Introduction

The Yangtze River (YR) is the world’s third-longest river, having a total length of 6380 km (Shen et al., 2019). The upper reaches of YR refer to the mainstream section from Yibin to Yichang, which includes a variety of geomorphological types, such as mountains, hills, and plains, with a substantial elevation variation in the riverbed of more than 1500 m, alternating slow and fast currents and complex flow patterns (He et al., 2010; Wei, 2012; Liu et al., 2019; Meng et al., 2019). It is home to several rare and endemic fish species, as well as being a critical location for the conservation of the genetic germplasm and biodiversity of fish (Xie & Chen, 1999; Qin et al., 2008). The intricate topography and heterogeneous climate of the YR basin have engendered a wide variety of aquatic habitats, thereby fostering a diverse array of aquatic communities, comprising over 400 fish species (approximately 360 freshwater fish species), with endemic fish constituting up to 42% of the overall fish number (Cao, 2011; Liu et al., 2019). Meanwhile, there are 286 fish species in the upper YR, including 124 endemic fish, accounting for 79.49% of all endemic fish in the YR (He et al., 2010; Cao, 2011; Wei, 2012; Meng et al., 2019). With a total length of nearly 200 km, the section of the YR from Jiangjin to Fuling is located in the middle of the upper reaches. It is a crucial section in the upper reaches that includes the National Aquatic Germplasm Resources Conservation Area of the four major Chinese carps in the Chongqing Section of the Yangtze River (the reserve of the four major Chinese carps (Ctenopharyngodon idella (Valenciennes 1844), Hypophthalmichthys molitrix (Valenciennes 1844), Hypophthalmichthys nobilis (Richardson 1845), and Mylopharyngodon piceus (Richardson 1846)) and the connecting section of the national nature reserve of rare and endemic fish in the upper YR and the reserve of the four major Chinese carps.

As an essential part of the water ecosystem, fish impact the water environment and numerous aquatic organisms via upstream and downstream effects (Reis et al., 2020). However, due to anthropogenic factors, including dam construction, dredging and quarrying, and channel regulation, the endangerment of fish in the YR is increasing, and the number of threatened fish species is highest in the upper reaches of the YR basin (Jiang et al., 2016b). According to recent research, miniaturization of fish species is a serious problem in the upper YR, and endemic fish communities have been drastically reduced (Gao et al., 2015; Yang et al., 2017; Wang et al., 2019; Zhou et al., 2022). Meanwhile, the number of some fish species that are not native to the upper mainstream area of the YR, such as Neosalanx taihuensis (Chen 1956) and Pseudolaubuca engraulis (Nichols 1925), which prefer slow or still water habitats, have increased in the upper reservoir area (Gao et al., 2015; Yang et al., 2017; Wang et al., 2019; Zhou et al., 2022). According to previous research, the development of the upper YR cascade hydropower will fragment the habitats of 134 species of fish, restrict the migratory paths of 35 species, and impact the reproduction of 26 species which reproduce by drifting eggs fish (Cheng et al., 2015). Hence, the fish communities in the upper YR must be protected immediately.

Fish diversity is an essential aspect of biodiversity in aquatic ecosystems and a critical indicator that can be used to assess the health of aquatic ecosystems (Holmlund and Hammer 1999; Zou et al., 2020). Several researchers have explored and analyzed the upper reaches of the YR, which is a hotspot for fish diversity research. Traditional fishing methods (e.g., gill nets, ground cages, etc.) are still widely employed in the YR to monitor the fish diversity (Wu et al., 2011; Gao et al., 2013; Xing et al., 2021), which is not only time-consuming and the results are contingent, but is also hazardous to fish and to the environment. Further, it is challenging to discover rare or small-sized fish with this technique (Becker et al., 2015; Sales et al., 2018). Furthermore, acoustic detection has been utilized to monitor YR fish (Shi 2019), however, this approach focuses more on the density of species than on species-specific identification. Simultaneously, in a non-invasive manner, eDNA technology has been used extensively due to its ability to gather information on the presence of species and even produce quantitative estimations of species (Doi et al., 2017).

