Water Quality, Toxicity and Diversity of Planktonic and Benthic Cyanobacteria in Pristine Ancient Lake Khubsugul (Hövsgöl), Mongolia

For the first time, microcystin-producing cyanobacteria have been detected in Khubsugul, which is ancient, pristine and one of the world’s largest lakes. The microcystin synthetase genes belonged to the genera Nostoc, Microcystis and possibly Snowella spp. No microcystins were found in the water of the lake. Using the HPLC-HRMS/TOF, five microcystin congeners were identified in biofilms from stony substrates sampled in the coastal zone. The concentration of microcystins in biofilms was low: 41.95 µg g−1 d. wt. by ELISA and 55.8 µg g−1 d. wt. using HPLC. The taxonomic composition of planktonic and benthic cyanobacterial communities was determined by means of microscopy and high-throughput sequencing of 16S rDNA amplicons. Nostocales cyanobacteria dominated benthos of Lake Khubsugul and Synechococcales—plankton. The abundance of cyanobacteria was low both in plankton and benthos; there was no mass development of cyanobacteria. Hydrochemical and microbiological analyses showed that the water in the lake was clean; the number of faecal microorganisms was significantly below the acceptable guideline values. Hydrochemical and hydrophysical parameters, and the concentration of chlorophyll a, were low and within the range of values recorded in the 1970s to 1990s, and corresponded to the oligotrophic state of the lake. There were no signs of anthropogenic eutrophication of the lake and no conditions for the cyanobacterial blooms.


Introduction
Cyanobacteria are a successful group of morphologically diverse prokaryotes that are widespread in aquatic ecosystems with different trophic statuses, where they play an important role as primary producers, nitrogen-fixing bacteria and as a food resource. Cyanobacteria are one of the most ancient organisms on the Earth: their fossils are found in the Precambrian deposits (about 3.5 billion years old) [1,2]. are used as drinking water sources and for recreation by a great number of people. The Laurentian Great Lakes are such an example: Lake Michigan [48], Lake Huron [49], Lake Ontario [50] and Lake Erie [51,52], as well as the largest freshwater lakes in China: Poyang Lake [53], Dong Ting Lake [54], Lake Taihu [11,12] and Lake Victoria in Africa [55].
We aimed to revise the diversity of cyanobacteria and search for toxigenic and toxinproducing cyanobacteria in ancient oligotrophic Lake Khubsugul located in a sparsely populated highland part of Northern Mongolia. Lake Khubsugul is ranked 16th in the area among all freshwater lakes on the Earth and the second among Mongolian lakes, containing about 70% of all freshwater reserves in the country. The severe climate in the region is unfavourable for versatile agricultural and industrial activities, and nomadic pastoralism is the main occupation of the population (Figure 1). negative consequences of anthropogenic eutrophication in water bodies can manifest as an "environmental disaster" within several years or decades. The key factors of anthropogenic eutrophication include: an influx of mineral and organic substances from the atmosphere polluted by industrial activities, hydraulic engineering and surface runoff from agricultural land and urban areas as well as domestic and industrial wastewater.
Toxic bloom events are especially relevant in urbanised regions with developed industries, as well as in countries with a warm climate and scarce drinking water. Toxic cyanobacterial blooms have a major negative impact on the large water bodies which are used as drinking water sources and for recreation by a great number of people. The Laurentian Great Lakes are such an example: Lake Michigan [48], Lake Huron [49], Lake Ontario [50] and Lake Erie [51,52], as well as the largest freshwater lakes in China: Poyang Lake [53], Dong Ting Lake [54], Lake Taihu [11,12] and Lake Victoria in Africa [55].
We aimed to revise the diversity of cyanobacteria and search for toxigenic and toxinproducing cyanobacteria in ancient oligotrophic Lake Khubsugul located in a sparsely populated highland part of Northern Mongolia. Lake Khubsugul is ranked 16th in the area among all freshwater lakes on the Earth and the second among Mongolian lakes, containing about 70% of all freshwater reserves in the country. The severe climate in the region is unfavourable for versatile agricultural and industrial activities, and nomadic pastoralism is the main occupation of the population (Figure 1). Despite the remoteness and poor accessibility of the lake, the first data on phytoplankton in Lake Khubsugul appeared at the beginning of the 20th century (reviewed by [56]). Over a century of study, the authors emphasised the absence of cyanobacteria in the Despite the remoteness and poor accessibility of the lake, the first data on phytoplankton in Lake Khubsugul appeared at the beginning of the 20th century (reviewed by [56]). Over a century of study, the authors emphasised the absence of cyanobacteria in the lake many times, along with predominance of diatoms and green algae [57][58][59][60][61][62][63]. Cyanobacteria were first identified in the lake in the late 1980s. Seven of the 76 species found in plankton belonged to cyanobacteria and one species was identified in benthos [64,65]. The latest list contains 97 taxa of planktonic algae, while the taxonomic composition and number of cyanobacteria remain the same [56]. The morphotypes and abundance of picoplanktonic cyanobacteria and bacteria have been determined quite recently, using electron and fluorescent microscopy [66]. Studies of the microbial diversity of aquatic ecosystems in Mongolia using high-throughput sequencing (HTS) have not yet been carried out.
In this study, we assessed the water quality based on hygiene-microbiological and hydrochemical indicators, investigated microbial communities including opportunistic and pathogenic bacteria using amplicon-based sequencing targeting the V3-V4 region of the 16S rRNA gene, detected cyanobacteria containing the microcystin synthetase gene E (mcyE) by PCR-based methods, identified microcystin congeners via mass spectrometry and determined MC concentration by enzyme-linked immunosorbent assay (ELISA). At the same time, we tried to carry out a comprehensive assessment of the current state of the lake and forecast the development of cyanobacterial communities in the nearest future, taking into account the obtained results and the not readily available literature data from previous decades, as well as the literature data on the evaluation of global climate change in Mongolia.

Hydrophysical and Hydrochemical Parameters
From 11 to 14 July, the surface water temperature in the northern part of Lake Khubsugul ranged from 8.9 • C to 9.5 • C. Mean water transparency measured by Secchi disk was 20 m (SD = 0.5) in the pelagic zone. The pH value in the surface water layer was 8.4 to 8.5 at all stations, slightly decreasing with depth (to pH = 8.1 at a depth of 25 m).
Hydrophysical and hydrochemical parameters at different depths and in the 0-25 m layer are presented in Table 1. The water in the lake was well aerated throughout the water column; the oxygen saturation reached 100 to 105%. The concentrations of nutrients in the lake water were low and similar at all the stations. Silica (Si) and nitrate ion (NO 3 -) were rather evenly distributed in the upper 25 m layer. Nitrite (NO 2 − ) and ammonium (NH 4 + ) ions were detected only in the surface layer at trace amounts. The concentration of inorganic phosphorus (PO 4 − ) changed insignificantly with depth and was 2.3 to 2.7 µg L −1 . The ratio of the total nitrogen to total phosphorus (TN/TP) in the lake water averaged 32, indicating a phosphorus deficiency and high probability of limiting plankton growth. The maximum concentrations of organic nitrogen (N org ) and phosphorus (P org ) was determined at a depth of 15 m. The chemical oxygen indices (COD Mn and COD Cr ) in the pelagic zone of Lake Khubsugul were low during the study period and their ratio (100 * COD Mn /COD Cr ) averaged approximately 19%, indicating the predominance of easily hydrolysable organic matter in the water. The concentration of organic carbon in the water of Lake Khubsugul on average was 1.0 mg C L −1 .

