Comparative analysis of bottom trawl and nanopore sequencing in fish biodiversity assessment: The sylt outer reef example

.


Introduction
Preserving biodiversity and protecting key habitats and species are at the heart of the European Natura 2000 network of protected areas.The Natura 2000 network comprises Special Protection Areas (SPAs) and Special Areas of Conservation (SACs) under the Birds and Habitats Directive and aims to safeguard and ensure the long-term survival of the most valuable and threatened species and ecosystems (European Commission. (n.d.).Natura 2000).The Sylt Outer Reef (SOR) is one of Germany's Natura 2000 sites in the Exclusive Economic Zone (EEZ) of the North Sea and harbours a rich marine fauna, including a diverse fish community.In order to effectively conserve and manage these resources, a comprehensive understanding of fish diversity within the SOR is of vital importance.
To achieve the objectives outlined in the Natura 2000 initiative, a regular monitoring of these protected areas is essential to understand and safeguard their diversity.Trawl sampling, while being a classical tool to perform marine fish biodiversity monitoring, can inflict harm on sensitive habitats due to its invasive and possibly destructive nature.Furthermore, trawling comes with a number of other disadvantages, namely its relatively high costs, labor-intensive nature, time-consuming processing, and limited spatial and temporal coverage (Greenstreet, 2008;Rodríguez-Mendoza and Saborido-Rey, 2023).This underlines the need to adopt non-invasive monitoring techniques that are in line with the Natura 2000 objectives of conserving threatened species and ecosystems without causing damage or disturbance.
Indeed, the use of remotely operated underwater vehicles (ROVs), satellite tracking, and sonar systems are among the non-invasive techniques that allow monitoring and assessment of marine biodiversity without disturbing its natural habitat (Busnel and Fish, 1980;Connolly et al., 2021).These tools, while effective, have shown certain limitations in their application, such as the inability to detect smaller or less mobile species, or the difficulty in covering large areas in a short amount of time.
To address these limitations, the emerging field of environmental DNA (eDNA) analysis offers a promising and innovative approach (Adams et al., 2019;Petit-Marty et al., 2023;Schwentner et al., 2021).By analysing genetic material in water samples, eDNA allows for the identification of species presences, even when traditional survey methods may be challenging or impractical (Maiello et al., 2022;Petit-Marty et al., 2023;Stoeckle et al., 2021).Therefore, it doesn't only circumvent invasive sampling methods, but even has the potential to provide a more comprehensive biodiversity assessment (Dickie et al., 2018;Schwentner et al., 2021).
Since its first application by Ficetola et al. (2008), eDNA has revolutionised the way marine biodiversity can be studied in a wide range of aquatic ecosystems (Adams et al., 2019;Coble et al., 2019;Rourke et al., 2021).This breakthrough has been made possible by the rapid development of molecular biology and sequencing tools and techniques.However, the majority of eDNA applications have been carried out using short-read technologies, particularly MiSeq, due to their low error rate (Stoler and Nekrutenko, 2021).
On the other hand, long-read sequencing technologies are still in very limited use (Doorenspleet et al., 2021;van der Reis et al., 2023).While these techniques promise potential benefits such as the ability to resolve complex genomic regions and provide a more complete picture of genetic diversity, they are still in their infancy in the eDNA field.Studies have shown that in addition to their ability to sequence long reads, these technologies are able to achieve a read quality of up to 99% (Banerjee et al., 2024;P. Chen et al., 2023).As a result, entire genes or even the mitochondrial genomes can be sequenced in a single read without the need for assemblies (Davidov et al., 2020;Doorenspleet et al., 2021;Latorre-Pérez et al., 2020).This makes it possible to improve the sensitivity of eDNA to distinguish between genetically closely related species.For example, MiFish primers are unable to distinguish among species of the scombrid genus Thunnus, the freshwater eel genus Anguilla, the salmonid genus Oncorhynchus, the puffer fish genus Takifugu, and the rockfish genus Sebastes, due to their genetic proximity (Miya, 2022).Long-read sequencing technologies can play a crucial role here, allowing more accurate species or even stock identification.
The objective of this study was to compare the efficacy of eDNA metabarcoding using MinION sequencer (Nanopore Technology) against traditional survey methods, specifically bottom trawling (i.e., beam trawling), in terms of species detection and diversity assessment.By providing a comprehensive assessment of fish diversity, the study aimed at showcasing the potential of eDNA as an additional tool for future monitoring in European Commission sites.By employing cutting-edge eDNA metabarcoding methods, the study aimed at providing a comprehensive and accurate assessment of fish diversity at the sampling sites.

