The deep ocean is the largest biome on Earth and faces increasing anthropogenic pressures from climate change and commercial fisheries. Our ability to sustainably manage this expansive habitat is impeded by our poor understanding of its inhabitants and by the difficulties in surveying and monitoring these areas. Environmental DNA (eDNA) metabarcoding has great potential to improve our understanding of this region and to facilitate monitoring across a broad range of taxa. Here, we evaluate two eDNA sampling protocols and seven primer sets for elucidating fish diversity from deep sea water samples. We found that deep sea water samples (> 1400 m depth) had significantly lower DNA concentrations than surface or mid-depth samples necessitating a refined protocol with a larger sampling volume. We recovered significantly more DNA in large volume water samples (1.5 L) filtered at sea compared to small volume samples (250 mL) held for lab filtration. Furthermore, the number of unique sequences (exact sequence variants; ESVs) recovered per sample was higher in large volume samples. Since the number of ESVs recovered from large volume samples was less variable and consistently high, we recommend the larger volumes when sampling water from the deep ocean. We also identified three primer sets which detected the most fish taxa but recommend using multiple markers due the variability in detection probabilities and taxonomic resolution among fishes for each primer set. Overall, fish diversity results obtained from metabarcoding were comparable to conventional survey methods. While eDNA sampling and processing need be optimized for this unique environment, the results of this study demonstrate that eDNA metabarcoding can facilitate biodiversity surveys in the deep ocean, require less dedicated survey effort per unit identification, and are capable of simultaneously providing valuable information on other taxonomic groups.
Citation: McClenaghan B, Fahner N, Cote D, Chawarski J, McCarthy A, Rajabi H, et al. (2020) Harnessing the power of eDNA metabarcoding for the detection of deep-sea fishes. PLoS ONE 15(11): e0236540. https://doi.org/10.1371/journal.pone.0236540
Editor: Hideyuki Doi, University of Hyogo, JAPAN
Received: July 8, 2020; Accepted: September 25, 2020; Published: November 4, 2020
Copyright: © 2020 McClenaghan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All DNA sequence data files are available from on NCBI sequence read archive (project number: PRJNA643526).
Funding: This project was supported in part by a grant from the Canadian Ocean Supercluster, a grant from Genome Canada through Genome Atlantic, NSERC grant RGPIN-2014-04826, NSERC ship time grant RGPST, the ArcticNet Hidden Biodiversity Project, the Canadian Foundation for Innovation, the Marine Conservation Targets Program (Fisheries and Oceans Canada), a Petroleum R&D Grant from InnovateNL (contract number 5405.2121.101), an award from the Atlantic Canada Opportunities Agency’s Atlantic Innovation Fund (project number 781-37749-207993), and a grant from Petroleum Research Newfoundland and Labrador. Samples used for this study were collected from the Canadian research icebreaker CCGS Amundsen with the support of the Amundsen Science program funded by the Canada Foundation for Innovation (CFI) Major Science Initiatives (MSI) Fund. Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the funders or their members. B.M., N.F., H.R., A.M., and G.S. are employees of eDNAtec Inc. and M.H. is the founder and Chief Scientific Officer of eDNAtec Inc. The compensations for authors B.M., N.F., H.R., A.M., and G.S. were supported in part by the funders, but the funders had no role additional in study design, data collection and analysis, decision to publish, or preparation of the manuscript. J.C. was financially supported by the Ocean Frontier Institute funded through the Canada First Research Excellence fund, the Atlantic Fisheries Fund, and the ArcticNet project ArcticFish. The specific roles of all the authors are articulated in the ‘author contributions’ section.
Competing interests: B.M., N.F., H.R. A.M. and G.S. are employees of eDNAtec Inc. M.H. is the founder and Chief Scientific Officer. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
The deep ocean is the largest biome on Earth by volume and also one of the planet’s most understudied environments . The biodiversity of the deep ocean has not been fully explored nor is the distribution and biology of many deep-water species well understood [1–3]. Despite our limited knowledge of deep-water fauna, several species are commercially targeted and, along with many other taxa, face increasing pressure from climate change [4–6]. Monitoring and managing the impacts of commercial fishing and climate change in this environment is difficult due to logistic constraints and the high cost of sampling such challenging environments . Despite these impediments, documenting the biodiversity of this region is integral to sustainable management and ecosystem monitoring.
Deep ocean biodiversity surveys are often done using a combination of methods, each targeting a particular taxonomic group. For fish and micronekton, trawling, long-lining, and acoustic monitoring are often used. Small nets and filtration systems can target small zooplankton and phytoplankton and autonomous video camera systems can capture a range of macrofauna . Each of these methods have limitations in their ability to capture a community based on morphological and behavioral selectivity as well as taxonomic resolution. Additionally, not all of these methods can be employed equally well in all areas of the ocean. For example, bottom trawling is ineffective for surveying along steep slopes and rocky surfaces and is undesirable in areas with sensitive epifauna, such as deep-water corals and sponges [9, 10]. The need to employ multiple sampling methods to assess the biodiversity of the deep sea increases the sampling effort required, complicates the interpretation of data, and thereby adds to the challenges of surveying this environment.
Metabarcoding using environmental DNA (eDNA) is a relatively new approach to biodiversity analysis that can facilitate surveys by reducing the sampling effort and taxonomic expertise required and thus far metabarcoding has been underutilized in the deep ocean . Marine eDNA studies have primarily surveyed coastal and/or surface water (e.g. [12, 13]), with very few studies sampling water at depths > 1000m for eukaryotic eDNA (but see ). Much of the eDNA work on deep-sea communities has focused on sediment sampling to study benthic communities (e.g. [15–17]) as opposed to fish and pelagic communities.