Chlorophyll a
In July, the concentration of chlorophyll a (Chl a) in the plankton of Lake Khubsugul was low (Table 1), which is typical of oligotrophic waters [67]. The data on the Chl a concentration in biofilms covering the surface of stones in Lake Khubsugul varied significantly depending on the sample type. Chl a content in cyanobacterial biofilms averaged 324 (SD = 97) µg g −1 wet weight; it was higher in biofilms dominated by green algae.

Microscopy Analysis of Planktonic and Benthic Cyanobacterial Communities
Microscopy analysis of plankton samples from Lake Khubsugul indicated that the diatom Cyclotella ocellata dominated phytoplankton; its abundance reached 170 thousand cells L −1 , and its contribution to the total abundance of phytoplankton was 80% ( Figure 2). Chrysophycean algae Dinobryon sociale and green algae Mychonastes minusculus and Monoraphidium contortum were subdominant; other algae were minor. The autotrophic picoplankton was numerous and was mainly represented by cells belonging to the morphotypes of the genera Synechococcus and Cyanobium. Dolichospermum lemmermannii and Limnococcus limneticus represented nanoplanktonic cyanobacteria with an abundance of up to 689 cells L −1 and 70 cells L −1 , respectively ( Figure 3A,C). The total abundance of nanoplanktonic algae and cyanobacteria reached 200 thousand cells L −1 (Figure 2). The cyanobacterial species, Coelosphaerium kuetzingianum, Planktothrix sp. and Trichodesmium lacustre were found in net samples ( Figure 3B,D,E).  Examination of benthic samples revealed two types of foulings on the stony substrates. Ulothrix zonata dominated the samples, with all-over green foulings from the northern margin of the lake. The diatom Hannaea arcus and cyanobacteria Leibleinia epi-     Examination of benthic samples revealed two types of foulings on the stony substrates. Ulothrix zonata dominated the samples, with all-over green foulings from the northern margin of the lake. The diatom Hannaea arcus and cyanobacteria Leibleinia epiphytica were epiphytes on the filaments of U. zonata; cyanobacteria Calothrix parietina was found among the filaments. The second type of foulings identified near the Khankh settlement was in the form of separate or confluent small hemispherical brown colonies ( Figure 4A). Diatoms and cyanobacteria of the orders Synechococcales, Chroococcales and Nostocales dominated by Rivularia coadunata prevailed in their composition ( Figure 4B-D). The number of R. coadunata reached 58 million filaments m −2 , and the biomass-4458 g m −2 . Nostoc sp. with the average biomass of 160 g m −2 was present in benthos (Table S1).
Foulings on biogenic substrates were mainly dead filamentous green algae, dominated by six species of cyanobacteria: Calothrix parietina, Nostoc sp., Leibleinia epiphytica, Chamaesiphon subglobosus, Chamaesiphon polonicus, Chroococcus sp. and Pseudanabaena galeata ( Table 2). The total number of benthic cyanobacteria on the stones averaged 82 million ind. m −2 (Table S1). Moss thickets were seen on the sandy bottom at the northern margin of the lake, while cyanobacterial biofilms and other foulings were absent.

High-Throughput Sequencing of 16S rRNA Gene of Bacterial Community
In plankton and benthic biofilms, we obtained 206,809 sequences of 16S rRNA gene fragment with an average read length of 450 bp. Figure S1 shows the number of obtained sequences, identified OTUs and diversity indices (Chao1, ACE, Shannon and Simpson) calculated for all OTUs, except for singletons and duos ( Figure S1). Sequences belonging to the phyla Actinobacteria   Table 2. Cyanobacteria of Lake Khubsugul based on light microscopy, HTS and the literature data. Planktonic species are highlighted in bold; grey colour mark species detected previously using microscopy [56,65].
Contrarily, the contribution of cyanobacteria increased to 13% in the green biofilm sample ( Figure 5A). OTUs of filamentous and coccoid cyanobacteria of the order Synechococcales were the most numerous: up to 87% and 47%, respectively ( Figure 5B). The phylotypes of Synechococcus (99% identity with Synechococcus sp. MW6C6) dominated the coccoid forms. The diversity of phylotypes was almost two times higher than the number of species based on microscopy data ( Table 2). The greatest diversity of cyanobacterial OTUs (17) was observed in the samples taken from stones near the Khankh settlement. We detected eight cryptic genera of Synechococcales with similar morphology: thin filamentous cyanobacteria of the phylotypes Jaaginema, Leptolyngbya, Nodosilinea, Phormidesmis, Pseudanabaena, Shackletoneilla, Stenomitos and Timaviella. Their contribution to the total number of cyanobacterial phylotypes ranged from 28 to 87% ( Figure 5B). We identified Nostocales cyanobacterial phylotypes of the genera Calothrix, Nostoc, Rivularia and Scytonematopsis, which amounted to less than 2% of the total number of cyanobacterial phylotypes. For the first time, we found a rare member of the order Gloeobacterales, the oldest branch of cyanobacteria without thylakoids, the phylotype Gloeobacter kilaueensis.

Microbial Water Quality Assessment
Water quality in all samples from the lake corresponded to the sanitary and hygiene standards and requirements established by the sanitary rules and norms, imposed in Russia for assessing water use objects of I and II categories [68][69][70] (Table S2). The number of faecal derived bacteria and phages as sensitive indicators of recent faecal contamination was significantly below the guideline values. We detected total coliforms in six coastal and two pelagic samples, ranging from 2 ± 1 to 54 ± 2 colony-forming units (CFU) in 100 mL (with a safety standard ≤ 1000 and 500 CFU in 100 mL for the two categories of water use objects, respectively). Thermotolerant coliforms and Escherichia coli were identified in four coastal and two pelagic samples, the number of which did not exceed 4 ± 0 CFU in 100 mL (with a safety standard ≤ 100 CFU in 100 mL). Enterococci (bacteria of the genus Enterococcus) were present in five coastal samples and two pelagic samples (1-6 ± 1 CFU in 100 mL, with a safety standard of ≤ 100 and 10 CFU in 100 mL for each category, respectively). In 12 samples, we did not find total coliforms, thermotolerant coliforms and E. coli, eight of which were pelagic samples. No coliphages were detected in the water from Lake Khubsugul. The coastal stations near the Khankh settlement had higher values of total and thermotolerant coliforms compared to the northern stations.