Trawl and catch
In the Natura 2000 site Sylt Outer Reef (SOR) 12 stations (Fig. 1) were sampled between the 30th June to the July 11, 2022 with a 3 m beam trawl, rigged with a sole net and five tickler chains.The mesh size in the cod end was 40 mm.Haul duration was fixed to 15 min with a towing speed of approximately 3 kn.Total catches were sorted and caught fish species were determined to the highest possible taxonomic level.

Environmental DNA sampling
24 water samples from the 12 fishing stations (Table 1) plus 2 negative control samples were taken.Two 5-Liter samples were collected from each station.Water was collected in Niskin bottles and immediately filtered through 0.45 μM PES membranes in sterile funnel units (Sartorius, Göttingen, Germany) with a 2-port filtration system.Typically, 4 to 5 filters were used per sample, depending on the degree of filtration saturation.The filters were then stored in clean, sterile tubes at − 20 • C until DNA extraction.All filters from a single water sample were combined and processed as a single unit during eDNA extraction.
All Niskin bottles were thoroughly cleaned prior to water sample collection by rinsing first with fresh water and then with ultrapure water.Negative controls for all Niskin bottles used were performed by filtration of 0.5 L of ultrapure water rinsed through the cleaned bottles prior to the first sample collection.All filtration steps were performed in a pre-cleaned UV sterilization cabinet (Cleaver Scientific, Rugby, UK).Clean gloves were worn during each sampling step and all materials used, e.g.water sample containers or membrane tweezers, were sterilized by washing with 3% bleach and/or UV sterilization where possible.

Environmental DNA extraction
eDNA was extracted from water according to Renshaw et al. (2015).In brief, filtered and frozen PES membranes were placed in a 2 ml tube and then supplied with 700 μL of CTAB buffer (2% CTAB (w/v), 1.4 M NaCl, 100 mM Tris, 20 mM EDTA) and incubated at 65 • C for 10 min with agitation at 650 rpm.Subsequently, 900 μL of PCI Phenol/chloroform/isoamyl alcohol (25:24:1) was added and vortexed for 5 s, Table 1 The geographical coordinates of water and fish sampling locations used in the present study.Trawling was performed directly after water sampling.before centrifugation at 15,000×g for 5 min 700 μL of supernatant was transferred to a new 2 ml tube, prior to adding 700 μL of chloroform, followed by centrifugation at 15,000×g for 5 min.After transferring 500 μL of the supernatant to a new 2 ml tube, the Monarch® cleaning kit (New England Biolabs, Frankfurt, Germany).The elution was transferred into 40 μL of elution buffer.DNA extracts were stored at − 20 • C until analysis.All DNA extraction steps were performed under a chemical hood in a dedicated area for eDNA analysis at the Thünen Institute in Bremerhaven.