Using eDNA from deep sea water samples to characterize biodiversity has the potential to provide critical insight into deep ocean biodiversity however, eDNA sampling protocols need to be optimized for this environment. Abiotic factors, such as the reduced light levels and comparatively low variability in temperature and salinity in deep ocean water , affect the persistence of eDNA while biotic factors, such as the predominant life histories and/or metabolism of the organisms living in the deep ocean (e.g. slower metabolism; ), may affect the amount of eDNA released into the water. Therefore, the optimal protocols for eDNA sampling in the deep ocean must be determined separately from coastal and surface marine water sampling and furthermore, sample processing should be optimized for the particular target groups of deep-sea organisms (e.g. fish).
The objectives of this study were to develop an eDNA metabarcoding sampling protocol for the deep sea, evaluate the performance of multiple primer sets for the detection of deep-sea fishes and compare eDNA results to conventional fish surveys. We collected seawater samples over two sampling years and refined the sampling and lab protocols in the second season to improve the detection of deep-sea fishes. While fishes were the target group, we also report general biodiversity results that were detected concurrently.
We surveyed fish communities in the Labrador Sea, in the Northwest Atlantic Ocean, in the summer (June-August) over three sampling years. Surveys using conventional sampling techniques (2017–2019) and eDNA water sampling (2018–2019) were conducted along three transects each covering a water depth gradient of approximately 500 m to 3000 m (see S1 Fig for map and S1 Table for GPS coordinates). All field sampling was conducted under experimental licenses from Fisheries and Oceans Canada.
Conventional fish surveys
Harvester logbooks and research vessel (RV) surveys using Campelen trawls are typically used to monitor and manage demersal fish communities in Canadian waters but these collections are restricted to waters less than 1500m and are relatively sparse for northern areas . In deeper waters (>1500 m) of the Labrador Sea, there is very limited information on demersal or pelagic fish communities. Therefore, to augment species lists from RV surveys and logbooks, targeted sampling of demersal (baited hooks and cameras; ) and pelagic (Isaac Kidd Midwater Trawls (IKMT) ) fish was conducted in the study area. Demersal fish sampling was conducted along two transect lines in 2017 and 2019, whereas pelagic fish communities were sampled across three transect lines in 2018 and 2019. Baited hooks and cameras were deployed on the ocean bottom whereas IKMT samples were collected from the mesopelagic deep-scattering layer (an area of concentrated pelagic biomass ; sampled depths ranged from 360–536 m) as detected by hull-mounted echosounders. Fish captured using both methods were identified morphologically. While the exact sampling sites differed for pelagic and demersal sampling sites, pelagic sampling was conducted over the same transects as the demersal sampling but was restricted to a maximum water depth of 2500 m (versus a maximum depth of ~3000 m for demersal sampling).
eDNA water sample collection
eDNA water samples were collected from seven stations along one transect in 2018. In 2019, two of these stations were resampled and water samples were collected from an additional eight stations along the two other transects. At each station, samples were collected from the surface, the deep scattering layer and just above the bottom up to a depth of ~2,500 m depth (n = 144, S1 Table). Water samples were co-located in time and space with pelagic fish (IKMT) sampling. Samples were collected using a Niskin-style rosette sampler. Rosette bottles were assigned to eDNA sampling for the duration of the field mission and were decontaminated prior to sampling and between stations using ELIMINase (Decon Labs, Inc., King of Prussia, PA, USA). At each sampling station, a field blank was collected using distilled water to control for potential contamination. In 2018, we employed a sampling strategy adapted from previous coastal surface water sampling in the North Atlantic , where triplicate 250 mL samples were collected at each sampling depth. Water samples were then frozen at -20°C and shipped frozen to the lab for subsequent processing. Water filtration took place in a clean lab, thereby reducing the potential for sample contamination, however cold storage space was required on the vessel to store water samples. Based on the results of 2018 sampling, the sampling strategy was modified for 2019. We increased the water volume collected by a factor of 6, collecting triplicate 1.5 L water samples in 2019, however the larger volume of water collected could not be kept in cold storage on the vessel due to space limitations. As such, water samples in 2019 were filtered on the vessel. Filter cartridges (requiring less storage space) were stored at -20°C for the duration of the expedition.
All water samples were filtered through 0.22 μm PVDF Sterivex filters (MilliporeSigma, Burlington, MA, USA) using a peristaltic pump. Filtration on the vessel took place in a dedicated lab space that included a positive pressure ventilation system. Before each filtration session, surfaces and equipment were all decontaminated with ELIMINase and rinsed with deionized water. Filtration began immediately after sample collection (average volume filtered 1.35 ± 0.15 L). For samples filtered in the lab, filtration took place in a PCR clean lab under a laminar flow hood (AirClean Systems, Creedmoor, NC, USA) which was decontaminated using ELIMINase, lab-grade water and 70% ethanol prior to each sample set. Water samples were thawed at 4°C and immediately filtered. DNA was extracted from all filter membranes using the DNeasy PowerWater Kit (Qiagen, Hilden, Germany). DNA extracts were quantified using the Quant-iT PicoGreen dsDNA assay with a Synergy HTX plate fluorometer (BioTek, Winooski, VT, USA).
Seven DNA markers from three gene regions (cytochrome c oxidase I (COI), 12S and 18S) were selected to assess eukaryotic biodiversity in the 2018 samples (Table 1A), including three primer sets specifically for bony fish. The 2019 samples were analyzed with only these three fish-targeting primer sets. Each PCR reaction contained 1X reaction buffer, 2 mM MgCl2, 0.2mM dNTPs, 0.2 μM of each of the forward and reverse Illumina-tailed primers, 1.5U Platinum Taq (Invitrogen, Carlsbad, CA, USA) and 1.2 μL of DNA in a total volume of 15 μL. Due to the higher concentration of DNA recovered from 2019 samples, diluted DNA was used for 2019 samples (1/10 and 1/2 for surface samples and samples at depth, respectively). The mean concentration of template DNA used was 0.44 ± 0.96 ng/μL. See Table 1B for PCR conditions for all primer sets. Three PCR replicates were performed for each primer set from each sample and then pooled for a single PCR cleanup with the QIAquick 96 PCR purification kit (Qiagen).