High-Throughput Sequencing of 16S rRNA Gene of Pathogenic and Opportunistic Bacteria
Analysis of microbial communities using HTS revealed the presence of pathogenic and opportunistic bacteria of the families Enterobacteriaceae, Moraxellaceae, Legionellaceae and Vibrionaceae (class/phylum Gammaproteobacteria) in the water and in biofilms. Enterobacteriaceae that include the greatest number of bacteria harmful to human health were not found in the water; they were present in biofilms in small amounts. The biofilms contained sequences identified as Escherichia coli (99.8%), a well-known member of the human and animal gut microbiome, and Pantoea agglomerans (99.3%), the most commonly isolated species in humans [71] (Figure 6). We detected sequences similar to the widely distributed species Buttiauxella gaviniae (99.3%) isolated from molluscs, mainly snails and slugs, which is possibly pathogenic to humans [72].
The sequences of the phylum Firmicutes were mainly present in the coastal water; bacteria of the order Lactobacillales predominated among them being largely represented by the species Lactococcus piscium (99.3%) isolated from diseased rainbow trout before [74]. Notably, bacteria associated with fish were found in the studied samples, for example, Epulopiscium sp., gram-positive bacteria that had a symbiotic relationship with surgeonfish [75]. as a human pathogen associated with "sepsis, mixed infections and wound infections", as well as with meningitis. Most members of Bacillus species are distributed in natural environments, but some of these species are able to cause severe to self-limited disorders as an actual or opportunistic pathogen, such as B. circulans [77]. Among staphylococci, there was Staphylococcus warneri (99.8%), a member of normal microflora of the skin and mucous membrane of some organs in humans, primates and domestic animals [78]. Among Firmicutes, there were numerous bacteria of the order Clostridiales in the water and biofilms. Most species of this order are saprophytes inhabiting many We determined sequences closely related to Streptococcus sp. (99.5%) inhabiting the human respiratory and digestive tracts. Bacteria of the order Bacillales were numerous and diverse; they had more than 97% identification with bacteria from other habitats. Representatives of the genus Bacillus turned out to be mainly soil species, while some of them were the inhabitants of the gastrointestinal tract of ruminants and humans, for example B. subtilis [76]. In the samples, we identified the sequences of Bacillus circulans, described as a human pathogen associated with "sepsis, mixed infections and wound infections", as well as with meningitis. Most members of Bacillus species are distributed in natural environments, but some of these species are able to cause severe to self-limited disorders as an actual or opportunistic pathogen, such as B. circulans [77]. Among staphylococci, there was Staphylococcus warneri (99.8%), a member of normal microflora of the skin and mucous membrane of some organs in humans, primates and domestic animals [78].
Among Firmicutes, there were numerous bacteria of the order Clostridiales in the water and biofilms. Most species of this order are saprophytes inhabiting many environments, primarily the soil. They are found in healthy people and comprise a substantial part (several tens of percent) of the normal microflora in the oral cavity, intestines and human urogenital system. Among the sequences that were 98 to 99% identical to the representatives of the human microflora, there were bacterial genera belonging to the family Christensenellaceae, which inhabit the human gut and are important for human health [79].
The genus Clostridium was diverse, accounting for over 10 OTUs. Bacteria belonging to the genus Clostridium are anaerobic Gram-positive rods with the ability to form endospores [80]. Clostridium species are ubiquitous in the environment. Clostridium species are a part of the normal flora of the intestinal tract of humans and other animals, and may also be isolated from the female genital tract and the oral mucosa [80]. In our case, we detected potentially pathogenic Clostridium beijerinckii and Clostridium ventriculi (formerly Sarcina ventriculi), as well as Cellulosilyticum ruminicola isolated from the rumen of a yak in Mongolia [81].

Microcystin Synthetase Genes
PCR analysis yielded a positive result for net phytoplankton samples collected in the coastal zone of Lake Khubsugul near the Khankh settlement. No cyanobacteria with MC synthetase genes were found in the plankton from pelagic and benthic samples.
We obtained 30 sequences from PCR positive samples representing three different genotypes. Three sequences of the mcyE gene had the greatest identification (98.3%) with the sequences of the genus Microcystis, in particular, with the strain Microcystis aeruginosa NIES-88, isolated from Lake Kawaguchi (Yamanashi, Japan). Fifteen sequences were closely related (99.6%) to the sequences of uncultured cyanobacteria isolated from Lake Baikal biofilms [85] and were 92% identical to the Snowella sp. CHAB strains 6601, 6602, 6604, 6605 and 6606, isolated from a brackish water lake in Qinghai-Tibetan Plateau (China) (unpublished data). Cultured strains of Nostoc, for example, Nostoc sp. 152 (93%) and Nodularia spumigena UHCC0039, CCY9414, NSOR10 (91.1-91.5%) were the closest relatives of 12 sequences.
The phylogenetic tree showed amino acid sequences of the mcyE gene from Lake Khubsugul, forming stable clusters with closely related sequences ( Figure S2). It was previously shown that branching of mcyE genes corresponds to highly supported diversification according to Microcystis, Anabaena, Nodularia, Planktothrix, Nostoc and Phormidium genera [86].
Within the genus Microcystis (order Chroococcales) cluster, genotypes from Lake Khubsugul were grouped together with the sequences of M. aeruginosa, M. flos-aquae, M. viridis and M. wesenbergii, mainly from the Asian water bodies, including Lake Baikal. Amino acid sequences of the order Nostocales from Lake Khubsugul formed a stable cluster with Nostoc sp. "Peltigera membranacea cyanobiont" 232 within the clade formed by clusters of nodularin-producing Nodularia spumigena and microcystin-producing Nostoc sp. 152 and Nostoc sp. "Peltigera membranacea cyanobiont" 213.
Sequences of members of the order Synechococcales from Lake Khubsugul were included in the cluster, together with those of the Snowella sp. CHAB strains and uncultured sequences were isolated from Lake Baikal benthos.

Microcystin Concentration and Microcystin Congeners
The initial screening of MCs was performed using the Microcystins-Adda ELISA kit. All samples and standards were analysed in duplicate. There were no MCs in water samples and net samples from Lake Khubsugul, in the biofilms from stones sampled in the coastal zone of the northern margin of the lake (LODs for the Adda ELISA < 0.10 ppb). We detected MCs in biofilms from stony substrates sampled in the coastal zone near the Khankh settlement (eight biofilms, station "Stones"); mean MC concentration was 41.95 µg g −1 d. wt. (SD = 5.57) ( Figure S3).
Integrated samples of eight biofilms of stone substrates from the coastal zone off the Khankh settlement contained five MC congeners, according to the high performance liquid chromatography-high resolution mass spectrometry results (HPLC-HRMS/TOF): MC-YR, MC-LAba, MC-LY, [ADMAdda 5 ]MC-LR and [ADMAdda 5 ]MC-XR (where X is a fragment, its structure has not yet been determined) ( Figure S4). The content of these MCs was estimated using a calibration curve for MC-LR standard solution (Biosense Laboratories, Norway), as follows: 50 (SD = 7), 0.4, 0.2, 0.7 and 4.5 (SD = 1.4) µg g −1 d. wt., respectively.
A weak positive correlation was noted between dissolved inorganic phosphorus (DIP) and C. ocellata number. Strong negative correlations were indicated between DIP and green algae Mychonastes minusculus (r = −0.49, p < 0.05), Monoraphidium contortum (r = −0.72, p < 0.01) and other algae (r = −0.85, p < 0.001) biomasses. It seems likely that dissolved inorganic phosphorus limited the growth of these algae groups. The DIP concentration in the water of the lake was low; organic phosphorus prevailed over inorganic phosphorus at the depths of 10-15 m, indicating the active production of organic matter in these horizons. Based on the data obtained, it can be concluded that the water of Lake Khubsugul was characterised by the lack of inorganic phosphorus for phytoplankton growth.
Most of the biological and hydrochemical parameters were significantly distinct by depth after non-parametric Kruskal-Wallis (KW) test. Boxplots for each factor are arranged in Figure S5 and statistics for KW test are presented in Table S4. According to the implemented Principal Component Analysis (PCA), most of the variations in the data were covered by the two Principal Components (PCs)-81.6%: 1st Principal Component explained 64.8% of the variance and the 2nd Principal Component-16.8% ( Figure 8). As shown on the PCA graph, the greatest variability at depths from 0 to 5 m was marked for "Dinobryon_sociale_B". At depths from 10 to 25 m, "Total_B", "Mychonastes_minusculus_B", "Cyclotella_ocellata_B", "P org ", "TP" and "Chl_a" had the highest variability. Moreover, the last three indicators had a strong positive correlation, because their vectors were oriented in the same direction and had a rather small angle in relation to each other. Most of the biological and hydrochemical parameters were significantly distinct by depth after non-parametric Kruskal-Wallis (KW) test. Boxplots for each factor are arranged in Figure S5 and statistics for KW test are presented in Table S4. According to the implemented Principal Component Analysis (PCA), most of the variations in the data were covered by the two Principal Components (PCs)-81.6%: 1st Principal Component explained 64.8% of the variance and the 2nd Principal Component-16.8% ( Figure 8). As shown on the PCA graph, the greatest variability at depths from 0 to 5 m was marked for "Dinobryon_sociale_B".
At depths from 10 to 25 m, "Total_B", because their vectors were oriented in the same direction and had a rather small an relation to each other. The other indicators, such as "TN", "Ec_25C", "Si", etc. had reliable differen depth according to the KW test, but due to their low variability, made minimal con tion to the overall distribution.

