Environmental DNA Metabarcoding as a Promising Conservation Tool for Monitoring Fish Diversity in Dongshan Bay, China

Abstract

Dongshan Bay is a typical subtropical semi-enclosed bay characterized by abundant fish resources. We aimed to assess fish diversity and its seasonal variation in Dongshan Bay and to provide a scientific basis for the sustainable management and conservation of the fishery’s resources. In this study, we employed environmental DNA (eDNA) metabarcoding technology to analyze fish diversity in the bay during winter 2023 and summer 2024. A total of 76 fish species were detected across 12 sampling sites, with 43 species identified in summer and 45 species seen in winter. Overall, 13 species were detected in both the winter and summer. Non-significant differences were observed in Alpha diversity among the sampling sites. Fish species richness at the HXH2 site was the lowest among all the sampling sites for the reason that this sampling site was near to the effluent outlet of the Zhangzhou nuclear power plant and notably influenced by the thermal discharge. In general, fish diversity and abundance were higher in winter than in summer. RDA test analysis revealed that water temperature and dissolved oxygen were the primary environmental factors influencing fish distribution in summer. In winter, the influence of various factors is relatively balanced, with chlorophyll and Blue Green Algae Phycoerythrin (BGA PE) having a relatively greater impact than other factors. Our results offer valuable insights into enhancing fish diversity management in Dongshan Bay.

1. Introduction

Fish is the largest group of vertebrates, constituting nearly half of all vertebrate species, and plays an important role in global biodiversity protection [1]. Fish contribute substantially to human society by providing valuable fishery resources. Conducting comprehensive and systematic evaluations of biological and ecosystems is central to ecological research and serves as a vital step in safeguarding global biodiversity [2]. Fish diversity plays a pivotal role in biodiversity, while the assessment and monitoring of fish diversity serve as the foundation for ecosystem monitoring and health evaluations [3]. Traditional methods for fish diversity monitoring, such as electrofishing, netting, and trapping, typically involve collecting samples and then determining the abundance and biomass of fish through the morphological identification, counting, and weighing of the catch [4]. However, these conventional techniques are often environmentally invasive, time-consuming, labor-intensive, require advanced expertise in morphological identification, and face challenges in capturing low-density species [5]. The emergence of environmental DNA technology has introduced a novel approach to biodiversity monitoring [6]. Environmental DNA refers to the genetic material left behind by organisms in the environment, which can be derived from mitochondrial or nuclear DNA, including shed cells from tissues such as the intestines and skin, as well as bodily fluids such as urine, mucus, eggs, and sperm [7]. This technology enables researchers to detect the presence or recent presence of species without directly observing or capturing the organisms [5,8]. Environmental DNA metabarcoding allows for the identification of multiple target species in the environmental samples (such as water, sediments, or soil) by extracting DNA from these samples, using universal primers for the target taxa and performing PCR amplification using high-throughput sequencing [9]. This method does not require the capture of the organisms, and this non-invasive, efficient, and highly sensitive approach overcomes the limitations of traditional morphological surveys, presenting significant potential for biodiversity assessments [10].

Since DNA metabarcoding’s initial application in 2008, when Ficetola et al. (2008) used environmental DNA technology to monitor the invasive Rana catesbeiana in ponds, eDNA metabarcoding has seen rapidly development [11]. It is widely used in fishery management and in fish diversity monitoring in both freshwater and marine ecosystems, with particularly widespread use in biodiversity studies [12]. Thomsen et al. employed eDNA technology to evaluate fish diversity in Danish harbors, identifying 15 species. The funding included several economically significant fish and a rare migratory species [13]. Sigsgaard conducted a year-long investigation, combining water sampling and snorkeling observations along the Danish coastline [14]. As result, the seasonal dynamics of fish community were identified via eDNA metabarcoding. When compared with historical data, the eDNA metabarcoding results showed partial consistency with snorkeling observations. Notably, most fish species detected through snorkeling were identified by eDNA metabarcoding. This research highlights the utility of eDNA metabarcoding in capturing seasonal shifts in the diversity of marine fish communities. eDNA metabarcoding technology has recently emerged as a powerful tool for biodiversity monitoring [15].

Dongshan Bay is situated in a subtropical latitude zone and is influenced by the Min-Zhe coastal current and the Taiwan Strait thermocline in the autumn, and by the South China Sea in the summer, presenting the typical characteristics of a subtropical bay ecosystem [16]. The region serves as a vital habitat for many aquatic species, particularly fish species, providing essential areas for their habitat, growth, fattening, and reproduction [16,17]. However, excessive human development has led to the scattering of aquaculture areas, worsening the eutrophication of seawater. Reclamation activities have resulted in the destruction of wetland habitats, and tidal volumes have consequently diminished [18]. Dongshan Bay is also subject to various impacts from the Zhangzhou Nuclear Power Plant located within the bay. Therefore, there is an urgent need to establish a rapid, effective, and environmentally friendly monitoring method for the protection and ecological restoration of fish diversity in Dongshan Bay, providing scientific support for fishery management and the development and implementation of ecological protection policies.

