Environmental DNA Analysis in a River Detected a Possible Distribution of Fish Species Difficult to Capture

Abstract

Environmental DNA (eDNA) analysis is a biological survey method that has drawn much attention in recent years. However, the results of eDNA analysis and capture surveys often do not completely match, and the validity of the eDNA analysis needs to be verified. Verification of eDNA metabarcoding was conducted in a river in Fukushima Prefecture, Japan, in comparison with capture survey data. Most of the captured species were detected, and 13 uncaptured lineages (two genera and 11 species) were detected in the eDNAs. Some rare species detected in the eDNAs were also identified, including exotic eels and an endangered yet out-of-range bitterling fish. To confirm the validity of the exotic Anguilla spp. detected in eDNAs, mitochondrial Cytb sequencing was performed on captured eel specimens. All eel specimens were sequenced as the native Anguilla japonica, indicating a small biomass of the exotic species difficult to capture. Our results clearly indicated the eDNA analysis as a powerful tool for detecting possible habitats of rare fish species without disturbance to the natural environment.

1. Introduction

Environmental DNA (eDNA) analysis is a rapidly developing biodiversity monitoring technique that uses polymerase chain reaction (PCR) to amplify DNA fragments released into the environment from plants, fish and animals, and helps inferring their distribution [1]. The detection of eDNAs has been possible in a variety of carriers, including water, soil, feces, air, or ancient sediments and ice cores [2,3,4]. Most of the previous eDNA studies have used water samples collected from aquatic environments such as oceans, ponds and rivers [5,6,7,8,9].

Conventional aquatic monitoring is highly invasive to captured organisms because it uses fishing gears such as cast nets, hand nets and electrofishers [10]. Furthermore, capture surveys require significant effort, time and cost [11]. Additionally, there is also the issue of data bias due to differences in the investigators’ experience, knowledge and other skills. On the one hand, eDNA analysis is a non-invasive technique that can save the time and cost of surveys per site compared to capture surveys. As such, the eDNA analysis is an important survey method that is not only used for detecting early invasions or rare species, but also for conventional monitoring in aquatic organisms [12,13,14].

The analysis of eDNAs includes two methods: one is the species-specific detection, which can be achieved by the presence/absence of the target species-specific signals [11,15,16], and the other is the eDNA metabarcoding that can comprehensively detect organisms in the sampling site [5,17,18]. The latter method has been used for many taxa, including fish [5], birds [19] and mammals [8], and is expected to complement conventional methods. Comparisons of the eDNA analysis with capture survey results have been reported, but the results of sampling surveys and the eDNA analysis are often not in complete agreement (e.g., detected by eDNA but not appeared in capture). These false-positive results might be attributable to the transport of eDNAs by water flow and the difficulty in discriminating between dead and alive individuals, which is a problem that needs to be solved in eDNA analysis [2,20]. On the other hand, false-negative results might be due to the low number of PCR replicates and the influence of PCR inhibitors.

This study deals with ichthyofaunal monitoring via eDNA analysis and capture survey in a river in the Fukushima Prefecture, Japan (Figure 1). The geography of the freshwaters of the target area of this study can be grouped into three areas: (1) the eastern Pacific Ocean side, (2) the central and (3) the western inland parts. The eastern part accommodates numerous shorter rivers originated in the Abukuma Mountains with short stretches of their middle and lower reaches. The central part consists of a single river basin (Abukuma River) from its origin on the Abukuma and the Nasu Mountains. to the middle reach. The Abukuma River is one of the large rivers in the Japanese Archipelago (6th in length and 11th in basin coverage). The river has numerous tributaries from the mountains and a large alluvial basin. The western part consists of areas with an upper reach of a large river, Agano (10th in length and 8th in basin coverage). The Agano River has diverse environments with mountain streams from the Nasu Mountains and the Oze Highland Peat Swamp, as well as springs and lakes such as Lake Inawashiro and Lake Hibara. The freshwater fish fauna of the Fukushima Prefecture is characterized by a mixture of eastern Siberian cold-water and China–Korean warm-water fish faunas. The cold-water fishes typically include salmonids, Salvelinus leucomaenis, Oncorhynchus masou and Oncorhynchus keta, stickleback Gasterosteus aculeatus, cyprinid leuciscine Pseudaspius sachalinensis, etc. They have extended their ranges from the north via coastal migration, depending on the climatic conditions [21] except for Misgurnus chipisaniensis, which extended its range via a land connectivity from the Siberian region [22]. The Pacific coast of the Fukushima Prefecture is in the transition area of warm and cold currents [23]. Their occurrence in the Fukushima Prefecture is hence at or near the southern limit. Some of the warm-water fishes, such as Plecoglossus altivelis altivelis, Lateolabrax japonicus, Mugil cephalus, Rhinogobius spp., etc., also have migratory life cycles. Their distribution also has a climatic basis, and is near the northern limit. The other component of warm-water fishes, such as Carassius sp., Acheilognathus spp., Misgurnus anguillicaudatus, Cobitis spp., etc., is primary freshwater fish, extending their ranges through geological events [24].