Discussion
For the first time, we detected cyanobacteria that contain genes encoding microc synthetase and produce microcystins in the high-altitude, deep, ancient Lake Khubs which is located in a region with a severe climate and widespread permafrost. The d sity and abundance of planktonic cyanobacteria in the lake is extremely low. Be The other indicators, such as "TN", "Ec_25C", "Si", etc. had reliable differences in depth according to the KW test, but due to their low variability, made minimal contribution to the overall distribution.

Discussion
For the first time, we detected cyanobacteria that contain genes encoding microcystin synthetase and produce microcystins in the high-altitude, deep, ancient Lake Khubsugul, which is located in a region with a severe climate and widespread permafrost. The diversity and abundance of planktonic cyanobacteria in the lake is extremely low. Benthic cyanobacteria are richer in species composition and are more abundant on stone substrates. Our measurements of the water transparency, concentration of nutrients, total nitrogen and phosphorus and chlorophyll a in Lake Khubsugul were consistent with the parameters typical of oligotrophic water bodies [67]. Based on the concentration of organic matter, the water in Lake Khubsugul can be classified as "extremely clean" [87].
In the 1970s to 1990s, extensive information about chemical composition of Lake Khubsugul water was obtained as a result of joint Russian-Mongolian [57,[88][89][90] and other international expeditions [91] that served as a basis for assessing the changes that had been occurring in recent decades. In 2017, the concentrations of mineral and organic phosphorus and organic carbon in Lake Khubsugul were almost the same as the values determined previously [57,[88][89][90][91].
We registered a silica concentration drop in Lake Khubsugul compared to recent decades, which was likely associated with a prolonged low water level in the region. For example, Lake Baikal has recently shown a reduction in the river runoff and a decrease in silica concentration [92]. The water contained insignificant amounts of ammonium and nitrite ions that had not been found in the 1960s to 1990s. The appearance of these inorganic nitrogen compounds-which were first detected in 2011, according to [89]-was due to permafrost melting and/or livestock overgrazing. The concentrations of nitrates, total nitrogen and chlorophyll a during our survey were slightly higher than in the previous years [57,66,89,91]. Water pollution with ammonium ions was detected in the coastal zone of the lake because livestock waste was transferred to the lake by rainwater, as shown in recent studies [93]. At the same time, concentrations of nitrite ion and organic nitrogen in Lake Khubsugul were lower than those in other large oligotrophic water bodies, e.g., in Lake Baikal [94,95].
The established relationships and PCA analysis indicated that phytoplankton biomass and production in the euphotic layer of water was probably limited by inorganic phosphorus; inorganic nitrogen concentration was sufficient to increase phytoplankton biomass during the study period. We still note that diatom Cyclotella ocellata and two species of green algae, especially Mychonastes minusculus, were the main phytoplankton of Lake Khubsugul. Green and diatom algae contributed most to the primary production of the lake; their amount was strictly positively correlated with chlorophyll concentration, the concentrations of total and organic nitrogen and phosphorus. A PCA analysis clearly demonstrated a non-uniform vertical distribution of phytoplankton; the diatoms and green algae preferred 10-25 m depth, where their abundance was strongly positively correlated with the main hydrochemical parameters. In the upper water column (0-5 m), the mixotrophic Chrysophycean algae Dinobryon sociale, not previously reported as an important phytoplankton species, appeared as a key species negatively correlated with chlorophyll a, organic and total phosphorus, organic and total nitrogen concentration.
Based on the data obtained, we can state that the ecosystem of Lake Khubsugul currently does not show signs of eutrophication; hydrophysical and hydrochemical parameters are unfavourable for the mass growth of cyanobacteria. We can also conclude that the trophic state of the lake has not changed over the past 50 years.
In recent decades, tourism has begun to actively develop in the Khubsugul region: recreation centres, tourist camps and holiday homes were constructed on the lake shores. People use the water from Lake Khubsugul for domestic purposes, recreation and drinking. Due to the lack of centralised sewage treatment systems, household wastewater enters the lake with rains or often directly from rustic toilets. However, the number of faecal indicator microorganisms in all studied samples did not exceed the guideline values for these two water use categories [68,69].
According to Kirschner's surface water hygiene classification, the water in Lake Khubsugul in 2017 to 2019 corresponded to "pollution level I-low", i.e., the number of total coliforms was ≤500 in 100 mL, and the number of faecal coliforms was ≤100 in 100 mL of water [93]. Slightly higher coliform counts compared to our values were determined in summer in the lake water near the southern and northern settlements and along the west coast of the lake, but they did not exceed the values typical of low pollution level [93]. The authors indicate that the improvement of sanitation and hygiene conditions in the tourist destinations along Lake Khubsugul contributed to the reduction of the E. coli and coliforms in the lake water. Therefore, the results of the microbiological water quality assessment testified to a low degree of faecal pollution of the water in Lake Khubsugul; in other words, the water quality in the lake can be described as near-pristine.
Our data suggest that faecal microorganisms reach the water body not only with household wastes, but also with the faeces of horses and large and small cattle grazing in large numbers on the lake shore. The results of high-throughput sequencing of the water and biofilm microbiomes indicated the presence of bacteria from the microbiota of the skin and gastrointestinal tract of both humans and animals, including fish (gut microbiome). Along with a low human population density in the Khubsugul region, a high number of animals was recorded: there are ten animals per one Mongolian citizen, and it is very likely that bacterial pollution is generally animal derived. Pastoralism occupies a major position in the agriculture of Mongolia. Previously, the number of livestock in the Khubsugul region was the highest in the country owing to the aimak water reserves; it amounted to 1.6 million animals, which was 7.1% of the total livestock in the country [57,88].
Despite excessive livestock grazing and tourism development, Lake Khubsugul does not experience significant anthropogenic load, partly due to adequately managed economic activities on its shores. In 1992, the lake and the adjacent territory were included in Khuvsgul Lake National Park (KLNP) in an area of 11,800 km 2 . The KLNP supports relatively small human populations: the area around the lake (Khubsugul region) is the most sparsely populated compared to other large lakes in the world [88]. KLNP serves as one of the largest centres of ecotourism in Mongolia [96].
Based on personal data and the literature, we estimate that anthropogenic stressors on the lake ecosystem are minimal.
Researchers are increasingly reporting that, in addition to the excessive supply of nutrients from watersheds, global climate change is another important factor that contributes to the intensive proliferation of cyanobacteria, leading to surface water warming and ensuing thermal stratification [3][4][5]7,8,22,[45][46][47].
Lake Khubsugul environments are evidently involved in global climate changes: warming rates in Mongolia even exceed the global average ones, while northern Mongolia is the fastest-warming region of the country [97,98]. Tree-ring-based reconstructions of heatwaves and soil moisture for the past 260 years revealed a trend toward warmer and drier conditions in inner East Asia, including Mongolia [98]. From 1963 to 2002, the mean annual air temperature near station Khatgal located at the southern end of the lake rose by 1.7 • C, specifically in winter-by 3.1 • C, in spring-by 2.1 • C, in summer-by 1.4 • C, and in autumn-by 0.9 • C [96,97,99]. From 1991 to 2017, the mean annual air temperature at station Khatgal increased by 0.80 • C and amounted to −3.7 • C, and over the past decade it rose by 0.24 • C [93]. In recent years, permafrost melting in the region has contributed to the lake area expansion [96].
We assume that present global climate change affects the ecosystem of Lake Khubsugul even more than regional anthropogenic processes, firstly caused by intensive pastoralism and secondly by human activity.
Microscopic examination and genetic analysis of photoautotrophic communities revealed, as in previous years, dominance in the composition of rare but not endemic diatoms and green algae species. For the first time over the past 20 years, we have assessed the diversity of benthic cyanobacteria in Lake Khubsugul as species-poor, including nine species and 18 phylotypes of benthic cyanobacteria. Rivularia coadunata and Calothrix parietina, potentially toxic species, predominated among them. R. coadunata is a cosmopolitan species, the strains of which can synthesise MCs [6]. MC-producing isolates of C. parietina were described in water bodies of Spain, Saudi Arabia, Egypt and New Zealand. Moreover, this species exhibited neurotoxic activity [6,100,101]. In our study, we did not detect the species of the genus Tolypothrix-including the endemic T. mongolica that was described previously-in the rocky littoral zone at depths of between 0.2 and 10 m [65]. Cyanobacteria belonging to the order Gloeobacterales were first identified in the benthos of Lake Khubsugul. The species composition of benthic cyanobacteria in Lake Khubsugul was similar to that in Lake Baikal in 2010 and 2011, before the deterioration of the environmental situation, when Rivularia, Tolypothrix and Chamaesiphon spp. prevailed in the biofilms of stony substrates [102]. In general, we found that the cyanobacterial communities in plankton and benthos of Lake Khubsugul did not demonstrate signs of uniqueness and/or endemism.
MC synthetase genes were identified only in three genera: Microcystis, Nostoc and, possibly, Snowella. The presence of toxigenic Microcystis sp. was anticipated in Lake Khubsugul: in the water bodies of Northern Mongolia, this cosmopolitan species was found in well-heated areas enriched in nutrients supplied by river and coastal runoff [103]. Species of the genus Microcystis, typically M. aeruginosa and M. flos-aquae, are the most common causative agents of algal blooms and the most well-known MC producers in highly productive water bodies throughout the world [27]. Previously, the members of the genus Microcystis were absent in the list of algae and cyanobacteria of Lake Khubsugul. We also did not identify the Microcystis species in the study period. I representatives of this genus can probably be found sporadically in the samples.
