Pilot study of a comprehensive resource estimation method from environmental DNA using universal D-loop amplification primers

Many studies have investigated the ability of environmental DNA (eDNA) to identify the species. However, when individual species are to be identified, accurate estimation of their abundance using traditional eDNA analyses is still difficult. We previously developed a novel analytical method called HaCeD-Seq (haplotype count from eDNA by sequencing), which focuses on the mitochondrial D-loop sequence for eels and tuna. In this study, universal D-loop primers were designed to enable the comprehensive detection of multiple fish species by a single sequence. To sequence the full-length D-loop with high accuracy, we performed nanopore sequencing with unique molecular identifiers (UMI). In addition, to determine the D-loop reference sequence, whole genome sequencing was performed with thin coverage, and complete mitochondrial genomes were determined. We developed a UMI-based Nanopore D-loop sequencing analysis pipeline and released it as open-source software. We detected 5 out of 15 species (33%) and 10 haplotypes out of 35 individuals (29%) among the detected species. This study demonstrates the possibility of comprehensively obtaining information related to population size from eDNA. In the future, this method can be used to improve the accuracy of fish resource estimation, which is currently highly dependent on fishing catches. Supplementary Information The online version contains supplementary material available at 10.1007/s10142-023-01013-3.


Introduction
The advent of high-throughput sequencing has enabled the assessment of biodiversity in the environment using the genetic information contained in environmental DNA (eDNA) in water, soil, and air (Rees et al. 2014;Thomsen and Willerslev 2015;Ruppert et al. 2019;Garlapati et al. 2019;Clare et al. 2021). Many studies have estimated abundance using eDNA (Knudsen et al. 2019;Salter et al. 2019;Stoeckle et al. 2020). However, biomass estimation using previously reported eDNA had some limitations. One of the major problems is that different quantities of cells are released into the environment from different individuals, even in the same species and in the same environment, due to variations in size and age-class distributions (Maruyama et al. 2014) as well as alterations in environmental factors such as water temperature and flow (Jane et al. 2015). We recently developed a novel method for the reliable and simple estimation of fish population abundance from eDNA using the D-loop haplotype count for Anguillids (Yoshitake et al. 2019) and Pacific bluefin tuna Thunnus orientalis (Yoshitake et al. 2021). This method, called HaCeD (haplotype count from eDNA)-Seq, was successfully adopted to noninvasively estimate the abundances of various eels or tunas inhabiting the same experimental tank using eDNA. There were several problems for conventional eDNA survey methods with 12S rRNA because the different quantities of cells are released into water due to variations in size and age-class distributions (Maruyama et al. 2014) as well as alteration in environmental factors such as water temperature and flow (Jane et al. 2015). On the other hand, HaCeD-Seq is a method that can directly measure the number of individuals in eDNA, taking advantage of the fact that the D-loop sequence differs from one individual to another in eels and tuna.
HaCeD-Seq is not limited to eels and tuna, but can be applied to any organism with high mitochondrial D-loop diversity. High haplotype diversity of D-loop has already been reported for some commercially important small pelagic fish such as jack mackerel (0.964-1.000) (Song et al. 2013), chub mackerel (0.505-0.967) (Zhu et al. 2016), and spotted mackerel (0.996) (Tzeng 2007) for which the total allowable catch (TAC) has been set in the fishery management system of Japan (Ichinokawa et al. 2017). The haplotype diversity of salmon Oncorhynchus masou returning to their home rivers is low (Okabe et al. 2020), but the haplotype diversity of many marine fishes is high (Sang et al. 1994;Ishikawa et al. 2001;Tzeng 2007;Song et al. 2013;Nomura et al. 2014;Zhu et al. 2016;Kumar et al. 2016).
HaCeD-Seq could be used for stock assessments of above major target species with high haplotype diversity in fisheries. However, unresolved issues remain. One problem is that primers for amplifying the D-loop are species-specific. Because it is laborious to design primers and count the D-loop haplotypes for each target species, we aimed to design generic D-loop primers. In addition, although the 12S rRNA gene sequences of many species have been sequenced and registered in public databases for metabarcoding analysis, such as MiFish primers (Miya et al. 2020), D-loop sequences have not sufficiently been sequenced. It is necessary to feed databanks with D-loop sequences as reference data. In this study, we efficiently sequenced whole mitochondrial genomes, including the D-loop, as reference sequences and registered the D-loop sequence in the GenBank database. Although universal D-loop primers designed based on D-loops from 161 fish species have been reported (Cheng et al. 2012), we redesigned universal primers for the D-loop based on complete mitochondrial sequences of 2854 fish species and validated a method to obtain population information for multiple fish species.