Amplicons were visualized using agarose gel (1.5% w/v) electrophoresis to verify amplification of DNA markers and to assess negative controls generated during PCR, extraction, filtration, and field collection. Negative controls were carried through to sequencing as an added level of verification. Amplicons were then indexed using unique dual Nextera indexes (IDT, Coralville, IA, USA; 8-bp index codes). Indexing PCR conditions were initiated for 3 mins at 95°C, followed by 12 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s, and a final extension at 72°C for 5 mins. Amplicons were quantified with Quant-iT PicoGreen dsDNA assay and pooled together in equimolar concentrations by DNA marker. Amplicon pools were cleaned using AMPure XP cleanups, quantified with a Qubit fluorometer (Thermo Fisher, Waltham, MA, USA) and the size distribution of each pool was verified with the DNA 7500 kit on the Agilent 2100 Bioanalyzer. The 2018 12SV5, COI Leray, COI MiniFishE, 18SV9M, and COI F230 amplicon pools were combined into one library. The 2019 12Steleo, 12S MiFishU and COI MiniFishE amplicons pools were combined with the 2018 12Steleo and 12S MiFishU amplicon pools in a second library. The libraries were sequenced with a 300-cycle S1 kit and a 500-cycle SP kit, respectively, on the Illumina NovaSeq 6000 following the NovaSeq standard workflow with a target minimum sequencing depth of 1 million sequences per sample per amplicon. Raw sequence reads are available in NCBI’s sequence read archive under project PRJNA643526.
Base calling and demultiplexing were performed using Illumina’s bcl2fastq software (v220.127.116.112). Primers were trimmed from sequences using cutadapt v1.16  and then DADA2 v1.8.015  was used for quality filtering, joining paired end reads (maxEE = 2, minQ = 2, truncQ = 2, maxN = 0) and denoising using default parameters to produce exact sequence variants (ESVs). Taxonomy was assigned to ESVs using NCBI’s blastn tool v1.9.0  and the nt database (downloaded: November 30, 2019) with an e-value cut-off of 0.001. In cases where a sequence matched multiple taxa with an equally high score, we only assigned taxonomy to the lowest common ancestor of the ambiguous hits. The resulting taxonomic hits were filtered using a selection criterion (% sequence similarity multiplied by % overlap between the query sequence and the reference sequence). Family-level matches were reported using a minimum of 95% selection criterion, genus-level matches were reported using a minimum of 98% selection criterion and species-level matches were reported using a 100% or perfect match. All taxa detected were verified using the WoRMS  and EOL  databases and spurious or irrelevant hits (e.g. terrestrial or domestic species) were omitted.
All statistical analyses were performed using R v3.5.1 . Sampling sites in 2018 (250 mL) and 2019 (1.5 L) did not overlap completely therefore samples with different volumes were collected in different locations and different years, meaning no direct comparisons between samples can be made. However, we made general comparisons across all small volume samples and all large volume samples. Additionally, previous studies in this region suggest that spatial differences in community structure are small compared to community changes by water depth . We used a robust two-way ANOVA (α = 0.05) implemented using the ‘Rfit’ package v0.24.2  to compare the DNA concentrations in each sample between sampling volumes and between sampling depths, categorized as shallow (<500 m), mid-depth (500–1400 m) or deep (>1400 m). Shallow includes samples from the surface and the deep scattering layer for some stations. Mid-depth includes samples from the deep scattering layer and the bottom for some stations. Deep includes only samples from the bottom. Depth categories were chosen based on the distribution of depths sampled at each site and preliminary data exploration (see S2 Fig). Using data from the three markers used on 2018 and 2019 samples (COI MiniFishE, 12Steleo, 12S MiFishU), we used a robust two-way ANOVA to compare the number of ESVs recovered in each sample between sampling volumes and between sampling depths. Post-hoc comparisons between groups were performed using the ‘rcompanion’ package v1.13.2 . We used Levene’s test to determine if the variance in DNA concentration and number of ESVs differed between years and water depths.
We assessed the performance of different primer sets by comparing the number of taxa detected and the resolution of taxonomic assignments for all markers, with a particular focus on the recovery and resolution of fishes. In addition, we used a multi-species, multi-scale occupancy modeling framework to compare the detection probabilities of all fish specific primer sets (12S MiFishU, 12Steleo, COI MiniFishE) across fish taxa while accounting for false negatives following McClenaghan et al. . We included water depth (meters) as a covariate at the level of occupancy and primer set as covariate at the level of detection probability (see S1 Text for model formulation and detailed methods). We ran two models using observations from different levels of taxonomic resolution: fish species and fish families.
We compared the fish taxa detected via eDNA metabarcoding to the fish taxa detected via conventional survey methods for a single sampling expedition (2019 eDNA and 2019 pelagic IKMT sampling). This represents approximately equal field sampling effort for both methods. We summarized the total number of taxa detected using each method at multiple taxonomic levels (family, genus, and species). Additionally, we summarize the total number of taxa recovered from multiple years and methods of conventional sampling.
General sequencing summary
The mean number of sequences recovered per sample per amplicon after bioinformatic filtering was 1,250,418 (range: 16–13,603,412) yielding an average of 706 (range: 1–6003) ESVs per sample per amplicon and a total of 148,339 ESVs. 77.8% of the ESVs matched a sequence in the reference database, although the taxonomic rank assigned to each ESV was variable and resolution differed between amplicons (Table 2).
