Seasonal Dynamics and Heavy Rain Effects on the Diversity of Microeukaryome in the Nakdonggang River

Abstract

Rapid and cost-effective eDNA-based approaches are valuable for understanding biological communities and monitoring the biodiversity of inaccessible areas, such as large rivers and lakes. To determine temporal dynamics and precipitation effects on micro-eukaryotic assemblages, we monitored upstream (SJ) and downstream (MG) sites of the Nakdonggang River, monthly or bimonthly, from March 2019 to April 2021 using an eDNA metabarcoding technique. We observed 775 unique operational taxonomic units during the entire sampling period using the V4 region of the 18S rRNA gene as a marker. In the cluster analysis and non-metric multidimensional scaling analysis based on the relative abundance ratio of the micro-eukaryotic assemblages at the SJ and MG sites, respectively, both regions were classified by four seasons, with dominant taxa for each season being different. Whereas Chlorophyta was relatively abundant in the MG site during summer, diatoms were most abundant at the SJ site during winter. Further, metazoan taxa were relatively abundant at the MG site. Total detection of metazoa, protozoa, and microalgae taxa significantly increased with heavy rainfall. In addition, the indicator taxa representative for heavy rainfall and tolerance ranges of the amount of precipitation were predicted at both sites. As the frequency of drought and heavy rain is expected to increase owing to climate change, efforts to standardize the relation between precipitation and eDNA analysis are required.

1. Introduction

The structure and distribution of biological communities in freshwater ecosystems are highly influenced by combined or interacting factors related to natural (e.g., heavy rain and drought) and anthropogenic (e.g., land use change and eutrophication) disturbances. Changes in hydrological regimes caused by heavy rain generally decrease species richness in freshwater habitats, such as streams, rivers, and lakes, because the species that are especially vulnerable to flow rate changes could flow downstream from their own habitats [1]. However, heavy rain can occasionally increase species richness by introducing new species from upstream or riparian areas via drift or inflow [2]. The influx of nitrogenous or phosphorus compounds from nonpoint source inputs is also a major factor driving changes in freshwater biodiversity [3,4,5]. The increase in the influx of nitrogenous or phosphorus compounds from the external environment due to land use change (e.g., agriculture, urbanization, etc.) near freshwater ecosystems [6] or precipitation sludge [7] could affect freshwater biodiversity by causing harmful algal blooms and hypoxic conditions [8,9,10].

The ecological community of micro-eukaryotic organisms, comprised of phytoplankton, microzooplankton, fungi, and even tiny metazoan taxa, plays a crucial role in maintaining the structural and functional diversity of freshwater ecosystems as the central linkages for energy flow and material circulation [11]. Some eukaryotic freshwater species, such as benthic macroinvertebrates, microzooplankton, and phytoplankton, have also been used as bioindicators to evaluate water quality and ecosystem health. Moreover, the importance of these indicator species has gradually been emphasized in various studies to investigate and predict changes in biodiversity due to natural or anthropogenic disturbances [12,13]. The composition of micro-eukaryotic assemblages has been reported to fluctuate according to variations in environmental factors, which are highly correlated with seasonal dynamics [14]. Phytoplanktons are the most fundamental producers in the aquatic food chain and closely related to heterotrophic organisms (e.g., zooplankton and invertebrates) through organic and inorganic material transfer [14,15]. Moreover, phytoplankton and heterotroph communities are usually repositioned or changed during seasonal variations, because conditions of their growth are strongly associated with environmental factors (i.e., the inflow of nitrogenous and phosphorus compounds, changes in discharge, etc.) of freshwater systems.

Field survey procedures for multiple taxa require taxon-specific sample collection methods (e.g., Surber net for macroinvertebrates, plankton net for microzooplankton and phytoplankton). However, these time-consuming processes can be surmountable by using environmental DNA (eDNA) metabarcoding techniques, in which genetic materials derived from multiple taxa can be extracted from an environmental sample (e.g., water or sediment) at once [16]. Thus, the eDNA-based metabarcoding approach to determine micro-eukaryotic assemblages is attracting attention as a convenient method for evaluating the current status of freshwater ecosystems [17]. This technique has the advantage of relatively rapid and easy sampling and minimizing habitat destruction compared to the traditional field survey method employing morphology classification [18,19,20]. This technology is expected to expand in the future owing to its high applicability to several pending ecological issues (e.g., biodiversity assessment and detection of endangered or invasive species), especially for micro-eukaryotic organisms that are not easily detected based on morphology-based methods and are related to issues such as food web and material circulations [21]. In addition, it is expected that high-frequency sampling (e.g., monthly, weekly, daily, real-time) of micro-eukaryotic assemblages in freshwater ecosystems will become more feasible with the development of this rapid and cost-effective investigation technique [22].

