Environmental DNA metabarcoding revealing fish species diversity in the Zhelin Reservoir

Abstract

Reliable fish biodiversity baselines are essential for monitoring and adaptive management in large subtropical reservoirs. Environmental DNA (eDNA) metabarcoding provides a non-invasive approach for assessing fish assemblages, but molecular baseline information remains limited for Zhelin Reservoir in the Poyang Lake Basin. In this study, eDNA metabarcoding targeting the mitochondrial 12S rRNA gene was used to characterize fish community composition, diversity patterns and exploratory environment-community associations across 14 sampling sites in December 2023. A total of 85 fish taxa were detected, belonging to 6 orders, 12 families and 59 genera, with Cypriniformes and Cyprinidae being dominant. The taxa with the highest relative read abundances were Xenocypris microlepis, Aristichthys nobilis, Carassius auratus, Cyprinus carpio and Ctenopharyngodon idella, indicating high relative detection intensity of several widespread cyprinids. After rarefaction to an equal depth of 388,125 reads, expected richness was higher upstream, whereas Shannon diversity, Simpson diversity and Pielou evenness did not differ significantly among reservoir sections. Bray-Curtis beta-diversity analysis showed no significant compositional separation among upstream, middle and downstream sections (PERMANOVA: R² = 0.223, p = 0.160), and PERMDISP indicated similar multivariate dispersion (p = 0.379). RDA after variance inflation factor screening suggested a significant but exploratory association between measured environmental variables and fish community composition (R² = 0.600, p = 0.045), with dissolved oxygen showing the strongest single-variable association. Overall, this study provides a molecular inventory and monitoring baseline for fish communities in Zhelin Reservoir, supporting future repeated surveys and adaptive reservoir management.

1 Introduction

Freshwater ecosystems support a disproportionately high share of global biodiversity but are increasingly affected by habitat degradation, overexploitation, hydrological alteration and climate change (Dudgeon et al., 2006; Strayer and Dudgeon, 2010; Reid et al., 2019; Sayer et al., 2025). Reservoirs are particularly complex coupled human-natural systems because dam operation, water-level fluctuation, longitudinal habitat gradients, stocking and fishing regulation may jointly shape fish assemblages (Tundisi and Straskraba, 1999; Miranda, 2017; Winemiller et al., 2016). These characteristics create a need for monitoring approaches that can describe present biodiversity patterns while avoiding unsupported attribution to a single driver.

Conventional methods, such as gill nets and electrofishing, are typically constrained by depth limitations, taxonomic selectivity, and insufficient spatial coverage in sampling, which frequently lead to biased estimates of species diversity and limited effectiveness in detecting rare, elusive, or deep-water species (Schloesser et al., 2012; Miya et al., 2020; Afzali et al., 2021; Pinna et al., 2023; Daun et al., 2025). These approaches can also be intrusive, imposing stress on target species and potentially altering their natural behavior (Danielsen et al., 2005; Cooke et al., 2012). Moreover, they are inherently limited in capturing transient ecological responses triggered by activities such as hydrological regulation. Such methodological shortcomings often result in biased biodiversity assessments and compromised ecological interpretations, ultimately hindering effective ecosystem management and conservation planning (Reid et al., 2019).

Environmental DNA (eDNA) metabarcoding provides a non-invasive and sensitive approach for surveying aquatic biodiversity. By capturing genetic material shed by organisms into water, eDNA can detect species traces without direct capture (Thomsen and Willerslev, 2015), and high-throughput metabarcoding enables simultaneous detection of multiple taxa from a single water sample (Deiner et al., 2017). Compared with conventional surveys, eDNA can improve detection of rare, elusive or difficult-to-sample species, but its results are also affected by DNA shedding, degradation, transport, primer bias, amplification efficiency and reference-database completeness (Goldberg et al., 2016; Miya et al., 2020). Therefore, eDNA is best used as a complementary monitoring tool rather than a direct substitute for conventional abundance or biomass estimates. It has been applied in large and complex water bodies such as Erhai Lake (Su et al., 2020) and the Three Gorges Reservoir (Yang et al., 2025), supporting its relevance for reservoir biodiversity assessment. Recent freshwater fish eDNA studies have further emphasized the importance of transparent reporting of controls, replication, filtering criteria and primer/database information, while reservoir applications have shown that eDNA metabarcoding can provide useful complementary biodiversity information (Liu et al., 2025; Lin et al., 2025).

