Investigating the impact of the construction of the Duliu River dam in China on the spatiotemporal changes of fish communities

Abstract

Dam construction alters river hydrology, influencing the temporal and spatial heterogeneity of fish communities. This study utilized environmental DNA (eDNA) sequencing to analyze seasonal variations in fish communities in the Duliu River, Guizhou Province, and to assess the impact of hydrological regulation on biodiversity. High-throughput sequencing revealed that sequence reads were significantly higher in the dry season than in the wet season (P < 0.05), while operational taxonomic unit (OTU) richness was greater in the wet season. A total of 69 fish species (6 orders, 18 families) were detected in the dry season, with only 23 species matching historical records. eDNA sequencing identified an additional 46 species, primarily from the order Cypriniformes. In the wet season, 82 species (7 orders, 17 families) were recorded, including 44 species not found in historical data. Cypriniformes dominated in both seasons, comprising 94% of the community in the dry season and 82% in the wet season. Spatial heterogeneity analysis using principal coordinates analysis (PCoA) indicated significant differences among river sections (PC1 = 65.2% in the dry season; 83.65% in the wet season). Beta diversity analysis (NMDS) confirmed significant temporal and spatial variation in fish communities (P < 0.05). This study highlights the ecological impact of dam construction on fish communities and underscores the need for conservation strategies to protect biodiversity in regulated mountain rivers.

Introduction

Biodiversity encompasses the totality of the ecological complex formed by organisms (animals, plants, microorganisms) and microorganisms and their environment, along with the environment and various ecological processes associated with them. Fish play a crucial role in the material cycling and energy flow within aquatic ecosystems. The spatial heterogeneity and diversity of fish community structures serve as vital indicators to evaluate the health status of aquatic ecosystems ^1,2,3,4,5. Spatial heterogeneity refers to the variability and complexity in the spatial distribution of ecological processes and patterns⁶. Research conducted on the Xiangxi River reveals that the spatial heterogeneity of river ecosystems is primarily reflected in the variations in fish community structure and their living environments⁷. Environmental protection strategies have influenced fish diversity and community structure, resulting in reduced differences in fish community structures and increased stability within ecosystems⁸.

Rivers serve as vital conduits linking highland and lowland ecosystems in mountainous regions. Due to their steep gradients, turbulent currents, abundant water energy, and variable natural environments, mountain rivers create optimal conditions for the development of cascade hydropower⁹. Dams alter river flow, diminish seasonal variations in the aquatic environment, and further disrupt the natural connectivity between water ecosystems and their surroundings^10,11,12. Moreover, the construction of dams hinders the exchange of genetic information among fish populations in different river sections, leading to a reduction in ecosystem biodiversity, stability, and resilience¹³.

The construction of dams has resulted in significant hydrological changes in rivers, impacting the life cycles of fish and the composition of fish communities, as well as influencing the biodiversity of river ecosystems¹⁴. Dam construction transforms the river from a "river-type" hydrodynamic state (shallow-water, rapid-current areas) to a relatively static hydrodynamic state (deep-water, slow-flowing areas) that is close to a "lake-type"¹⁵. The dominant species of "lake-type" were mainly widely adaptable species adapted to the slow flow habitat, while the "river-type" was mainly the water-loving species that thrive in flowing water conditions¹⁶. The flow characteristics near the dam are regulated by its operation, making the fish community structure in the upstream section of the reservoir more vulnerable to the impacts of cascade dam construction and the management of water storage in the downstream section than those in the downstream area itself¹⁷.

The Duliu River is the uppermost source of the Liujiang River and is part of the Pearl River water system. It originates in Dushan County, the basin covers an area of 11,326 square kilometers and is situated between 107° 30′ and 109° 25′ east longitude and 25° 30′–26° 30′ north latitude¹⁸. Prior to the construction of the dam, the river valley exhibited deep incision, steep slopes, and a vertical drop of up to 1214 m, numerous rapids and predominantly elevated terrain, with altitudes reaching 800–1000 m. The basin was rich in vegetative cover and boasted a remarkable diversity of natural resources and biodiversity¹⁹. The Duliu River is a typical stepped reservoir river, with 11 planned avionics hubs in Guizhou, and eight hydropower stations have been constructed to date²⁰. The implementation of these plans may lead to significant changes in the food chain, biodiversity, community structure, and resources of the Duliu River aquatic ecosystem. This has been reflected in studies of the cascade construction of the Wujiang River, particularly concerning fish communities²¹. The impacts of dam construction on fish habitats include habitat fragmentation, migration barriers, and biodiversity loss^22,23,24. Furthermore, dam construction results in alterations to river storage and flow, which in turn disrupts fish habitats and impedes interspecific communication²⁵. The increase in water storage and the obstruction caused by dams have led to changes in the fish composition of the Duliu River. In the vicinity of the avionics hub, there has been a noticeable decrease in fish species richness along the horizontal continuum of the river.