A total of 21 fish families, 23 genera and 15 species were identified using eDNA from 2018 and 2019 samples across all markers (Table 3). In the deep-water samples (>1400 m), 11 fish families, 11 genera and 8 species were identified. The fish species detected included several deep-water and demersal specialists, such as Bigelow’s Ray (Rajella bigelowi), Agassiz’ Slickhead (Alepocephalus agassizii), Greenland Dwarf Snailfish (Psednos groenlandicus), along with the Roundnose Grenadier (Coryphaenoides rupestris) and the Northern Wolffish (Anarhichas denticulatus), which are listed as Critically Endangered and Endangered, respectively on the IUCN Red List [39, 40]. Several globally important members of the mesopelagic community were also detected, including Glacier Lanternfish (Benthosema glaciale) and Veiled Anglemouth (Cyclothone microdon). In addition to the fishes, 13 metazoan phyla were detected, where 58 families, 39 genera, 25 species were assigned names (S2 Table).
Based on a two-way ANOVA, there was a significant increase in the total amount of DNA recovered (as measured by fluorometry of DNA extracts) from the 1.5-liter samples collected in 2019 compared to the 250 mL samples from 2018 (F = 219.32, df = 1, p < 0.001; Fig 1A). Water depth also had a significant effect on the amount of DNA recovered (F = 35.64, df = 2, p < 0.001), with a lower DNA concentration recovered from deep water samples (>1400 m) compared to mid-depth (500–1400 m) and shallow (<500 m) samples.
Comparison of (A) DNA concentration (pg/μL) in extracts and (B) number of ESVs recovered from small volume samples collected in 2018 and large volume samples collected in 2019 at various depths (shallow < 500 m, mid 500–1400 m, deep >1400 m). The lines inside the boxes represents the median values, the top and bottom of the boxes represent the 75% and 25% quartiles. The whiskers represent 1.5 times the inter-quartile range (IQR). Outliers (any data beyond 1.5*IQR) are shown by circles. Different letters indicate significant differences.
Based on a two-way ANOVA, there were significantly more ESVs detected in the 1.5-liter 2019 samples compared to the 250 mL 2018 samples (F = 88.28, df = 1, p < 0.001; Fig 1B). Additionally, there was significantly less variance in the number of ESVs recovered from the larger volume samples (F = 30.00, df = 1, p < 0.001). The different sampling volumes were collected in different years and at different locations so no direct comparisons of the biodiversity detected by volume could be made. There was a significant difference in the number of ESVs recovered between sampling depths (F = 6.53, df = 2, p = 0.002). Post-hoc comparisons revealed that large volume mid depth samples from 2019 recovered the most ESVs. The number of ESVs recovered from large volume deep samples in 2019 was not significantly higher than the number of ESVs detected in any of the small volume water samples. See S3 Fig for a comparison between DNA concentration and number of ESVs by sampling location (surface, deep scattering layer, bottom).
Of the seven primers sets tested on the 2018 samples, the fish-targeted 12Steleo, 12S MiFishU and COI MiniFishE primer sets were the most effective at detecting fish with 11, 8 and 3 families detected by each primer set respectively. 12Steleo also provided the highest resolution with 7 species identified. COI Leray and COI F230 failed to detect any fish families. The two primer sets that identified the most metazoan families other than fish, were 18SV9M and COI FishE, which detected 18 and 12 families in 2018 samples, respectively. The three primer sets which performed well for fishes were run on samples from 2019 (12Steleo, 12S MiFishU and COI FishE) to maximize fish detection while also identifying a range of metazoans. An additional three fish families and 22 metazoan families were detected in the 2019 samples.
For the three primer sets used across both years, no single fish species was detected by all primer sets and all three primer sets detected at least one species that was unique to that primer set. Occupancy modeling revealed taxa specific variability in probabilities of detection between primer sets at the species and family level (Figs 2 and 3), however when comparing the primer sets across the whole fish community, there was little difference in the community mean probability of detection for each primer set (Fig 4).
The lines inside the boxes represents the median values, the top and bottom of the boxes represent the 75% and 25% quartiles. The whiskers represent 1.5 times the inter-quartile range (IQR).
The lines inside the boxes represents the median values, the top and bottom of the boxes represent the 75% and 25% quartiles. The whiskers represent 1.5 times the inter-quartile range (IQR).
Similar results were seen from the fish species-level model. The lines inside the boxes represents the median values, the top and bottom of the boxes represent the 75% and 25% quartiles. The whiskers represent 1.5 times the inter-quartile range (IQR).
Morphology & eDNA comparison
Conventional surveys were conducted using multiple methods on three transects in the sampling area. These surveys were performed over multiple years and on multiple sampling expeditions and allowed us to assemble an inventory of species for the region. Overall, these morphological surveys identified 27 species, 25 genera and 18 families in the sampling area. To directly compare between conventional methods and eDNA, we considered only morphological data that was generated during the same sampling expedition as eDNA sample collections. There was a high degree of overlap in taxa detected between eDNA and identified via morphology (i.e. via IKMT pelagic trawls), but several taxa were unique to metabarcoding (Fig 5). A total of 14 fish species, 21 genera and 16 families were identified using eDNA while 10 fish species, 8 genera and 6 families were identified morphologically.
We demonstrated a successful protocol for the detection of deep-sea fishes using eDNA from seawater samples collected at depths down to 2500 m. Our results suggest that eDNA is less abundant in seawater from depths > 1400 m, a factor which should be considered for sampling designs of future deep-sea eDNA studies. The physical characteristics of the deep ocean (e.g. lower temperature, less sunlight) suggest DNA persists longer in this environment than at the surface [41, 42], however the lower DNA concentration may reflect the different biological community present, with less abundant plankton and more species with low metabolic rates living in the deep ocean [18, 43, 44]. We recommend the sampling protocol followed in 2019 where larger water volumes (≥ 1.5 L) were collected, particularly for sampling the deep marine environment where the amount of DNA recovered from samples was lower. While the number of ESVs recovered from large volume deep samples was not significantly higher than the small volume samples, the reduced variance in the number of ESVs recovered suggests a more robust sampling method. Increasing the sequencing depth may be a means to make up for low DNA recovery in samples such as this. Indeed, in this study, the samples were sequenced at much higher depth (~1,000,000 reads per sample per amplicon) than most metabarcoding studies . Despite an equally high sequencing depth in the small volume and low DNA concentration samples, the number of ESVs recovered was consistently higher in large volume and high DNA concentration samples. These results highlight the need for metabarcoding sampling methods to be tailored to the sampling environment and for further research into the origin, persistence, and degradation of eDNA in marine systems. Much of the research on the dynamics of eDNA has focused on freshwater systems (e.g. [45–47]) and much less is known about this cycle for eDNA in marine environments, particularly in the deep ocean (but see [48, 49]). As our understanding of eDNA dynamics in the deep ocean progresses, eDNA can be a reliable way of detecting deep sea organisms, provided appropriate sampling methods are used.