Sequences related to Snowella sp., the second group of toxigenic cyanobacteria in Lake Khubsugul, formed a separate cluster likely representing the order Synechococcales, which also included uncultured Baikal sequences [85] and the sequences of the Snowella sp. CHAB strains (China) (unpublished data). It is noteworthy that all toxigenic Snowella-like sequences were detected in Asia. We did not identify other registered sequences of microcystin synthetase genes that reliably belonged to the order Synechococcales, although the information about the MC production by the species of this order is available. For example, the presence of mcyB gene was detected in 12 strains of cyanobacteria: Synechocystis (three strains), Synechococcus (six strains) and Romeria (two strains), but the sequences in Genbank are missing [104].
The genus Snowella includes seven species of freshwater unicellular and colonial cyanobacteria, two of which (S. rosea and S. lacustris) were detected in Lake Khubsugul. There were no MC synthesis genes in the Snowella ULC335 genomes from microbial mat (Belgium) and Snowella Erken-D5_bin-0612 from lentic water body in Sweden. However, it should be taken into account that the sequences from Lake Khubsugul may also belong to the closely related genera of the family Coelosphaeriaceae, e.g., Coelosphaerium kuetzingianum living in the lake.
The Nostoc mcyE gene sequences from Lake Khubsugul were closely related to lichenassociated Nostoc strains isolated from the samples of a moss carpet at Keldur, Reykjavik, Iceland: mostly to Nostoc sp. 232 "Peltigera membranacea cyanobiont", a Peltigera membranacea lichen photobiont [105] and, to a lesser extent, to Nostoc sp. 113 "Peltigera membranacea cyanobiont". Nostoc sp. 113 was more identical to Nostoc sp. 152, a well-studied MC producer isolated from the water bloom sample in Lake Saaksjarvi, Finland [106,107].
It seems most likely that the aminotransferase domain of the microcystin-synthetase gene clusters of the Nostoc sequences from Lake Khubsugul belongs to symbiotic cyanobacteria. Nostoc is a cosmopolitan cyanobacterial genus occurring in both terrestrial and aquatic ecosystems, especially in the temperate and cold regions of the world. Strains of the genus Nostoc are the most common cyanobacteria in various symbioses [108]. The members of the order Nostocales and the family Nostocaceae are the most active in terms of general secondary metabolite production. They can synthesise about 26% of the total known metabolites produced by filamentous cyanobacteria [109].
Five MCs congeners were identified in the integrated sample of biofilms from Lake Khubsugul using the HPLC-HRMS/TOF equipment: MC-YR, MC-LAba, MC-LY, [ADMAdda 5 ]MC-LR and [ADMAdda 5 ]MC-XR. MC-YR is one of the most common MC congeners in freshwater ecosystems all over the world. It is produced-along with the MC-LAba and MC-LY congeners-by members of the genus Microcystis, mainly Microcystis aeruginosa. The toxicity of these three congeners is low [25,27].
The production of highly toxic [ADMAdda 5 ]MC-LR is typical for planktonic Nostoc sp. 152 [107] and terrestrial Nostoc sp. strain IO-102-I isolated from a lichen association inhabiting rocks in Finland. In the former strain, its proportion reaches 35%, and 80% of the total MCs in the latter strain [110]. The Nostoc sp. strain IO-102-I also synthetises the minor [ADMAdda 5 ]MC-XR that was also detected in the biofilms from Lake Khubsugul.
The [ADMAdda 5 ]MC-LR congener was found in microbial mat samples of the Svalbard Archipelago in the Arctic where Nostoc sp. was identified as a putative toxin producer using molecular methods that targeted genes involved in microcystin production [111]. Sequences of Nostoc sp. clones didn't join the cluster formed by Khubsugul isolates. Based on microscopic, genetic and mass spectroscopy analyses, we assume that Nostoc sp. may be one of the MC producers. The detection of MCs only in benthic samples may be associated with a greater biomass of cyanobacteria in biofilms, unlike planktonic samples, in which the biomass of cyanobacteria was low even in net samples.
The MC concentrations in biofilms of stone substrates from the coastal zone of Lake Khubsugul were low when analysed by ELISA (35.4 to 52.0 µg g −1 d. wt.), and by HPLC-HRMS/TOF (integrated sample, ca. 55.8 µg g −1 d. wt.). The total concentration was much lower than in cultured benthic species (1600 to 4100 µg g −1 d. wt.) [101] and in reservoirs of the Segura River basin, Murcia, Spain (300 to 3300 µg mg −1 , mean 1600 µg g −1 d. wt.) [100], but higher than in cyanobacterial mats from Antarctic water bodies (1 to 16 µg g −1 d. wt.) [112] and in Arctic freshwater ecosystems (0.10 µg g −1 d. wt.) [113]. It is impossible to assess the real threat of mass development of benthic toxic cyanobacteria due to the absence of international guideline values for the MC content in their biomass [112,114].
The microcystin synthetase gene cluster is ancient. Phylogenetic reconstructions revealed that housekeeping genes, 16S rDNA and rpoC1 genes, as well as microcystin synthetase genes, co-evolved throughout the whole evolutionary history of this toxin [44]. The sporadic distribution of microcystin synthetase genes in modern cyanobacteria suggests that the ability to produce MCs may have been lost in more advanced cyanobacterial lineages. On the other hand, cyanobacterial strains that have not previously been suspected of producing MCs may retain the genes necessary for their synthesis. Future investigations of different extreme and poorly accessible habitats, including low-productive ecosystems, will possibly detect toxic species there, updating the knowledge about their distribution. Lake Khubsugul is young (5.5 million years old) relative to the age of cyanobacteria, which is 3.5 billion years old. Apparently, cyanobacteria with their ancient toxin synthesis genes appeared in the area of modern Lake Khubsugul long before the appearance of the lake itself, which is also reasonable for other ancient lakes, Lake Baikal in particular. The Khubsugul and Baikal basins at an early stage of the Baikal Rift Zone formation (70 to 30 Ma) were a single Archeo-Baikal system [115,116]. The active uplift of the western side of the Khubsugul depression and the acquisition of its modern features took place between 3.3 and 2 Ma, and about 8.5 Ka, when the water level in the lake became close to the modern level [117].
Baikal is the largest, deepest and most ancient (20-30 million years old) freshwater lake on the planet and is older than Lake Khubsugul [118]. Many common features in these two geographically close lakes connected by a system of rivers, including old age and common origin, similar morphological and morphometric parameters, climatic conditions, relative isolation, etc., provoked the researchers to compare Baikal and Khubsugul biota [56,57,88,91]. The authors point out significant differences in the composition, abundance and seasonal dynamics of planktonic and benthic communities caused by hydrophysical and hydrochemical specifics: altitude above sea level, solar irradiance, water balance in the watershed basin, etc.
In our case, we found that, despite significant differences in the species composition of cyanobacteria from the two related lakes, the same cosmopolitan species represented toxigenic bacteria in them, as in most water bodies on the Earth. Previously, we reported on the Microcystis, Dolichospermum and Snowella-like cyanobacteria that contained MC synthetase genes [20,85], as well as on saxitoxin-producing Nostoc [119] in Lake Baikal. Since 2010, toxic Dolichospermum and Microcystis producing MCs in low concentration, presumably associated with the global temperature rise, were detected in the coastal zones and bays of Lake Baikal [20]. In other words, the toxigenic cyanobacteria in these lakes are cosmopolitan, which confirms the old age of the microcystin synthetase cluster.
The mass development of filamentous algae, death of endemic sponges, intensive growth and colonisation of toxic cyanobacteria on various substrates have been observed in Lake Baikal over the past decade. These features have become an "environmental crisis" caused by increased anthropogenic pressure, namely, the excessive supply of nutrients with rains and a deep influx of groundwater from the lake shores [120]. However, the present trophic state of the lake is oligotrophic based on hydrochemical parameters; signs of eutrophication were identified in the coastal zone of some areas of the lake [95]. Our observations and long-term monitoring of phytoplankton showed that the abundance of planktonic cyanobacteria in Lake Baikal was lower than 20 million cells L −1 not reaching the dangerous threshold [30]. At the same time, vegetation of the species Dolichospermum with a maximum abundance of up to 10 million cells L −1 has been regularly recorded in July and August since the 20th century [121]. Excessive growth of Dolichospermum cyanobacteria (D. lemmermannii as the dominant species) was first reported in the littoral zone on the western shore of the southern basin at the end of July and the beginning of August in 2019. The abundance of cyanobacteria in blooming spots varied within 7.2-72 million cells L −1 , with 0.73-7.20 g m −3 of biomass attained [16].
Information about cyanobacterial blooms in large oligotrophic ecosystems ("oligotrophic blooms") has appeared relatively recently; in some lakes, researchers observed mass development of mainly nontoxic diazotrophic cyanobacteria [14,15,[122][123][124][125]. Gloeotrichia echinulata blooms were recorded in oligotrophic and mesotrophic lakes in the northeast of the United States [122], and prolific growths of Dolichospermum, Aphanizomenon, Microcystis and Gloeotrichia were observed in low-productive lakes in the province of Ontario [123]. The largest lakes of the Alps (Garda, Iseo, Como, and Maggiore) were progressively colonised by the species D. lemmermannii [124,125]. In the oligotrophic Lake Superior, often referred to as the most pristine of the Laurentian Great Lakes, the mass development of D. lemmermannii [14,15] was experienced. Recently, the North Patagonian lakes, notwithstanding their low nutrient concentration and productivity, also demonstrated cyanobacterial blooms, mainly constituted by toxigenic Microcystis and Dolichospermum species [126].
In Lake Khubsugul, we detected D. lemmermannii that had not been previously found there. It seems an intriguing finding presaging mass development of this species in the lake. The present survey did not reveal toxic genotypes of this species, apparently due to its low abundance (689 cells L −1 ). In this respect, further search for toxigenic Dolichospermum in the lake is needed. The occurrence of toxigenic Dolichospermum in Lake Khubsugul may be regarded as a trend for toxic algal blooms similar to Baikal events.
Under favourable conditions, toxigenic cyanobacteria pose a potential risk of producing cyanotoxins in concentration hazardous for humans and animals. Identifying the diversity of bloom-causing cyanobacterial species and revealing associated factors that stimulate these blooms are useful tools to predict and control them.
In Lake Khubsugul, anthropogenic eutrophication is one of the most controlled factors contributing to the mass development of cyanobacteria, with the appropriate regulation of economic, tourist and recreational activities, as well as animal husbandry, but it should not remain a major issue in the near future. The prevailing influence of global warming allows us to use the ecosystem of Lake Khubsugul as a model for studying the effect of climate change on the lake biota. It is assumed that the unique ecosystem of Lake Khubsugul would be affected by long-term warming and drying with a possible succession of hydrobionts species composition.