In this study, we employed eDNA metabarcoding technology to analyze the relationship between fish diversity and environmental factors in Dongshan Bay during December 2023 and June 2024. The objective was to conduct an initial assessment of the fish community status in Dongshan Bay, with the goal of providing foundational data to support the conservation and management of fish diversity in this region.

2. Methods

2.1. Field Site and Sample Collection

Environmental DNA samples were collected in Dongshan Bay on 9 December 2023 and 21 June 2024. A total of 12 sampling sites that comprehensively cover the marine area and the distribution of the sampling sites are shown in Figure 1. At each site, a 1 L water sample was collected using a water sampler and then placed in a disposable sterile sample bag. If a site had a water depth greater than 5 m, two samples were taken from the surface and bottom, while for sites with a depth of less than 5 m, only one sample was taken from the surface. To avoid cross-contamination, each collection bag (Labshark, Changde, China) was rinsed twice with 150 mL of local seawater before sampling. The rinse water was discarded, and disposable gloves were replaced. The collected samples were stored in a cooling box and filtered within 24 h using a circulating water vacuum pump (Greatwall, Zhengzhou, China) and a six-way filter (PALL, Port Washington, NY, USA). Filtration was carried out using a diameter of 47 mm and a polycarbonate membrane with a pore size of 0.2 μm (Millipore, Darmstadt, Germany), and the filter was disinfected with a 75% ethanol spray. We waited for a few seconds and then wiped the filter dry with a cotton ball. We performed this both before and after filtering to eliminate residual DNA and prevent contamination [7]. The filters were then placed in 4.5 mL cryovials (Axygen, Union City, CA, USA) and stored at −20 °C until genomic DNA extraction was performed in the laboratory. DNA was extracted using the DNeasy PowerWater Kit (Qiagen, Hilden, Germany) to enrich and recover the eDNA [13,19]. The quality of the extracted eDNA was assessed using 1% agarose gel electrophoresis, and the DNA samples were stored at −20 °C for subsequent PCR amplification. Each sample was independently processed, with all equipment disinfected using 75% ethanol spray before and after each step to prevent exogenous contamination [20].

2.2. Measurement of Environmental Factors

The collection of environmental parameters was carried out simultaneously with sample collection. During the sampling process, a multi-parameter water quality meter (YSI, Yellow Springs, OH, USA) was employed to measure the following environmental parameters, including chlorophyll (chlorophyll), dissolved oxygen (DO), salinity (Sal), Blue Green Algae Phycoerythrin (BGA PE), pH (pH), and water temperature (WT).

2.3. PCR, Sequencing and Annotation

eDNA metabarcoding, conducted using universal MiFish primer pairs, has been shown to amplify short fragments of fish DNA in various taxa from environmental samples [21]. Our samples were analyzed using the universal fish primer pairs (Mifish-U-F: 5′-GTCGGTAAAACTCGTGCCAGC-3′; Mifish-U-R: 5′-GTTTGACCCTAATCTATGGGGTGATAC-3′) in order to amplify the mitochondrial 12S rRNA gene [22]. The multiplex PCR volume was 25 μL. This included 12.5 μL of Taq 2× Master Mix (Vazyme, Nanjing, China), 1 μL of each primer (10 μmol/L), 1 μL of the DNA solution, and 9.5 μL of sterile distilled H₂O. The thermocycler used was an ABI2720 model (ThermoFisher, Waltham, MA, USA). The thermal cycle PCR process included an initial 2 min denaturation at 94 °C, followed by 35 cycles of denaturation at 98 °C for 5 s each, and then annealing at 50 °C for 10 s, extension at 72 °C for 10 s, and completion with a final extension at 72 °C for 5 min [23]. The PCR products were analyzed using 1% agarose gels. The electrophoresis procedure was as follows: 2 μL of the DNA sample and 2 μL of DL2000 DNA Ladder Marker (Takara, Osaka, Japan) were loaded into the sample wells of an agarose gel. Electrophoresis was conducted in a 1% TAE (Biosharp, Beijing, China) agarose gel at 100 V with a constant voltage and an approximately 80 mA current for 20 min. DNA fragment size and concentration were assessed by comparing the band positions and intensities with those of the Marker. Samples with a bright main strip of 297 ± 25 bp were selected and all negative controls showed no target bands, confirming the purity of the samples and the absence of contamination.