Library preparation
All libraries were prepared with MiFish primers targeting the 12S rRNA gene according to the protocol of Miya et al. (2015).Briefly, the library preparation consists of two PCR steps.The first PCR cycle (1st PCR) aimed at specifically amplifying the fish 12S rRNA gene with adaptor sequences according to Nanopore recommendations.This PCR was performed in 8 replicates with a total reaction volume of 12 μL, including 2 μL of 1:5 diluted eDNA, 6.0 μL of 2 × KAPA HiFi HotStart ReadyMix (Roche, Germany), 2.8 μL of primer mix (Primer Mix: MiFish-E-F/R-v2: MiFish-U-F/R: MiFish-U2-F/R = 0.125:0.250:0.125μM), and 1.2 μl sterile distilled water.These primer pairs co-amplify a hypervariable region of the fish mitochondrial 12S rRNA gene (approximately 180 bp).
The 1st PCR product was purified using the GeneRead size selection kit (Qiagen, Germany) to remove all sequences shorter than 150 bp and primer dimers.The purified product was then analyzed using the TapeStation 4150 to quantify and check for the presence of the fish band (approximately 380 bp), then diluted to a concentration of 0.1 ng/μL.A second cycle PCR (2nd PCR) using the diluted 1st PCR product were performed to add Nanopore unique index to each sample as well as the Nanopore sequencer adapters.A total volume of 20 μL was used for the 2nd PCR, containing 7.5 μL of KAPA Master Mix, 5 μL of DNA, 0.5 μL of barcode primers and 7 μL of water.The thermal cycling profile after an initial denaturation at 95 • C for 3 min was as follows (12 cycles): denaturation at 98 • C for 20 s; annealing at 62 • C for 15 s (as recommended by Nanopore) and extension at 72 • C for 15 s with a final extension at 72 • C for 5 min.The second PCR pooled equal volumes from all samples, and therefore, the product was isolated by size at approximately 370 bp.The 'fish band' was purified using the MinElute gel extraction kit (Qiagen, Germany), then quantified using the Qubit fluorometer.
Pool preparation for sequencing was done using the Ligation Sequencing Amplicons -Barcode PCR protocol with EXP-PBC096 and SQK-LSK112 kits (Oxford Nanopore Technologies) (Version: NBE_9065_v109_revY_9Mar2021).Briefly, approximately 200 fmol of purified amplification products were subjected to DNA repair and final preparation using NEBNext DNA Repair Mix and NEBNext Ultra II End Repair/dA-Tailing Module (New England Biolabs).The purpose of which is to repair the DNA and pre-prepare the DNA ends for adapter attachment, followed by attachment of the sequencing adapters to the DNA ends using T4 Ligase (New England Biolabs).Each step was followed by DNA purification with AMPure XP magnetic beads (provided in the sequencing kit).
Sequencing was performed on a MinION 10.4 cell according to Oxford Nanopore Technologies (ONT) instructions.Libraries were loaded into the Nanopore MinION Spot-on Flow Cell (FLO-MIN112, version R10.4) and sequenced for 10 h.The baseline call was performed automatically by the MinKnow program.Raw reads were obtained in FAST5 and FASTQ formats from which 'pass' quality reads were subjected to further analysis.
2.4.2.Environmental DNA data analysis 2.4.2.1.Raw data pre-processing.Reads were demultiplexed and adapters trimmed using MinKNOW Nanopore software.Primers and adapters were removed using Cutadapt with default settings and a 10% error tolerance.Quality was checked using Epi2m and FASTQC.Therefore, sequences were read quality filtered to a minimum of 6, while read length was limited to 145 bp using MinKNOW.The FASTQs of each sample were assembled into a single file using a "cat" command in a Linux environment.All analyses were performed in a unix-ubuntu 21.04 environment.The FASTQs were converted to FASTA using a cut adapter.
2.4.2.2.Taxonomic assignment of consensus sequences.To taxonomically assign the sequences in the fasta files, they were BLAST-ed against the fish specific MARE-MAGE database (12S North and Baltic Sea subdatabase) (Kasmi et al., 2023).The unassigned ASVs were aligned to the NCBI database (Benson et al., 2013).A BLASTN (Basic Local Alignment Search Tool) search (Chen et al., 2015) was performed using the GenBank nucleotide database and MARE-MAGE locally installed via the BLAST package version 2.10.0.The complete nucleotide database was downloaded (Feb 1, 2023), version December 2022 was used for the MARE-MARE database.An R script was applied to select the OTUs assigned to species, genus and family levels according to the percentage of identity, based on Fishbase taxonomy (Froese, 1990;Froese and Pauly, 2019).At similarities higher than 98% sequences were assigned to a species level, between 90 and 98% to a genus and at less than 90% to a family level.

Alpha and beta diversity.
The assessment of fish biodiversity from metabarcoding eDNA was carried out using measures of alpha and beta diversity.Alpha diversity, which refers to the intrinsic diversity of a specific habitat, was measured using Shannon and Chao1 indices, which take into account the abundance, evenness and potential richness of species.These measures were calculated using the 'vegan' package in R. Beta diversity, which captures differences between fish communities in different environments, was analyzed using non-metric multidimensional scaling (NMDS) and Bray-Curtis dissimilarity.NMDS is a rankbased method that represents pairwise dissimilarity between samples in a low-dimensional space, while Bray-Curtis dissimilarity calculates compositional dissimilarity between different sites.The beta diversity analysis was performed using the 'vegan' package in R, and the results were visualised using 'ggplot2'.

