple species-specific assay
Good efficiency of the qPCR (98.8-109%) and standard replication were observed since all qPCR run parameters fell within slope values of -3.35<m<-3.12 and r 2 �0.99. The limit of detection (LOD) of all three assays at 95% or greater detection is below one copy μl -1 but could not be determined because no lower standards were tested, which corresponded to 10 fg μl -1 for the gDNA standards. Values of the C q for the gBlocks followed the expected shift for tenfold dilutions of 3.3 cycles (Fig 4). Eighty percent of the sampling negative controls from the validation subset had detectable signal of M. productus ranging from 0.2 to 2.2 copies μl -1, which sets the limit of quantification (LOQ). Some lamprey signal at low levels (24 and 2 copies μl -1 on the 1 and 0.1 ng μl -1 standards, respectively) was found in the T. pacificus genomic DNA standards in agreement with the same sample used for the in vitro validation above. Concentration values of M. productus found in all eDNA samples of this study (n = 50) ranged from 0 to 73.36 copies μl -1 and 44.64% were above the LOQ (n = 25) (S2 File). T. pacificus was also found in two samples out of the four biological replicates of the extraction test subset at <1 copy μl -1 from the site (120 m deep out of Cape Flattery) and absent in the validation subset. The finding of negligible amounts of T. pacificus among the extraction test subset is supported by the trawl survey conducted 30 h prior at this site, which documented 55.4% M. productus, 44.1% G. chalcogrammus, 0.2% T. pacificus, 0.2% Oncorhynchus gorbuschaI (Pink salmon), 0.1% Oncorhynchus tshawytscha (King Salmon), and 0.02% Clupea pallasii (Pacific herring) in a total of 935.3 kg. However, no formal quantitative analysis was attempted with these data because of the collection time differential.

PLOS ONE
The amplification of the species-specific gBlocks standards individually and in the same well showed negligible C q variation (Fig 4 and S2 File). Also, the addition of the other species primers and probes in the multiplex did not affect the amplification of the M. productus gBlocks when compared to M. productus gBlocks run individually and in a single reaction. Substantial difference in copy number was found for the genomic DNA standards (starting at 1 ng μl -1 ) depending on the species genome size (S2 File). The combination of the three gBlocks standards did not affect the reliability of the assay since it did not impact the exponential part of the amplification curve, nor the ΔRn of the plateau. However, it delayed the reach of the plateau phase most likely due to competition for the reagents in the master mix (Fig 5).
The IPC value for the highest gBlocks standards (100,000 copies), ran individually or combined, influenced the IPC amplification, most likely due to competition and exhaustion of reagents (Fig 5). The IPC amplification of the non-template controls was robust, with a standard deviation that ranged from 0.06 to 0.24 among PCR reactions (three to nine replicates,  PLOS ONE respectively). The threshold for the identification of inhibition (ΔC q > @C q ntc ) was thus set at 0.5. According to this parameter 17.85% of all eDNA samples (10 of 56 samples) showed inhibition with no apparent correlation with the chlorophyll a level (fluorescence values 0.05 to 8.98). The ΔCq of the IPC for the four most inhibited samples dropped from 17.19 to 0.71 after purification with inhibitor removal columns. None of the sampling (n = 10) and extraction negative controls (n = 3) showed signs of inhibition, hence the reagents in the PCI extraction method are ruled out as the source of inhibition. In addition, the vacuum grease used for the phase-lock is not a source of inhibitors since none of the samples of the phase-lock evaluation in S1 File showed signs of inhibition.

Discussion
Environmental DNA monitoring holds promise as a non-invasive tool that can be included in the stock monitoring toolset [4]. The number of marine monitoring surveys using eDNA is rapidly increasing globally. There is a need for optimizing, validating and standardizing the methodology around eDNA for large-scale oceanic studies especially given the constraints of variable vessels with limited storage space and the desire for high sampling effort. DNA extraction protocols must maximize the recovery of good quality and high quantity DNA. The general trend for fish-targeted eDNA studies is to collect 1 or 2 L of surface water and capture eDNA on 0.7 μm GF filters, followed by extraction with the DNeasy Blood and Tissue Kit or PowerWater DNA Isolation Kit (Qiagen) of frozen filters, which has shown to yield high-quality eDNA and reliable results [31]. A larger volume of water (2.5 L) and a larger filter pore size (1 μm) were used here to maximize the DNA yield while avoiding filter clogging. The Longmire's lysis buffer (see S3 File) is an inexpensive, safe and reliable lysis medium, favoring DNA preservation for prolonged periods of storage at ambient temperatures. It is also non-hazardous and convenient for transport. The 5 ml tube storage format for filters was chosen to facilitate the filter immersion in the buffer at sea with less manipulation by collectors and allows intube extraction of all suspended and filter bound materials. Commercial kits have their own proprietary lysis buffer to initiate the DNA extraction, but the samples used in this study were already stored in Longmire's buffer. This required an adjustment of the original kit protocols (S1 File) that most likely affected the ideal performance of the DNA purification, since both the amount of DNA available for purification and the concentration of salts determining the binding capacity of the magnetics beads was likely not optimal for the performance of the commercial kit. Four commercial kits that exploit the capability of the nucleic acids binding a solid phase (either silica matrix or magnetic beads) in order to isolate the DNA, as well as the