The Zhelin Reservoir, located in the Yangtze River Basin, covers a water area of 308 km² with a maximum water depth of 45 m. As a large subtropical canyon-type reservoir with multiple ecological and social functions, it is the largest artificial lake in Jiangxi Province (Dai et al., 2015; Chen et al., 2022). The reservoir serves as an important drinking water source for Jiujiang City and is part of the Elopichthys bambusa National Aquatic Germplasm Reserve, playing important roles in water conservation, flood regulation, tourism, fisheries and biodiversity conservation (Liu and Liao, 2013). Because it is located in the Poyang Lake Basin and has experienced fishing prohibition, stock enhancement and other management interventions, Zhelin Reservoir is an important system for establishing fish community monitoring baselines.

Recent work has substantially improved understanding of fish communities in Zhelin Reservoir. Conventional and functional-diversity studies have described fish community structure and conservation priorities (Chen et al., 2022; Meng et al., 2026), hydroacoustic surveys have assessed fish density and aggregation patterns (Meng et al., 2025), and a previous study explored fish species identification using eDNA (Chen, 2020). However, these studies do not provide a standardized site-scale eDNA baseline that links molecular species detection, spatial community similarity and measured environmental gradients under a consistent sampling design. Such a baseline is needed to complement conventional and hydroacoustic surveys and to support future repeated monitoring in Zhelin Reservoir.

Accordingly, this study used eDNA metabarcoding to characterize fish taxa detected across 14 sites, evaluate longitudinal differences in alpha and beta diversity, and explore associations between community composition and measured environmental variables. We expected that reservoir sections might differ because of longitudinal habitat gradients, while treating the study primarily as a molecular baseline for future repeated monitoring.

2 Method and material

2.1 Study area

This study was conducted in the Zhelin Reservoir (29°03′–29°27′N, 115°04′–115°40′E), the largest man-made reservoir in Jiangxi Province, China (Figure 1). Formed by damming the Xiuhe River (one of the five major tributaries of the Poyang Lake), the reservoir covers a water surface area of 308 km² with a total storage capacity of 7.92 × 10⁹ m³. It has an average water depth of 16.3 m and a maximum depth of 45 m. The region features a subtropical monsoon climate, with distinct seasons and an average annual precipitation of 1611.8 mm (Meng et al., 2023; Zhu et al., 2021). Based on the geographical distribution of Zhelin Reservoir, 14 fixed sampling sites were established to monitor abiotic and biotic factors. These sites were distributed across the reservoir, with S1–S4 located in the upstream region, S5–S9 in the midstream region, and S10–S14 in the downstream region (Figure 1).

FIGURE 1

2.2 Sample collection

Water samples were collected from 14 sites across Zhelin Reservoir in December 2023. At each sampling site, water was collected from three nearby points and three vertical layers, including the surface layer, middle layer approximately 1 m below the surface and bottom layer at least 2 m above the sediment, with 1 L collected for each water collection. These water samples were mixed to form one composite sample for each site. The composite sample was then divided into three subsamples, which were independently filtered, extracted and PCR-amplified. The resulting data were combined as site-level eDNA data for community analyses; therefore, the three subsamples were treated as technical process replicates rather than independent biological replicates.

Simultaneously with water sampling, key environmental variables were measured in situ. Water temperature (WT), pH, and dissolved oxygen (DO) were determined using a portable multi-parameter water quality analyzer (YSI ProQuatro, USA). Transparency (SD) was measured with a Secchi disk. Additional water samples for nutrient analysis were collected and stored at 4 °C. In the laboratory, total nitrogen (TN) was determined by the alkaline potassium persulfate digestion-ultraviolet spectrophotometric method, total phosphorus (TP) by the ammonium molybdate spectrophotometric method, ammonia nitrogen (NH₃-N) by Nessler’s reagent spectrophotometric method, BOD₅ by the dilution and seeding method, and COD by the potassium dichromate method. Chlorophyll a (Chl.a) concentration was determined via ethanol extraction and spectrophotometry, following the Standard Methods for the Examination of Water and Wastewater (APHA, 2017).