QIIME2 (v.2022.11) was employed, and the analysis workflow was modified and optimized following the official tutorial (https://docs.qiime2.org/2022.11/tutorials/) (accessed on 2 September 2024). The raw sequencing data were processed, using the demux plugin for decoding, the Cutadapt (v2.3) for primer removal, and the DADA2 (v1.26) for quality filtering, denoising, and merging. Sequences were clustered into Amplicon Sequence Variants (ASVs) using the uparse algorithm in VSEARCH (v2.7.1), and the resulting sequences were clustered at 100% similarity to generate ASVs. The ASVs feature sequences were, using the BLAST algorithm, compared against the reference sequences in the NCBI (https://www.ncbi.nlm.nih.gov/) (accessed on 10 September 2024) and MitoFish (http://mitofish.aori.u-tokyo.ac.jp) (accessed on 10 September 2024) database to obtain taxonomic information for each ASV. Rare ASVs were excluded from the abundance matrix for further analyses. Non-fish ASVs were excluded, and ASVs identified as belonging to the same species were merged. The relative abundance of each fish species’ valid sequences was calculated in Excel, and species identification was cross-verified and refined using the Fishbase database (https://www.fishbase.de/search.php) (accessed on 22 September 2024) and historical catch data from Dongshan Bay [24], along with additional information on fish classification and life history.

2.4. Data Analysis

Data preprocessing was performed using Excel, and the differences in fish species between the two seasons were visualized using a Venn diagram created through the online platform Venny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/) (accessed on 27 September 2024). Initially, QIIME2 (v.2022.11) was used to randomly subsample the total sequence counts of each sample in the ASV abundance matrix at varying depths. Rarefaction curves were generated based on the number of sequences sampled at each depth and the corresponding number of ASVs. This process was used to assess whether the current sequencing depth for each sample was adequate to capture the microbial diversity of the community. Subsequently, to facilitate the comparison of diversity across samples, the ASV abundance matrix was rarefied to 95% of the sequence count from the sample with the lowest sequence count, thereby correcting for the sequencing of depth-related diversity discrepancies between samples. Then, Alpha diversity was assessed using several indices—the Shannon index ($H^{\prime }$) and Simpson index (${\mathrm{D}}^{\prime }$) [25,26]—to represent inter-group diversity. We used the Chao1 index [27] to estimate inter-group richness, the Pielou_e index ($J^{\prime }$) [28] to measure species evenness, and the goods_coverage index to evaluate sample coverage [29]. The calculation formulas are as follows:

$$\mathrm{S}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{n}\mathrm{o}\mathrm{n} \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}: H^{\prime } = -{\sum }_{i=1}^s{p}_i\ast ln(p_i)$$

(1)

$$\mathrm{S}\mathrm{i}\mathrm{m}\mathrm{p}\mathrm{s}\mathrm{o}\mathrm{n} \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}: {\mathrm{D}}^{\prime } = 1 - \sum p_i^2$$

(2)

$$\mathrm{C}\mathrm{h}\mathrm{a}\mathrm{o}1 \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}: \mathrm{C}\mathrm{h}\mathrm{a}\mathrm{o}1 \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x} = \mathrm{S} + {\displaystyle \frac{n_1(n_1-1)}{2(n_2+1)}}$$

(3)

$$\mathrm{P}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{u}\_\mathrm{e} \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}: J^{\prime } = H^{\prime }/\mathrm{l}\mathrm{n}\mathrm{S}$$

(4)

$$\mathrm{g}\mathrm{o}\mathrm{o}\mathrm{d}\mathrm{s}\_\mathrm{c}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}: \mathrm{C}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{g}\mathrm{e} \mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x} = 1 - n_1/\mathrm{N}$$

(5)

In the formulas, S is the number of ASVs; p_i denotes the relative abundance of the ASV for the i-th fish species as a proportion of the total fish abundance; n₁ represents the count of ASVs with only one sequence; n₂ represents the count of ASVs with exactly two sequences; and N indicates the total sequence count in the sample. All statistical analyses and visualizations were conducted in R (v4.3.3) [30]. The vegan package (v2.6.6.1) was used to calculate Alpha diversity indices, including the Shannon–Wiener index, Simpson index, Pielou_e index, and goods_coverage index. The Shapiro–Wilk normality test was performed on each index, revealing that none of the indices conformed to a normal distribution. Thus, the non-parametric Wilcoxon test was applied. Differences in Alpha diversity indices between groups were evaluated using either a t-test or the Wilcoxon test. In order to assess the similarity between samples, we employed pheatmap package (v1.0.12) and used hierarchical clustering. For both rows and columns, Euclidean Distance was used as the distance metric, and clustering was performed with the Complete Linkage method. Complete Linkage merges clusters based on the maximum distance between them, generating a dendrogram to illustrate the similarity relationships between samples and species. The clustering results were visualized as a heatmap, where the color of each cell indicated the standardized abundance of the samples and species. Detrended Correspondence Analysis (DCA) was conducted by using the decorana function from the vegan package to decide whether to apply Redundancy Analysis (RDA), which is based on linear models, or Canonical Correlation Analysis (CCA), which is based on unimodal models, in order to investigate the primary environmental factors influencing the distribution of fish communities. If the gradient length of the ordination axes in DCA exceeded 4.0, CCA was chosen; otherwise, RDA was used [31]. The plots were created using the ggplot2 package (v3.5.1).