Data visualization.
Visualization of the results was performed in Krona and Rstudio using the ggplot and phylum packages.

Bottom trawl results
Fish species diversity and abundance were assessed through bottom trawl sampling at 12 designated stations in the Sylt Outer Reef (SOR).A total of 17 fish species were identified and quantified during the monitoring.The most abundant species observed were common dab (Limanda limanda), with 2427 individuals, followed by plaice (Pleuronectes platessa) with 1063 individuals.Scaldfish (Arnoglossus laterna) and solenette (Buglossidium luteum) were also relatively abundant with 798 and 620 individuals, respectively.Common dragonet (Callionymus lyra), hooknose (Agonus cataphractus), and sand goby (Pomatoschistus minutus) were present in moderate numbers, while the remaining species were observed in lower numbers, ranging from 19 to 1 individual (Fig. 2).

eDNA using metabarcoding
To reveal the ichthyodiversity in the Natura 2000 site Sylt Outer Reef, a metabarcoding method targeting the 12S rRNA gene was applied to 24 eDNA water samples from 12 stations.In total, two million reads were generated by the Nanopore Minion sequencer.0.5 million reads (28.54%) were discarded due to the presence of chimeras, quality score and copy number, while 1.5 million reads (72.55%) were retained for ASV (Amplicon Sequence Variants) analysis.Of these, 42% could be assigned to distinct fish species, while 58% were assigned as unknown fish sequences.All sample files were of good quality for reads up to 260 bp.The percentage of fish reads ranged from 50% to 83% per sample, with a median of 72%.Negative controls were all without fish reads.The BLAST algorithm was used to taxonomically assign the ASVs, using the GenBank database.
The BLAST results identified diverse fish species with differences in species composition among the 12 sites sampling sites.A total of 36 fish species were detected (Table S1 and Fig. S1 A and B), with the higher abundant reads: The Atlantic mackerel (Scomber scombrus), cod (Gadus morhua) common dab (Limanda limanda), common dragonet (Callionymus lyra), European pilchard (Sardina pilchardus), grey gurnard (Eutrigla gurnardus) and Raitt's sand eel (Ammodytes marinus) accounting for 70% of all assigned ASVs (Fig. 2 A).Common mola (Mola mola), common sole (Solea solea), European anchovy (Engraulis encrasicolus), European eel (Anguilla anguilla), Greenland halibut (Reinhardtius hippoglossoides), haddock (Melanogrammus aeglefinus), lemon sole (Microstomus kitt), lesser weever (Echiichthys vipera), short-snouted seahorse (Hippocampus hippocampus) and turbot (Scophthalmus maximus) has much less reads, each representing less than 1% of the total fish ASV.About 98% of the ASVs were classified as bony fishes (Actinopterygii).Nevertheless, some sequences of mammals, namely of the harbor porpoise (Phocoena phocoena), were also detected, representing on average 1% of the relative abundance.The BLAST-results also revealed two non-native, North Pacific species as false-positives: The Pacific herring (Clupea pallasii pallasii) (132 reads) and Alaska pollock (Gadus chalcogrammus) (93 reads).A phylogenetic analysis revealed that Mifish primer-based 12S-sequences are unable to distinguish b e tween these two species and their North Atlantic congeners (Clupea harengus and Gadus morhua) because the similarities among the two species-pairs is less than 2% (1-3 mutations in the 180 pb fragments).Considering the known error rate of the Nanopore technology, Pacific herring and Alaska pollock reads were therefore assigned to Atlantic herring and Atlantic cod, respectively.
The alpha and beta diversity indices were calculated in order to characterize the fish diversity at the different stations.Alpha-diversity analysis revealed a mean richness of 35 and a Shannon diversity of 2.9., whereas the beta-diversity analysis showed significant differences between samples (p < 0.05), indicating relative spatial variation in the fish community composition (Fig. 3a).