2.3 Environmental DNA extraction and contamination control

Water samples were filtered through 0.45 μm membranes, and DNA was extracted from the filters following the sequencing provider protocol. Field blank controls were included once per sampling batch to assess potential contamination during field handling and laboratory processing. The blank controls were carried through PCR screening and showed no visible target bands during agarose gel electrophoresis; therefore, they were not sequenced. No positive control or mock community was included.

2.4 PCR amplification and sequencing

Fish eDNA was amplified using the Tele02 12S primer pair Tele02F/Tele02R, which targets a short mitochondrial 12S rRNA fragment suitable for fish metabarcoding. PCRs were performed with FastPfu polymerase in 20 μL reactions, and three PCR replicates were used for each composite site sample. Field blank controls were included in PCR screening and showed no target bands on agarose gels. Amplicons were purified, pooled and sequenced on the Illumina MiSeq platform using paired-end sequencing by BIOZERON.

After demultiplexing, quality filtering and denoising were performed using DADA2 (v1.18.0). Across the 42 processed subsamples, 37,421,272 high-quality reads were retained in the ASV table, with 726,918-1,080,275 reads per subsample before site-level combination. The three subsamples corresponding to each site were combined before community-level statistical analyses.

2.5 Bioinformatics analysis

Raw sequence data were processed using an ASV-based workflow rather than 97% OTU clustering. DADA2 was used for quality filtering, error correction and chimera removal. Taxonomic assignment was conducted using blastn (v2.2.30; e-value 1e-10) against MitoFish/mitogenomes (updated July 2023, version 3.90; 811,468 mitochondrial sequences covering 42,015 species), supplemented by NT-euk and a curated fish reference database from the sequencing provider. After taxonomic assignment, low-abundance species-level records with ≤ 5 reads in a single sample were set to zero, and taxa with zero reads across all samples were removed.

Relative read abundance (RRA) was calculated as the proportion of fish reads assigned to each taxon within a sample. RRA was used as an index of relative detection intensity, not as a direct estimate of biomass, absolute abundance or individual number, because eDNA read counts can be affected by species-specific DNA shedding, degradation, transport, primer bias and amplification efficiency.

2.6 Statistical analysis

All community-level statistical analyses were conducted in R version 4.4.3 (2025-02-28 ucrt). The main packages used were vegan 2.6–10, ape 5.8–1, car 3.1–3, readxl 1.4–5 and permute 0.9–7. Species richness, Shannon diversity, Simpson diversity and Pielou evenness were calculated using the vegan functions specnumber and diversity. To examine whether uneven sequencing depth influenced richness estimates, rarefaction-standardized expected richness was calculated using rarefy after standardizing all samples to the minimum site-level sequencing depth (388,125 reads). Normality and homogeneity of variance were assessed using shapiro. test in stats and leveneTest in car, respectively. After confirming these assumptions, section-level differences among upstream, middle and downstream samples were tested using one-way ANOVA with aov in stats.

Beta-diversity analyses were conducted using vegdist in vegan. Bray-Curtis dissimilarity calculated from relative read abundance was used as the primary metric for community-composition analysis and PCoA visualization. PCoA was performed using pcoa in ape. Jaccard dissimilarity based on presence-absence data was used as a complementary sensitivity analysis to evaluate whether section-level patterns were consistent when read abundance was not considered. PERMANOVA was performed using adonis2 in vegan with 999 permutations to test for differences in community composition among reservoir sections. PERMDISP was performed using betadisper followed by permutest with 999 permutations to assess whether multivariate dispersion differed among sections.