In aquatic biological monitoring, eDNA technology collects DNA from fish skin cells, mucus, feces, and water samples from various sources to identify and quantify fish populations²⁶. It offers the advantages of being non-invasive, having a rapid speed, and providing extensive coverage, which allows for an effective assessment of fish diversity²⁷. However, this method also has certain limitations, including sample and DNA contamination, as well as inaccurate species information in reference databases. These issues may lead to false positive results and hinder the accurate identification of local biodiversity²⁸.To investigate the influence of dam construction on fish populations in mountain rivers. This study, which in November 2021 and June 2022, is monitoring the fish composition in the Duliu River during both the dry and wet seasons. The objective is to gain insights into the temporal and spatial variations in fish community structure influenced by the construction of the dam. The data collected will be utilized to evaluate the impact of the hydropower station construction on fish diversity and composition in the future.

Materials and methods

Sampling scheme design

The experimental samples were collected in November 2021 and June 2022 from the Duliu River basin, located in the Qiannan and Qiandongnan districts of Guizhou Province. The sampling area extended from Panjiawan in the Dushan section of the Duliu River, following the main stream to Nanjiangtun, which marks the border between Guizhou and Guangxi. A total of ten sampling sites were established along this section of the river (Fig. 1 and Table S1). Most of the sampling sites were strategically located at the confluence of dry and tributary streams, both upstream and downstream of reservoirs and dams, as well as near coastal counties, to facilitate the collection of a larger volume of eDNA. All sampling equipment was disinfected with a 10% bleach solution prior to sampling at various sites²⁹. A total of 15L of water samples were collected from each sampling point and combined, resulting in a final concentrated volume of 1 L. The samples were categorized into three layers: the surface layer (0 m), the middle layer (2.5 m), and the bottom layer (5 m). The collected water samples were refrigerated within 24 h and sent to the laboratory, where they were filtered using a vacuum pump and concentrated on a 0.45 μm mixed cellulose filter membrane. All equipment is disinfected and rinsed with distilled water before use to prevent cross-contamination. Finally, the membrane is stored at -80 °C in preparation for the subsequent DNA extraction. After each sample is extracted and filtered, the nozzle filter, tweezers, and water bottle must be soaked in sterile water for 5 min prior to use³⁰.

Fig. 1

Distribution map of sampling points in Duliujiang.

Historical data and reference database

Since 1989, the only publications documenting the fish species of the Duliu River are of the Pearl River Journal, which described 8 species (subspecies) of fishes of 2 orders, 4 families, 6 genera collected from the Duliu River, and The ichthyology of Guizhou, in which detailed 51 subspecies of fish from the Duliu River in 4 orders, 12 families, and 37 genera³¹, along with historical documents³². The mitochondrial genome sequences were obtained from the MitoFish (http://mitofish.aori.u-tokyo.ac.jp) and NCBI (https://www.ncbi.nlm.nih.gov) databases, serving as reference databases for eDNA in this study.

DNA extraction and PCR amplification

Total genomic DNA was extracted from the filter membrane using the PowerWater DNA Isolation Kit. To avoid contamination, each sample is extracted independently. The mitochondrial gene 12S rRNA is the most widely used molecular marker in fish diversity monitoring as an amplification target. The fish-specific primer “Tele02” (Tele02-F-5′-AAAC) was used as the amplification target. fish-specific primer “Tele02” (Tele02-F: 5′-AAACTCGTGCCAGCCACC-3′, Tele02-R: 5′-GGGTATCTAATCCCAGTTTG-3′), which was developed by Taberlet, was used in amplification with addition of the sample- specific barcode sequence. The total volume of the PCR reaction is 20 μL and prepared as follows: 4 μL of 5 × FastPfu buffer, 2 μL of dNTPs, 0.4 μL of FastPfu polymerase, 0.8 μL of FastPfu polymerase, 0.8 μL of FastPfu polymerase, and 0.8 μL of FastPfu polymerase. polymerase, 0.8 μL per 10 μM primer, 2–5 μL of genomic DNA (10 ng/μL). Finally, ddH2O was added to a total system volume of 20 μL. The PCR conditions included an initial denaturation step at 95 °C for 5 min, followed by 27 cycles The PCR conditions included an initial denaturation step at 95 °C for 5 min, followed by 27 cycles at 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 45 s, with a final duration of 10 min at 72 °C. PCR of the product was sent to Biozeron for sequencing. (The animal study protocol was conducted following the guidelines for the care and use of experimental animals of China (GB/T35892 2018), and approved by the ethics committee of Guizhou University (No.: EAE-GZU-2022-T136; Approval Date: 30 May 2022).)