PLOS ONE
phenol:chloroform:isoamyl (PCI) separation method were evaluated in this study. The DNeasy Blood and Tissue kit (Qiagen) had been previously tested in our laboratory starting with filters stored in Longmire's buffer with poorer results than the standard PCI (results not shown) and thus was disregarded in the search for a more cost-effective, reduced-labor method that yielded sufficient or more DNA. The best performing method in terms of total DNA and target DNA yield was the PCI method (16.13 and 5.17 × 10 −4 ng μl -1 respectively), followed by the DNeasy PowerWater with frozen filters (5.67 and 9.02 × 10 −5 ng μl -1 ) and GenFind v.3 (3.55 and 1.77 × 10 −4 ng μl -1 ) (more in S2 File). The PCI method is the cheapest among all methods tested (as far as supply and reagents costs) and less than half the price of PW (details in S2 File). The integrity of the genomic DNA extracted with the PW kit was compromised, likely sheared due to the mechanical lysis procedure. Although DNA integrity could not be analyzed for MagAttract PowerWater DNA due to low yield, it is expected to be affected in the same way.
The PCI method is the most labor intensive and chemically hazardous of all methods tested. However, it is the most efficient method for the purposes of targeting specific species in the open ocean as described here. The PCI behaved less stochastically (showed the lowest C q standard deviation) between biological and technical replicates, most likely due to the higher DNA concentration. As expected, the target DNA yield is a small fraction of the total DNA, hence the importance of enhancing the DNA concentration. The ratio of target DNA was not affected by the extraction method used indicating that the treatment does not affect the proportion of DNA of different taxa. The addition of the phase lock grease, although laborious, generates a higher yield and throughput because the aqueous layer containing the DNA can be quickly decanted into the next tube, reducing pipetting variation during recovery of the aqueous layer and exposure to fumes from the organic reagents. Mukhopadhyay and Roth [32] recovered 20% more total DNA with their phase-lock protocol; however, the improvement could be very variable between samples, replicates, number of steps in the DNA purification method, DNA concentration and the expertise of the person performing the protocol.
The exogenous internal control provides confidence on the identification of false negative or reduced quantification caused by inhibition and enables identification of samples that need further treatment. Equal levels of inhibition were observed in biological replicates showing consistency in the inhibition co-purification. Inhibition was observed in 18% of the eDNA samples. The origin was not ascertained here. However, this issue was circumvented by purifying the DNA with inhibitor removal columns with an additional cost per sample (S1 File). The shift of ΔC q � 3 is often applied in the eDNA field [33-35] as the cut-off to account for inhibition. That level is disregarded here and anticipated to skew the quantification results with almost a log-loss in copy number. The use of an inhibitor-tolerant master mix should be favored to avoid bias caused by purification or dilution treatments [36].
The use of multiple synthetic oligonucleotides (gBlocks) in the same well as standards of the qPCR enabled accurate quantification of target copy number instead of an inference from genomic DNA standards. It also reduced considerably the number of pipetting steps and standard preparation time. Synthetic oligonucleotides are a good substitute for genetic laboratories with no plasmid cloning capabilities. Merluccius productus contamination at very low levels was found in the sampling negative controls. Although sparse replication and low amplification was observed, it signals underlying contamination at collection. This is unsurprising given the operations of the survey typically target this species and daily trawls brought large amounts on board. Stricter contamination measures, further tests and improvements should be pondered in future sampling campaigns. Interspecific contamination was also found in the reference samples used to test the specificity of the triplex assay. This was most likely due to insufficiently clean collection measures at fishing trawl campaigns in which the specimens are handled and subsampled swiftly. These findings show the sensitivity of the assays and call for future reference material to be decontaminated prior to the DNA extraction by submerging the samples in bleach [37]. Despite the above-mentioned findings, the triple assay was found to be specific, confirmed by the fact that the extraction test subset of samples was taken at a site where a mixed school of M. productus and G. chalcogrammus was observed. Moreover, negligible levels of T. pacificus were found and no lamprey was detected for the extraction test subset as expected from the trawl observations. M. productus was present in 44% of the survey eDNA samples analyzed (validation subset samples) while the other two species were absent in agreement with the total trawl catch in de Blois [38].
In summary, we describe an approach for the quantification of multiple species of fish in oceanic water that can be used in large fishery surveys. In addition to the logistics for collection and filtration used, the use of synthetic oligonucleotides as qPCR standards in a multiplex in combination with an exogenous internal positive amplification control made this assay practical, quantitative, sensitive and reliable. This represents a validated method for the isolation and analysis of environmental DNA (Fig 1 and S3 File) with promising prospects for fish monitoring.