For environment-community association analysis, the fish community matrix was Hellinger-transformed using decostand in vegan before RDA. RDA was conducted using rda in vegan. Environmental variables were screened for multicollinearity using vif. cca, and variables with VIF >10 were sequentially removed before final model fitting. The final RDA model retained temperature, dissolved oxygen, NH₃-N, BOD₅, chlorophyll a and transparency. Overall model significance was evaluated by permutation testing using anova. cca in vegan with 999 permutations. To characterize the individual explanatory signal of each retained environmental variable, single-variable RDA tests were conducted separately. Full statistical outputs are provided in Supplementary Tables S1, S2.

3 Results

3.1 Composition of fish species

Based on eDNA metabarcoding targeting the 12S rRNA marker gene, a total of 85 fish species were identified across the 14 sampling sites, belonging to 6 orders, 12 families, and 59 genera (Supplementary Table S3). The taxonomic composition was dominated by the order Cypriniformes, accounting for 70.59% (60 species) of the total species, followed by Perciformes (16.47%, 14 species) and Siluriformes (9.41%, 8 species). This dominance of Cypriniformes was further reflected at the family level, where Cyprinidae was overwhelmingly predominant, comprising 54 species and accounting for 63.53% of the total assemblage (Table 1).

To identify taxa with high relative detection intensity, we analyzed relative read abundance (RRA). The five taxa with the highest RRA were Xenocypris microlepis, Aristichthys nobilis, Carassius auratus, Cyprinus carpio and Ctenopharyngodon idella. All five taxa belong to Cyprinidae and were detected at all 14 sampling sites. Together, they accounted for 80.24% of the total fish reads. Because RRA is affected by DNA shedding, transport, degradation and amplification bias, these taxa are described as dominant detected taxa rather than as species with confirmed highest biomass or individual abundance.

3.2 Analysis of fish species diversity

Alpha diversity indices were calculated to characterize the fish community diversity, including the Shannon index, Simpson index, and Pielou evenness index (Table 2).

The results indicated that the Shannon index ranged from 1.5599 to 2.0738, with a mean value of 1.8635. The Simpson index varied between 0.6637 and 0.7944, averaging 0.7434, while the Pielou evenness index ranged from 0.4122 to 0.5578, with a mean of 0.4819.

After rarefaction to an equal depth of 388,125 reads, expected richness differed significantly among reservoir sections (ANOVA: F = 6.942, p = 0.011; Supplementary Table S1), with the upstream section showing higher expected richness than the middle and downstream sections. In contrast, Shannon diversity, Simpson diversity and Pielou evenness did not differ significantly among upstream, middle and downstream sections (Shannon: p = 0.624; Simpson: p = 0.261; Pielou: p = 0.364; Figure 2; Supplementary Table S1). Thus, richness and evenness-related indices showed different spatial signals, and alpha diversity alone was insufficient to infer community homogenization.

FIGURE 2

3.3 Beta diversity and community turnover

Bray-Curtis PCoA showed partial overlap among upstream, middle and downstream samples, indicating limited compositional separation among reservoir sections (Figure 3). PERMANOVA based on Bray-Curtis dissimilarity detected no significant difference in fish community composition among sections (R² = 0.223, F = 1.581, p = 0.160), and PERMDISP showed no significant difference in multivariate dispersion among sections (F = 1.069, p = 0.379). The complementary presence-absence-based Jaccard analysis yielded a similar conclusion (PERMANOVA: R² = 0.149, F = 0.965, p = 0.515; PERMDISP: F = 0.335, p = 0.683; Supplementary Table S1). These results suggest broad community similarity among reservoir sections during the sampling period, but do not support a strong claim of pronounced spatial homogenization.

FIGURE 3

3.4 Relationship between environmental variables and fish community composition

After VIF screening, temperature, dissolved oxygen, NH3-N, BOD5, chlorophyll a and transparency were retained in the RDA model, whereas variables with high collinearity were excluded before final model fitting. The final RDA explained 60.0% of the variation in Hellinger-transformed fish community composition and was statistically significant (R² = 0.600, adjusted R² = 0.257, F = 1.749, p = 0.045; Figure 4). Single-variable RDA tests indicated that dissolved oxygen had the strongest individual association with community composition (R² = 0.208, F = 3.148, p = 0.013), whereas BOD5 (R² = 0.148, F = 2.092, p = 0.074) and chlorophyll a (R² = 0.142, F = 1.988, p = 0.076) showed weaker trends (Supplementary Table S2). These results suggest that local water conditions were associated with variation in fish community composition.