On the other hand, non-native species disturb the integrity of the fish fauna of each river [25,26], such as the increase in population due to the propagation of Zacco platypus, Candidia temminckii, Pseudorasbora parva, Gnathopogon elongatus and Rhodeus ocellatus ocellatus, which were introduced from accompanying stocking projects. Furthermore, the outlaw stocking of non-native Micropterus spp. and Lepomis macrochirus into natural lakes, dam reservoirs and ponds threatens native fish populations, especially of warm-water primary freshwater fishes. The faunal monitoring of fish in the Fukushima Prefecture is thus needed because of its biogeographical importance yet threats by non-native species and other factors like civil engineering.

We conducted an eDNA metabarcoding study in a river in the Fukushima Prefecture in 2018 to confirm the validity of the eDNA analysis for monitoring fish assemblages in rivers of faunal mixing. In such places, the abundance of climate-driven species might attenuate toward the distribution limit. Some other species might decline due to various threats. The eDNA analysis is expected to detect those rare species. By comparing the results of the eDNA analysis and a year-round capture survey data in the same river, we evaluated the validity of the analysis results and verified the effectiveness of the eDNA analysis for biological monitoring surveys in the river.

2. Materials and Methods

2.1. Field Survey

Water samples were collected at three points (right and left banks, and the center) of a sampling site in a small river in the Fukushima Prefecture, Japan (ca. 25 km from its origin to drainage) in October 2018. The sampling site was about 16 km from the origin in an alluvial cone with 7.7‰ of slope. The river width (distance between levees) was ca. 50 m, and the flow width was ca. 10 m at the site. Approximately 1 L of water samples with two replicates were collected using polyethylene bottles, filtered through Sterivex filter cartridges (pore size 0.45 μm; Merck Millipore, Darmstadt, Germany) and stored at −30 °C. As a negative control (NC), 500 mL of purified water (KENEI Pharmaceutical Co., Ltd., Osaka, Japan) was filtered in the same way as the field water sample.

2.2. DNA Extraction

DNA was extracted from the filters using the DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany). The Sterivex filter was incubated at 37 °C for 30 min with 180 μL of lysozyme (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan). Then, a mixture of 220 μL of PBS, 200 μL of buffer AL (a part of the kit) and 20 μL of proteinase K (a part of the kit) was added and incubated at 56 °C for 30 min. The DNA was subsequently purified following the manufacturer’s protocol. After purification, DNA was eluted using 200 μL of buffer AE, which was provided with the kit.

2.3. PCR and Sequencing

The extracted DNA was subjected to 12SrRNA amplicon analysis via Illumina MiSeq sequencing (Illumina, San Diego, CA, USA) [27]. The first-round of PCR with 28 cycles was carried out with a 20 μL reaction volume containing 10 μL of 2 × KAPA HiFi^TM HotStart ReadyMix (KAPA Biosystems, Wilmington, WA, USA), 0.6 μL of each MiFish primer, excluding the random hexamer (10 μM primer F/R) [5], 6.8 μL of sterile distilled water and 2.0 μL of DNA template. The thermal cycle profile after an initial 3 min denaturation at 95 °C was as follows: denaturation at 98 °C for 20 s; annealing at 65 °C for 15 s; and extension at 72 °C for 15 s, with a final extension at the same temperature for 5 min. The PCR product was purified using AMPureXP (PCR product: AMPure XP beads = 1:1; Beckman Coulter, Brea, CA, USA) and eluted into 20 µL of buffer EB (Qiagen). The second-round of PCR with 8 cycles was carried out in an identical reaction mixture as that of the first round, except for each 1 μL of index primers (10 μM of overhanging primers with indexed sequencing adaptor), 6.0 μL of sterile distilled H₂O and template (the first-round PCR product). The thermal cycle was carried out in an identical profile with the first round except for annealing at 72 °C for 7 s and extension at 72 °C for 8 s. The second-round PCR product was then purified in the same way as the first round. Sequencing was performed according to the manufacturer’s protocol using the MiSeq Reagent Kit v2 for 2 × 250 bp PE (Illumina). The raw read obtained with MiSeq was analyzed using the method described previously [27] (Table S1).

For species barcoding, those filtered reads with no more than two nucleotide differences with the reference sequence or clustered with the references from a species were adopted. Sequences equal to or equally close to a number of species, or with three to six nucleotide differences were assigned to genera. Sequences with less than 10 reads were excluded. The species identification was first based on the MiFish pipeline [28], and then species assignment of the consensus sequences was further confirmed with Blast search on the GenBank database.