Conclusions
Based on our data on transparency, concentrations of mineral and organic compounds of phosphorus and nitrogen, and chlorophyll a, Lake Khubsugul may be attributed to oligotrophic lakes. The water in Lake Khubsugul is classified as clean in terms of faecal indicator microorganisms and the concentration of organic matter. The analysis of microbiomes showed small amounts of pathogenic and opportunistic bacteria of anthropogenic origin in the lake. In microbial communities, we identified the sequences of pathogenic and opportunistic bacteria of animal origin. The genetic and taxonomic diversity of cyanobacteria in the plankton and benthos of Lake Khubsugul is small, and the abundance of planktonic cyanobacteria is low. No unique and endemic species were encountered, mostly just cosmopolitan species. Diatoms and green algae dominated phytoplankton; picoplanktonic genera of the order Synechococcales prevailed among cyanobacteria, and members of the order Oscillatoriales predominated in the benthos. Synechococcales were subdominant, while Chroococcales and Nostocales were minor. We detected cyanobacteria containing MC synthetase genes, which belonged to the genera Microcystis, Nostoc and possibly Snowella in the plankton from the coastal zones of Lake Khubsugul. Cyanobacteria inhabiting the lake bottom produced five MC congeners, both common and rare. Toxins were not found in the pelagic plankton. MC concentration in benthic cyanobacteria from Lake Khubsugul is much lower than in other water bodies. At present, we suggest that anthropogenic loads on the lake ecosystem are minimal with no signs of cyanobacterial blooms. In the future, with slow changes in the global climate, considering the preservation of the current trophic state of the water body and relatively low anthropogenic activity, the risk of cyanobacterial blooms and toxin production is minimal, even in the presence of toxigenic genotypes and toxin-producing members. Apparently, there is always a potential source of occurrence and spread of toxic blooms in ancient oligotrophic lakes due to the existence of cyanobacteria containing toxin synthesis genes. In the future, under conditions favourable for mass vegetation of cyanobacteria, toxic blooms are possible. The presence of toxigenic and toxic cosmopolitan cyanobacteria in the ancient and oligotrophic Lake Khubsugul confirms the paleo origin of MC synthesis genes that appeared even before the formation of the lake itself.