When field sampling protocols are optimized for a particular system, there are often logistical constraints that must be considered in addition to the biological factors. In this study, the cold storage of large volume water samples on the sampling vessel was a limitation and therefore large volume samples were filtered in situ on the vessel rather than in a dedicated pre-PCR lab where downstream processing occurred. While this allowed larger water volumes to be collected, it required additional personnel time on the vessel and there may have been an increased risk of contamination for filtering in situ on an operational vessel at sea. In this case, precautions were taken to minimize the contamination risk including decontaminating the lab and the addition of negative controls at every step in the field (sample collection, filtration) and subsequent laboratory steps (extraction, PCR amplification). The adaptability of the filtering process was essential for allowing the collection of large volume water samples in this study. We acknowledge that these additional changes to the protocol may have contributed to the different results seen in 2018 and 2019. Additionally, it is possible that the different results obtained between years were due, in part, to changes in the fish diversity present in the sampling area each year. Ideally, samples would have been collected during a single survey however, the logistics of sampling in this remote region prevented this. Based on data from previous conventional surveys in the area, patterns of fish diversity in this area are thought to be relatively homogenous over space and time [19, 50]. Furthermore, water sampling volume is known to affect the biodiversity recovered from metabarcoding samples . Therefore, water volume was likely the primary factor contributing to the observed differences between study years. The specific logistical constraints of sampling will be unique to each sampling mission and depend on the resources available, but they are an important consideration when optimizing sampling protocols.
We identified multiple primers sets that performed well for deep-sea fishes, but we also determined that these primer sets vary considerably in their detection probabilities within the fishes. It should also be noted that the fish-specific 12S primers used in this study (12Steleo, 12S MiFishU) were designed to target bony fish and not cartilaginous fish. While we did detect one species of cartilaginous fish (Rajella bigelowi) using 12Steleo, alternative primers should be considered for studies targeting cartilaginous fish (e.g. 12S MiFishE ). The fish primer sets used in this study recovered many fish taxa, however the species-level resolution was not always consistently high. For example, for the 12Steleo primer set, 110 fish ESVs were recovered and only 7 (6.4%) were identified to the species level (as seen in Table 2). The low resolution is due to a combination of low sequence diversity between species (where query sequences matched multiple species in the reference database) and poor reference database coverage (where query sequences did not match any reference sequences at our species-level threshold) . Other studies comparing primer sets for the detection of fish have found similar taxonomic biases and showed that a lack of reference database coverage negatively affects the resolution of several targeted primer sets [53, 54]. This reinforces the importance of marker selection and highlights the need to use multiple markers to maximize detection and taxonomic resolution even within a relatively narrow target group, such as fish. Integrating data from multiple primer sets from multiple marker regions is often recommended for metabarcoding-based biodiversity surveys [55–58]. This also highlights the need for improved species coverage in reference databases. We identified a primer set (COI MiniFishE ) that performs well for a range of metazoan taxa in addition to fishes, suggesting this would be a useful primer set for comprehensive biodiversity assessments in marine environments. Conversely, one of the primer sets that has been used in a number of marine metabarcoding papers (COI Leray; mlCOIintF/ jgHCO2198 ) did not detect any fish taxa and hence is not recommended for analyses of fish biodiversity. Using deep sequencing with multiple primer sets is a simple strategy that can capture deep sea biodiversity especially for less abundant and elusive fish taxa.
The fish taxa detected using eDNA metabarcoding were comparable to those identified via conventional fish survey methods, although several taxa were unique to each method. This is consistent with other studies comparing eDNA to other methods of biodiversity assessment (e.g. [59, 60]). When looking at a single sampling expedition, eDNA captured more fish diversity than conventional methods, and did so from rosette deployments that were used to fulfil other mission objectives (e.g. obtaining water for chemical analyses). Given the expense and time constraints associated with large research vessels, achieving such efficiencies is noteworthy. Furthermore, the relative simplicity of eDNA sample collection allows for synchronous usage of hydroacoustics and in-situ sensors. Metabarcoding also has the added benefit of potentially detecting species outside the target taxa. While this is dependent on the primer sets selected, the ability to detect species from all trophic levels and life histories from the same sample drastically increases the efficiency of biodiversity assessments by minimizing the number of different sampling methods required to holistically survey an ecosystem (e.g. ). Furthermore, various marine habitats (e.g. pelagic, demersal) can be sampled using the same methods compared to conventional surveys where various capture methods, each with their associated biases, are used in separate habitats. And finally, eDNA samples, once collected, can be used for subsequent analyses with other primer sets to generate biodiversity data for other groups or to target specific species or their populations without the need for additional sampling campaigns. For example, while the samples used in this study were collected and processed with the goal of detecting fishes, these same water samples could be processed with primers targeting corals to provide insight into deep-sea coral diversity without the need for additional sampling effort.