Data analysis

DNA analysis and species identification

A window of 10 base pairs (bp) is established. If the average mass value within this window is lower than 20, the back-end bases are removed, and sequences shorter than 50 bp after quality control (using Trimmomatic v0.39) are filtered out. Based on the overlap from paired-end sequencing, the paired sequences are merged into a single sequence. The minimum overlap length is set at 10 bp, and the maximum allowed mismatch ratio in the overlap region is 0.2. Non-conforming sequences are eliminated using FLASH v1.2.7, and the sequence orientation is adjusted according to the labels and primer sequences to obtain high-quality sequences for each sample. OUT clustering is performed on the high-quality sequences at a 98% similarity threshold³³. Chimeric sequences are removed during the clustering process, and OTU tables are generated using Perl v5.18.2, Usearch v10, and QIIME v1.9.1. The generated OTUs will be annotated in a self-built freshwater fish database based on the NCBI nt database (to be established in October 2021) to obtain information on fish species, with classification based on the Fishbase database.

Statistical analysis

After removing the highly identical sequence data for non-fish organisms (such as bacteria, birds, amphibians, mammals, etc.), the remaining filtered data were compared with the fish sequences, showing a ≥ 97%, an E-value ≤ 10, and the OTUs corresponding to the same species were combined. data were compared with the fish sequences, showing a ≥ 97%, an E-value ≤ 10, and the OTUs corresponding to the same species were combined. If any OTUs do not show sufficient similarity at the species level for identification, statistical analysis would be performed. show sufficient similarity at the species level for identification, statistical analysis is performed for genera and families with higher taxonomic levels. If any OTUs do not show sufficient similarity at the species level for identification, statistical analysis is performed for genera and families with higher taxonomic levels. Use Excel to calculate the proportion of sequence numbers for each species in each sample. Improved fish classification information by referring to the fish-based database “https://www.fishbase.in/home.htm”. Finally, use R software (version 3.1.3 to plot histograms of fish composition at each sample point; packages: gg plot 2, pile, scatter plot 3d, ellipse, map tool, vegetarian and ape. map tool, vegetarian and ape).

Alpha diversity analysis

For this study, the Chao1 index, Observed_species index, Shannon index, Simpson index, and Pielou_J index were selected to reflect community richness, species diversity, and species evenness³⁴.

Chao1:

$${\text{Chao1}} = {\text{Sobs}} + \frac{{{\text{n}}1\left( {{\text{n}}1 - 1} \right)}}{{2\left( {{\text{n}}2 + 1} \right)}}$$

Chao1: the estimated number of OTUs; Sobs: the observed number of OTUs; n₁: the number of OTUs with only one sequence; n₂: the number of OTUs with only two sequences.

Observed_species:

$${\text{S}} = {\text{n}}$$

S: the Observed_species index; n: the total number of species types with an individual number (abundance) greater than 0.

Shannon:

$${\text{H}} = - \mathop \sum \limits_{{{\text{i}} = 1}}^{{{\text{Sobs}}}} \frac{{{\text{ni}}}}{{\text{N}}}{\text{ln}}\frac{{{\text{ni}}}}{{\text{N}}}$$

H: the Shannon index; Sobs: the total number of species; n_i: the number of individuals of the i-th species; N: the total number of individuals of all species.

Simpson Index:

$${\text{D}} = \frac{{\mathop \sum \nolimits_{{{\text{i}} = 1}}^{{{\text{Sobs}}}} {\text{ni}}\left( {{\text{ni}} - 1} \right)}}{{{\text{N}}\left( {{\text{N}} - 1} \right)}}$$

D: the Simpson index; Sobs: the total number of species; n_i: the number of individuals of the i-th species; N: the total number of individuals of all species.

Pielou_J:

$${\text{J}} = \frac{{\text{H}}}{{{\text{lnS}}}}$$

J: the Pielou_J index; H: the Shannon diversity index; S: the number of species.