Site Description
Lake Khubsugul is located in Central Asia on the territory of Mongolia. It lies between 50 • 30 and 51 • 35 N, 100 • 15 and 100 • 40 E. The basin of Lake Khubsugul is an extreme southwestern wing of the Baikal Rift Zone system, one of the largest rift zones on the Earth. Khubsugul is an ancient lake: the age of the sedimentary cover of the lake depression is 5.5 million years old [117]. The lake is situated at an altitude of 1645 m above sea level; its basin is trough-shaped with an area of 2760 km 2 , average width of 20.3 km, maximum width of 36.5 km, length of 136 km and water volume reaching 383.3 km 3 [88,91]. The average depth of the lake is 139 m, and the maximum depth is 262 m. The pelagic zone occupies about 85% of the lake surface, and the littoral zone with a depth down to 50 m-15%.
The climate is continental, with extreme ranges of daily and annual air temperatures and a small amount of precipitation. The average annual air temperature is below zero (−4. . . −5 • C); the annual range of average monthly temperatures on the lake coast is 30-35 • C. In summer, the air temperature rises to 11-14 • C and changes during the day from 5 to 16 • C. The ice cover breaks in June, and the lake freezes in late November or early December; the ice thickness reaches 1.5 m [88]. The water in Lake Khubsugul is low-mineralised (210 to 230 mg L −1 ) and belongs to the bicarbonate class of calcium group, type 1. It has an alkaline reaction (pH 8.1 to 8.5), high oxygen concentration from the surface to the bottom in all seasons (8.10 to 11.43 mg L −1 ) and low CO 2 in any season.
Forty-six rivers and streams flow into the lake, and only the Egiin Gol River flows out of the lake, which is a tributary of the Selenga River flowing into Lake Baikal. Therefore, there is a direct water connection between the two largest lakes in Central Asia. The amount of precipitation in the Khubsugul region is low: the maximum precipitation is in summer (100 to 300 mm), which is 60 to 70% of the annual amount.

Sampling
Water and biofilms from stones were sampled from 11 to 14 July 2017 in the northern part of Lake Khubsugul: on the eastern coast off the Khankh settlement and close to the northern margin of the lake (Figure 9). The water was sampled at four stations in the pelagic zone from the surface to 25 m depth (1N and 11-13Kh) and at ten stations in the littoral zone at 0 m (1H to 10H). The surface water was sampled with sterile bottles; water from the depths of 5, 10, 15 and 25 m was sampled with a Ruttner bathometer. Net sampling was also carried out using an Apstein net with a mesh size of 16 µm. The net was towed three to five times from depths of 1 to 5 m until a concentrated sample was obtained.
Biofilms formed on the stones were taken together with the stones from the lake bottom at a depth of 2 to 3 m (stations 2N and "Stones") and transferred to the laboratory. The biofilms were then scraped from the surface of the stones with a sterile spatula and placed into sterile test tubes.
For the microscopic examination of phytoplankton, including cyanobacteria, samples were prepared by two methods. First, the samples were fixed with Lugol solution, concentrated by sedimentation and observed using light microscopy. Second, the samples were fixed with 4% formaldehyde (final concentration), filtered through polycarbonate filters with a pore size of 0.4 µm (Millipore, Burlington, MA, USA) and examined via epifluorescent microscopy. The net plankton and biofilms for light microscopy were also fixed with 4% formaldehyde. For ELISA, mass spectrometry and genetic analysis, net samples were frozen in liquid nitrogen. For genetic studies and determination of the chlorophyll concentration, unfixed samples of 1 L each from each layer were filtered in duplicate through polycarbonate filters with a pore size of 0.2 µm (Millipore, Burlington, MA, USA) and frozen. For microbiological analysis, the water samples were taken according to the ISO standard [127]. Simultaneously with sampling, water transparency was determined using a Secchi disk, and the temperature and pH were determined using a portable pH meter (Hanna Instruments, Woonsocket, RI, USA).

Sanitary-Microbiological Analysis
An assessment of the water quality in Lake Khubsugul was carried out according to the hygienic standards established by the sanitary rules and norms and methodological recommendations of the Russian Federation [68][69][70]. We determined the number of total coliforms, thermotolerant coliforms, enterococci, Escherichia coli and coliphages (Table S2).
The water quality was assessed based on the category of the water body. According to sanitary rules and regulations, water use category I includes water bodies or their areas used for drinking and domestic purposes, as well as for the water supply of food industry enterprises [69]. Water use category II includes water bodies or their areas used for recreational purposes, as well as localities within the settlement boundaries (Table S3).

Microscopy
Quantitative and qualitative analysis of phytoplankton and cyanobacteria was carried out using an Axio Imager M1 light microscope (Carl Zess, Jena, Germany), equipped with a BO 50 mercury lamp, three sets of filters (G 365, BP 450-490 and BP 546/12) and an AxioCam MRm camera. Algae and cyanobacteria were counted in duplicate in a 0.1 mL Nageotte chamber with magnifications of ×400 and ×1000. Biomass was calculated from algal abundance and individual cell volumes. To determine the biovolume, 30 cells of each species were measured. In case of using the filtration method and epifluorescent microscopy, cells were counted on 20 photomicrographs with the Image-Pro Plus 6.0 software (Media Cybernetics Inc., Silver Spring, MD, USA). Cyanobacteria were identified based on the autofluorescence of phycobilins, additional photosynthetic pigments, which were observed under a green optical filter (wavelength 540 nm). Identification was carried out according to the manuals [128][129][130].