To understand the seasonal dynamics and precipitation effects on micro-eukaryotic assemblages in freshwater ecosystems, we monitored the upstream and downstream sites of the Nakdonggang River monthly (or bimonthly) from March 2019 to April 2021 using an eDNA metabarcoding technique. We performed three major studies using eDNA assessment. First, we examined the relationship between the micro-eukaryotic assemblages and the amount of precipitation. Second, we attempted to determine whether seasonal differences occur in the micro-eukaryotic assemblages. Lastly, we observed the effect of heavy rainfall on eDNA signals for each taxon. We expected that our results can be applied to establish management policies for streams. As eDNA metabarcoding has many advantages, it is necessary to improve the analytical procedure in the future by accumulating data obtained under various environmental conditions. Finally, we tried to select indicator species for each season and analyze the relationship between the appearance of indicator species and precipitation level. Therefore, we examined whether these indicator species could change according to the variations in precipitation patterns in response to climate change.

2. Materials and Methods

2.1. Study Area

The study sites were located in the Nakdonggang River, which is the longest river (510.4 km) in South Korea. The watershed area of the Nakdonggang River is approximately 23,609.3 km², which accounts for one-fourth of the entire area of South Korea (The Ministry of Environment in Korea; https://www.me.go.kr/, accessed on 15 April 2021). Eight multifunctional weirs (Sangju, Nakdan, Gumi, Chilgok, Gangjeong-Goryeong, Dalseong, Hapcheon-Changnyeong, and Changnyeong-Haman) have been constructed and are currently operational.

The average annual precipitation of the study area for 10 years (2009~2018) was 1156.7 (800.3–1563.7 mm) and 1510.1 mm (720.5–1892.5 mm) at the upstream of Sangju Weir (SJ) and downstream of Changnyeong-Haman Weir (MG), respectively (Korea Meteorological Administration, http://www.kma.go.kr, accessed on 28 May 2021). More than 50% of the annual precipitation (631.3 mm and 774.3 mm) occurs from July to September, whereas less than 7% of the precipitation (78.1 mm and 109.6 mm) occurs during the winter season (December to February) [23]. The SJ site is located in a relatively upstream region (i.e., 37.2°N latitude) of the Nakdonggang River, whereas the MG site is located in the downstream area (i.e., 35.1° N latitude). In addition, regional differences (e.g., land use) between the two sites can affect the differences in the micro-eukaryotic assemblage composition. The relative ratio of land use in the riparian area between the SJ and MG sites is highly distinct (Figure S1). At the SJ site, the proportion of the forest was the highest at 44.69%, followed by that of agriculture (24.53%), bareland (22.82%), wetland (2.29%), and grassland (10.03%), whereas at the MG site, the proportion of agriculture was the most dominant at 71.70%, followed by that of forest (18.14%), bareland (1.26%), and grassland (0.66%).

2.2. Sampling and Ecological Data

We collected water samples, monthly or bimonthly, for conducting eDNA metabarcoding analysis at two sampling sites (i.e., the upstream of Sangju Weir: SJ, and the downstream of Changnyeong-Haman Weir: MG) in the Nakdonggang River from March 2019 to April 2021 (Figure 1).

Using sterile bottles at each site, we collected 6 L of water samples from the river. The water samples were vacuum filtered immediately using a Supor^® 200 Membrane Filter (pore size: 0.2 μm; Pall Corporation, Ann Arbor, MI, USA), and then the filters were kept at –80 °C until DNA extraction. Total environmental DNA was extracted from the resulting filters using a PowerWater^® DNA Isolation Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Subsequently, the V4 region of the 18S rRNA gene (small ribosomal subunit) was PCR-amplified with the universal primer sets, Uni18SF and Uni18SR [24], to examine the microeukaryome, comprising metazoa, fungi, microalgae (i.e., autotrophic microeukaryotes), and protists. Sequencing analysis was conducted using Illumina MiSeq (Illumina, San Diego, CA, USA), and bioinformatic analysis was performed as described previously [25].

Eighteen environmental factors, such as water quality variables, discharge, and the amount of precipitation, were obtained from the open database of the Water Environment Information System supported by the Ministry of Environment and the National Institute of Environmental Research (http://water.nier.go.kr/, accessed on 20 June 2021) and the Korea Meteorological Administration (http://www.kma.go.kr, accessed on 28 May 2021) in South Korea (

Table 1. Average values and standard deviation of the environmental variables at the sampling sites.