3. Results

3.1. Species Composition

A total of 1,399,397 raw MiFish sequences were obtained. After initial quality filtering, low-quality sequences and chimeras were removed, leaving a total of 1,309,228 sequences and 30,881 chimeric sequences. High-quality sequences accounted for 93.56% of the total raw sequences.

A total of 76 fish species were detected across the winter and summer seasons, with 43 species identified in winter. These belonged to 23 orders, 30 families, and 41 genera. Among these, 13 species were commonly detected in both winter and summer, representing 17.1% of the total species detected, as shown in Figure 2. These species included Acanthopagrus schlegelii, Pseudobalistes fuscus, Plectorhinchus cinctus, Acanthopagrus latus, Halichoeres notospilus, Decapterus maruadsi, and so on. In winter, based on the ASVs detected, the five families with the highest relative abundance were Gobiidae, Dasyatidae, Cynoglossidae, Clupeidae, and Sparidae. In the summer, 45 fish species were identified, belonging to 19 orders, 24 families, and 40 genera. The top five families with the highest relative abundance in the summer were Gobiidae, Labridae, Clupeidae, Balistidae, and Carangidae (

Table 1. Detection of fish species in Dongshan Bay based on eDNA metabarcoding technology.

Order	Family	Species	Season
			Winter	Summer
Mugiliformes	Mugilidae	Planiliza affinis (Günther, 1861)	*	*
		Valamugil speigleri (Bleeker, 1858)	*
		Moolgarda engeli (Bleeker, 1858)	*
		Mugil cephalus (Linnaeus, 1758)	*	*
Clupeiformes	Clupeidae	Sardinops melanostictus (Jenyns, 1842)	*	*
		Clupanodon thrissa (Linnaeus, 1758)	*	*
		Sardinella albella (Valenciennes, 1847)		*
		Sardinella hualiensis (Chu & Tsai, 1958)		*
	Engraulidae	Encrasicholina punctifer (Fowler, 1938)	*
		Engraulis japonicus (Temminck & Schlegel, 1846)	*
		Thryssa hamiltonii (Gray, 1835)		*
Centrarchiformes	Girellidae	Girella punctata (Gray, 1835)	*
	Terapontidae	Pelates quadrilineatus (Bloch, 1790)		*
Chaetodontiformes	Leiognathidae	Leiognathus brevirostris (Valenciennes, 1835)	*
		Leiognathus ruconius (Hamilton, 1822)
Acropomatiformes	Banjosidae	Banjos banjos (Richardson, 1846)	*
Anguilliformes	Congridae	Uroconger lepturus (Richardson, 1845)	*
	Moringuidae	Moringua javanica (Kaup, 1856)	*
	Muraenidae	Uropterygius makatei (Bleeker, 1864)	*
	Nemichthyidae	Nemichthys curvirostris (Strömman, 1896)	*
Atheriniformes	Atherinidae	Hypoatherina valenciennei (Bleeker, 1854)	*
Beloniformes	Belonidae	Strongylura strongylura (van Hasselt, 1823)	*
Blenniiformes	Blenniidae	Parablennius yatabei (Jordan & Snyder, 1900)		*
Caproiformes	Caproidae	Antigonia rubicunda (Ogilby, 1910)		*
Carangiformes	Carangidae	Decapterus maruadsi (Temminck & Schlegel, 1843)	*	*
		Selaroides leptolepis (Cuvier, 1833)	*
		Seriola dumerili (Risso, 1810)	*	*
Gadiformes	Macrouridae	Coelorinchus hubbsi (Matsubara, 1936)	*
Gobiiformes	Gobiidae	Amblychaeturichthys hexanema (Bleeker, 1853)	*
		Drombus triangularis (Weber, 1909)	*	*
		Parachaeturichthys polynema (Bleeker, 1853)		*
		Rhinogobius giurinus (Rutter, 1897)	*
		Trypauchen vagina (Bloch & Schneider, 1801)		*
Kurtiformes	Apogonidae	Ostorhinchus pleuron (Fraser, 2005)		*
Labriformes	Labridae	Cheilinus oxycephalus (Bleeker, 1853)	*
		Halichoeres notospilus (Günther, 1864)	*
		Halichoeres hartzfeldii (Bleeker, 1852)		*
		Hologymnosus doliatus (Lacepède, 1801)	*
		Labroides bicolor (Fowler & Bean, 1928)	*
		Labroides alleni (Randall, 1981)		*
		Leptojulis cyanopleura (Bleeker, 1853)	*	*
		Parajulis poecilepterus (Temminck & Schlegel, 1845)		*
		Coris pictoides (Randall & Kuiter, 1982)		*
Lutjaniformes	Haemulidae	Plectorhinchus cinctus (Temminck & Schlegel, 1843)	*	*
Perciformes	Platycephalidae	Platycephalus indicus (Linnaeus, 1758)	*
	Cottidae	Trachidermus fasciatus (Heckel, 1837)		*
	Sebastidae	Sebastiscus marmoratus (Cuvier, 1829)		*
	Serranidae	Epinephelus akaara (Temminck & Schlegel, 1842)		*
		Epinephelus bruneus (Bloch, 1793)		*
		Epinephelus coioides (Hamilton, 1822)		*
	Stichaeidae	Alectrias benjamini (Jordan & Snyder, 1902)		*
Pleuronectiformes	Cynoglossidae	Cynoglossidae quadrilineatus (Bleeker, 1851)	*
		Paraplagusia japonica (Temminck & Schlegel, 1846)		*
Scombriformes	Scombridae	Scomber japonicus (Houttuyn, 1782)	*	*
		Auxis thazard (Lacepède, 1800)		*
	Trichiuridae	Trichiurus japonicus (Temminck & Schlegel, 1844)	*
Spariformes	Sparidae	Acanthopagrus latus (Houttuyn, 1782)	*	*
		Acanthopagrus schlegelii (Bleeker, 1854)	*	*
		Rhabdosargus sarba (Forsskål, 1775)		*
Mulliformes	Mullidae	Upeneus japonicus (Houttuyn, 1782)	*
Myctophiformes	Myctophidae	Benthosema fibulatum (Gilbert & Cramer, 1897)		*
		Benthosema pterotum (Alcock, 1890)		*
		Myctophum aurolaternatum (Garman, 1899)		*
Syngnathiformes	Syngnathidae	Hippichthys penicillus (Cantor, 1849)	*
		Hippichthys spicifer (Rüppell, 1838)	*
Tetraodontiformes	Balistidae	Pseudobalistes fuscus (Bloch & Schneider, 1801)	*	*
	Monacanthidae	Monacanthus chinensis (Osbeck, 1765)		*
		Paramonacanthus otisensis (Whitley, 1931)		*
		Stephanolepis setifer (Bennett, 1831)		*
Trachichthyiformes	Monocentridae	Monocentris japonicus (Houttuyn, 1782)	*
Zeiformes	Parazenidae	Cyttopsis cypho (Fowler, 1934)	*
	Zeidae	Zeus faber (Linnaeus, 1758)	*
Myliobatiformes	Dasyatidae	Brevitrygon walga (Müller & Henle, 1841)	*
		Hemitrygon akajei (Müller & Henle, 1841)		*
Osmeriformes	Salangidae	Neosalangichthys ishikawae (Wakiya & Takahashi, 1913)		*
		Salangichthys microdon (Bleeker, 1860)		*