Molecular Genetic Analysis
Total DNA was extracted from the samples using the DNA-sorb B kit, according to the manufacturer's protocol (Central Research Institute of Epidemiology of Rospotrebnadzor, Moscow, Russia).

Microscopy
Quantitative and qualitative analysis of phytoplankton and cyanobacteria was carried out using an Axio Imager M1 light microscope (Carl Zess, Jena, Germany), equipped with a BO 50 mercury lamp, three sets of filters (G 365, BP 450-490 and BP 546/12) and an AxioCam MRm camera. Algae and cyanobacteria were counted in duplicate in a 0.1 mL Nageotte chamber with magnifications of ×400 and ×1000. Biomass was calculated from algal abundance and individual cell volumes. To determine the biovolume, 30 cells of each species were measured. In case of using the filtration method and epifluorescent microscopy, cells were counted on 20 photomicrographs with the Image-Pro Plus 6.0 software (Media Cybernetics Inc., Silver Spring, MD, USA). Cyanobacteria were identified based on the autofluorescence of phycobilins, additional photosynthetic pigments, which were observed under a green optical filter (wavelength 540 nm). Identification was carried out according to the manuals [128][129][130]. To search for and identify cyanobacteria containing microcystin synthesis genes, we used primers on the genes encoding microcystin synthetase (mcy). HepF and hepR primers were applied to detect the AMT domain of the mcyE gene [86]. The DNA of the toxinproducing strain, Microcystis aeruginosa CALU 972 (Botanical Institute collection, St. Petersburg State University, St. Petersburg, Russia), was used as a positive control. The amplicon length was 470 bp. The sequences were deposited in GenBank under accession numbers OM810332-OM810346.
Amplification and cloning were performed as described previously [19,20]. Determination of the nucleotide sequences of the cloned fragments (sequencing) according to the Sanger method was performed on a Beckman CEQ 8800 automatic sequencer (Beckman Coulter Inc., Brea, CA, USA). Unique sequences assigned as the closest relatives were included in the phylogenetic analysis. Amino acids were aligned in the Mega X software using the ClustalW algorithm [131]. A phylogenetic tree was constructed through Bayesian analysis using the MrBayes software (v. 3.2.6) [132]. Two independent Markov chain Monte Carlo (MCMC) analyses were launched for 25 million generations with 25% burn-in (rejection of initial generations) and four chains (one cold and three hot ones); the amino acid model was blosum62. The analysis was completed with an average standard deviation of the split frequencies: 0.02; the parameter potential scale reduction factor (PSRF+) was 1.000.

Enzyme-Linked Immunosorbent Assay
The concentration of toxins in samples was employed by ELISA using the Abraxis Microcystins-ADDA ELISA kit (Abraxis LLC, Warminster, PA, USA), according to the manufacturer's instructions. Prior to ELISA testing biofilms collected from stony substrates (14 samples) and water with phytoplankton (20 samples) were lyophilised and then resuspended in water. Cyanobacterial material was lysed with QuickLyse (Abraxis LLC) and centrifuged to provide clarified extracts for ELISA. The results were processed with the RIDA ® SOFT Win software.

High Performance Liquid Chromatography-High Resolution Mass Spectrometry
Cell disintegration was carried out by a freeze-thaw cycle method, followed by grinding in a mortar after lyophilisation. MCs fraction was extracted from lyophilisate twice with 75% methanol for 60 min under sonication. The resulting extracts were combined, evaporated to dryness on a rotary evaporator at a temperature of 45 ± 2 • C and redissolved in methanol [18].
MCs were identified using the HPLC-HRMS/TOF system (Agilent 1200 HPLC chromatograph coupled with Agilent 6210 mass-spectrometer, Santa Clara, CA, USA). MCs were separated on a 150 × 2 mm Zorbax 300 SB-C18 column, using aqueous 0.1% formic acid (mobile phase A) and 0.1% formic acid in acetonitrile (mobile phase B). Gradient elution from 10% to 100% organic mobile phase for 40 min at 0.12 mL min −1 flow rate was applied. Detection was carried out in ESI+ mode, the m/z range from 100 to 3000 Da [137].

Hydrochemical Analysis
Concentration of the dissolved oxygen was measured using the method of Winkler as described previously [94]. Organic matter, total P and total N were determined in unfiltered water. Si, PO 4 3− , NO 3 − , NO 2 − and NH 4 + were analysed in water samples filtered through mixed cellulose ester membrane filters (Advantec, Tokyo, Japan), with a pore diameter 0.45 µm. Concentration of the nutrients was determined using a spectrophotometer UNICO-2100 (Dayton, NJ, USA) according to [138]: PO 4 3− was identified by the Denigès-Atkins method in modification with tin chloride, NH 4 + was detected by the indophenol method [139] and silica with a method based on measuring the intensity of the yellow colour accompanying the formation of the silicomolybdic acid complex. NO 3 − and NO 2 − concentration was measured by high performance liquid chromatography (EcoNova, Novosibirsk, Russia) with UV detection on an inverse-phase column modified with octadecyltrimethylammonium bromide [94,95]. Total N and P were measured spectrophotometrically after oxidation with persulfate. Permanganate oxidation index [140] and bichromate oxidation demand were used to assess organic substances content [141]. Chlorophyll a concentration was determined as described previously [66].

Statistical Analysis
The correlation matrix among biological and hydrochemical parameters was calculated using Spearman's correlation coefficient. All p-values were adjusted for multiple testing with Benjamini-Hochberg (BH) false discovery rate (FDR) [142]; associations with a BH p-value < 0.05 were considered significant.
The Kruskal-Wallis test with p-value adjustment by the BH method was used to assess statistical difference between depth layers of water samples. Principal Component Analysis was performed on a log10 transformed data matrix of nutrients concentrations. Only factors with p-value < 0.05 were chosen for PCA.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/toxins15030213/s1, Figure S1: number of pair-end reads, OTUs and alpha-diversity indices for bacterial communities in the samples from Lake Khubsugul; Figure S2: unrooted phylogenetic tree obtained with MrBayes, based on the alignment of mcyE gene sequences of cyanobacteria. Sequences from Lake Khubsugul are in bold. Bayesian posterior probabilities are indicated near their nodes. The scale bar shows 0.02 estimated substitutions per site; Figure S3: total MC concentration detected by ELISA in biofilms from stony substrates sampled in the coastal zone near Khankh settlement (st. "Stones"). Error bars represent standard deviation. Figure S4: extracted ion current chromatogram of microcystins from the Lake Khubsugul, integrated sample "Stones"; Figure S5: boxplots for nutrient concentrations in different depth water layers in Lake Khubsugul. Non-parametric Kruskal-Wallis test was executed to assess if there is any significant difference between the average nutrient concentrations in different depth layers of water. Abbreviations: A-abundance, B-biomass; Table S1: cyanobacteria species composition, abundance and biomass in biofilms on stones (n = 8) from Lake Khubsugul; Table S2: faecal indicator microorganisms in the water from Lake Khubsugul sampled on 11 to 14 July 2017; Table S3: guideline values of water safety indicators in surface water according to normative documents of the Russian Federation [68][69][70]; Table S4: results for the Kruskal-Wallis test based on depth factor dependence. Funding: This study was funded by the national government and was carried out within the framework of State Project No. 0279-2021-0015 "Viral and bacterial communities as the basis for the stable functioning of freshwater ecosystems. . . ".