Method comparison
The eDNA-analysis of water samples included almost all species caught by bottom trawling (Fig. 3), except for three species: the scaldfish (Arnoglossus laterna), reticulated dragonet (Callionymus reticulatus) and solenette (Buglossidium luteum) which at the time of the analysis were not available in any public genetics databases with regard to their mitochondrial 12S gene.However, considerably more species were detected through eDNA analyses compared to bottom trawling.Fig. 3b A total of 22 species were exclusively found by eDNA analysis, among them various pelagic species known to be underrepresented in beam trawls (Fig. 4).With the Harbour porpoise (Phocoena phocoena), even the presence of one marine mammal was confirmed.

Discussion
In the quest to understand and preserve marine biodiversity, the utilization of advanced scientific methods has proven pivotal.Environmental DNA analysis could be a game-changer in this context.However, case studies showing the general efficiency of the use of eDNA as compared to standardized reference methods in the marine realm are still scarce.The present study employed metabarcoding technology to assess fish diversity at 12 sampling stations in a Natura 2000 site in the North Sea.
The results show a significant difference in the number of detected fish species using eDNA metabarcoding compared with traditional bottom trawling.The metabarcoding approach detected a total of 36 species of fish as well as one marine mammal in the Sylt Outer Reef (SOR), whereas the beam trawl caught only 17 species.Many of the additional species detected with eDNA metabarcoding were pelagic and are -by nature -not caught with a beam trawl, which underpins the high efficiency of the method.The high concordance rate of 87% between trawling and eDNA metabarcoding further emphasizes the reliability and accuracy of the genetic approach.This suggests that eDNA metabarcoding can serve as a robust and reliable complement to traditional survey methods, depending on the study objectives.Using short reads technology (Illumina NovaSeq 6000) on the same area, Barco et al. (2022) detected 24 fish species at 16 stations in the SOR using eDNA metabarcoding, which may be due to a higher threshold (99%) for alignment with reference data.The eDNA outputs of that study were compared to the online Database for Trawl Surveys (DATRAS) of the International Council for the Exploration of the Sea (ICES), in which it was found that 37% of the species were detected with both methods, which is in good agreement with our results (Barco et al., 2022).

Method comparison
A major advantage of eDNA analyses is undoubtedly its non-invasive nature.The use of classical fisheries methods potentially poses a risk especially for rare or even protected species, like marine mammals, and can further have a negative impact on habitats.Zhang et al. (2023) demonstrated that eDNA metabarcoding based on Mifish primers originally targeting fish diversity is also capable to detect whales, verified by the parallel use of species-specific primers and the visual observation.Therefore, the detection of the Harbour porpoise in the present study must be considered a valid observation.Valsecchi et al. (2021) already proposed primers for marine mammals designed on the basis of Mifish.In the case of cartilaginous fishes (sharks, skates and rays), Miya (2022) and Miya et al. (2015) proposed a slightly adapted primer set (Mifish-E).Despite using Mifish-E in this study, no cartilaginous species were detected.
One of the main obstacles for a standardized use of eDNA metabarcoding in biodiversity monitoring is the incomplete availability of quality-assured reference sequences in DNA databases.This can result in false statements about species absences and therefore negative findings.However, the available databases do often also not reflect genotypic plasticity of single species and, as a consequence, possible overlaps between closed related species.This often results in false positive results, so the erroneous detection of non-native species to a certain region, in our case of Pacific herring (Clupea pallasii) and Alaska pollock (Gadus chalcogrammus), highlighting the limitations of the Mifish primers in distinguishing between some North Sea and Pacific Ocean species due to   high similarity across the target region.
An eDNA advantage is its long half-life compared to reference methods: several studies on the kinetics of eDNA degradation in fresh and salt water show that eDNA has a half-life of 24 h, whereas trawling only covers a few minutes, with a maximum of 1 h (Collins et al., 2018;Harper et al., 2019).However, the eDNA decay can vary from taxon to another or depending to the biotic and abiotic factors (Scriver et al., 2023).Therefore, more research should be addressed to two fundamental questions: i) the persistence of eDNA in the water column dependent on physical and chemical parameters; and ii) the influence of hydrodynamic features, such as currents, on eDNA dispersal.Understanding these factors is important for an accurate interpretation of eDNA data and for inferring about the targeted area of interest.
However, it is crucial to acknowledge some limitations of eDNA.While eDNA metabarcoding offers a presumably rather comprehensive overview of fish diversity, it provides limited information on species abundance and individual sizes.Additionally, accurately identifying certain fish species based solely on their DNA sequences, especially for closely related taxa, remains a challenge.To address these limitations, future research efforts should focus on refining and expanding reference databases to enhance the accuracy of species identification.It is also important to note that eDNA analyses have a detection limit for each species, although it is rarely mentioned in eDNA research.This detection limit refers to the minimum amount of DNA from a particular species that must be present in a sample for it to be reliably detected.When the amount of DNA falls below this limit, the species may not be detected in the sample, leading to potential false negatives in the analysis.Therefore, understanding and accounting for the detection limit is crucial for accurately interpreting eDNA results and determining the presence or absence of species in a given environment.This may also explain the absence of three species in eDNA output, despite being present in the trawl catches.
Despite these limitations, the present results demonstrate the effectiveness of eDNA metabarcoding as a powerful tool for assessing fish diversity in the North Sea.The advantages of this approach over traditional survey methods are evident, providing a more complete understanding of species composition and enabling the detection of rare species.Integrating eDNA analysis into conservation and management strategies holds immense promise for the preservation and sustainable use of marine resources in the Natura 2000 network and beyond.