Since eDNA metabarcoding detected non-native Anguilla anguilla and Anguilla rostrata, an upstream part of mitochondrial cytochrome b (Cytb) from 18 eel specimens collected using the electrofisher (LR-20B; Smith-Root, Inc., Vancouver, WA, USA) on 31 July 2019 (n = 6) and 26 October 2021 (n = 12) in the river was sequenced and compared with published sequences [29]. Primers were L14695 on the L-strand (AATTYTTGCTCRGACTCTAACC) [30] and H15560 on the H-strand (TAGGCRAATAGGAARTATCA) [31]. Methods for obtaining high-quality reads through touch-down PCR and sequencing were based on a previous study [30].

2.4. Comparison of eDNA with Capture Data

We applied the capture survey data published by the Ministry of the Environment [32] to compare the eDNA results with the capture data in 2018, the same year as the eDNA sampling. When comparing with the capture survey results, even if a species was determined as a genus by eDNA barcoding, it was regarded to correspond, if any, to the congeneric species obtained from the capture data.

3. Results

The MiSeq paired-end sequencing yielded a total of 1,804,039 raw reads. Of these, 1,708,609 reads passed the merging process, 1,707,507 reads passed the quality filter process and 1,643,156 reads passed the denoise process (Table S1).

The filtered reads from three field samples were converged to 41 haplotypes mapping onto the database for MiFish pipeline [28] (Table S2). These haplotypes were assigned to a total of 28 lineages (eight genera and 20 species,

Table 1. Correspondence between reads obtained from eDNA metabarcoding and capture data. In the capture data, (+) indicates species captured and (−) indicates species not captured.

	eDNA				Capture Data
Scientific Name	Left Bank	Center of Flow	Right Bank	NC *1
Lethenteron sp. N	0	0	11	0	+
Anguilla japonica	7417	12579	11837	0	+
Anguilla anguilla	66	569	281	0	−
Anguilla rostrata	0	36	0	0	−
Cyprinus carpio	13897	11239	28390	0	+
Carassius cuvieri *2	0	578	1158	0	+
Carassius langsdorfii *2	3825	590	0	0	+
Pseudorhodeus tanago	3723	1664	399	0	−
Zacco platypus	45853	37922	39856	0	+
Candidia temminckii	22889	28735	39263	0	+
Rhynchocypris lagowskii steindachneri	0	312	246	0	−
Pseudaspius hakonensis	302217	211997	205452	24	+
Pseudaspius sachalinensis	283	29	0	0	−
Misgurnus anguillicaudatus	2356	442	1048	0	+
Misgurnus chipisaniensis	111	82	163	0	−
Cobitis sp. BIWAE type C	10929	8201	9772	0	+
Lefua echigonia	57406	26880	24161	40	−
Hypomesus nipponensis	135	348	108	0	−
Plecoglossus altivelis altivelis	1128	1943	3188	0	+
Oncorhynchus keta	1057	2534	2245	0	+
Oncorhynchus masou *2	2787	9594	9542	0	+
Salvelinus sp.	0	110	0	0	−
Lateolabrax japonicus	978	0	0	0	−
Lepomis macrochirus *2	351	288	723	0	−
Cottus reinii *2	605	171	0	0	+
Cottus pollux	0	0	0	0	+
Tridentiger sp.	1045	2811	2359	0	−
Rhinogobius fluviatilis *2	48110	110493	106468	0	+
Rhinogobius nagoyae	0	0	0	0	+
Takifugu alboplumbeus	0	262	0	0	−

*¹ Negative control. *² Species assigned to generic level via the eDNA analysis, and confirmed through the capture data.

). Some species (Cobitis sp. BIWAE type C and Cottus reinii) were identified by clustering with known sequences (Figure S1). A high sensitivity might cause false-positive results in the negative control on two commonly detected species from samples, but read counts were small (<0.2% of the samples).

Compared to the capture survey results, 16 out of 18 species and genera captured were detected via the eDNA analysis (Table 1). The eDNA analysis detected 13 lineages (two genera and 11 species) not identified in this river through the capture survey. Anguilla anguilla and Anguilla rostrata, species outside the distribution range, and Pseudaspius sachalinensis, whose southernmost distribution is in the Fukushima Prefecture, were detected [33]. Sequences of upstream 759 bp of Cytb from eel specimens tightly clustered with Anguilla japonica (Table S3, Figure S2). Pseudorhodeus tanago, an endangered and legally protected species, was also detected, but the current location was further north from the northernmost known locality [30,34]. Other species that appeared only in the eDNAs included several primary freshwater (Rhynchocypris lagowskii steindachneri, Misgurnus chipisaniensis and Lefua echigonia), migratory (Salvelinus sp., Lateolabrax japonicus, Tridentiger sp. and Takifugu alboplumbeus) and introduced (Hypomesus nipponensis and Lepomis macrochirus) elements.

