Flourishing in Darkness: Protist Communities of Water Sites in Shulgan-Tash Cave (Southern Urals, Russia)

Abstract

Karst caves, formed by the erosion of soluble carbonate rocks, provide unique ecosystems characterized by stable temperatures and high humidity. These conditions support diverse microbial communities, including wall microbial fouling, aquatic biofilms, and planktonic communities. This study discloses the taxonomic diversity of protists in aquatic biotopes of Shulgan-Tash Cave, a culturally significant site and popular tourist destination, by 18S rRNA gene metabarcoding. Our findings reveal the rich protist communities in the cave’s aquatic biotopes, with the highest diversity observed in Blue Lake at the cave entrance. In contrast, Distant Lake in the depth of the cave was inhabited by specific communities of plankton, mats, and pool fingers, which exhibited lower richness and evenness, and were adapted to extreme conditions (cold, darkness, and limited nutrients). High-rank taxa including Opisthokonta, Stramenopiles, and Rhizaria dominated all biotopes, aligning with observations from other subterranean environments. Specific communities of biotopes inside the cave featured distinct dominant taxa: amoeboid stramenopile (Synchromophyceae) and flagellates (Choanoflagellatea and Sandona) in mats; flagellates (Choanoflagellatea, Bicoecaceae, Ancyromonadida) and amoeboid protists (Filasterea) in pool fingers; flagellates (Ochromonadales, Glissomonadida, Synchromophyceae), fungi-like protists (Peronosporomycetes), and fungi (Ustilaginomycotina) in plankton. The specificity of the communities was supported by LEfSe analysis, which revealed enriched or differentially abundant protist taxa in each type of biotope. The predominance of Choanoflagellatea in the communities of cave mats and pool fingers, as well as the predominance of Synchromophyceae in the cave mats, appears to be a unique feature of Shulgan-Tash Cave. The cold-tolerant yeast Malassezia recorded in other caves was present in both plankton and biofilm communities, suggesting its resilience to low temperatures. However, no potentially harmful fungi were detected, positioning this research as a baseline for future monitoring. Our results emphasize the need for ongoing surveillance and conservation efforts to protect the fragile ecosystems of Shulgan-Tash Cave from human-induced disturbances and microbial invasions.

1. Introduction

Karst caves are often formed by the erosion of soluble carbonate rocks. On the one hand, the stable temperature and high humidity make karst caves attractive to various organisms. On the other hand, the cold environment and lack of nutrients present challenges that require special adaptations. Wall microbial fouling, aquatic biofilms, and planktonic communities are commonly found in karst caves [1,2,3,4]. The invasion of underground cavities can occur by microorganisms living on the surface and in the epikarst zone [5]. However, most cave-dwelling species are not found on the surface, suggesting that underground ecosystems form cave-specific communities [6]. To date, several prokaryotic cave communities that support subterranean ecosystems and participate in the primary production of organic matter have been described in detail [4,7,8,9]. As for protists, which are a crucial part of the food chains [10,11], data on their diversity in karst caves are very limited. The first descriptions of protist diversity in cave communities have been presented by Gittleson and Hoover [12,13] and Hill et al. [14], using culture methods followed by microscopy. In recent years, in addition to classical methods [15,16,17], the use of DNA metabarcoding and metagenomics has become increasingly popular [18,19,20] for studying environmental communities. Although the taxonomic resolution of protists remains challenging in metabarcoding, the V4 region of the 18S rRNA gene is considered a rather reliable taxonomic marker for the discrimination of many protist taxa, especially high-rank ones [21,22].

Historical artifacts and the exhilarating atmosphere of the caves make them the subject of tourist attention. However, tourism-related anthropization leads to changes in microclimate and intense growth of allochthonous microorganisms forming spots and biofilms, e.g., in Petralona Cave (Chalkidiki, Greece) and Lascaux Cave (France) [20,23]. Moreover, the equipment of tourist routes with artificial lighting encourages the development of lampenflora, which is uncharacteristic for caves [20]. The term “lampenflora” (also known as lamp-flora or lamp flora) was introduced by a botanist Klaus Dobat in the 1960s [24] and refers to all phototrophic organisms (cyanobacteria, algae, plants) growing in permanently illuminated areas. Frequent visits to caves cause an increase in air temperature and the introduction and development of extraneous microorganisms, including potential pathogens. The development of bacteria and fungi in Paleolithic caves can result in stained spots formation on the cave walls and damage Paleolithic art due to microbial alterations [23,25]. The conservation of these fragile, unique cave ecosystems requires their close study and monitoring.