By extracting DNA from the environment, selecting appropriate primers, and integrating molecular biology techniques such as PCR and DNA sequencing, eDNA techniques are utilized for the qualitative or quantitative analysis of species. All DNA present in environmental samples, including mixed DNA from inside and outside the cells of the target organism, such as epidermal cells, secretions, germ cells, and so on, are referred to as eDNA. This allows us to determine the presence of the species in the approximate range without collecting organisms (Taberlet et al., 2012; Thomsen et al., 2015; Balasingham et al., 2018). The technology is also commonly used in the YR basin (Qu et al., 2020; Li et al., 2021; Wang et al., 2022). Even though eDNA is easily affected by many environmental factors, including the temperature, ultraviolet light, and pH (Bohmann et al., 2014; Strickler et al., 2015), studies have shown that the detection efficiency and sensitivity of this technology can be higher than those of traditional monitoring methods (Dejean et al., 2012; Jerde et al., 2013; Valentini et al., 2016), indicating that eDNA technology is an effective complementary tool for traditional monitoring methods (Hinlo et al., 2017).

On January 1, 2021, the “10-year ban on fishing in the YR basin” came into effect. Traditional resource survey approaches have been constrained in the context of YR preservation. In this study, the eDNA technique was used to analyze the fish species composition in the upper reaches of the YR from Jiangjin to Fuling to explore its applicability for fish monitoring and provide data for the “10-year ban on fishing the YR.”

Materials and methods

Study area

The river section from Jiangjin (106° 26′ 44.43″ E, 29° 21′ 9.72″ N) to Fuling (107° 30′ 30.35″ E, 29° 53′ 6.02″ N) is roughly 200 km long and is located in the upper reaches of the YR. A total of 12 sampling sites were established in this river section, with four points in the non-protected areas, Luo Huang (LH), Da Dukou (DDK), Chao Tianmen (CTM), and Xia Kou (XK), and eight points in the protected areas, Guang Yangdao (GYD), Mu Dong (MD), Ma Liu (ML), Chang Shou (CS), Shi Tuo (ST), Li Du (LD), Qing Xi (QX), and Nan Tuo (NT) (Fig. 1).

Fig. 1
figure 1

Information on sampling locations created by ArcGIS 10.7

Full size image

eDNA sampling and processing

In early August 2021, water collectors gathered 8 l of surface water within 50 m of each sampling point along the riverbank, which was further mixed, and then 6 l of the water samples were divided into three polyethylene bottles (3 replicate samples), a total of 36 samples. Before collecting water samples from several sampling sites, all equipment and sampling bottles were rinsed with a 10% bleach solution and replaced with disposable gloves to prevent exogenous DNA contamination (Dibattista et al., 2017). Within 24 h, the water samples were immediately stored under refrigeration and pumped through a vacuum peristaltic pump onto a 0.45 µm mixed cellulose membrane (Whatman, UK). Water samples with sediments were pre-filtered using sterile medical gauze before collection (Stewart et al., 2017). Similarly, all filtering apparatus were sterilized before sample filtration to avoid cross-contamination between samples. A negative control was also set up with distilled water at the time of filtration (Zhang et al., 2020). Finally, the membranes were frozen at − 80°C until use in the next DNA extraction step.

The total DNA from the water samples was extracted from the filter membrane using the PowerWater DNA Isolation Kit (Qiagen) in accordance with the instructions of the manufacturer. Next, 1% gel electrophoresis was used to detect the quality of the extracted eDNA. Each sample was extracted separately, and a blank filter membrane was established as a negative control simultaneously. Finally, the extracted DNA samples were stored at − 80°C until the next step of PCR amplification.