FIGURE 4

4 Discussion

4.1 Species composition of fish communities detected by eDNA

The systematic eDNA sampling design detected 85 fish taxa across 14 sites in Zhelin Reservoir, belonging to 6 orders, 12 families and 59 genera. At the order and family levels, the assemblage was dominated by Cypriniformes and Cyprinidae, a pattern broadly consistent with recent taxonomic and functional diversity research on Zhelin Reservoir fish communities (Meng et al., 2026). In contrast to hydroacoustic assessment, which is well suited to evaluating fish density and aggregation patterns but has limited taxonomic resolution (Meng et al., 2025), eDNA metabarcoding provides direct molecular evidence for the taxonomic composition of the detected assemblage. This consistency supports the basic reliability of the eDNA survey for describing the main taxonomic structure of the reservoir fish assemblage.

Compared with recent conventional taxonomic and functional-diversity surveys, eDNA metabarcoding provides several complementary advantages. It can be applied with a standardized site-scale water-sampling design, is non-invasive, and can provide taxon-level molecular detections across reservoir sections where capture efficiency may vary with depth, habitat type and fish behavior. These features are particularly useful for establishing a repeatable molecular baseline for future monitoring. However, eDNA detections are not equivalent to captured individuals, biomass or population size, and the method is influenced by DNA transport, degradation, primer performance and reference-database completeness. Therefore, the present eDNA inventory should be interpreted as a molecular biodiversity baseline that complements, rather than replaces, recent conventional and hydroacoustic investigations in Zhelin Reservoir (Meng et al., 2025; Meng et al., 2026).

4.2 Community structure inferred from dominant eDNA detection signals

At the community-structure level, the five taxa with the highest relative read abundance in this study were Xenocypris microlepis, Aristichthys nobilis, C. auratus, C. carpio and C. idella, all of which belong to Cyprinidae and were detected at all sampling sites. A useful comparison is provided by the capture-based survey of Meng et al. (2023), which used the index of relative importance (IRI) to evaluate dominance in Zhelin Reservoir. In that survey, bighead carp (A. nobilis, reported as Hypophthalmichthys nobilis by Meng et al.; IRI = 6857), H. leucisculus (IRI = 6417) and Xenocypris davidi (IRI = 1646) were dominant species (IRI >1000), whereas silver carp, C. idella, C. auratus, C. carpio and several other taxa were important species (100 < IRI <1000).

The two datasets show agreement at the family and reservoir-community level, but not a one-to-one match among dominant species. Both approaches indicate a cyprinid-dominated assemblage, and bighead carp was prominent in both datasets. In addition, C. idella, C. auratus and C. carpio, which were among the highest-RRA taxa in this study, were also listed as important species in the capture-based survey. However, the capture-based dominant species H. leucisculus and X. davidi were not among the five highest-RRA taxa in the eDNA dataset, whereas X. microlepis showed the strongest molecular detection signal in this study. This difference is informative because Meng et al. (2023) calculated IRI from numerical percentage, biomass percentage and occurrence, whereas the present eDNA ranking is based on relative read abundance.

Thus, the comparison supports the scientific value of eDNA for identifying the major cyprinid signal and widely distributed taxa in Zhelin Reservoir, but it also illustrates the main limitation of using RRA as a measure of dominance. eDNA can reduce some capture-related selectivity and provide standardized molecular detection across sites, but RRA may be influenced by species-specific DNA shedding, degradation, transport, primer amplification efficiency and reference-database resolution (Goldberg et al., 2016; Deiner et al., 2016; Strickler et al., 2015; Miya et al., 2020; Liu et al., 2025). Therefore, the high-RRA taxa in this study should be interpreted as dominant detected taxa or strong molecular signals rather than as direct evidence of numerical abundance or biomass dominance. For Zhelin Reservoir, eDNA is most defensible as a standardized molecular baseline for community-structure assessment, while capture-based IRI and hydroacoustic density estimates remain necessary for abundance, biomass and size-structure evaluation (Meng et al., 2023; Meng et al., 2025).