While there is a lot to be gained by applying metabarcoding tools to surveying the deep ocean, there are also limitations to this method. Since the biodiversity of this environment is not well-known, the reference database coverage for deep sea species is unlikely to be as comprehensive as coastal or freshwater systems. Low reference database coverage can reduce the taxonomic resolution of eDNA studies . This limitation can be dealt with by generating a reference library for key fish species in the deep ocean alongside eDNA metabarcoding monitoring efforts. Metabarcoding is also limited in its quantitative ability  and most studies use a presence/absence approach (e.g. ). This method is very useful for assessing species richness and community structure , and determining species distributions , but the current methodology cannot be used to infer absolute abundance. Age structure, reproductive stage, and contaminant load are other examples of data that cannot be determined via eDNA. These factors will still rely on the capture of specimens, however eDNA can significantly increase our understanding of spatial and temporal distribution of species, which can be used to guide more detailed sampling where conventional sampling is required.
eDNA metabarcoding is a powerful approach for surveying biodiversity in the deep ocean. While future work will continue to improve these methods, such as increasing the taxonomic coverage in reference databases and refining sampling designs, this methodology can be employed immediately to complement ongoing biodiversity monitoring efforts in the deep ocean. Given the vastness of the deep ocean environment, our limited knowledge of this region’s biodiversity and the increasing anthropogenic pressures facing this fauna, there is huge potential for eDNA metabarcoding to revolutionize biodiversity monitoring and environmental stewardship in these areas.
S1 Fig. Map of the sampling area in the Labrador Sea showing sampling sites along three transects that follow a depth gradient of approximately 500 m to 3000 m.
Colours of sampling sites indicate the year of eDNA sampling. Inset map shows the location of the sampling area on a global map. Map data source: Esri. Ocean Reference [basemap]. 1:6000000. Ocean Basemap. February 10, 2012. www.arcgis.com/home/item.html?id=5ae9e138a17842688b0b79283a4353f6. (Accessed: June 15, 2020).
S2 Fig. Scatterplot plot comparing water sampling depth and DNA concentration for eDNA water samples collected in the Labrador Sea in 2019.
The blue line represents the predicted values based on a generalized linear model with 95% confidence intervals shown in gray.
Comparison of (A) DNA concentration (pg/μL) in extracts and (B) number of ESVs recovered from small volume samples collected in 2018 and large volume samples collected in 2019 at various depth sampling locations (surface, deep scattering layer, bottom). Different letters indicate significant differences.
S1 Text. Detailed occupancy modeling methods and model structure.
S1 Table. Sampling summary table listing the sampling stations, their location, date of collection, sampling depths and approximate water depth.
Triplicate water samples were collected at each station and date listed.
S2 Table. Summary of all metazoan taxa identified in seawater samples, indicating whether or not the taxa was detected at each depth (shallow < 500 m, mid 500–1400 m, deep > 1400 m) and the total number of samples in which the taxa was detected.
We would like to thank Kerry Hobrecker for her assistance in sample processing at CEGA and Jasmin Schuster, Andrew Murphy, Amy McAllister, Thibaud DeZutter, Sheena Roul, Jennica Seiden and Catie Young for their assistance processing samples at sea.
- 1. Webb TJ, Vanden Berghe E, O’Dor R. Biodiversity’s big wet secret: The global distribution of marine biological records reveals chronic under-exploration of the deep pelagic ocean. PLOS ONE. 2010;5: e10223. pmid:20689845
- 2. Bergstad OA. North Atlantic demersal deep-water fish distribution and biology: present knowledge and challenges for the future: deep-water fish distribution and biology. J Fish Biol. 2013;83: 1489–1507. pmid:24298948
- 3. Costello MJ, Chaudhary C. Marine biodiversity, biogeography, deep-sea gradients, and conservation. Curr Biol. 2017;27: R511–R527. pmid:28586689
- 4. Koslow J. Continental slope and deep-sea fisheries: implications for a fragile ecosystem. ICES J Mar Sci. 2000;57: 548–557.
- 5. Levin LA, Le Bris N. The deep ocean under climate change. Science. 2015;350: 766–768. pmid:26564845
- 6. Clark MR, Althaus F, Schlacher TA, Williams A, Bowden DA, Rowden AA. The impacts of deep-sea fisheries on benthic communities: a review. ICES J Mar Sci. 2016;73: i51–i69.
- 7. Pais RT, Pastorinho MR. Chapter 2: Sampling Pelagic Marine Organisms. Marine Ecology: Current and Future Developments. Bentham Science Publishers; 2019.
- 8. Woodall L, Andradi-Brown D, Brierley A, Clark M, Connelly D, Hall R, et al. A multidisciplinary approach for generating globally consistent data on mesophotic, deep-pelagic, and bathyal biological communities. Oceanography. 2018;31.
- 9. Pusceddu A, Bianchelli S, Martin J, Puig P, Palanques A, Masque P, et al. Chronic and intensive bottom trawling impairs deep-sea biodiversity and ecosystem functioning. Proc Natl Acad Sci. 2014;111: 8861–8866. pmid:24843122
- 10. Pham CK, Murillo FJ, Lirette C, Maldonado M, Colaço A, Ottaviani D, et al. Removal of deep-sea sponges by bottom trawling in the Flemish Cap area: conservation, ecology and economic assessment. Sci Rep. 2019;9. pmid:31676767
- 11. Ruppert KM, Kline RJ, Rahman MS. Past, present, and future perspectives of environmental DNA (eDNA) metabarcoding: A systematic review in methods, monitoring, and applications of global eDNA. Glob Ecol Conserv. 2019;17: e00547.
- 12. Andruszkiewicz EA, Starks HA, Chavez FP, Sassoubre LM, Block BA, Boehm AB. Biomonitoring of marine vertebrates in Monterey Bay using eDNA metabarcoding. Doi , editor. PLOS ONE. 2017;12: e0176343. pmid:28441466
- 13. Lacoursière‐Roussel A, Howland K, Normandeau E, Grey EK, Archambault P, Deiner K, et al. eDNA metabarcoding as a new surveillance approach for coastal Arctic biodiversity. Ecol Evol. 2018;8: 7763–7777. pmid:30250661
- 14. Everett MV, Park LK. Exploring deep-water coral communities using environmental DNA. Deep Sea Res Part II Top Stud Oceanogr. 2018;150: 229–241.