The alpha diversity index analysis utilized in this experiment was the mothur software (version v.1.30.1 http://www.mothur.org/wiki/Schloss_SOP# Alpha_diversity) and the OTU similarity level for exponential evaluation was 97% (0.97).

Beta diversity analysis

In this study, to identify potential principal components that affect differences in community composition in the sample by reducing dimensions. PCoA analysis of community compositions at different sample points was based on the Bray–Curtis distance algorithm to explore differences or similarities in community composition among different groups of samples (PCoA uses the ellipse package and the maptools package)³⁵.

Results

sequence and OTUs information

Enrichment of eDNA was performed by filtration, and then the samples were sequenced to obtain the corresponding number of reads and OTUs in dry and wet seasons. The eDNA of water samples collected from 10 sampling sites in the Duliu River. High-throughput sequencing for the dry season, and a total of 1,696,708 reads and 4237 OTUs were obtained, while a total of 848,636 reads and 4,601 OTUs were obtained during the wet season (Table S2). Based on the habitat and hydrological characteristics, they were grouped into three subgroups: SR, SF and DF (Table S1), and ANOVA analysis was performed on the number of reads in each season respectively, and the results of both the wet and dry seasons showed no significant differences among the three subgroups of SR, SF and DF (P > 0.05). Seasonal analysis of OTUs and reads measured in the Duliu River showed that the number of reads was significantly different (P < 0.05) between the wet and dry seasons, and the number of OTU sequences of the species they interpreted was not significantly different (P > 0.05). The number of reads in the dry season was twice as high as that in the wet season, but the number of interpreted OTUs was higher in the wet season.

species of fish comparing eDNA with historic records

After sequencing the eDNA of the Duliu River during the dry season, the resulting sequences were analyzed and validated using the Nucleotide Sequence Database (Nt) for species annotation. A total of 69 species of fishes were obtained during the dry season in 6 orders and 18 families, as shown in Table 1. Of these, only 23 species were recorded in the historical data compiled by Yinggui Dai¹⁹. Compared with their records, eDNA sequencing detected 46 species of exotic fish, including 31 species of Cypriniformes, 5 species of Siluriformes, 5 species of Perciformes, 1 species of Acipenseriformes, 2 species of Cyprinodontiformes, 2 species of Beloniformes, and 65 species of fishes not detected by eDNA sequencing, primarily concentrated in the Cobitidae family, which is adapted to shallow and fast-flowing waters. Notable small Cypriniformes, such as: Leptobotia guilinensis, Botia (Sinibotia) pulchra, Micronemacheilus pulcher, Parabotia bimaculata, Discogobio brachyphysallidos, Microphysogobio kiatingensis, and Saurogobio dabryi. During the wet season, a total of 82 species of fishes from 7 orders and 17 families were identified, with only 34 species appearing in the historical records compiled by Yinggui Dai. Compared to the original records, the eDNA sequencing results revealed an increase of 44 species of fishes, including 30 species of Cypriniformes, 3 species of Siluriformes, 5 species of Perciformes, 1 species of Cyprinodontiformes, 2 species of Beloniformes, 1 species of Clupeiformes, and 2 species of Salmoniformes. The eDNA sequencing results for the wet season indicated that the loach family included 9 species, four of which were indigenous. Eleven indigenous fish species were added to the monitoring results for the wet season, including Botia (Sinibotia) pulchra, Botia (Hymenophysa) robusta, Parabotia fasciata, Schistura fasciolata, Rhodeus ocellatus, Zacco platypus, Siniperca undulata, Sarcocheilichthys nigripinnis, Spinibarbus hollandi,Acrossocheilus iridescens longipinnis, and Acrossocheilus parallens. Cypriniformes accounted for 82%, Perciformes for 11%, and Siluriformes for 5% during the wet season. In contrast, during the dry season, Cypriniformes accounted for 94%, Perciformes for 3%, and Siluriformes for 3% (Fig. 2A,B). A stacked histogram of species presence and absence and their abundance at each sampling point (Fig. 3) showed that species abundance was much higher in the dry period than in the rich period.

Table 1 Dry Season and Wet Season and documented species and their number statistics table of Duliujiang river.

Fig. 2

Percentage of OTUs in Wet Season (A) and Dry Season (B).

Fig. 3

Histogram of species clustering at each sampling point in Dry Season (W) and Wet Season (S) of Duliujiang.