Regarding the number of species detected among sampling points, 18 out of 28 commonly appeared in all three points, while four lineages were shared by two of three points (Figure 2). Coherent with the river topology, none of the species were shared in two opposite banks only. Six lineages appeared in either of one point.

4. Discussion

In this study, eDNA was able to detect almost all the species captured. Several species were detected only in eDNA samples. Lineages commonly detected on right and left banks were always also shared in the stream center (18 species in the gray-colored area in Figure 2), indicating that a distribution of free DNAs relates to the running water along river topology. This may be supporting evidence for the reliability of our eDNA analysis. The different species detected in the same sampling site at different sampling points suggests that eDNAs were not evenly distributed in the river. This may be related to the ecology of the organisms or may be an effect of false negative due to the transport of eDNAs in complex water flow in the river.

For species that were only detected through eDNA, as for native species, Rhynchocypris lagowskii steindachneri has been commonly reported [25,26], suggesting a decline in this species at the collecting site. Hypomesus nipponensis and Salvelinus sp. (Salvelinus leucomaenis) have been reported in past surveys to either be naturally distributed in rivers or to be introduced by release [25]. These species are popular as angling and the fishing cooperative had been releasing eggs or juveniles [35]. Habitats of these species are far from the water sampling site, indicating a long distance retention of river water eDNAs. This also addresses the caution of using eDNA data as some species could not occur at the exact site. Misgurnus chipisaniensis has been reported in the Fukushima Prefecture [36]. It is difficult to identify Misgurnus chipisaniensis from its external appearance, but eDNAs indicate that such species can be detected. Lefua echigonia is known to be naturally distributed in rivers from past surveys [25]. Moreover, Lefua echigonia is thought to inhabit irrigation canals and ditches around paddy fields, and such environments exist in the vicinity. Reportedly, Anguilla anguilla and Anguilla rostrata may have been mixed and released into the river during past stocking activities of Anguilla japonica seedlings [37]. A limited number of reads of these non-native eels in eDNAs and their absence among capture survey samples revealed by sequencing indicate their uncommon occurrence, if any, at the site. Lepomis macrochirus has been identified in the Fukushima Prefecture [25,26,38]. Lepomis macrochirus is expanding its habitat throughout Japan and there is a good possibility of distribution in the river. The confirmation of eDNAs of this species in the present study indicates that it is useful for the early detection of exotic species in their low biomass stage. Regarding saltwater and brackish water species, Lateolabrax japonicus and Takifugu alboplumbeus are known to migrate upstream to brackish and freshwater areas [25,39,40]. As for Tridentiger sp. detected in eDNAs, Tridentiger brevispinis has been confirmed in rivers in the Fukushima Prefecture [41]. Because the sampling site were far from the sea across a number of weirs and rapids, the highly sensitive eDNA analysis is especially useful for finding fishes without past capture records. As for rare species, Pseudaspius sachalinensis, which appeared in eDNAs yet absent from our capture survey, was collected in the past [25,26,33,42]. Although Pseudorhodeus tanago has not been confirmed in the past, its presence cannot be completely ruled out because of the existence of streams with spring water in the vicinity that Pseudorhodeus tanago prefers. Several habitats of this legally protected species have been found which locations are difficult to explain phylogeographically [30]. Illegal stocking was suspected on those newly found habitats. Further capture survey and genetic typing, if collected, are needed for confirmation. The absence of eDNAs of two species (Cottus pollux and Rhinogoboius nagoyae) indicates a limitation of this technology. It might be because of low biomass or an insufficient number of reads.

It should also be noted that non-habitant fish DNAs sometimes appear because of food consumption within the vicinity [28]. Alien eels, Hypomesus nipponensis and Lateolabrax japonicus are appreciated as foods, but the sampling site does not receive sewage from residential areas around. These areas are located over distributaries with water from the river on the alluvial cone, and have lost many of the residents after the nuclear accident. Takifugu alboplumbeus is highly toxic and can never be consumed. The presence of eDNAs of non-captured fish is thus unlikely to have been from a food source, but rather likely to have been a natural occurrence.

5. Conclusions

The eDNA analysis covered most of the species in the capture survey results, suggesting that it is useful for biological monitoring. Furthermore, it could be an effective tool for the quick detection of non-native species and rare species. On the other hand, there are some species that can be determined as a species via capture survey but are difficult to determine as a species via eDNA analysis. Therefore, the evaluation of the survey results requires further verification based on the results of previous surveys.