- 15. Guardiola M, Uriz MJ, Taberlet P, Coissac E, Wangensteen OS, Turon X. Deep-sea, deep-sequencing: Metabarcoding extracellular DNA from sediments of marine canyons. Duperron S, editor. PLOS ONE. 2015;10: e0139633. pmid:26436773
- 16. Dell’Anno A, Carugati L, Corinaldesi C, Riccioni G, Danovaro R. Unveiling the biodiversity of deep-sea nematodes through metabarcoding: Are we ready to bypass the classical taxonomy? Bianchi CN, editor. PLOS ONE. 2015;10: e0144928. pmid:26701112
- 17. Fonseca VG, Sinniger F, Gaspar JM, Quince C, Creer S, Power DM, et al. Revealing higher than expected meiofaunal diversity in Antarctic sediments: a metabarcoding approach. Sci Rep. 2017;7. pmid:28733608
- 18. Drazen JC, Seibel BA. Depth-related trends in metabolism of benthic and benthopelagic deep-sea fishes. Limnol Oceanogr. 2007;52: 2306–2316.
- 19. Coté D, Heggland K, Roul S, Robertson G, Fifield D, Wareham V, et al. Overview of the biophysical and ecological components of the Labrador Sea Frontier Area. Fisheries and Oceans Canada Canadian Science Advisory Secretariat; p. 59. Report No.: 2018/067.
- 20. Amundsen Science. Field Report: Integrated Studies and Ecosystem Characterization of the Labrador Deep Sea Ocean (ISECOLD) 2019. Amundsen Science; 2020.
- 21. Johnson M. Sound as a tool in marine ecology, from data on biological noises and the deep scattering layer. J Mar Res. 1948;7: 443–458.
- 22. Singer GAC, Fahner NA, Barnes JG, McCarthy A, Hajibabaei M. Comprehensive biodiversity analysis via ultra-deep patterned flow cell technology: a case study of eDNA metabarcoding seawater. Sci Rep. 2019;9: 5991. pmid:30979963
- 23. Leray M, Yang JY, Meyer CP, Mills SC, Agudelo N, Ranwez V, et al. A new versatile primer set targeting a short fragment of the mitochondrial COI region for metabarcoding metazoan diversity: application for characterizing coral reef fish gut contents. Front Zool. 2013;10: 34. pmid:23767809
- 24. Shokralla S, Hellberg RS, Handy SM, King I, Hajibabaei M. A DNA Mini-Barcoding System for Authentication of Processed Fish Products. Sci Rep. 2015;5. pmid:26516098
- 25. Gibson JF, Shokralla S, Curry C, Baird DJ, Monk WA, King I, et al. Large-Scale Biomonitoring of Remote and Threatened Ecosystems via High-Throughput Sequencing. Fontaneto D, editor. PLOS ONE. 2015;10: e0138432. pmid:26488407
- 26. Stoeck T, Bass D, Nebel M, Christen R, Jones MDM, Breiner H-W, et al. Multiple marker parallel tag environmental DNA sequencing reveals a highly complex eukaryotic community in marine anoxic water. Mol Ecol. 2010;19: 21–31. pmid:20331767
- 27. Riaz T, Shehzad W, Viari A, Pompanon F, Taberlet P, Coissac E. ecoPrimers: inference of new DNA barcode markers from whole genome sequence analysis. Nucleic Acids Res. 2011;39: e145–e145. pmid:21930509
- 28. Valentini A, Taberlet P, Miaud C, Civade R, Herder J, Thomsen PF, et al. Next-generation monitoring of aquatic biodiversity using environmental DNA metabarcoding. Mol Ecol. 2016;25: 929–942. pmid:26479867
- 29. Miya M, Sato Y, Fukunaga T, Sado T, Poulsen JY, Sato K, et al. MiFish, a set of universal PCR primers for metabarcoding environmental DNA from fishes: detection of more than 230 subtropical marine species. R Soc Open Sci. 2015;2: 150088. pmid:26587265
- 30. Martin C. Cutadapt removes adapter sequences for high-throughput sequencing reads. EMBnet.journal. 2011;17: 10–12.
- 31. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13: 581–583. pmid:27214047
- 32. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215: 403–410. pmid:2231712
- 33. WoRMS Editorial Board. World Register of Marine Species. 2020 [cited 2 Jun 2020]. Available: http://www.marinespecies.org at VLIZ
- 34. Encyclopedia of Life. 2014 [cited 2 Jun 2020]. Available: http://eol.org
- 35. R Core Team. R: A language and environment for statistical computing. R Found Stat Comput Vienna Austria. 2018. Available: https://www.R-project.org/
- 36. Kloke JD, Mckean JW. Rfit: Rank-based estimation for linear models. R J. 2012;4: 57–64.
- 37. Mangiafico S. rcompanion: functions to support extension education program evaluation. Available: https://CRAN.R-project.org/package=rcompanion
- 38. McClenaghan B, Compson ZG, Hajibabaei M. Validating metabarcoding-based biodiversity assessments with multi-species occupancy models: a case study using coastal marine eDNA. PLOS ONE. 2020.
- 39. Collette B, Heessen H, Fernandes P. Anarhichas denticulatus. IUCN Red List Threat Species 2015. 2020; e.T18155990A44739291.
- 40. Iwamoto T. Coryphaenoides rupestris. IUCN Red List Threat Species 2015. 2020; e.T15522149A15603540.
- 41. Eichmiller JJ, Best SE, Sorensen PW. Effects of Temperature and Trophic State on Degradation of Environmental DNA in Lake Water. Environ Sci Technol. 2016;50: 1859–1867. pmid:26771292
- 42. Strickler KM, Fremier AK, Goldberg CS. Quantifying effects of UV-B, temperature, and pH on eDNA degradation in aquatic microcosms. Biol Conserv. 2015;183: 85–92.