Universal D-loop primer design
We downloaded the MitoFish database (2020/2/20) (Iwasaki et al. 2013), containing 2854 species of complete mitochondrial sequences. The D-loop sequence and the surrounding 1000 bp sequence of the Japanese eel Anguilla japonica (Accession ID: AB038556) were extracted, and a homology search was performed using blastn (2.6.0 +) (Camacho et al. 2009) to MitoFish database with default parameters. Sequences of the hit regions were extracted, and multiple alignments were performed using Mafft (v7.402) (Katoh et al. 2002). The consensus sequences of the positions that were aligned in more than 90% of the species were extracted, and the percent identity of 2854 species and the consensus sequences was calculated. Since a simple percent identity for each position did not tell us whether the region was continuously conserved or not, we calculated the gap-considered identity score by subtracting the bases in the region that were aligned less than 90% of the species from the identity bases following the gap. Multiple alignments were displayed in Geneious Prime 2020.1 (Biomatters), and highly conserved regions were extracted. Sites with less than 85% conservation were designed as mixed bases in the primers. Sequence conservation was visualized using WebLogo (Crooks et al. 2004). To test the designed primers in silico, primer sequences covering all patterns of mixed bases were prepared in a FASTA file, and blastn was performed with the options of word_size 4 and max_target_seqs 10,000 against the MitoFish database.

Library preparation and sequencing for mitochondria
We decided to sequence mitochondria from 45 species of fish kept at Tokyo Sea Life Park and whose DNA is preserved at the Chiba Prefectural Museum. DNA concentrations were quantified by electrophoresis and absorbance, and DNA extracts were obtained from 45 fish (Table 1) and mixed to achieve equal DNA concentrations. After library preparation using the Nextera DNA Flex Library Preparation Kit (Illumina), one-lane sequencing was performed using a HiSeqX sequencer at Macrogen (Seoul, Korea).

Mitochondrial full-length sequence analysis
Whole genome shotgun sequencing data from a mixture of 45 species were assembled using CLC Genomics Workbench 8 (QIAGEN) and Megahit v1.1.1 (Li et al. 2015), respectively. The assembled contigs were subjected to a BLASTN homology search against MitoFish, and contigs with > 70% homology and > 10 kb length were extracted. Extracted contigs were annotated using MitoFish Annotator (Iwasaki et al. 2013) to remove noisy contigs without 12S rRNA and D-loop. Subsequently, the contigs were compared with the MitoFish and NCBI nt databases to determine the species of the contigs based on some of the registered information, such as 12S rRNA, and the mitochondrial sequences were cleaned up manually by removing chimeras. A total of 45 mitochondrial sequences were obtained and registered in GenBank except for five that were almost exact matches to known sequences (accession numbers: OK554506-OK554545).

Water sampling and eDNA extraction
Two liters of water was collected from tank 36 at Tokyo Sea Life Park and filtered through a 0.45 µm Sterivex filter (Merck) following the eDNA Society Manual (https:// ednas ociety. org/ en/ manua ls/). Briefly, after filtering the water with a syringe manually, the filter was stored with RNAlter (Thermo Fisher), and the next day, after removing the RNAlter, 200 µL of Buffer AL of Dneasy Blood and Tissue kit (QIAGEN), 220 µL of phosphate-buffered saline (PBS) (-), and 20 µL of proteinase K (20 mg/mL) were added. The mixture was sealed with Parafilm and rotated at 56 °C for 20 min. The inlet of the Sterivex filter was inserted into a 2.0-mL tube and centrifuged in a 50-mL tube at 6000 g for 1 min. The eluted solution was added with 200 µL of 100% ethanol and transferred to a column. After centrifugation at 6000 g for 1 min, 500 µL of Buffer AW1 was added to the column and centrifuged at 6000 g for 1 min. Next, 500 µL of Buffer AW2 was added to the column, which was centrifuged at 2000 × g for 3 min. Finally, the column was added with 200 µL of Buffer AE, which was left at room temperature for 1 min, centrifuged at 6000 × g for 1 min, and eDNA was collected.