The samples were amplified using universal primers (tele02-F: 5′-AAA CTC GTG CCA GCC ACC-3′; tele02-R: 3′-GGG TAT CTA ATC CCA GTT TG-5′), which amplify a short fragments of the 12S rRNA mitochondrial gene (Taberlet et al., 2018). The 20 µl TransStart Fastpfu DNA polymerase amplification system contains 4 µl of 5 × FastPfu buffer, 2 µl of dNTPs (2.5 mM), 0.4 µl of FastPfu polymerase, 1–2 µl of template DNA (10 ng), 0.8 µl of each of the upstream and downstream primers (10 µM in total). It was made up to 20 µl with ddH2O. The following PCR reaction conditions were used: predenaturation at 95°C for 5 min, 35 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 45 s, and a final extension at 72°C for 10 min. The ddH2O was used as a template for the PCR negative control to determine whether contamination occurred during the PCR amplification. The PCR amplification process was repeated three times for each sample, and their PCR products were mixed and detected by 2% agar gel electrophoresis. The target bands in this investigation were obtained by electrophoresis for 36 samples and were found to be around 167 bp in length, but none of the negative controls had a target band. Finally, the purified PCR products were recovered, then the Illumina PE250 library preparation was constructed and sequenced via paired-end sequencing by the Illumina sequencing facility.

Bioinformatic analyses and taxonomic assignment

Firstly, some poor-quality sequences, such as those less than 100 bp in length, were filtered out using Trimmomatic v.0.36 (Bolger et al., 2014), and pairs of reads were merged into a sequence using FLASH (v.1.2.11) (Magoč and Salzberg 2011). Then, the chimeras were removed using Usearch software combined with the reference sequences of the GOLD database and denovo sequences. Using Usearch (version 10 http://drive5.com/uparse/), high-quality sequences were combined as “parent–child” sets with 97% similarity (Zhang et al., 2022), and the unique sequences were obtained. Later, the comparison and taxonomic annotation were done on the public database, NCBI (https://www.ncbi.nlm.nih.gov/) with the Blastn tool using the uclust algorithm. A preliminary taxonomic (i.e., molecular operational taxonomic unit (MOTU)) annotation table was obtained by searching the unique sequences in the public database using criteria of ≥ 97% similarity (Sales et al., 2020), e ≤ 10–5, and coverage ≥ 0.9. The results obtained from the annotations were manually filtered to remove MOTUs sequences of non-fishes, and then combined with historical data from the Yangtze River basin to screen out MOTUs sequences of fish that are unlikely to belong to the region. Finally, MOTUs with sequence numbers of greater than ten were retained (Shu 2022; Zhang et al., 2022).

Reads from each sample were randomly selected using QIIME v.1.9.0 to normalize all eDNA sample data by the smallest number actually sequenced in all samples (Caporaso et al., 2010). After standardization, the relative sequence abundance (read ratio) of each species in each sample was kept constant. The averaged values of three parallel samples were taken for subsequent analyses such as species composition, alpha diversity, and beta diversity. The species composition mapping was based on the species sequence abundance ≥ 1%, where parts with an abundance of less than 1% can be merged with others (Wang et al., 2022). The alpha diversity, often expressed as the species richness, is the diversity of species within a relatively small area and reflects the results within a sample (Tuomisto 2010). The beta diversity analysis shows the variation in species characteristics across sites, reflecting the relationships among samples (Anderson et al., 2011). Based on the Bray–Curtis distance, the principal co-ordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) analysis were selected for the beta analysis.

Based on the relative sequence abundance of the species, the Shannon index (Shannon 1948), Simpson index (Simpson 1949), and Pielou index were calculated for each site, and the McNaughton index was used to determine the dominant species (Ling et al., 2022; Woodland et al., 2019). The Shannon index measures the community diversity by taking the richness and evenness of the community into consideration, with a higher index values indicating higher community diversity (Shannon 1948). The Simpson index commonly represents an area biodiversity, where higher values indicate a lower community diversity (Simpson 1949). The index is calculated as follows:

$$ {\text{Shannon-Wiener}}\,{\text{index}}: \, H^{\prime} = \sum {{{P_{i} \log P_{i} , \, P_{i} = \, n_{i} } \mathord{\left/ {\vphantom {{P_{i} \log P_{i} , \, P_{i} = \, n_{i} } N}} \right. \kern-0pt} N}} $$
(1)
$$ {\text{Simpson}}\,{\text{index}}: \, D = {{{1} - \left[ {\sum {n_{i} \left( {n_{i} - 1} \right)} } \right]} \mathord{\left/ {\vphantom {{{1} - \left[ {\sum {n_{i} \left( {n_{i} - 1} \right)} } \right]} {N\left( {N - 1} \right)}}} \right. \kern-0pt} {N\left( {N - 1} \right)}} $$
(2)
$$ {\text{Pielou}}\,{\text{index}}: \, J = H/H_{\max } $$
(3)
$$ {\text{McNaughton}}\,{\text{index}}:{{Y_{i} = \, n_{i} } \mathord{\left/ {\vphantom {{Y_{i} = \, n_{i} } {N^{\prime}f_{i} }}} \right. \kern-0pt} {N^{\prime}f_{i} }} $$
(4)