Note: * indicates that the fish was detected in the season.

).

Regarding species richness, the HX3 station had the highest number of species in winter, with 18 species detected, while the HXH1 station had the lowest, with only 5 species detected. In summer, the DS04 station had the highest number of species, with 22 species detected, and the HXH2 station had the lowest, with only 3 species detected, as shown in Figure 3. The top 10 dominant species by relative sequence abundance across the sampling sites are shown in Figure 3. Gobius sinensis and Sphaeramia nematoptera were detected at all sites during both winter and summer.

3.2. Alpha Diversity Analysis

To comprehensively assess fish species richness and diversity across sampling sites, 5 Alpha diversity indices were calculated. As shown in

Table 2. Alpha diversity index.

Site	Shannon Index		Simpson Index		Chao1 Index		Pielou_e Index		goods_coverage Index
	W	S	W	S	W	S	W	S	W	S
HX3	1.979	1.151	0.768	0.414	18.333	19.200	0.475	0.295	0.998	0.931
DS03	2.093	1.823	0.807	0.783	17.000	12.500	0.512	0.527	1.000	0.967
DS04	1.645	2.545	0.756	0.878	11.000	25.500	0.476	0.571	0.999	0.931
DS05	1.953	0.591	0.828	0.268	12.000	5.000	0.545	0.254	0.998	0.988
HXH2	1.550	0.506	0.706	0.255	7.000	3.000	0.552	0.319	1.000	1.000
HXE1	1.716	0.977	0.719	0.494	13.000	5.000	0.464	0.421	1.000	0.982
DS06	1.408	1.396	0.623	0.682	11.000	7.000	0.407	0.497	1.000	1.000
HX1	1.232	1.218	0.608	0.678	7.000	4.000	0.439	0.609	1.000	1.000
HX2	1.978	1.311	0.822	0.678	12.000	5.000	0.553	0.565	1.000	1.000
HX5	0.861	1.643	0.377	0.715	9.000	13.000	0.287	0.458	0.994	0.972
HXH1	0.701	0.681	0.354	0.415	5.000	3.000	0.302	0.430	1.000	0.972
HX4	1.646	2.234	0.765	0.876	10.000	11.000	0.495	0.646	1.000	1.000

Note: W: winter; S: summer.