Amplification of D-loop region from eDNA for Nanopore sequencing
The adapter design of the primers for Nanopore sequencing with UMI followed the previous paper (Karst et al. 2021), adding N13 to both ends of forward and reverse primers. The first PCR to introduce the unique molecular identifiers (UMI) was performed under the following conditions: We mixed 10 µL of repliQa HiFi ToughMix (Quantabio), 0.6 µL of 10 µM forward primer (D-loop-F-UMI: CTC TTC CGA TCT GTCNNNNNNNNNNNNNARA GCR YCG GTC TTG TAA ), 0.6 µL of 10 µM reverse primer (D-loop-R-UMI: CTC TTC CGA TCT CAGNNNNNNNNNNNNNCG-GAKACT TGC ATG TRT AA), 1 µL of template DNA, and 7.8 µL of nuclease-free water for a total of 20-µL reaction mixture. After denaturation at 95 °C for 3 min, five cycles of amplification were carried out at 98 °C for 10 s, 40 °C for 5 s, and 68 °C for 5 s, followed by extension at 68 °C for 1 min. The samples were then loaded onto 1.7% agarose gel and subjected to electrophoresis at 50 V for 30 min. The gel mold for electrophoresis was printed using an ELEGOO MARS PRO 3D printer (Elegoo, Shenzhen, China) with the model of Supplementary Information 1, and 3D printer resin SK01F (SK HONPO, Tokyo, Japan). The 3D model was designed using a Fusion 360 (Autodesk). A disposable electrophoresis chamber was created to prevent contamination ( Supplementary Fig. 1). A disposable electrophoresis chamber was made using knock-off pencil refills 2.0 HB (Takumi, Tokyo, Japan) and a silicon lunch square cup C-4974 (PEARL METAL, Tokyo, Japan). After electrophoresis, the gel with the size of 800-1500 bp was cut out and DNA was purified using the FastGene Gel/PCR Extraction Kit (NIPPON Genetics) and extracted with 20 µL of nuclease-free water.
A second PCR to amplify the D-loop with UMI was performed under the following conditions: We mixed 10 µL of repliQa HiFi ToughMix, 0.6 µL of10 µM forward primer (UMI-2nd-F: TAC ACG ACG CTC TTC CGA TCT GTC ), 0.6 µL of 10 µM reverse primer (UMI-2nd-R: AGA CGT GTG CTC TTC CGA TCT CAG ), 2 µL of template DNA, and 6.8 µL of nuclease-free water for a total of 20 µL. After denaturation at 95 °C for 3 min, 45 cycles of amplification were carried out at 98 °C for 10 s, 40 °C for 5 s, and 68 °C for 5 s, followed by extension at 68 °C for 1 min. The PCR product was then loaded onto a 1.7% agarose gel in a disposable electrophoresis chamber, and after electrophoresis at 50 V for 30 min, the band with a size of 800-1500 bp was cut out and purified using a FastGene Gel/PCR Extraction Kit with the elution of 20 µL of nuclease-free water.

Sequencing with Oxford Nanopore
Nanopore sequencing was performed using a MinION flow cell R10.3 and Ligation kit SQK-LSK110 (Oxford Nanopore Technologies). Fast5 files were created by Nanopore, transferred to a Linux computer with GPU, and basecalled using Guppy v5.0.11 (Oxford Nanopore Technologies) with the super-accuracy basecalling model.

Amplification of D-loop region from eDNA for LoopSeq
Using the same eDNA samples sequenced by Nanopore, we mixed 0.2 µL of Ex Taq (5 U/µl) (Takara), 2 µL of 10 × Ex Taq Buffer, 1.6 µL of 2.5 mM dNTP Mixture, 0.5 µL of Template DNA, 1 µL of 10 µM forward primer (D-loop-F-UMI), 1 µL of 10 µM reverse primer (D-loop-R-UMI), and nuclease-free water up to 20 µL. After denaturation at 94 °C for 2 min, 40 cycles of amplification were carried out at 94 °C for 30 s, 50 °C annealing temperature for 30 s, and 72 °C for 80 s, followed by extension at 72 °C for 5 min. The PCR product was then loaded onto a 1.7% agarose gel, and after electrophoresis, the band with a size of 1000-1500 bp was cut out and purified using the FastGene Gel/PCR Extraction Kit with the elution of 20 µL of nuclease-free water. LoopSeq libraries were prepared using the LoopSeq PCR Amplicon 24 sample Kit (Loop Genomics) and sequenced with HiSeqX 1 lane at Macrogen. Sequenced data were uploaded to the LoopSeq website, and synthetic long reads were generated.