In the formulas, N is the total number of fish sequences detected; ni is the number of sequences of the i-th fish species; H is the Shannon index; Hmax is the maximum Shannon index that can be achieved with the same species richness (i.e., when the abundance of all species in the community is identical); Yi is the dominance index of the i-th fish species (species with Yi > 0.02 are dominant); and fi is the frequency of occurrence of species i. The statistical analysis was performed and figures were generated using the Biozeron Cloud Platform (http://www.cloud.biomicroclass.com/CloudPlatform).

Results

Sequence information and taxonomic assignment

A total of 5,067,423 valid sequences were collected from 12 sample locations. The sequenced data are published in the NCBI sequence Read Archive (SRA) database (accession numbers: SRR19217018-SRR19217053). A total of 62 MOTUs were shared by all sampling sites, which accounted for 59.62% of the total number of MOTUs in the fish with this river section (Fig. 2). A comparative annotation analysis of the database the fish in the Jiangjin to Fuling section of the upper YR included eight orders, 24 families, 72 genera, and 104 fish species based on recovered MOTUs (Table 1).

Fig. 2
figure 2

Petal map of MOTUs with fish in this river section contained in each sampling site. (The number in the middle of the petal map represents the number of MOTUs shared by all sampling points, and the number at the edge of the petal represents the total number of MOTUs owned by a sampling point.)

Full size image
Table 1 Species lists and sequence numbers of fish detected by eDNA at 12 sampling sites from Jiangjin to Fuling in the upper reaches of the Yangtze River
Full size table

Fish species composition based on historical information and eDNA

Over the last 10 years, we have collected data from the YR section from Yibin to Wushan (Li et al., 2013; Xiong et al., 2015; Wei et al., 2021). In this section, there are 175 fish species belonging to 11 orders, 28 families, and 98 genera, included 11 national-level protected fish species, 12 municipal-level protected fish species, and 79 endemic fish species (Table S1). The Cyprinidae represent the most species-rich family in the region, accounting for 56.57% of the population, followed by the Cobitidae at 9.14% and Bagridae at 8.57%, while the specie number of all other families are low (Fig. 3b).

Fig. 3
figure 3

Family level composition map based on eDNA (A) and traditional methods (B) (Li et al., 2013; Xiong et al., 2015; Wei et al., 2021). (The boxed families are unique to this result)

Full size image

In the eDNA data, based on number of species, Cyprinidae (n = 54) was shown to have the highest share of 51.92%, followed by Cobitidae (n = 11) with 10.58% and Bagridae (n = 10) with 9.62%. The remaining families contained no more than three species each (Fig. 3a). The results of this study also included six national protected fish species Acipenser sinensis (Gray 1835), Coreius guichenoti (Sauvage et Dabry 1874), Rhinogobio ventralis (Sauvage et Dabry 1874), Schizothorax davidi (Sauvage 1880), Leptobotia rubrilabris (Dabry 1872), Myxocyprinus asiaticus (Bleeker 1864), three municipal-level protected fish species Ochetobibus elongatus (Kner 1867), L. rubrilabris, Leptobotia microphthalma (Fu et Ye 1983), and 19 species of fish that are endemic to the upper YR, along with 17 exotic species, accounting for 16.35% of the total number of species.