, the goods_coverage index for winter exceeded 0.99 across all sites, reflecting high sample coverage [32]. The average Shannon index for winter was 1.563, peaking at 2.093 at DS03 and dipping to 0.701 at HXH1. The average Simpson index was 0.678, ranging from 0.828 at DS05 to 0.354 at HXH1. Although the Shannon and Simpson indices showed general consistency, site-specific differences were evident. The Chao1 index was averaged at 11.028, with the highest value of 18.333 recorded at HX3 and the lowest value of 5.000 recorded at HXH1. The Pielou_e index averaged 0.459, with the highest value of 0.553 seen at HX2 and the lowest value of 0.287 seen at HX5.

In summer, the average Shannon index was 1.340, with the highest value of 2.545 observed at DS04 and the lowest value of 0.506 seen at HXH2. The Simpson index averaged 0.595, ranging from 0.878 at DS04 to 0.255 at HXH2, following a similar trend. The Chao1 index averaged 9.433, with a maximum value of 25.5 seen at DS04 and a minimum value of 3.000 seen at HXH2. The Pielou_e index was 0.466 on average, with the highest value of 0.646 recorded at HX4 and the lowest of 0.254 at DS05.

The fish species composition heatmap at the species level reveals differences in species composition across various sampling sites, as shown in Figure 4. Clustering analysis of the sites indicates that winter sites HXE1 and DS03 share the greatest similarity. Regarding species clustering, the distributions of Strongylura strongylura and Leiognathus brevirostris are most similar across different sites. In terms of species abundance, DS03 demonstrates higher species richness and diversity, while HXH1 shows lower diversity, which is consistent with the Alpha diversity index results. In the summer, sites HXH2 and DS05 exhibit the greatest similarity. Species clustering shows that the distribution of Pelates quadrilineatus and Scomber japonicus across different sites is most similar. In terms of species diversity and richness, the DS04 and HX5 sites show higher levels of both, while HXH2 exhibits lower diversity, which corresponds closely to the Alpha diversity index findings.

Figure 5 presents a comparison of Alpha diversity between the two seasons. Both the Shannon and Simpson indices in winter are higher than those in summer, suggesting that the fish diversity in winter is generally higher than that in summer. The Chao1 index for winter is also greater than that in summer, although summer shows two notable larger values at the HX3 and DS04 sites. The species richness at these two summer sites is significantly higher than that at the other sites, but overall, fish richness remains lower in summer compared to winter. There is no significant difference in the Pielou_e index between the two seasons (p > 0.05). The uniformity between sites in summer shows more variation, while the winter uniformity is relatively consistent, with the Pielou_e index predominantly around 0.5.

3.3. Correlation Analysis of Environmental Factors

Figure 6 presents the environmental factors determining the fish distribution. In winter, the arrows for WT and Sal are relatively short, suggesting that these factors have a limited capacity to distinguish between samples during winter. In contrast, chlorophyll and BGA PE exhibit longer arrows, indicating that they have a more significant influence on sample distribution in winter. HX1 and HX2 are located closer to chlorophyll, suggesting that these sites are more strongly influenced by this environmental factor. DS05 appears as an isolated point, showing a weak correlation with most environmental variables. In summer, the arrows for WT and DO are notably extended, indicating that these factors have a significantly stronger influence on sample distribution in summer. Sal continues to have a significant effect in summer, consistent with the winter findings. HX3 and HX4 are positioned closer to WT and DO, suggesting that these sampling sites are located in areas with high temperature and high oxygen levels. In winter, the primary factors are chlorophyll and BGA PE, indicating that photosynthesis-related variables play a dominant role during this season. In contrast, WT and DO are the key factors in summer, highlighting that WT and DO are the decisive influences. The sample distribution in winter is more dispersed, suggesting the environmental factors have a lesser impact on the fish community structure across the sites. In summer, the sample distribution becomes more concentrated, indicating that environmental factors may have a stronger gradient effect.

4. Discussion

4.1. Fish Species Composition in Dongshan Bay During Winter and Summer

Due to the extensive area of cage aquaculture in Dongshan Bay, traditional sampling methods, such as bottom trawling, are difficult to implement in these areas [24]. As a result, we utilized eDNA metabarcoding technology to analyze the fish diversity in Dongshan Bay during winter and summer. The eDNA metabarcoding analysis of samples from 12 sites identified a total of 76 fish species. At the class level, the majority of species belonged to Actinopteri, while only one species from Chondrichthyes, Brevitrygon walga—was detected in winter. The Mifish-U primers used in the analysis are optimized for amplifying Actinopteri and show limited effectiveness for Chondrichthyes, which contribute to these phenomena [33]. A total of 43 species were identified in winter and 45 were found in in summer: the total number of species detected showed little variation. Overall, 13 species were detected in both winter and summer, representing 27.9% of the winter species and 26.7% of the summer species, highlighting distinct seasonal differences in species composition. In a gillnet survey conducted by our laboratory in November 2014 (unpublished), 115 fish species were captured, exceeding the 76 species identified in this study. A comparison shows that the 2014 survey involved more sampling sites and a longer duration, with a broader spatial distribution of sites. The differences in temporal and spatial factors were likely the main reasons for the discrepancy in the number of fish species identified between the two studies. Previous fishing surveys in Dongshan Bay and its adjacent waters identified 43 fish species in spring and detected 45 in autumn [34]. This was similar to the number of species identified through eDNA, but there was a notable difference in terms of dominant species composition. According to past fishing surveys, Thryssa mystax was the dominant species, with other key species including Sardinella zunas, Leiognathus brevirostris, and Secutor ruconius. In contrast, our eDNA survey revealed that the dominant species in Dongshan Bay were smaller species such as Drombus triangularis, Rhinogobius giurinus, and Clupanodon thrissa. This results suggests a clear trend towards smaller dominant species in the fish community of Dongshan Bay, as well as a marked decrease in economically significant species. As a crucial fishery resource base in Fujian Province, Dongshan Bay has been experiencing increasing fishing pressure due to its growing development. An assessment in 2018 indicated that the overall carrying capacity of Dongshan Bay’s fishery resources is reaching a critical threshold, and the impact of fishing may be a significant driver of the observed changes in fish species composition and diversity in this region [35].