Differences in fish community among river sections

The data collected during the dry season were categorized into three subgroups SR, SF, and DF based on habitat and hydrographic characteristics. PCoA analysis was conducted using the R software, focusing on the distribution of sampling points as well as the presence-absence and abundance of species at each sampling point. PCoA analysis was performed using the Bray–Curtis algorithm was employed to calculate the distance metrics, which were subsequently analyzed and visualized using the ggplot2 package in R (Fig. 4a). The results indicated that PC1 accounted for 65.2% of the variance, while PC2 accounted for 20.61%. The analysis revealed a notable distance gap between the DF and SR groups along the x-axis, with the SR group clustering on the right side and the DF group distributed on the left. In contrast, the data collected during the wet season, which grouped the cj and xjz river segments together while categorizing the other river segments separately, demonstrated that PC1 accounted for 83.65% and PC2 for 8.63% (Fig. 4b). This analysis indicated a significant distance difference between the cj river section and the other river segments along the x-axis, with less variation observed on the y-axis and minimal differences between the cj river section and the other segments.

Fig. 4

(a) PCoA analysis of various sites based on species presence and abundance in Dry Season. (b) PCoA analysis of various sites based on species presence and abundance in Wet Season. (a) The red circle is the shallow water section group; The green square is the group of moderate deep-water sections; The orange triangle is the deep-water. (b) The red circle (A) is the cj from the river; The green triangle (B) shows the other river sections.

The fish community structure analysis of DF, SF and SR groups was carried out, and the data of dry season were displayed (Fig. 5A), from DF to SF to SR, the number of fish in Spinibarbus and Chanodichthys is gradually increasing. The number of fish in Cypriniformes_norank and Hemiculter gradually decreased, and the proportion of other fish was small, and the difference was not significant. Data on the wet season (Fig. 5B). In the three different hydrological basins of DF, SF and SR, there was little difference in the dominant species. Carassius, Procypris, Ctenotrypauchen and Hemiculter accounts for a large proportion, and Xenocyprididae_norank accounts for the largest proportion in the flowing environment.

Fig. 5

Plot of species differences between groups Dry Season (A) and Wet Season (B).

The results of the Chao1 index and the Observed_species index indicated significant differences between the SR group and both the DF and SF groups (P < 0.05). However, no significant difference was observed between the DF and SF groups (P > 0.05). Additionally, the results of the Pielou_J index, Shannon index, and Simpson index revealed no significant differences in the fish community among the groups (Fig. 6). NMDS ranking analysis (stress = 0.09 < 0.2) indicated significant differences among the components (P < 0.05) by Adonis (P = 0.006) and MRPP (P = 0.005) (Fig. 7).

Fig. 6

Plot of between-group differences in the alpha diversity index (DF, SF, SR).

Fig. 7

Fish species detection NMDS differences between groups (DF, SF, SR).

Differences in fish community between wet and dry seasons

According to the Chao1 index, Observed_species index, Shannon index, Simpson index, and Pielou_J index were positively correlated with species diversity and the number of OTUs in the watershed. Furthermore, the variations in the Chao1 and Observed_species index corresponded with the changes observed during the wet and dry seasons. The Shannon index also exhibited a positive correlation with the biodiversity of the watershed. In contrast, the values of the Simpson index were negatively correlated with the Shannon index was positively correlated with the biodiversity of the watershed, and the values of Simpson index were negatively correlated with the biodiversity of the watershed, and the changes in the values of the points coincided with those of the wet and dry seasons. The results showed that Chao1 and Observed_species index were significantly different in the seasonal analysis (P < 0.05), while the Pielou_J, Shannon, and Simpson indices did not demonstrate significant differences (P > 0.05) (Fig. 8; Tables S1 and S2).

Fig. 8

Comparison histogram of Alpha diversity index in Dry season(W) and Wet season(S). (a) Chao1 index; (b) Observed_species index; (c) Pielou_J index; (d) Shannon index; (e) Simpson index.

Based on the “presence or absence” of species and their abundance at each sampling point during the wet and dry seasons, a correlation heat map was generated using Origin (Fig. 9). The results indicated that the pjw sampling point during the dry season was significantly different from the xjz sampling point during the wet season (P < 0.05). Additionally, the xjz sampling point during the dry season was significantly different from the xjz sampling point during the wet season (P < 0.05). Furthermore, a significant difference (P < 0.05) was observed between the xjz sampling site in the dry season and the xjz sampling site in the wet season. Apart from these findings, all sampling points in the dry season were significantly different from those in the wet season (P < 0.05).

Fig. 9

Correlation heat map of sampling points in Dry Season (W) and Wet Season (S).