4.3 Spatial variation and environmental associations of fish community composition

The beta-diversity results provide the third level of evidence for evaluating the scientific applicability of eDNA in this reservoir. Bray-Curtis PCoA, PERMANOVA and PERMDISP showed no significant compositional separation or dispersion difference among upstream, middle and downstream sections, and the complementary Jaccard analysis led to the same general conclusion. These results indicate broad spatial similarity and weak section-level differentiation during the December 2023 survey. This pattern is more consistent with hydrological connectivity, habitat continuity within the reservoir and the widespread occurrence of generalist cyprinids than with strong longitudinal community turnover.

The RDA analysis adds an exploratory environmental layer to this spatial interpretation. After VIF screening, the retained environmental variables were significantly associated with Hellinger-transformed community composition, with dissolved oxygen showing the strongest single-variable association. This suggests that local oxygen conditions may be related to among-sample variation in eDNA-inferred fish composition, while BOD5 and chlorophyll a may reflect weaker gradients associated with organic-matter demand and primary productivity. However, these associations should not be interpreted as direct causal mechanisms because the study was based on a single December sampling campaign and did not explicitly model hydrodynamics, habitat structure, fish movement or stocking locations.

From a monitoring perspective, the combination of beta-diversity analysis and RDA demonstrates that eDNA can be used not only to list detected taxa, but also to evaluate spatial similarity and screen candidate environmental correlates under a unified sampling framework. This is a key advantage for reservoir-scale baseline assessment. At the same time, the weak section-level differentiation and exploratory nature of the RDA results indicate that repeated seasonal sampling, local reference-database improvement, conventional fish surveys and hydroacoustic validation are needed before inferring recovery dynamics or causal drivers. The strongest contribution of the present study is therefore the establishment of a standardized eDNA molecular baseline for future repeated surveys of fish diversity in Zhelin Reservoir.

5 Conclusion

This study detected 85 fish taxa in Zhelin Reservoir using eDNA metabarcoding and provides a molecular baseline for reservoir fish community monitoring. Rarefaction-standardized richness was higher upstream, whereas Shannon diversity, Simpson diversity and Pielou evenness did not differ significantly among reservoir sections. Beta-diversity analyses showed no significant compositional separation or dispersion difference among upstream, middle and downstream sections, indicating broad spatial similarity and weak section-level differentiation during the December 2023 survey. RDA after VIF screening suggested significant associations between measured environmental variables and community composition, with dissolved oxygen showing the strongest single-variable association. Overall, eDNA metabarcoding is a useful non-invasive tool for reservoir-scale biodiversity monitoring. Future management-oriented monitoring should combine fixed seasonal eDNA surveys, field blanks with sequenced negative controls, conventional fish surveys, hydroacoustic assessment and local stocking or habitat-management records, so that changes in fish community composition can be evaluated more robustly through time.

References

• 1

  AfzaliS. F.BourdagesH.LaporteM.MérotC.NormandeauE.AudetC.et al (2021). Comparing environmental metabarcoding and trawling survey of demersal fish communities in the gulf of st. Lawrence, Canada. Environ. DNA3 (1), 22–42. 10.1002/edn3.111

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  APHA (2017). Standard Methods for the Examination of Water and Wastewater. 23rd ed.Washington, DC: American Public Health Association, American Water Works Association, and Water Environment Federation.

  • Google Scholar
  • View reference in article

• 3

  ChenX. (2020). Characteristics of fish assemblage in Zhelin Reservoir and preliminary study on fish species identification using eDNAMaster’s thesis. Nanchang: Nanchang University.