eDNA: comparison between long and short read sequencing technologies
The essential advantage of analysing eDNA with long-read technology, such as Nanopore, is their increased sensitivity for discriminating between closely related species (Davidov et al., 2020;De Vivo et al., 2022;Fonseca, 2018;Franco-Sierra and Díaz-Nieto, 2020;Gaonkar and Campbell, 2024;Stevens et al., 2023;Toxqui Rodríguez et al., 2023) and for reducing false positives as well as taxonomic artefacts as compared to short reads (Ramírez-Amaro et al., 2022).Thanks to continued progress in the development of long-read sequencers and chemistry, the reduction in error associated with these techniques promises to address more complex issues in biodiversity monitoring (Bogaerts et al., 2024;Stevens et al., 2023;T. Zhang et al., 2023), including quantitative assessments based on random short repeat sequences or micro-satellites (Adams et al., 2019;Andres et al., 2021Andres et al., , 2023;;Couton et al., 2023;Hellström et al., 2023).By applying a ligation approach, long-reads also open up the possibility of PCR-free library preparations to prevent any PCR bias (Fonseca, 2018;Manu and Umapathy, 2023).

Conclusion
The application of metabarcoding in eDNA analysis has provided valuable insights into species distribution and relative abundance.However, to further enhance our understanding, it is crucial to combine eDNA approaches with other non-invasive methods.This integrated approach holds great potential in supporting sustainable fisheries management and conservation strategies in the future.
The findings from the metabarcoding approach have revealed a significantly higher number of fish species compared to bottom trawling with a beam trawl, highlighting the limitations of conventional sampling techniques in capturing the full spectrum of species present.However, limitations have been observed in accurately quantifying species abundance using eDNA metabarcoding alone.These results contribute to the expanding body of knowledge on the application of eDNA in monitoring fish populations, forming a foundation for informed decision-making and the development of conservation strategies.
Further research is necessary to refine species identification, address limitations in quantification methods, and validate the models developed across different regions and fish species.The integration of eDNA analysis into conservation and management frameworks, such as the Natura 2000, holds great promise for the sustainable use of marine resources and the preservation of biodiversity.

Funding
This work is funded by the German Federal Ministry of Education and Research (BMBF) as part of the DAM Pilot Mission (MGF-Nordsee) (03F0847C).

Fig. 1 .
Fig. 1.Sampling sites in the Sylt Outer Reef (SOR; green line), German Bight North Sea.Dashed line is the German Exclusive Economic Zone.

Fig. 2 .
Fig. 2. Percentage of fish species by station sampled by a 3m beam trawl.

Fig. 3a .
Fig. 3a.Chao1 and Shannon indices as measures for Alpha diversity.The error bars in the left panel represent the variability or uncertainty of the Chao1 index.On the right represents the alpha diversity measured using the Shannon index for different samples.

Fig. 4 .
Fig. 4. Venn diagram showing the intersection between eDNA and catch results (per fish).In global, 22 species were detected only by eDNA, 14 species were detected by eDNA and Trawl, while 3 species were detected only by the trawl.