Environmental Factors	Abbreviation	SJ	MG
Water temperature (°C)	WT	16.53 (7.32)	18.93 (7.77)
Dissolved oxygen (mg/L)	DO	10.09 (2.32)	10.38 (2.24)
Biochemical oxygen demand (mg/L)	BOD	1.45 (0.58)	1.85 (0.57)
Chemical oxygen demand (mg/L)	COD	5.06 (0.90)	6.13 (1.29)
Suspended solid	SS	8.30 (5.96)	12.84 (24.55)
Total nitrogen (mg/L)	TN	2.16 (0.54)	2.54 (0.67)
Total Phosphorus (mg/L)	TP	0.034 (0.017)	0.040 (0.021)
Total organic carbon	TOC	3.96 (0.60)	4.48 (0.81)
pH	-	7.79 (0.56)	7.98 (0.43)
Electric conductivity (µS/cm)	Conductivity	220.96 (24.91)	290.11 (110.34)
Dissolved total nitrogen (mg/L)	DTN	2.09 (0.56)	2.40 (0.69)
Ammonia (mg/L)	NH3	0.04 (0.02)	0.06 (0.03)
Nitrate (mg/L)	NO3–	1.69 (0.49)	1.87 (0.65)
Dissolved total phosphorus (mg/L)	DTP	0.030 (0.015)	0.030 (0.019)
Orthophosphate (mg/L)	PO43–	0.011 (0.015)	0.014 (0.019)
Chlorophyll-a (mg/L)	Chla	15.35 (10.57)	23.8 (16.46)
Precipitation (mm)	-	121.18 (139.83)	162.47 (170.78)

). Water temperature, dissolved oxygen, pH, and electric conductivity were measured during the field survey using YSI ProDSS (ESRI, Redlands, CA, USA). Other physicochemical factors such as biological oxygen demand (BOD), chemical oxygen demand (COD), suspended solids (SS), total nitrogen (TN), total phosphorus (TP), and total organic carbon (TOC) were measured in the laboratory using the method described by the American Public Health Association [26].

2.3. Data Analysis

First, as descriptive measures, we compared the relationships between taxon richness and the amount of precipitation. Second, we characterized the temporal variation in micro-eukaryotic assemblages using non-metric multidimensional scaling (NMDS) with the Bray–Curtis distance measure. We applied multi-response permutation procedures to evaluate the significant differences among the four seasons (A = 0.01, p < 0.05). We used the ‘metaMDS’ and ‘envfit’ functions in the R program to find the lowest stress value (i.e., the best solution) and figure out the relationship between the axes of NMDS and environmental factors, respectively. Third, indicator species analysis was performed to identify the representative taxa in each season at the SJ and MG sites. In this analysis, we computed the indicator values (IndVal) of each taxon based on the relative abundance and occurrence frequency in the defined group (here, each season). The IndVal range from 0 (no indication) to 100 (all individuals of taxa are only observed in one group) [27]. The indicator taxa were determined if IndVal was higher than 25% and showed a significant difference (p < 0.05). We used a Monte Carlo permutation (9999 random permutations) to determine the significance of IndVal for each taxon. Indicator species analysis was performed using the function “indval” in the Labdsv package [28]. Lastly, Threshold Indicator Taxa Analysis (TITAN) [29], which has largely been used to evaluate the tolerance range according to the degree of natural disturbances [2] and anthropogenic disturbances [30,31,32,33], was applied to detect the change points in the response of eukaryote assemblages according to the precipitation gradient. Here, TITAN computes the thresholds based on the changing point along with the amount of precipitation for each taxon of micro-eukaryote [29,34,35,36]. TITAN discriminates between negative (z−) and positive (z+) species responses at change points in response to the amount of precipitation. To confirm the thresholds of the amount of precipitation for each taxon, we measured the purity (i.e., the matching proportion between the observed value and resampling value at changing points) and reliability (i.e., the proportion of significant p-values in the indicator values computed from resampling) based on the bootstrap technique (i.e., 500 resamples with replacement). The assemblage-level threshold (i.e., the coincident changing point of the entire assemblage structure [sum(z)]) was also computed after identifying the indicator species. We only used the taxa observed in more than three samples, based on the recommendations of Baker and King (2010). The TITAN was performed using the TITAN2 package [29] in R.