PCoA analysis for the spatio-temporal distribution characteristics of the Duliu River (Fig. 10), the sampling points were clustered on the left side in the dry season and the clustering range was small, and the sampling points were clustered on the right side in the wet season, but their clustering range was large.

Fig. 10

PCoA analysis based on species presence and abundance in Dry Season (W) and Wet Season (S).

Deliberations

eDNA technology enriches historical monitoring data

eDNA technology can facilitate effective large-scale surveillance, while providing a more accurate reflection of the composition of fish species at sampling sites³⁶. In this study, eDNA technology was used to monitor the fish composition of the Duliu River, and 98 fish species were detected, including 69 species in the dry season and 82 species in the wet season. The results showed that there were 53 species in the two, accounting for 54%. Because of the construction of the dam, the hydrological characteristics have changed, and the results of this study have reduced the number of fish that prefer to live in a flowing water environment and habitually inhabit the bottom compared to historical data, most of which are Cobitidae and Cyprinidae. In addition, a number of exotic fish species have also appeared in the monitoring of eDNA technology, and the current survey results have increased by a total of five orders of magnitude compared with historical data. namely Acipenseriformes, Cyprinodontiformes, Beloniformes, Clupeiformes and Salmoniformes. Furthermore, certain fish species exhibit a preference for still water and slow currents, while others are typically found in the upper layers of the water. This observation aligns with the ecological changes in aquatic ecosystems resulting from dam construction.

Seasonal variation in the structure of fish communities

There are temporal differences in the structure of fish communities in the Duliu River Basin, which are mainly divided into differences between the dry and wet seasons. The water flow was low during the dry season, and some sections of the river became still water after the dam was constructed. During the wet season, the water flow is high and the tributaries are interconnected³⁷. We compared the differences in fish community structure and α diversity between the dry and wet seasons in the Duliu River Basin in Guizhou. From the perspective of differences in species composition, comparing the abundance proportions of each order in the dry and wet seasons, the proportion of Cypriniformes accounted for 94%, Perciformes accounted for 3%, and Siluriformes accounted for 3% in the dry season. In the wet season, Cypriniformes, Perciformes, and Siluriformes accounted for 82%, 11%, and 5%, respectively. ANOVA seasonal analysis of the OTUs and reads measured in the Duliu River revealed a significant difference in the number of reads between the wet and dry seasons (P < 0.05). However, no significant difference was observed in the number of OTU sequences between the species in question (P < 0.05). The results demonstrated that there were no notable differences in the diversity of the fish species. However, there were discernible variations in the abundance of each species, which are indicative of biological diversity.

The fish composition of the Duliu River has the characteristics of typical inland water bodies, particularly reservoir fish composition. (1) Cypriniformes was the main species, and the number of cypriniformes accounted for 72.45% of the 71 fish species collected in the reservoir, which was basically the same as the proportion of cypriniformes surveyed by Dai Yinggui, and 9 species of 5 orders, 5 families, and 9 species have been added¹⁹. (2) During the dry season, deep-water fish were the dominant species, and limited water flow resulted in an increased abundance of fish that preferred still water. Among these, Gambusia affinis is particularly prevalent, exhibiting a tendency to swim in clusters on the surface of the water and displaying characteristics of a warm and temperate small fish³⁸. It inhabits a variety of aquatic habitats, including reservoirs, lakes, dams and depressions³⁹. Cirrhinus mrigala, which prefers still water, has a fast growth rate, usually lives in the middle and lower layers of water bodies, can live in reservoirs, lakes, rivers, and other water bodies, and has a strong ability to tolerate hypoxia and diseases⁴⁰. It is not tolerant to low temperatures because the construction of the dam weakens the seasonal variation in the water environment, resulting in a small difference in water temperature in different seasons, which is conducive to the growth of Cirrhinus mrigala and is detected more downstream of the dam^10,41. Ictalurus punctatus is active at the deep bottom⁴², is adaptable and easy to reproduce, and is an economically important fish that can be cultivated intensively in rivers, lakes, and pools⁴³. Most of these fish species are found in the still water bottom, and the adaptability of fish to changes in the aquatic environment is closely related to the composition of fish ecological types⁴⁴. (3) The dominant flow-type fish during the wet season, owing to the large water flow during this season, are the interconnected fish of each tributary, Zacco platypus, a rapid-flow fish in mountain streams⁴⁵. It is predominantly found in sandy and gravel shoals with strong currents on river tributaries⁴⁶. In recent years, the germplasm resources of wild Zacco platypus have decreased owing to overfishing and the construction of water projects⁴⁷. Hemibarbus labeo often inhabits the middle and lower layers of the upper reaches of rivers where there is flow⁴⁸. It prefers low temperatures and clear water basins and has less activity in lakes and reservoirs⁴⁹and more activity in the tributaries of the Duliu River during the wet season. Botia pulchra is a small demersal fish that prefers to live in flowing water with sandy substrate⁵⁰. During the wet season, most of the water-loving fish were detected, and the creek was a natural channel without the impact of dam construction. Its upstream habitats are diverse, and the diversity of benthic fauna is high; therefore, the biological density is also high⁵¹.