The species composition differed between sampling sites at the genus level, as shown in Fig. 4. The McNaughton index was used to determine the species dominance, and the eight dominant fish species in the YR section were shown to be C. idella, Hemiculter tchangi (Fang 1942), Rhodeus sinensis (Günther 1868), Pseudorasbora parva (Temminck et Schlegel 1846), Cyprinus carpio (Linnaeus 1758), Triplophysa cf. rosa (Chen et Yang 2005), Tachysurus fulvidraco (Richardson 1846), and Rhinogobius cliffordpopei (Nichols 1925), which all exhibited extremely high levels of relative sequence abundance at all sampling sites. In contrast, M. asiaticus, Elopichthys bambusa (Richardson 1845), O. elongatus, Acheilognathus cf. rhombeus (Temminck et Schlegel 1846), and 16 other fish species were found exclusively at a few sampling points (< 5 sampling points) with extremely low relative sequence numbers. Moreover, the overall relative sequences abundance of the “Four major Chinese carp” varied considerably in this river section. C. idella was dominant in terms of its relative sequence abundance among the “Four major Chinese carps,” followed by H. molitrix, then H. nobilis and finally M. piceus.

Fig. 4
figure 4

Species composition of fish at the genus level based on the relative sequence abundance

Full size image

On one hand, the division of fish by habitat stratum at each sampling site revealed that demersal fish accounted for more than half of all fish species, whereas benthopelagic fish accounted for 40% to 54%, based on the relative sequence abundance. In contrast, the spawning types of results demonstrated, based on the species number and relative sequence abundance, that adhesive eggs accounted for more than 50% of all egg types at each sampling site. Between the sampling points, the percentages of different habitat layers or different spawning types were relatively similar (Fig. 5).

Fig. 5
figure 5

Habitat stratigraphy classification (A1 based on the number of species; A2 based on the number of sequences) and spawning types (B1 based on the number of species; B2 based on the number of sequences B2)

Full size image

Alpha diversity analysis

Each sampling point had a coverage of between 0.999959 and 0.999908, suggesting that the sequencing depth included all sequences and reflects the true situation of the samples (Chen, 2020). The mean Shannon index value for the 12 sampling sites ranged from 2.346796 to 2.721591. The LH had the highest Shannon index value among all sampling sites, indicating the highest community diversity. QX had the lowest value, indicating that it had the lowest community diversity. Despite the fact that the Simpson index showed the opposite trend to the Shannon index, it had roughly the same meaning. Furthermore, the Pielou index of each sampling site was not significantly different, suggesting that the species distribution was consistent across the sampling sites.

Beta diversity analysis

The X-axis PC1 accounted for 21.56% of the variance in the sample composition, the Y-axis PC2 explained 18.537%, and the Z-axis PC3 accounted for 11.264%. These values were calculated by the PCoA of each sampling point (Fig. S1). Except for four sampling sites, LH, MD, ML, and XK, which were far apart from each other, all sampling points were closer to each other, indicating that the fish compositions of the LH, MD, ML, and XK locations varied more than at the other sampling points.

We divided the 12 sampling sites into 12 groups (three parallel samples from each sampling site were grouped) and used the Bray–Curtis distance matrix to conduct an NMDS analysis of the species composition at the species level. We obtained a stress score of 0.10969, indicating that the results have some explanatory significance. The findings revealed that the fish composition was dissimilar between the subgroup samples (nmds1, R2 = 0.21, P = 0.02; nmds2, R2 = 0.58, P = 0.30). In Fig. S2, MD, XK, and GYD are scattered and far apart from other sampling sites, implying that the fish compositions at these three sampling points is more differed from that of other sampling points. This result is similar to that obtained in the PCoA.

Discussion

Species composition

In the upper reaches of the YR from Jiangjin to Fuling, a total of 104 species of freshwater fish were found using eDNA technology, with a coverage rate of 70%. This is compared to the 149 species of fish contained in the most recent reference in historical data. The composition results at the family level based on eDNA revealed proportions of 51.92% Cyprinidae, 10.58% Cobitidae, and 9.62% Bagridae. These results are similar to those obtained with traditional methods in the previous decade, indicating that eDNA technology has some applicability in this watershed and could be a significant complementary tool for conventional investigation methods (Jiang et al., 2016a; Wang et al., 2022).