3. Results and Discussion

Through 18S rRNA gene-based meta-amplicon sequencing analysis, 775 unique operational taxonomic units (OTUs) were observed during the entire sampling period, ranging from 35 OTUs (May 2019 in the MG site) to 313 OTUs (July 2020 in the SJ site). Seasonal changes (i.e., spring, summer, autumn, and winter), as well as spatial differences in the micro-eukaryotic assemblages based on eDNA (e.g., the changes in relative abundances between the SJ and MG sites), were clearly noticeable in the Nakdonggang River basin between the SJ and MG sites. When comparing the cumulative values over the 2 years, at the SJ site, the relative ratio of microalgae was the highest at 47.43%, followed by that of protists (25.86%), metazoans (24.61%), and fungi (2.10%). However, at the MG site, the ratio of metazoans was the highest at 59.95%, followed by that of microalgae (22.97%), protists (14.38%), and fungi (2.70%).

Environmental DNA signals can reflect changes in the relative abundance of the micro-eukaryotic assemblages in response to climate factors, including seasonal variation and precipitation amount [37]. For example, up to 71% of the total annual precipitation from March 2019 to April 2021 was concentrated from summer to early autumn (i.e., from July to September). During these rainy seasons, the relative abundance of Chlorophyta in phytoplankton increased at both the SJ and MG sites (Figure 2). In contrast, during the winter season (i.e., from December to February), diatoms were highly abundant among the phytoplankton at the SJ site. Moreover, the relative abundance of metazoan taxa was higher in most seasons at the MG site than at the SJ site (Figure 2).

Physicochemical factors are generally considered to be one of the primary causes of compositional changes in micro-eukaryotic assemblages in freshwater ecosystems [38]. Particularly, at both the SJ and MG sites, the concentrations of total phosphorus (TP) increased drastically to 0.067 mg/L (hypereutrophic) and 0.051 mg/L (eutrophic concentration), respectively, when the level of accumulated precipitation within 30 days became higher (i.e., more than 100 mm) during the rainy seasons (Figure S2). These observations suggest that phosphorus components in the soil around riparian areas could penetrate as nonpoint source input into the river water during the rainy seasons with high precipitation [39]. In contrast, total nitrogen (TN) showed no significant correlation with precipitation at the SJ site, whereas TN showed a tendency to be inversely proportional at the MG site (Figure S2). This indicated that the nitrogen sources were continuously accumulated in river water rather than being rapidly introduced by rainfall, as in the case of phosphorus, so that the concentration of TN in river water could be diluted when precipitation is concentrated over short periods.

Interestingly, at both the SJ and MG sites, the number of taxa determined using the eDNA metabarcoding assessments increased significantly with the increment in monthly precipitation (i.e., the accumulated precipitation within 30 days before the sampling time), irrespective of the taxonomic groups assessed (i.e., metazoan, fungi, microalgae, and protozoa) (Figure 3, Figure S3 and Figure S4). In particular, the number of detectable taxa was especially elevated when the monthly precipitation was higher than 200 mm. These results could be attributed to various reasons during the rainy season. First, heavy rainfall can induce the inflow of a higher quantity of eDNA from riparian areas, including soil-based eDNA, into rivers or catastrophic drift of organisms. Therefore, more diverse taxa could be detected through meta-barcoding analyses of the sampled water [40,41,42]. Second, heavy rainfall abruptly changes water discharge and depth; therefore, even the eDNA derived from benthic species could be more easily detectable in the sampled water [43].

At both the SJ and MK sites, the micro-eukaryotic assemblages were related to water temperature and the concentration of nitrogenous compounds, such as nitrate (

Table 2. Relationships between environmental factors and the non-metric multidimensional scaling (NMDS) ordination of micro-eukaryotic assemblages from eDNA metabarcoding. The meaning of abbreviations for the environmental variables is presented in Table 1.