In addition, the α diversity index reflects species diversity in the region at different times. The results showed that the α diversity index of fish community diversity in the wet season was higher than that in the dry season, and the species diversity in the wet season was also higher than that in the dry season, which is similar to the results of a study in the Shishou section of the Yangtze River⁵².

spatial heterogeneity of the mountain river

The construction of dams, which affects approximately 63% of the world’s rivers, is one of the major factors contributing to the decline in global freshwater fish diversity⁵³. The construction of the Duliu Water Conservancy Project led to a large spatial difference in the structure of fish communities. In this study, we focused on comparing different hydrological cross-sections in the Duliu River Basin in Guizhou, and through the histogram of species composition and PCoA analysis, the fish community structure in the basin was divided into 3 groups according to hydrological characteristics.

(1) In the deep-water and slow-flowing areas (DF), the construction of the dam resulted in synchronous changes to the fish composition structure of the Duliu River due to the hydrological alterations caused by the cascade project. These alterations also led to an overall uplift of the river water level and flow relationship during the dry season⁵⁴. This section of the river detected more Rhodeus ocellatus, a small freshwater native ornamental fish that is distributed in all major water systems in China⁵⁵. Rhodeus ocellatus is adapted to low-altitude slow-flowing or stationary waters, and is more often found in still waters with low transparency and slightly higher eutrophication, and often moves in groups⁴³. (2) In the shallow water and slow-flowing areas (SF), which is the transitional section between the reservoir area and the tributaries, the natural hydrological situation of the downstream river channel subsequently affects the habitat environment of aquatic organisms in the river⁵⁶. The dominance of fish species changes owing to increased water velocity and changes in habitat conditions⁵⁷. it was detected that Discogobio yunnanensis, which preferred to live in the upper reaches of rivers and mountain streams, and was mostly distributed in the Jinsha River, Nanpan River and Yuanjiang River systems⁵⁸. The results showed that it also occurred in the Duliu River Basin. It mostly inhabits the flowing water at the bottom of the gravel and is a small economic fish with small individuals and high fat content⁵⁹. However, owing to the construction of cascade power stations, overfishing, water environment pollution, and other reasons, wild Yunnan panfish resources have become increasingly scarce⁶⁰. (3) In the shallow water and rapids area (SR), the river section in question is relatively shallow, and there are more water-loving fish. As the river flow increases, the distribution of suitable habitats for certain fish species expands from the middle of the river to both sides⁶¹. A special Onychostoma lini was detected, which is similar to flowing waters and is mainly distributed in the Qingshui River, which belongs to the Yuanjiang River system, and the Duliu River, which belongs to the Pearl River system³¹. Over time, its population has been declining, germplasm resources are on the verge of disappearing⁶². In recent years, Onychostoma lini has basically disappeared from the Qingshui River in Guizhou, and only appears in the rj section of the sampling point in the Duliu River, which has become an almost isolated population because of the construction of dams at the downstream hydropower station, and the number is extremely rare³².

In the dry season, the species diversity in SR waters was high, while the species diversity in DF waters was the lowest in the Observed_species, Chao1, and Pielou_J indices. The SF waters were similar to the SR waters, the water flow of these 8 sampling points was faster, and the dominant species was Barbodes sinensis, which led to differences in fish community structure due to differences in environmental factors such as pH and dissolved oxygen between the sampling points^63,64. The species composition of DF waters was quite different from that of other sampling sites, and the fish community structure of pjw was located near the reservoir site, which was more suitable for fish adapted to slow-flowing still water and relatively high-water temperatures⁶⁵. The blc sampling site has a large amount of water, is located near the shipping hub, and has slow water flow; therefore, its community composition is similar to that of the pjw sampling site. In the wet season, the 3 groups of cross-sectional layers divided by hydrological characteristics in the Duliu River basin had less influence on fish community structure than in the dry season, and the Siniperca scherzeri OUT at the two sampling points xjz and cj was 10 times higher than that of the other sampling points, suggesting that there may be Siniperca scherzeri in these two waters, or that there are other human factors affecting the abundance of Siniperca scherzeri DNA in the basin⁶⁶. Due to the real-time nature of eDNA technology and the limitations of sampling, it may be that the DNA distribution at these two sites is uneven, and Siniperca scherzeri is more active, resulting in an increase in the number of OTUs collected⁶⁷.