  • Google Scholar
  • View reference in article

• 4

  ChenK.MengZ. H.LiX. M.HuF. F.DuH. C.LiuL.et al (2022). Community structure and functional diversity of fishes in zhelin reservoir, Jiangxi province. Acta Ecol. Sin.42 (11), 4592–4602. 10.5846/stxb202106031466

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  CookeS. J.PaukertC.HoganZ. (2012). Endangered river fish: factors hindering conservation and restoration. Endanger. Species Res.17 (2), 179–191. 10.3354/esr00426

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  DaiG. F.LiuH. L.ZhangW.LiaoB.ZhongJ. Y.ChenH. W. (2015). Spatio-temporal variation of algae and eutrophic status of Lake zhelin, Jiangxi province. J. Lake Sci.27 (2), 275–281. 10.18307/2015.0211

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  DanielsenF.JensenA. E.AlviolaP. A.BaleteD. S.MendozaM.TagtagA.et al (2005). Does monitoring matter? A quantitative assessment of management decisions from locally-based monitoring of protected areas. Biodivers. Conserv.14, 2633–2652. 10.1007/s10531-005-8392-z

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  DaunN.FearingA.LehmanR. N.GauciC. L.HansenM.JonesJ. D.et al (2025). Rare or rarely sampled? Electrofishing and environmental DNA surveys of the imperilled freckled darter Percina lenticula. Aquat. Conserv.35, e70165. 10.1002/aqc.70165

  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  DeinerK.FronhoferE. A.MächlerE.WalserJ. C.AltermattF. (2016). Environmental DNA reveals that Rivers are conveyer belts of biodiversity information. Nat. Commun.7, 12544. 10.1038/ncomms12544

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  DeinerK.BikH. M.MachlerE.SeymourM.Lacoursiere-RousselA.AltermattF.et al (2017). Environmental DNA metabarcoding: transforming how we survey animal and plant communities. Mol. Ecol.26 (21), 5872–5895. 10.1111/mec.14350

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  DudgeonD.ArthingtonA. H.GessnerM. O.KawabataZ. I.KnowlerD. J.LévêqueC.et al (2006). Freshwater biodiversity: importance, threats, status and conservation challenges. Biol. Rev.81 (2), 163–182. 10.1017/S1464793105006950

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 12

  GoldbergC. S.TurnerC. R.DeinerK.KlymusK. E.ThomsenP. F.MurphyM. A.et al (2016). Critical considerations for the application of environmental DNA methods to detect aquatic species. Methods Ecol. Evol.7 (11), 1299–1307. 10.1111/2041-210X.12595

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  LinY. T.DaiY. F.YangH. Y.LinY. H.HsuY. T.LinY. I.et al (2025). Freshwater fish monitoring in a subtropical reservoir: integrating eDNA metabarcoding and traditional methods in biodiversity resolution, invasive species detection, and environmental correlates. Environ. Sustain. Indic.27, 100831. 10.1016/j.indic.2025.100831

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  LiuH. L.LiaoB. (2013). The major eco-environmental problems and its protection countermeasures in zhelin Lake. Jiangxi Sci.31 (1), 48–52. 10.3969/j.issn.1001-3679.2013.01.013

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  LiuS. S.SahuS. K.QuK. Y.LiH.GaoZ. X.NieQ. (2025). Environmental DNA metabarcoding: current applications and future prospects for freshwater fish monitoring. J. Environ. Manage.376, 124531. 10.1016/j.jenvman.2025.124531

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  MengZ. H.YangD. G.HuF. F.ChenK.LiuL.XiangM.et al (2023). Fish resources spatio-temporal distribution patterns and controlling factors in zhelin reservoir, Lake poyang basin during early fishing ban period. J. Lake Sci.35 (6), 2071–2081. 10.18307/2023.0635

  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  MengZ. H.HuF. F.XiangM.FuX. J.LiX. M. (2025). Spatiotemporal dynamics of fish density in a deep-water reservoir: hydroacoustic assessment of aggregation patterns and key drivers. Animals15 (7), 1068. 10.3390/ani15071068

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  MengZ. H.BrosseS.XiangM.HuF. F.WangX. Y.FuX. J.et al (2026). Assessing fish taxonomic and functional beta diversity to inform effective conservation strategies in a large reservoir of the yangtze river basin. Water Biol. Secur5, 100407. 10.1016/j.watbs.2025.100407

  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  MirandaL. E. (2017). Reservoir Fish Habitat Management. Totowa, New Jersey: Lightning Press.