SJ	NMDS1	NMDS2	R2	p	MG	NMDS1	NMDS2	R2	p
WT	−0.61	−0.79	0.61	0.001	WT	0.99	0.12	0.71	0.001
DO	0.43	0.90	0.55	0.001	DO	−0.98	−0.18	0.43	0.002
BOD	−0.71	0.71	0.28	0.02	BOD	0.77	0.64	0.03	0.672
COD	−0.26	−0.97	0.05	0.553	COD	0.62	0.79	0.37	0.004
SS	0.44	−0.90	0.13	0.171	SS	0.22	0.98	0.22	0.028
TN	0.98	−0.20	0.58	0.001	TN	−0.87	−0.50	0.63	0.001
TP	0.03	−1.00	0.23	0.048	TP	0.54	0.84	0.37	0.005
TOC	−0.25	−0.97	0.05	0.492	TOC	0.64	0.77	0.29	0.017
pH	−0.08	1.00	0.16	0.108	pH	0.69	−0.73	0.03	0.731
Conductivity	−0.02	1.00	0.30	0.015	Conductivity	−0.69	−0.72	0.73	0.001
DTN	0.98	−0.19	0.59	0.001	DTN	−0.86	−0.51	0.64	0.001
NH3	0.52	−0.85	0.04	0.637	NH3	1.00	−0.06	0.01	0.925
NO3-	0.96	−0.27	0.47	0.002	NO3-	−0.87	−0.49	0.61	0.001
DTP	−0.05	−1.00	0.21	0.056	DTP	0.38	0.92	0.35	0.003
PO43–	0.11	−0.99	0.31	0.014	PO43−	0.40	0.92	0.42	0.003
Chla	−0.11	0.99	0.11	0.216	Chla	−0.11	0.99	0.12	0.2
Precipitation	−0.06	−1.00	0.60	0.001	Precipitation	0.59	0.81	0.41	0.005

). In addition, it also showed a higher significance with the amount of precipitation and concentration of orthophosphate.

NMDS based on data from the micro-eukaryotic assemblages at the SJ and MG sites highly reflected the seasonal differences (Figure 4). In the NMDS ordination from the SJ site, the first two axes (stress value = 10.4) had the strongest relationship with water temperature (R² = 0.61, p < 0.001), followed by the amount of precipitation (R² = 0.60, p < 0.001), DTN (R² = 0.59, p < 0.001), TN (R² = 0.58, p < 0.001), and DO (R² = 0.55, p < 0.001) (Table 2). Most of the samples surveyed in autumn and winter were ordinated on the right side of NMDS1 with high values of TN, DTN, and nitrate (NO₃⁻), whereas samples from summer with high values of precipitation, water temperature, TP, and orthophosphate (PO₄³⁻) were mainly ordinated on the lower part of NMDS2. In the NMDS ordination from the MG site, the first two axes (stress value = 12.2) had the strongest relationship with conductivity (R² = 0.72, p < 0.001), followed by water temperature (R² = 0.71, p < 0.001), DTN (R² = 0.64, p < 0.001), TN (R² = 0.63, p < 0.001), and nitrate (R² = 0.61, p < 0.001). Samplings in spring and winter with high values of TN, nitrate, DTN, and conductivity were ordinated in the left and lower left parts of NMDS1 and NMDS2, respectively. In contrast, the samples from summer with high values of precipitation, orthophosphate, DTP, discharge, and TP were mainly ordinated in the right and upper-right parts of NMDS1 and NMDS2, respectively.

Thirty-eight taxa from the SJ site and forty-four taxa from the MG site were selected as indicator species representative of the four different seasonal groups based on IndVal (p < 0.05) (

Table 3. Results of the indicator species analysis in four groups (i.e., seasons) at the SJ site. Only taxa with significant values are shown (2: summer, 3: autumn and 4: winter).