The impact of dam construction on mountain rivers

Mountain rivers are often regarded as relatively pristine ecosystems characterized by steep gradients, high hydraulic roughness of banks, high turbulence, high spatial and temporal variability of river flows, and high connectivity of landscapes, hydrology, and sediments⁶⁸. The construction of dams is an effective method for intercepting sediment and retaining soil and water in mountain river catchments because of the natural characteristics of these environments and the hydro-geomorphological conditions that occur in their channels. Furthermore, the construction of dams affects the structure of fish communities in a number of ways²⁵, and the effects of dams on fish (including aquatic mammals) include blocking migration routes, habitat fragmentation, changing from lotic to lentic water in the impounded area, release of hypolimnetic cold water from the reservoir, and changes in water flow in downstream reaches⁶⁹. The construction of dams in the Duliu River Basin has led to changes in the living space of fish, spatial heterogeneity in river ecosystems, and changes in fish community structure⁷. Near the dam, owing to increased water storage and blocked dikes, the flow slows down and fish abundance increases, which decreases as the reservoir ages⁷⁰. Some fish species adapted to flowing water and gravelly sand environments decreased or disappeared in this area, while the abundance of fish adapted to deep and slow water increased, resulting in changes in the composition of reservoir fish communities⁷¹.

Reservoirs are used in human supply, irrigation, and power generation. Although dams are important for economic development, they contribute to the severity and irreversibility of the natural hydrological regime of rivers and alter the quality of habitats and dynamics of the biota as a whole⁷². Changes in natural flow regimes after reservoir formation have far-reaching implications, including changes in the original hydrological dynamics, historical patterns of biological production, spatial and temporal distribution of biodiversity, and changes in the functions and services provided by aquatic ecosystems⁷³. In the reservoir area, the main impact is the change from the terrestrial environment to the perspective environment, which greatly affects aquatic animals, including fish. These effects are associated with temporal and spatial variation⁷⁴. However, because of changes in seasonal water flows, with the most affected species being rheophilic species and those that make long migrations and require different types of habitats to complete their life cycles^75,76. In addition to paying attention to dam construction, more attention should be paid to the impact of cascading dams. The impact of cascading dams on aquatic ecosystems is more complex and severe than that of a single dam. However, only a few studies have focused on the effects of cascading dams⁶⁹.

The construction of the dam resulted in alterations to the structural diversity of fish communities. This change was also evident in the introduction of non-native species, which exhibited significant differences in behavior and were phylogenetically distinct from native species. These findings align with the results of this study⁷⁷. The study of the largest step dam in China also shows that the number of native fish in each section of the dam area has seriously decreased, while the number of invasive species has increased, and the ecological function of the river has changed¹¹. Environmental protection strategies changed fish diversity and community structure, which led to the reduction of spatial heterogeneity of fish community structure and the stabilization of ecosystems⁸.

Due to the construction of the Duliu River Water Conservancy Project, the hydrological characteristics of the river, the circulation of the river, and the stratification of water temperature have changed, resulting in changes in the fish composition of the river. Therefore, based on the current situation of the Duliu River, we put forward the following protection suggestions: establish a nature reserve in situ; increase species diversity by identifying and establishing new spawning grounds and habitats along river banks or tributaries; Enforce fishing bans in accordance with fish management regulations and establish a database of species diversity.

Conclusions

• 1. In terms of seasonality, fish diversity was greater during the wet season, and the Chao1 and Observed_species indices indicated significant seasonal differences. Although the number of reads during the dry season was twice that of the wet season, the number of OTUs identified during the wet season was higher.

• 2. Spatially, PCoA and NMDS revealed significant spatial differences in fish community structure in the Duliu River. This indicates that the construction of the dam has led to alterations in the spatial distribution of fish communities. From the perspective of species composition, the diversity indices DF, SF, and SR showed significant variations between the wet and dry seasons. This suggests that the ecological changes resulting from dam construction have impacted the composition of fish species.

• 3. eDNA analysis indicates a rise in exotic fish populations, particularly within the orders Cypriniformes and Perciformes. Additionally, the number of newly detected native fish species has also increased.