Universal HaCeD-Seq analysis
Only Nanopore reads that retained at least 80% homology were extracted from sequences that annealed around the D-loop region of the 1st PCR primer and the 2nd PCR primer sequence. If the UMI tag regions matched more than 80% identities with the more abundant tags, they were grouped as originated from the more abundant UMI. If a UMI pair shared one forward or reverse UMI with another UMI pair and had a different UMI on the opposite side, the UMI pair with the highest number of reads was adopted, and the other UMI pairs were removed as UMI chimeras. One hundred reads were extracted for each extracted UMI pair, and multiple alignments were performed using MUS-CLE (Edgar 2004) to create consensus D-loop sequences. Using these consensus D-loop sequences and all Nanopore reads, we created a more accurate consensus D-loop sequence using MEDAKA v1.3.2 (https:// github. com/ nanop orete ch/ medaka). We also created a BLAST database by adding our additional 45 mitochondrial sequences to the MitoFish database (2020/2/20) (Iwasaki et al. 2013) and performed blastn on the consensus D-loop sequences to determine species names. Since MEDAKA did not appear to completely remove errors, we decided to check whether the haplotypes that were polished as different haplotypes were really different haplotypes. When reads derived from a haplotype are compared to another haplotype, the mutation should be called if the haplotype is really different. Mutations were called by MEDAKA_Haploid_variant using the haplotype of the cluster with the highest number of reads for each species as the reference genome, and the reads of haplotypes with smaller clusters for correction. If no mutation was called for the reads from a small cluster, the actual haplotype of the small cluster was considered identical to the highest one, so the haplotypes for which no mutation was called were removed as noise. Once the calculation was performed for the haplotype of the cluster with the highest number of reads, the process was repeated for the haplotype with the largest number of reads among the unmerged haplotypes. Finally, the number of haplotypes was calculated for each species. All the above steps were published as the eDNA ~ HaCeD-Seq_with_UMI_using_Nanopore script in the Portable Pipeline (https:// github. com/ c2997 108/ OpenP ortab lePip eline/ blob/ master/ Porta blePi peline/ scrip ts/ eDNA~HaCeD-Seq_ with_ UMI_ using_ Nanop ore).

Design of universal primers
D-loop sequences were extracted from 2854 full-length mitochondria registered in the MitoFish database (Iwasaki et al. 2013), and multiple alignments were generated. The tRNA-Thr, tRNA-Phe, and 12S rRNA were the most highly conserved and existing regions at both ends of the D-loop (Fig. 1A, B). To compare which region is more conserved between tRNA-Phe (Fig. 1B, 579-598 bp) and 12S rRNA (Fig. 1B, 671-689 bp), we designed tRNA-Phe-rev: CAT CTT CAG TGY YAT GCT TT as a primer candidate sequence for tRNA-Phe region and 12S rRNA-rev: CGGAKACT TGC ATG TRT AA as a primer candidate sequence for 12S rRNA region. The results of blast search considering the degenerate bases of each primer showed that 2747 (96.3%) of 2853 species were hit by tRNA-Phe-rev, while 2817 (98.7%) of 2853 species were hit by 12S-rRNA-rev. Therefore, we decided to adopt the 12S-rRNA-rev sequence. Primer sequences were designed for each of these highly conserved together with existing regions (Fig. 2A), with N13 as the UMI tag and adapter sequence for the 2nd PCR. The average length of the amplified DNA fragments was 1238 bp, the length of the lower 25% was 1107 bp, and the length of the upper 25% was 1261 bp (Fig. 1C). A schematic representation of the final PCR products after 2nd PCR is shown in Fig. 2B.

Construction of mitochondrial reference sequences
The 12S rRNA region in mitochondria is often partially sequenced, and there are many reference sequences, but validated D-loop reference sequences are currently insufficient to support species assignment. The copy number of mitochondria is tens to hundreds of times greater than that of the nuclear genome (Hartmann et al. 2011;Fan et al. 2022), and full-length mitochondria can be easily assembled by lowcoverage shotgun whole genome sequencing. Therefore, we attempted to add the D-loop sequence of fish species living in the seas around Japan to the database by sequencing the complete mitochondrial genome using thin shotgun wholegenome sequencing ( Table 1). As a result, we were able to determine mitochondrial genome sequences of more than 10 kb, including the whole D-loop for 45 species. Although only partial sequences of the mitochondrial genome were registered in the public database for 18 of the 45 species and D-loop sequences were unknown, we were able to determine D-loop sequences with our sequencing data. We registered 40 mitochondrial sequences that were novel or had many polymorphisms compared to the GenBank database, excluding five species for which the sequences were nearly identical to known sequences (Accession ID: OK554506-OK554545).