Group	Taxonomic Group	Phylum	Class	Genus	IndVal	p-Value
2	Fungi	Chytridiomycota	Lobulomycetaceae	Alogomyces	0.50	0.018
2	Fungi	Chytridiomycota	Aquamycetaceae	Aquamyces	0.47	0.022
2	Fungi	Chytridiomycota	Undefined	Delfinachytrium	0.40	0.046
2	Fungi	Oomycota	Lagenidiaceae	Lagenidium	0.87	0.002
2	Metazoa	Rotifera	Brachionidae	Keratella	0.69	0.001
2	Metazoa	Rotifera	Filinidae	Filinia	0.40	0.021
2	Microalgae	Chlorophyta	Sphaeropleaceae	Atractomorpha	0.98	0.001
2	Microalgae	Chlorophyta	Dunaliellaceae	Hafniomonas	0.93	0.001
2	Microalgae	Chlorophyta	Golenkiniaceae	Golenkinia	0.80	0.025
2	Microalgae	Chlorophyta	Haematococcaceae	Gungnir	0.76	0.002
2	Microalgae	Chlorophyta	Chlamydomonadaceae	Gloeomonas	0.52	0.018
2	Microalgae	Chlorophyta	Bracteacoccaceae	Bracteacoccus	0.47	0.033
2	Protist	Cercozoa	Undefined	Gymnophrys	0.40	0.034
2	Protist	Ciliophora	Urostylidae	Perisincirra	0.80	0.001
2	Protist	Ciliophora	Bryometopidae	Bryometopus	0.56	0.036
2	Protist	Ciliophora	Spirofilidae	Hypotrichidium	0.44	0.016
3	Microalgae	Bacillariophyta	Acanthocerataceae	Acanthoceras	0.75	0.016
3	Microalgae	Bacillariophyta	Stephanodiscaceae	Cyclostephanos	0.50	0.012
3	Microalgae	Bacillariophyta	Thalassiosiraceae	Thalassiosiraceae_F	0.50	0.013
3	Microalgae	Bacillariophyta	Thalassiosiraceae	Stephanocyclus	0.44	0.037
3	Microalgae	Chlorophyta	Dunaliellaceae	Dunaliellaceae_F	0.65	0.005
3	Microalgae	Chlorophyta	Chlamydomonadaceae	Dangeardinia	0.50	0.007
3	Microalgae	Chlorophyta	Volvocaceae	Gonium	0.50	0.006
3	Protist	Ciliophora	Spathidiidae	Epispathidium	0.47	0.02
3	Protist	Haptista	Raphidiophryidae	Raphidiophrys	0.48	0.014
4	Fungi	Basidiomycota	Mrakiaceae	Udeniomyces	0.57	0.006
4	Fungi	Chytridiomycota	Spizellomycetaceae	Gaertneriomyces	0.72	0.001
4	Microalgae	Chlorophyta	Chlorococcaceae	Spongiococcum	0.40	0.03
4	Microalgae	Chlorophyta	Chlorellaceae	Carolibrandtia	0.40	0.029
4	Microalgae	Chlorophyta	Undefined	Fernandinella	0.40	0.027
4	Microalgae	Streptophyta	Closteriaceae	Closterium	0.46	0.027
4	Protist	Cercozoa	Undefined	Glissomonadida_O	0.68	0.031
4	Protist	Ciliophora	Nassulidae	Obertrumia	0.74	0.004
4	Protist	Ciliophora	Climacostomidae	Climacostomum	0.40	0.029
4	Protist	Ciliophora	Trachelophyllidae	Enchelyodon	0.40	0.017
4	Protist	Ciliophora	Orchitophryidae	Orchitophryidae_F	0.40	0.028
4	Protist	Ciliophora	Undefined	Stichotrichida_O	0.40	0.03
4	Protist	Imbricatea	Undefined	Thaumatomonadida_O	0.91	0.003

and

Table 4. Results of the indicator species analysis in four groups (i.e., seasons) at the MG site. Only taxa with significant values are shown (1: spring, 2: summer, 3: autumn, and 4: winter).