  • Google Scholar
  • View reference in article

• 20

  MiyaM.GotohR. O.SadoT. (2020). MiFish metabarcoding: a high-throughput approach for simultaneous detection of multiple fish species from environmental DNA and other samples. Fish. Sci.86, 939–970. 10.1007/s12562-020-01461-x

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  PinnaM.ZangaroF.SaccomannoB.ScaloneC.BozzedaF.FaniniL.et al (2023). An overview of ecological indicators of fish to evaluate the anthropogenic pressures in aquatic ecosystems: from traditional to innovative DNA-based approaches. Water15 (5), 949. 10.3390/w15050949

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  ReidA. J.CarlsonA. K.CreedI. F.EliasonE. J.GellP. A.JohnsonP. T.et al (2019). Emerging threats and persistent conservation challenges for freshwater biodiversity. Biol. Rev.94 (3), 849–873. 10.1111/brv.12480

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  SayerC. A.FernandoE.JimenezR. R.MacfarlaneN. B. W.RapacciuoloG.BöhmM.et al (2025). One-quarter of freshwater fauna threatened with extinction. Nature638, 138–145. 10.1038/s41586-024-08375-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  SchloesserJ. T.PaukertC. P.DoyleW. J.HillT. D.SteffensenK. D.TravnichekV. H. (2012). Heterogeneous detection probabilities for imperiled missouri river fishes: implications for large-river monitoring programs. Endanger. Species Res.16, 211–224. 10.3354/esr00399

  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  StrayerD. L.DudgeonD. (2010). Freshwater biodiversity conservation: recent progress and future challenges. J. North Am. Benthol. Soc.29, 344–358. 10.1899/08-171.1

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  StricklerK. M.FremierA. K.GoldbergC. S. (2015). Quantifying effects of UV-B, temperature, and pH on eDNA degradation in aquatic microcosms. Biol. Conserv.183, 85–92. 10.1016/j.biocon.2014.11.038

  • CrossRef
  • Google Scholar
  • View reference in article

• 27

  SuL.LinJ. Y.XuY.CaoT.FengJ. M.PengZ. G. (2020). Investigation the fish diversity in erhai Lake based on environmental DNA metabarcoding. Acta Hydrobiol. Sin.44 (5), 1080–1086. 10.7541/2020.125

  • CrossRef
  • Google Scholar
  • View reference in article

• 28

  ThomsenP. F.WillerslevE. (2015). Environmental DNA – an emerging tool in conservation for monitoring past and present biodiversity. Biol. Conserv.183, 4–18. 10.1016/j.biocon.2014.11.019

  • CrossRef
  • Google Scholar
  • View reference in article

• 29

  TundisiJ. G.StraskrabaM. (1999). Theoretical reservoir ecology and its applications. São Carlos, Brazil: International Institute of Ecology; Rio de Janeiro: Brazilian Academy of Sciences; Leiden: Backhuys.

  • Google Scholar
  • View reference in article

• 30

  WinemillerK. O.McintyreP. B.CastelloL.Fluet-ChouinardE.GiarrizzoT.SoN.et al (2016). Balancing hydropower and biodiversity in the Amazon, Congo, and mekong. Science351 (6269), 128–129. 10.1126/science.aac7082

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 31

  YangW.ShuZ. B.ZhangX. B.WangL.YangS. F.LiW. J. (2025). Fish diversity in the backwater area of three gorges reservoir based on eDNA metabarcoding. J. Hydroecol.46 (2), 122–133. 10.15928/j.1674-3075.202211210467

  • CrossRef
  • Google Scholar
  • View reference in article

• 32

  ZhuT. B.HuF. F.MengZ. H.LiX. M.HeY. F.FanC. L.et al (2021). Aquatic organism resources of elopichthys bambusa national aquatic germplasm reserves of the lushanxihai Lake. Biot. Res.43 (2), 188–193. 10.14188/j.ajsh.2021.02.012

  • CrossRef
  • Google Scholar
  • View reference in article