Shulgan-Tash Cave, located in the Southern Urals (the Republic of Bashkortostan, Russia), is a well-known cultural heritage site and a popular tourist destination [26,27]. Studying and monitoring the microbial communities of Shulgan-Tash Cave is an important task not only to characterize the underground ecosystems but also to assess human-induced threats to them. We have previously described the biodiversity of prokaryotic communities of wall fouling in Shulgan-Tash Cave [4,28,29]. In this work, we applied the method of metabarcoding based on the 18S rRNA gene to describe the general taxonomic diversity of protists in aquatic biotopes of the cave. We also performed a comparative analysis of the described subterranean communities with communities of Blue Lake, a similar water site on the surface, in the immediate vicinity of the cave.

2. Materials and Methods

2.1. Cave Site Description and Sampling

Shulgan-Tash Cave (also known as Kapova Cave) is located within the territory of Shulgan-Tash Nature Reserve (53°02′ N, 57°03′ E) in the Tirmentau massif (the height of the massif is 420 m.a.s.l., and the entrance to the cave is 280 m.a.s.l.) (Figure 1).

Distant Lake (Russian name “Bepxнee Дaльнee”, spelled “Verhnee Dalnee”) is a low-flow groundwater body located 700 m from the cave entrance, at an elevation of 305 m.a.s.l. and lying 90 m below the earth’s surface (Figure 1A). The lake extends 29 m from south to north, is 11 m wide, and has a maximum depth of 1.65 m [30]. The lake is recharged by karst water from a vadose zone. A permanent inflow comes from a dead-end fissure passage located 10 m from the northern shore (Figure 1B). The magnitude of the inflow ranges from 50 to 200 mL/sec during dry periods and from 1 to 2 L/sec during the snowmelt season. A small creek flows out of the lake at the southern shore and is absorbed by loose sediments after 20 m (Figure 1B). The vadose water of Distant Lake, when infiltrated by the outflowing stream, recharges the phreatic water of the Blue Lake (Russian name “Гoлyбoe”, spelled “Goluboe”) hydrosystem located at the entrance of the cave [31]. Vaucluse-type karst spring Blue Lake is characterized by a flow rate ranging from 230–240 L/sec during low-water periods to more than 3000 L/sec during rain floods and snowmelt. The dimensions of this lake-spring are 37 by 10 m, and the maximum depth of the phreatic zone channels reached by divers exceeds 80 m.

Figure 1. Map of the research area and Shulgan-Tash Cave sampling sites: (A) Overview map showing the global position of the research area. (B) The position of the cave in relation to the Southern Urals. (C) Digital model of the cave combined with the relief of the Tirmentau massif. Visualization was created using Cloud Compare 2020 software with laser scanning point clouds [32]. The cavities of the phreatic zone are shown schematically (by Snetkov E., unpublished data). (D) Sampling scheme of Distant Lake and its watercourses. PF—pool finger biofilm; CEW—lake water at the point of inflow into Distant Lake; WPF—water surrounding pool fingers; WDL—water in the middle of Distant Lake; MDL—mats of Distant Lake; BL—water from Blue Lake.

Figure 1. Map of the research area and Shulgan-Tash Cave sampling sites: (A) Overview map showing the global position of the research area. (B) The position of the cave in relation to the Southern Urals. (C) Digital model of the cave combined with the relief of the Tirmentau massif. Visualization was created using Cloud Compare 2020 software with laser scanning point clouds [32]. The cavities of the phreatic zone are shown schematically (by Snetkov E., unpublished data). (D) Sampling scheme of Distant Lake and its watercourses. PF—pool finger biofilm; CEW—lake water at the point of inflow into Distant Lake; WPF—water surrounding pool fingers; WDL—water in the middle of Distant Lake; MDL—mats of Distant Lake; BL—water from Blue Lake.