Group	Taxonomic Group	Phylum	Class	Genus	IndVal	p-Value
1	Metazoa	Arthropoda	Hexanauplia	Thermocyclops	0.60	0.01
2	Fungi	Chytridiomycota	Chytridiomycetes	Chytriomycetaceae_F	0.60	0.008
2	Fungi	Chytridiomycota	Chytridiomycetes	Aquamyces	0.44	0.049
2	Fungi	Chytridiomycota	Chytridiomycetes	Staurastromyces	0.40	0.05
2	Fungi	Oomycota	Undefined	Lagenidium	0.60	0.027
2	Metazoa	Arthropoda	Hexanauplia	Mesocyclops	0.66	0.016
2	Metazoa	Mollusca	Bivalvia	Aculamprotula	0.59	0.029
2	Metazoa	Platyhelminthes	Rhabditophora	Mesostoma	0.40	0.037
2	Microalgae	Chlorophyta	Chlorophyceae	Chlamydopodium	0.87	0.001
2	Microalgae	Chlorophyta	Chlorophyceae	Golenkinia	0.67	0.005
2	Microalgae	Chlorophyta	Chlorophyceae	Planktosphaeria	0.56	0.042
2	Microalgae	Chlorophyta	Chlorophyceae	Chlorococcaceae_F	0.40	0.05
2	Protist	Cercozoa	Undefined	Bodomorpha	0.53	0.017
2	Protist	Cercozoa	Undefined	Paracercomonas	0.50	0.041
2	Protist	Ciliophora	Nassophorea	Obertrumia	0.67	0.015
2	Protist	Ciliophora	Oligohymenophorea	Ichthyophthirius	0.67	0.018
2	Protist	Ciliophora	Spirotrichea	Perisincirra	0.60	0.01
2	Protist	Ciliophora	Colpodea	Cyrtolophosis	0.58	0.031
2	Protist	Ciliophora	Oligohymenophorea	Ophryoglena	0.50	0.015
2	Protist	Ciliophora	Heterotrichea	Stentor	0.47	0.029
2	Protist	Haptista	Centroplasthelida	Choanocystis	0.43	0.048
2	Protist	Tubulinea	Echinamoebida	Echinamoeba	0.68	0.002
3	Fungi	Chytridiomycota	Chytridiomycetes	Pendulichytrium	0.67	0.009
3	Fungi	Chytridiomycota	Chytridiomycetes	Uebelmesseromyces	0.59	0.004
3	Fungi	Chytridiomycota	Chytridiomycetes	Dangeardia	0.48	0.029
3	Metazoa	Arthropoda	Hexanauplia	Microcyclops	0.73	0.004
3	Metazoa	Arthropoda	Hexanauplia	Eucyclops	0.48	0.038
3	Metazoa	Mollusca	Bivalvia	Corbicula	0.53	0.037
3	Metazoa	Rotifera	Eurotatoria	Trichocerca	0.63	0.025
3	Metazoa	Rotifera	Eurotatoria	Ascomorpha	0.49	0.042
3	Metazoa	Rotifera	Eurotatoria	Collotheca	0.47	0.05
3	Microalgae	Streptophyta	Zygnemophyceae	Closterium	0.73	0.004
3	Protist	Cercozoa	Undefined	Cercomonadida_O	0.41	0.034
4	Fungi	Chytridiomycota	Chytridiomycetes	Gaertneriomyces	0.60	0.003
4	Fungi	Chytridiomycota	Chytridiomycetes	Rhopalophlyctis	0.49	0.03
4	Metazoa	Rotifera	Eurotatoria	Notholca	0.40	0.045
4	Microalgae	Bacillariophyta	Fragilariophyceae	Ulnaria	0.87	0.004
4	Microalgae	Bacillariophyta	Fragilariophyceae	Asterionella	0.79	0.016
4	Microalgae	Bacillariophyta	Fragilariophyceae	Tabellaria	0.33	0.033
4	Microalgae	Chlorophyta	Chlorophyceae	Microglena	0.76	0.005
4	Microalgae	Chlorophyta	Chlorophyceae	Dangeardinia	0.40	0.038
4	Microalgae	Chlorophyta	Chlorophyceae	Basichlamys	0.39	0.033
4	Microalgae	Streptophyta	Zygnemophyceae	Tortitaenia	0.48	0.047
4	Protist	Imbricatea	Undefined	Esquamula	0.40	0.031

, respectively). At the MG site, a downstream area of the Nakdonggang River, some OTUs in the metazoan phyla, including Arthropoda, Mollusca, Platyhelminthes, and Rotifera, were observed as indicator species. This result is distinct from that at the SJ site, where only an OTU belonging to the phylum Rotifera was included as an indicator species among the metazoan taxa. Because the predators higher in the food web could be relatively abundant in the downstream regions of the large river than in the upstream regions (Table 3 and Table 4), they may act as indicator species to represent seasonal variations and responses to ecological conversion due to climate change. In particular, the highest number of indicator species was selected from the summer group at both the SJ (16 taxa) and MG sites (21 taxa).

Detected OTUs may vary depending on the frequency and intensity of rain. For example, detected eDNA can be decreased with large amounts of precipitation because eDNA may be washed away into the downstream area with a high frequency of precipitation (e.g., [44,45]). High river flow from heavy rain may dilute eDNA concentration and reduce the detectability of some species as well [46]. These results suggest that rainwater could dilute the genetic material of species (e.g., fish); therefore, even common species that were commonly detected in the past could be undetected due to dilution of eDNA concentration [46,47]. However, the dilution effect also could be compensated by the increased influx of genetic material entering the riparian water and flowing into the rivers [44]. For example, the MG site could represent a cumulative list of taxa from upstream and/or the bottom area of rivers especially during the rainy season. Because their large frustules are frequently washed away from their bottom habitats, the diatoms (e.g., Nitzschia) included in phytobenthos, were even found in MG sites where its discharge was over 600 cubic meters [48]. In addition, Sales et al. (2021) showed that biodiversity decreased through the second rainfall that followed rather than the first heavy rain, because the second rainfall can reduce the detectability of some species such as rare, low-abundance, and endangered species [49]. In the present study, many OTUs were detected during the rainy seasons. A large amount of precipitation may initially increase the number of OTUs due to the influx of organisms (or genetic materials) from the riparian and upstream regions. However, there were turning points when the detected OTU decreased as the rainy periods are prolonged. For example, in most cases, rainfall increased the number of OTUs detected, but on 20 August 2020, after it had rained continuously for two months, the detected OTU in both the SM and MG sites showed a tendency to decrease.