According to the survey findings, the “Four major Chinese carp” sequence abundance vary considerably in this river section. It may be related to their living habitat, in terms of the relative sequence abundance, the results showed a ranking of C. idella > H. molitrix > H. nobilis > M. piceus. Based on this, C. idella is the most active and lives closest to the surface, followed by H. molitrix and H. nobilis, M. piceus is the least active (Xian et al., 2010; Xu et al., 2017). Our survey was primarily conducted in surface water, leading to more mixed DNA of C. idella and H. molitrix, which have active habits. Hence, they have higher relative sequence abundances than the other two fish species. Furthermore, the amount and rate of eDNA released into the water was shown to vary among the species, and the activity of an organism was demonstrated to influence the amount of eDNA released into the water column (Geerts et al., 2017; Minamoto et al., 2017). This suggests that we should use a mixture of upper, middle, and lower water samples for eDNA sampling.

The current status of fish communities

Twenty-two species included in the present survey results, such as Misgurnus cf. mizolepis (Günther 1888), T. rosa, and Sinobdella cf. sinensis (Bleeker 1870), are not contained in historical data. It is possible that the scarce numbers or specific habits of these species resulted in them not being observed in this river section by conventional methods [e.g., O. elongatus, Hemiculterella sauvagei (Warpachowski 1888)]. In addition, it has been shown that traditional DNA macrobarcoding is unable to distinguish some sibling species (Shen et al., 2019). In the eDNA method, the barcodes are shorter, which would result in a low taxonomic resolution and therefore a failure to distinguish some sibling species [e.g., Siniperca cf. obscura (Nichols 1930) and Siniperca chuatsi (Basilewsky 1855)], which would have to be further validated by traditional methods (Sales et al., 2020, 2021). This is a limitation of the current eDNA method. Meanwhile, the Yangtze River has an extremely rich fish diversity, and there is also a possibility that some species could not be assigned to the correct species because their information was not included in the reference database (Li et al., 2019). However, these are the cases of a very small fraction of species and need to be handled with caution when making relevant descriptions.

Previous studies have shown that fish in the upper YR tend to undergo species miniaturization (Wei et al., 2021), and our study results support this conclusion. The study revealed that the dominant fish species in the Chongqing section of the YR mainstream in 2019 included H. nobilis, Saurogobio dabryi (Bleeker 1871), and H. molitrix (Wang et al., 2021). However, our results showed that small fish species, such as H. tchangi, R. sinensis, and P. parva had the highest relative sequence abundances, which may be related to the overfishing that occurred before the fishing ban (Chen, 2016). However, it is also possible that smaller fish are more active and shed more DNA (Minamoto et al., 2017), or the results may be related to the sampling time and frequency. Therefore, it is important to perform multi-season and multi-frequency sampling to obtain more accurate results from the eDNA method.

Meanwhile, the number of endemic fish species was shown to account for 18% of all fish species contained in the eDNA results and 20% of the endemic fish species in the historical data, which implies a possible declining trend in the number of endemic fish species in the upper YR, while other studies have also shown a decline in the number of endemic fish in the Three Gorges reservoir area in the upper YR (Wei et al., 2021), possibly owing to habitat destruction (Yang et al., 2017). The study also discovered a high proportion of demersal fishes and adhesive egg fish in the upper reaches of the YR from Jiangjin to Fuling, which may be related to the formation of the Three Gorges Reservoir area (the nearest sampling site is about 492 km away from the Three Gorges Dam), leading to slower water flow, an increased water depth, sediment deposition, and increased bait organisms for demersal fish. The formation of the reservoir area is more favorable for the survival of demersal fish and adhesive egg fish (Cao, 2019). In summary, the current state of degradation of fish communities in the upper YR should be seriously considered, and necessary measures must be taken to protect them.

Fish diversity

Table 2 depicts the alpha diversity index of the fish community abundance. The Shannon and Simpson indices synthetically reflect the diversity and evenness of the species (He, 2016). The Shannon index and Simpson index values were the highest for LH, revealing that this sampling site contains the highest fish community diversity, which may be related to the richness of the fish species and the evenness distribution of sequences at this sampling site. Based on the above two indicators, QX was found to have the least number of fish species, which could be attributed to the sampling site proximity to a freight terminal and the high influence of human factors at this site, resulting in a reduced fish abundance. Nevertheless, different indices for assessing the fish diversity have different emphasis, and the community richness index is not a perfect indicator of high community diversity (Ling et al., 2021).