PCR and sequencing of D-loop
To evaluate the designed primers, we conducted a study in tank 36 of Tokyo Sea Life Park, where 14 species of fish were kept (Fig. 3). A two-step PCR was performed to add UMI to eDNA extracted from water in tank 36, resulting in a broad band around 1.2 kbp (Supplementary Fig. 2A). PCR products were sequenced using a Nanopore MinION flow cell. The sequenced nanopore reads contained 1,533,199 reads. The obtained read-length and quality distributions are shown in Fig. 4. In addition to Nanopore, we used LoopSeq, which is capable of synthetic long-read sequencing using Illumina, for validation. The use of custom UMI tags is not recommended for LoopSeq because the library conditioning kit for LoopSeq uses another UMI. Therefore, we performed a normal one-step PCR without UMI for LoopSeq and obtained a band around 1.2 kbp (Supplementary Fig. 2B). The PCR product was sequenced for the full-length D-loop using a LoopSeq library preparation kit. The reads obtained were 50,663 reads.

Data analysis
The analysis flow of nanopore-sequenced reads is shown in Fig. 5. The sequenced reads were reduced to 12% when searching for UMI tags. The accuracy of the Nanopore sequencer is as low as 90% (read quality: 10) (Fig. 4), so many reads were lost, but this could be improved in future Nanopore updates with improved read quality (Luo et al. 2022). When reads with chimeric UMIs at both ends were removed, the remaining reads were reduced to 2%. Forty-nine consensus sequences were created by collecting sequences with the same UMI. These 49 consensus sequences were supposed to be unique molecules and were corrected with Nanopore reads using Medaka. The UMI tags were then removed, and the same haplotypes were merged. The 29 candidate haplotypes were obtained and annotated using the species names. Although Medaka eliminated many sequencing errors, several errors remained when comparing raw Nanopore reads. Therefore, we mapped raw Nanopore reads of other haplotypes of the same species to the most abundant Medaka-corrected consensus sequence of the species, and checked again whether other haplotypes of the species were corrected by Medaka to be the original haplotypes. Out of 29 haplotypes, 10 duplications were found, and 19 haplotypes were finally obtained (Table 2; Supplementary Table 1). Finally, 5 species (Alectis ciliaris, Evistias acutirostris, Labracoglossa argentiventris, Parupeneus ciliates, and Trachurus japonicus) were detected in 19 haplotypes. We found 5 out of 15 species (33%) for the detection sensitivity of fish species and 10 haplotypes out of 35 individuals (29%) for the detection sensitivity of populations among the detected species (Table 3). Eight of the 19 haplotypes were also confirmed by LoopSeq, another long-read sequencing method using an Illumina sequencer ( Table 2). The most abundant haplotypes of each species identified by Nanopore were confirmed by LoopSeq sequenced reads. The detection of both Nanopore and LoopSeq suggested that