To determine the taxa influenced by changes in precipitation, we applied TITAN to the micro-eukaryotic assemblages (Figure 5 and

Table 5. TITAN assemblage-level thresholds estimated from the response of micro-eukaryotes to the amount of precipitation (mm). Sum(z−) and sum(z+) indicate a decrease (−) or increase (+) along the gradient of the precipitation; CP indicates the assemblage change point; 5, 10, 50, 90, and 95% are bootstrap quantile intervals.

Sites	Sum (z)	CP	0.05	0.1	0.5	0.9	0.95
SJ	sum (z−)	75.6	27.89	30.3	75.6	77.05	81.15
	sum (z+)	80.35	77.05	80.35	91.25	164.95	173.9
MG	sum (z−)	123.3	100.75	102.65	123.3	127.95	137.88
	sum (z+)	139.3	121.9	124.9	136.3	253.3	284.35

). It was found that 55 of the 334 OTUs (i.e., the taxa observed in more than three samples) showed a robust response (purity and reliability > 0.95). At the SJ site, the changing point was 42.4 mm/month for the sum (z−) (i.e., the taxa decreased with the increase in precipitation), whereas it was 164.95 mm/month for the sum (z+) (i.e., the taxa increased with decrease in precipitation). In contrast, at the MG site, the changing points were 23.95 and 316.05 mm/month for the sum (z−) and (z+), respectively. Because the total number of detected taxa showed a tendency to increase in the context of an increase in the precipitation level, it would be necessary to cross-check the taxa that showed a positive correlation with heavy rainfall, particularly through the traditional field survey, for further study.

The annual compositions of micro-eukaryotic assemblages, determined using an eDNA metabarcoding technique (i.e., via meta-amplicon sequencing tools), were clearly affected by seasonal changes in the freshwater ecosystem. Originally, the rainy season in the area (Nakdonggang River basin) ranged from the end of June to July, and more than half of the total precipitation was concentrated from June to August. However, during the study period, there was heavy rainfall even in early autumn, and approximately 71% of the total precipitation was concentrated from July to September. Thus, owing to the variance, we speculate that the clear fluctuations in temperature and precipitation levels in the area could be two major environmental factors that induce significant changes in the composition of microeukaryomes.

Temporal study is crucial for predicting biodiversity patterns and upcoming changes. The time-based fluctuations in population, assemblage, and genetic structure can provide warning signals for species at a high risk of extinction and the potential for sudden changes in ecosystem function [50]. Because temporal data are unidirectional, they may provide the best way to identify causal relationships between environmental changes (e.g., climate change and disturbances) and ecological dynamics. [51,52]. Studies based on eDNA metabarcoding can provide persistent temporal results on biodiversity over long periods for a wider range of taxonomic units and broader geographic coverage than most other temporal data sources [37]. Thus, the eDNA technology enables the time-series analysis of species in different trophic levels and functional groups of organisms with species-level resolution [18,37]. However, to realize a more accurate monitoring procedure for micro-eukaryotic assemblages, the behavioral properties (i.e., migration and degradation) of eDNA in freshwater ecosystems have to be considered [53,54]. Despite the many advantages and substantial potential of eDNA technology in assessing freshwater biodiversity, several research topics such as analyzing hydrological factors that can affect the spatial rearrangement of eDNA [53], calculating the quantitative decay kinetics (half-life) of eDNA materials [55], as well as normalizing the abundance of sensed signals to real population (or biomass) [56,57] need to be researched further.

4. Conclusions

As a result of analyzing and categorizing micro-eukaryotic assemblages for 2 years at the upstream and downstream sites of the Nakdonggang River, the biota of each site were clearly classified according to the seasonal variations. At the two sites, environmental factors (e.g., water quality) fluctuated according to seasonal differences, and these major environmental factors and changes in micro-eukaryotic biota were highly correlated. In addition, we observed the statistically significant increment in the number of taxa of all taxonomic groups analyzed using eDNA metabarcoding with heavy precipitation during the rainy season. Finally, lists of taxa that can be an indicator among seasonally typed organisms were derived, and the taxa highly influenced by responding to increases and decreases in precipitation level were presented through TITAN analysis. As the frequency of drought and heavy rain is expected to increase in the future due to climate change, efforts to standardize the relation between precipitation and eDNA analysis could be obviously required. Because micro-eukaryotes play a central role in food webs and material circulation, it is necessary to conduct further long-term monitoring to carefully investigate the changes in the micro-eukaryotic assemblages according to the alteration in environmental conditions caused by climate change.