Table 2 Alpha diversity index of each sample
Full size table

The PCoA and NMDS analyses of each sampling site revealed that the fish species compositions at the LH, GYD, MD, ML, and XK sampling sites differed from those at other sampling sites, which could be related to the river topography and anthropogenic activities carried out near the sampling sites. It has become clear that the abundance and density of fish species are linked to the horizontal area and dimensions of river basins. The species composition and diversity of these fish communities are directly influenced by the longitudinal slope drop and bed substrate of rivers, among other factors (Platts, 1979; Walters et al., 2003; Liao 2021). Anthropogenic activities such as cultivation, laundering of clothes, and machine operations can alter the nitrogen and phosphorus elemental contents and sediment the content of the water, affecting the distribution and diversity of fish (Liu et al., 2004; Shi et al., 2018). Other sampling sites showed comparable results close, implying that they had more similar fish species composition, probably because they are continuous rivers with no obstacles in the center, allowing fish to swim back and forth between nearby sampling sites. Meanwhile, eDNA can be transported downstream at least 50 km in large rivers. Thus it is possible that the similarities in the fish species composition are also due to the transmission of eDNA by rivers (Pont et al., 2018).

Exotic species

This survey discovered Acipenser schrenckii (Brandt 1869), Coilia brachygnathus (Kreyenberg et Pappenheim 1908), Neosalanx cf. tangkahkeii (Wu 1931), Protosalanx chinensis (Basilewsky 1855), and 17 other exotic species, accounting for 16.35% of the total number of eDNA fish species. In a recent study that investigated the fish diversity in the Three Gorges Reservoir area using traditional fishing methods, 20 exotic fish species were identified, accounting for 13.42% of all fish in the study (Wei et al., 2021). We detected a higher proportion of exotic species, indicating that the invasion risk in the Three Gorges Reservoir area is continuously increasing. Exotic species, such as A. schrenckii, Micropterus salmoides (Lacepède 1802), and N. tangkahkeii, which are commonly farmed in the Three Gorges Reservoir area, originate primarily from fish escaping from surrounding farms, blind introduction, and human release (Wu et al., 2007; Qiao et al., 2010; Ba & Chen, 2012). The lack of necessary escape prevention measures on the farms led to the invasion of these species into the YR. The impoundment of water in the Three Gorges Reservoir area has resulted in drastic changes in the water environment, destruction of structure and function of the original ecosystem, the severe vacancy of ecological niches, and resource enrichment, which have facilitated the settlement and spread of exotic species (Ba & Chen, 2012).

The invasion of exotic species can result in the simplification of community composition of ecosystem and the harshness of the original ecological environment. Exotic species can also compete with native species for resources, reduce the genetic diversity of native species, and cause the decline or even extinction of indigenous fish species. For example, the introduction of species such as N. taihuensis into the Dianchi lake has resulted in the depletion of endemic fish stocks (Xiong et al., 2006). Since the 1950s, C. idella, H. molitrix, and other fish from the middle and lower reaches of the YR have been introduced into the water bodies of the Yunnan–Guizhou Plateau on a large scale, resulting in the endangerment and extinction of many indigenous and endemic fish on the Plateau (Xie & Chen, 2001). Therefore, we should mitigate against the invasive hazards of exotic species by establishing a scientific selection and breeding assessment system for exotic species, developing a comprehensive set of laws and regulations related to exotic species, monitoring and alerting existing exotic species, and improving the protection of indigenous species (Xu et al., 2004; Qiao et al., 2010; Ba & Chen, 2012).

Conclusion

The eDNA results imply that fish communities in this area have tended to undergo species miniaturization, and the endemic fish may also be gradually declining as well as the exotic fish species are increasing in the upper reaches of the YR from Jiangjin to Fuling. However, the species composition at the family level has remained stable over the last decade. Simultaneously, eDNA technology detected 104 species of freshwater fish from 72 genera and 8 orders, and 24 families were detected in this river section, indicating that this technology has particular relevance in this river section and can be used as an auxiliary tool to the traditional method.