Discussion
In this study, we designed D-loop universal primers and demonstrated a comprehensive method for estimating fish population using eDNA. We named this version of the protocol Universal HaCeD-Seq (ver. 1). We found 5 out of 15 species (33%) for the detection sensitivity of fish species and 10 haplotypes out of 35 individuals (29%) for the detection sensitivity of populations among the detected species. It has been known that the population size of Anguillids (Yoshitake et al. 2019) and Thunnus orientalis (Yoshitake et al. 2021) can be estimated from environmental DNA using HaCeD-Seq. This study shows the possibility of simultaneously determining the population size of multiple species by a single experiment. The MiFish primer (Miya et al. 2020), which is commonly used in environmental DNA analysis, can determine the presence or absence of a target organism, but Universal HaCeD-Seq can also provide information on the number of individuals. However, while the MiFish primers were able to detect 93.3% of the fish species in the tank (Miya et al. 2015), Universal HaCeD-Seq was only able to detect 33%. This could be because the longer eDNA fragments degrade more quickly (Jo et al. 2017). The PCR fragment length of Universal HaCeD-Seq is 1 kbp which is longer than 200 bp of MiFish primers. In the future, it is expected to improve the accuracy of the estimation of the abundance of whole fish species to combine the advantages of the species detection of MiFish primers and the population size estimation of Universal HaCeD-Seq.
We also showed that thin whole-genome sequencing can efficiently determine the full-length mitochondrial sequences of 45 species. Only the 12S rRNA gene sequences Fig. 4 Read length and quality distribution. The read length and quality distribution of the sequenced Nanopore reads are shown. This graph was drawn using NanoPlot version 1.32.1 The mean phred quality score of the Nanopore reads was 11.9 (Fig. 4), so the average quality of the Nanopore reads was 94%, which was lower than that of the Illumina reads. Only 181,661 reads (12%) could be used for analysis using one Nanopore MinION flow cell because we eliminated 88% of the reads that did not have clear adapter sequences   (Fig. 5). This is expected to improve dramatically with the use of a more accurate nanopore flow cell that will become available in the near future (Luo et al. 2022). PacBio's HiFi sequencing is also available as a highly accurate long read sequence (Wenger et al. 2019;Patat et al. 2022), but is still relatively expensive. We also used LoopSeq as a synthetic long-read method using Illumina, but this method uses UMI tags in the kit, so it is not possible to use custom UMI tags. Therefore, errors during sequencing can be eliminated, but not during PCR using LoopSeq. Nanopore sequencing with UMI seems to be more accurate than LoopSeq because twostep PCR allows for PCR error removal using UMI tags. However, because the cost per base for nanopore sequencing is much higher than that of Illumina, a sequencing method using Illumina is desirable. However, the LoopSeq PCR Amplicon Sample Kit was discontinued in 2020 and is currently unavailable. In the future, it will be necessary to go back to BAsE-Seq (Hong et al. 2014) or other published synthetic long-read methods (Stapleton et al. 2016) which are the bases of LoopSeq, to establish a library preparation method for HaCeD-Seq. The 1st PCR is the step of adding UMI, and only 2 cycles are ideal to add unique UMI per molecule. However, in this study, the number of cycles of 1st PCR was 5, taking into account that DNA was lost during the gel purification after 1st PCR. These additional 3 PCR cycles might increase the number of apparent molecules up to 2 3 = 8 times. The increase in the number of apparent molecules decreases the number of reads per UMI, making it difficult to create a consensus sequence by reads derived from the same UMI. However, 29 UMIs were remained after removal of the UMI chimeras from the 49 UMIs, and 19 independent D-loop haplotypes were detected. Therefore, there were not so many chimeric UMIs, but if the 1st PCR could be performed in fewer cycles, the UMI chimeras would be suppressed, and the detection sensitivity of D-loop haplotypes would be improve.
The fish species detected in this study swim near the surface of the tank (Fig. 3). Because the sampled water was taken from the surface layer of the tank using a bucket, it is reasonable to detect many fish species near the surface layer. However, we were not able to detect fish swimming in the lower layers of the tank, and future work is needed to improve the detection sensitivity. We believe that the reason for the low detection sensitivity was the low number of valid reads (181,661 reads) (Fig. 5), which meant that fish species with low eDNA concentrations could not be detected. We expect that an increase in the number of valid reads in the future will increase the detection sensitivity and enable the detection of more species.
Fish collected from the natural environment of the sea around Japan were kept and displayed in tank 36. Because of the restriction of not being able to collect DNA from fish on display, we were unable to confirm the D-loop sequence of each individual fish, so we do not know about haplotype overlap, but seven haplotypes out of 25 fish were identified in the Labracoglossa argentiventris, indicating that at least seven Labracoglossa argentiventris were present based on eDNA. The species of fish detected in this study was the first handled in our laboratory, and the risk of contamination is considered low because mitochondria were sequenced to determine the reference sequence after eDNA sequencings were performed. This method has promise as a method for determining the lower limit of the number of individuals present in the environment, which is not possible with conventional eDNA research methods that target 12S rRNA genes. As data on D-loop haplotype diversity for each fish species becomes available, it will be possible to estimate the number of individuals more accurately from the number of haplotypes detected, taking into account the degree of overlap.
This study demonstrates the possibility of comprehensively obtaining information related to population size from eDNA. In the future, we believe that this method can be used in combination with 12S rRNA-based survey methods, such as the MiFish primer, which has high detection sensitivity, to greatly improve the accuracy of estimating methods of fish resources that are currently highly dependent on fishing catches.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.