4.2. Alpha Diversity Analysis and Correlation with Environmental Factors

In winter, the minimum values for the diversity and richness indices, including Shannon, Simpson, and Chao1, were observed at the HXH1 site, indicating that this site exhibited the lowest levels of fish diversity and richness. The HXH1 site is located at the intake of the Zhangzhou nuclear power plant, which generates significant heat during its operation and uses a cooling water system to dissipate this heat [36]. The intake can become obstructed due to the accumulation of aquatic organisms and sediment, and intercepting nets are installed to drive swimming species. Dispersal agents are also regularly used to drive away nearby aquatic life, which may be a significant factor contributing to the lower fish diversity and richness at HXH1 [37,38]. The DS03 site exhibited high diversity, while HX3 had the highest species richness. DS03 and HX3 are located in coral and aquaculture areas, respectively. These areas provide optimal ecological conditions and abundant food resources [39], attracting fish to inhabit and forage in these environments. Human activities in the aquaculture area also have negative effects, one of which is the onset of red tides caused by water eutrophication. The secretions from red tide algae can obstruct the gills of fish, impairing their ability to breathe and leading to suffocation and death. Moreover, the respiratory processes of the algae, along with the breakdown of their dead cells, significantly deplete the amount of DO in the seawater, resulting in severe hypoxia that further contributes to fish suffocation. Additionally, certain toxic algae produce substances that, once ingested by fish, accumulate in their bodies, ultimately leading to poisoning and death. Dongshan Bay, which is typically prone to red tide events, did not experience any during the two seasons observed in this study.

In summer, the highest and lowest diversity and richness values, as indicated by the Shannon, Simpson, and Chao1 indices, were found at the DS04 and HXH2 sites, respectively. The DS04 site recorded 22 fish species, while HXH2 recorded only 3 species, including Labropsis alleni, Drombus triangularis, and Coris pictoides. Despite the proximity of these two sites, there was a considerable difference in fish composition and diversity. This disparity could be explained by the fact that DS04 is located in an aquaculture area. Human activities in the aquaculture area, such as feeding and the discharge of domestic wastewater, significantly contribute to the increase in nutrients in the water. This nutrient enrichment fosters the growth of phytoplankton, which in turn boosts primary productivity at the site and attracts fish to congregate [40]. Another contributing factor may be the escape of cultured fish from nearby farms, with species such as Acanthopagrus schlegelii, Decapterus maruadsi, and Mugil cephalus commonly being cultured. In contrast, HXH2 is located near the discharge outlet of the nuclear power plant, where fish are impacted by thermal pollution from the warm discharge water [41], which may explain the lower diversity and richness at this site.

Fish community is influenced by both biotic and abiotic factors, and the relationship between environmental factors and fish communities varies significantly across different scales and ecosystem types [42,43]. In winter, environmental factors have a relatively balanced influence, with photosynthesis-related factors having a greater effect. Chlorophyll is strongly associated with primary productivity [44], and phytoplankton, which form the foundation of the aquatic food web, play a crucial role in providing food for fish. On the other hand, BGA PE is an important indicator in predicting red tide events. The massive proliferation of red tide algae can deplete nutrients in the water and even generate toxins, posing a threat to fish survival [45]. In the summer, WT and DO have a great impact on fish diversity. The influence of WT on fish communities is multifaceted, involving both physiological adaptation and ecological interactions [46]. An appropriate water temperature supports the stability and health of fish communities, while extreme temperatures, either too high or too low, can destabilize community structure, impact fish reproduction, food availability, and habitat conditions, ultimately leading to shifts in fish populations. DO supports the healthy circulation of substances in the water and reflects the suitability of the habitat for fish [47]. Dongshan Bay is a eutrophic bay, and during summer, rising temperatures lead to large cyanobacteria blooms, which can reduce DO levels in the water. However, according to the results of the environmental parameters, DO did not exhibit a negative correlation with temperature. This suggests that the summer survey may have occurred during the early stages of a red tide, when the oxygen produced by cyanobacterial photosynthesis outweighed the oxygen consumed through respiration. The level of DO in seawater affects fish respiration, swimming speed, and metabolic rate. Furthermore, chlorophyll and DO levels also influence zooplankton size and composition, which in turn affects fish species that predominantly feed on zooplankton, such as Halichoeres notospilus and Decapterus maruadsi. This could help explain the relatively higher sequence abundance of these species in our summer samples.