- 43. Hansen BK, Bekkevold D, Clausen LW, Nielsen EE. The sceptical optimist: challenges and perspectives for the application of environmental DNA in marine fisheries. Fish Fish. 2018;19: 751–768.
- 44. Weikert H. The vertical distribution of zooplankton in relation to habitat zones in the area of the Atlantis I1 Deep, central Red Sea.: 15.
- 45. Dejean T, Valentini A, Duparc A, Pellier-Cuit S, Pompanon F, Taberlet P, et al. Persistence of environmental DNA in freshwater ecosystems. Gilbert JA, editor. PLoS ONE. 2011;6: e23398. pmid:21858099
- 46. Barnes MA, Turner CR, Jerde CL, Renshaw MA, Chadderton WL, Lodge DM. Environmental conditions I\influence eDNA persistence in aquatic systems. Environ Sci Technol. 2014;48: 1819–1827. pmid:24422450
- 47. Sansom BJ, Sassoubre LM. Environmental DNA (eDNA) shedding and decay rates to model freshwater mussel eDNA transport in a river. Environ Sci Technol. 2017;51: 14244–14253. pmid:29131600
- 48. Andruszkiewicz EA, Sassoubre LM, Boehm AB. Persistence of marine fish environmental DNA and the influence of sunlight. Doi H, editor. PLOS ONE. 2017;12: e0185043. pmid:28915253
- 49. Collins RA, Wangensteen OS, O’Gorman EJ, Mariani S, Sims DW, Genner MJ. Persistence of environmental DNA in marine systems. Commun Biol. 2018;1. pmid:30417122
- 50. Murua H, de Cárdenas E. Depth-distribution of deepwater species in Flemish Pass. J Northwest Atl Fish Sci. 2005;37: 1–12.
- 51. Cantera I, Cilleros K, Valentini A, Cerdan A, Dejean T, Iribar A, et al. Optimizing environmental DNA sampling effort for fish inventories in tropical streams and rivers. Sci Rep. 2019;9. pmid:30816174
- 52. Porter TM, Hajibabaei M. Over 2.5 million COI sequences in GenBank and growing. Arthofer W, editor. PLOS ONE. 2018;13: e0200177. pmid:30192752
- 53. Collins RA, Bakker J, Wangensteen OS, Soto AZ, Corrigan L, Sims DW, et al. Non‐specific amplification compromises environmental DNA metabarcoding with COI. Yu D, editor. Methods Ecol Evol. 2019;10: 1985–2001.
- 54. Schenekar T, Schletterer M, Lecaudey LA, Weiss SJ. Reference databases, primer choice, and assay sensitivity for environmental metabarcoding: Lessons learnt from a re‐evaluation of an eDNA fish assessment in the Volga headwaters. River Res Appl. 2020; rra.3610.
- 55. Alberdi A, Aizpurua O, Gilbert MTP, Bohmann K. Scrutinizing key steps for reliable metabarcoding of environmental samples. Mahon A, editor. Methods Ecol Evol. 2018;9: 134–147.
- 56. Zhang GK, Chain FJJ, Abbott CL, Cristescu ME. Metabarcoding using multiplexed markers increases species detection in complex zooplankton communities. Evol Appl. 2018;11: 1901–1914. pmid:30459837
- 57. Hajibabaei M, Spall JL, Shokralla S, van Konynenburg S. Assessing biodiversity of a freshwater benthic macroinvertebrate community through non-destructive environmental barcoding of DNA from preservative ethanol. BMC Ecol. 2012;12: 28. pmid:23259585
- 58. Gibson J, Shokralla S, Porter TM, King I, van Konynenburg S, Janzen DH, et al. Simultaneous assessment of the macrobiome and microbiome in a bulk sample of tropical arthropods through DNA metasystematics. Proc Natl Acad Sci. 2014;111: 8007–8012. pmid:24808136
- 59. Thomsen PF, Møller PR, Sigsgaard EE, Knudsen SW, Jørgensen OA, Willerslev E. Environmental DNA from seawater samples correlate with trawl catches of subarctic, deepwater fishes. Mahon AR, editor. PLOS ONE. 2016;11: e0165252. pmid:27851757
- 60. Closek CJ, Santora JA, Starks HA, Schroeder ID, Andruszkiewicz EA, Sakuma KM, et al. Marine vertebrate biodiversity and distribution within the central California current using environmental DNA (eDNA) metabarcoding and ecosystem surveys. Front Mar Sci. 2019;6.
- 61. Stat M, Huggett MJ, Bernasconi R, DiBattista JD, Berry TE, Newman SJ, et al. Ecosystem biomonitoring with eDNA: metabarcoding across the tree of life in a tropical marine environment. Sci Rep. 2017;7. pmid:28947818
- 62. Cristescu ME. From barcoding single individuals to metabarcoding biological communities: towards an integrative approach to the study of global biodiversity. Trends Ecol Evol. 2014;29: 566–571. pmid:25175416
- 63. Lamb PD, Hunter E, Pinnegar JK, Creer S, Davies RG, Taylor MI. How quantitative is metabarcoding: A meta‐analytical approach. Mol Ecol. 2019;28: 420–430. pmid:30408260
- 64. Civade R, Dejean T, Valentini A, Roset N, Raymond J-C, Bonin A, et al. Spatial Representativeness of Environmental DNA Metabarcoding Signal for Fish Biodiversity Assessment in a Natural Freshwater System. Garcia de Leaniz C, editor. PLOS ONE. 2016;11: e0157366. pmid:27359116
- 65. Yamamoto S, Masuda R, Sato Y, Sado T, Araki H, Kondoh M, et al. Environmental DNA metabarcoding reveals local fish communities in a species-rich coastal sea. Sci Rep. 2017;7. pmid:28079122
- 66. Muha TP, Rodríguez-Rey M, Rolla M, Tricarico E. Using environmental DNA to improve species distribution models for freshwater invaders. Front Ecol Evol. 2017;5.