4.3. Impact of Thermal Discharge from Zhangzhou Nuclear Power Plant on Fish Communities

Water temperature plays a critical role in marine ecosystems and the biological activities of marine organisms. It significantly influences individual growth, metabolism, the maturation of reproductive cells, and the overall life cycle of species [48]. Compared to terrestrial and freshwater environments, ocean water temperature fluctuations are generally smaller, and marine organisms, including fish, have relatively low tolerance to temperature changes, making them more susceptible to the effects of thermal pollution [49,50]. Thermal discharge can alter the normal distribution of aquatic organisms, leading to shifts in community structure and abnormal developmental occurrences, and can also significantly impact migratory species. Thus, the impact of thermal discharge on fish is an issue that should not be underestimated [51].

DO levels in the aquatic environment significantly influence the life activities of aquatic organisms as DO is one of the essential factors required for metabolic processes. There is a strong negative correlation between water temperature and dissolved oxygen content. As the water temperature rises from 0 °C to 40 °C, the dissolved oxygen concentration decreases. However, in general, temperature increases in non-polluted water bodies do not lead to a reduction in dissolved oxygen levels below the minimum threshold required for fish survival [52]. However, at the discharge site during both seasons, no negative correlation was observed between DO and WT. The effect of thermal effluent diffusion on the environment is a complex issue. Given the elevated chlorophyll levels at the HXH2 site, we hypothesize that the higher water temperature at this site facilitated the growth of thermophilic algae [53], which in turn boosted the DO concentration in the water, counteracting the direct impact of WT on DO.

At the HXH2 site, located near the discharge outlet of the Zhangzhou Nuclear Power Plant, the water temperature is notably influenced by thermal discharge. The temperature at HXH2 (30.57 °C) is the highest among the 12 sampling sites, being 5 °C higher than the lowest temperature observed at HX4 and 3 °C above the average temperature at neighboring sites. Under high summer temperatures, thermal discharge further raises the water temperature at HXH2, surpassing the optimal living temperature for many fish species. This results in the avoidance of the area by fish during peak temperature periods. The three species detected at HXH2 are all warm-water species that prefer temperatures above 29 °C. Dongshan Bay, being a semi-enclosed bay, experiences seasonal variations in temperature, with winter temperatures primarily influenced by the Taiwan warm current [54]. During this period, freshwater input from the Zhangjiang River is minimal, and the warm water from the Taiwan current dominates, bringing additional warmth to the bay, which has limited water exchange. As a result, the environmental data collected during winter showed a temperature gradient, decreasing from south to north. The inflow of warm water during the winter months did not create a noticeable heating effect, and the water temperature at HXH2 did not differ significantly from surrounding areas. Moreover, there was no evidence of fish aggregating around thermal discharge areas during the colder months.

5. Conclusions

Traditional methods for fish resource surveys are time-consuming and labor-intensive. In contrast, eDNA metabarcoding technology outperforms traditional techniques in sensitivity, standardization, and species identification. It is also simple to use and has great potential for monitoring and conserving fish diversity. Despite the many advantages of eDNA metabarcoding technology, it cannot yet fully replace traditional fish survey methods. One limitation is that eDNA metabarcoding can only identify the presence of species based on genetic information from environmental samples, without providing key details such as population size, age structure, physiological condition, and the developmental stages of the target species. Additionally, the effectiveness of eDNA metabarcoding depends on the completeness of molecular databases; missing sequences for target species in the database can lead to false-negative results. Moreover, due to the complex mechanisms underlying eDNA presence in aquatic environments, further research is needed to improve the accuracy of estimating species relative biomass from eDNA sequence abundance.

In Dongshan Bay, where bottom trawl surveys are difficult to conduct, we applied eDNA metabarcoding technology to detect a total of 76 fish species over the winter and summer seasons, with 43 species detected in winter and found 45 in summer. Thirteen species were shared between the two seasons, revealing significant differences in species composition. We also analyzed the distribution of fish diversity across sampling sites, and the alpha diversity indices showed no significant seasonal differences, suggesting that the fish community in the region is relatively stable. Furthermore, we provide preliminary evidence of the impact of thermal discharge from the Zhangzhou Nuclear Power Plant on local fish communities. Fish exhibited a clear avoidance of high-temperature areas during the summer, while no significant changes were observed in the winter. Our findings can provide valuable scientific support for the conservation of fish diversity and the sustainable development of fishery resources in Dongshan Bay.