The waters of Distant Lake and its watercourses have HCO₃-Ca major ions, a stable TDS (Total dissolved solids) of 4000–4450 ppm, a pH close to neutral (6.5–7.5), and an almost constant temperature of 6.2–6.4 °C. A notable feature of the atmosphere over the lake is the elevated CO₂ concentration, ranging from 0.8–1.2% near the northern shore to 2.2% near the southern shore throughout the year. The waters of the Blue Lake karst spring are characterized by HCO₃-Ca composition, TDS 0.26–3.16 g/L, and a slightly alkaline reaction (pH 6.5–7.9). The water temperature varies from 2 °C to 9 °C (unpublished data).

Active biogenic-mineral subaquatic stalactoids, known in the literature as pool fingers [33], are found 5–10 cm below the water surface along the west and east shores. Additionally, cascades of calcite gyre dams are formed along the outflowing creek, within which filamentous littoral microbial mats develop (Figure 2).

The samples for metabarcoding were collected in July 2018. Sampling sites are marked in Figure 1 and included the following locations: PF—biofilm on the terminal part of pool fingers (stalactoids immersed in water); MDL—littoral microbial mats of Distant Lake; CEW—lake water at the point of inflow into Distant Lake; WPF—water surrounding pool fingers; WDL—water in the middle of Distant Lake; BL—water from Blue Lake (Figure 2). MDL biofilm samples were collected using a mechanical single-channel pipette (100–1000 µL) with sterile filtered pipet tips and placed in sterile tubes (1.5 mL) with 1 mL RNAlater solution (Invitrogen, Carlsbad, CA, USA) (Figure 2b). PF samples were collected by plucking the tip of subaquatic stalactoids. Biofilm samples were collected in quadruplicate. All water samples were collected from a depth of 10 cm. Water samples of 1.5–3 L were collected in sterile bottles, brought to the surface, and filtered through a 0.22 µm membrane immediately after collection. The membranes were cut into four pieces equal in size, which were then placed in RNAlater solution. Two pieces of each filter were used for downstream analysis. All 16 samples (4 PF, 4 MDL, 2 CEW, 2 WPF, 2 WDL, 2 BL) were transported to the laboratory and stored at −24 °C until DNA extraction.

2.2. DNA Extraction, PCR Amplification, and Sequencing

DNA was extracted using the DNeasy PowerBiofilm Kit (Qiagen, Germany) with a FastPrep-24 homogenizer (MP Biomedicals, Solon, OH, USA), according to the protocol provided by the manufacturer. The concentration of DNA was assessed using Qubit (Invitrogen, Carlsbad, CA, USA) and the Qubit dsDNA High Sensitivity Kit (Invitrogen, Carlsbad, CA, USA). The V4 region of the 18S rRNA gene was amplified with the universal eukaryotic UNonMet primers EK-565F (5′-GCAGTTAAAAAGCTCGTAGT-3′) [34] and EUK-1134R-UNonMet (5′-TTTAAGTTTCAGCCTTGCG-3′), which are biased against Metazoa [35] with Illumina i5 and i7 adapters (https://support-docs.illumina.com/SHARE/AdapterSequences/Content/SHARE/AdapterSeq/Nextera/SequencesNextera_Illumina.htm (accessed on 15 may 2024). A negative sterile water control was always performed to confirm the absence of DNA contamination. The libraries were sequenced using a MiSeq v3 Reagent Kit 600 cycles (Illumina, San Diego, CA, USA) on the MiSeq platform (Illumina, San Diego, CA, USA). All DNA manipulations and sequencing were performed at the Joint KFU-Riken Laboratory, Kazan Federal University (Kazan, Russia).

2.3. Bioinformatic Treatment and Data Analysis

Bioinformatic treatment and data analysis were performed at the “Persistence of Microorganisms” Science Resource Center of the Institute for Cellular and Intracellular Symbiosis UrB RAS (Orenburg, Russia).

The quality of sequenced reads was checked using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/, accessed on 1 December 2022). Primer sequences were removed using Cutadapt v. 4.1 [36]. DADA2 v. 1.28 [37] was used for further sequence analysis. Reads were sequentially filtered by quality with a maximum expected error threshold of 4.0 (maxEE = 4) and a minimum length of 250 bp, then merged with a minimum length of 12 bp overlap. Amplicon sequence variants (ASVs) were generated with the pseudo-pooling option, and chimera sequences were removed. ASVs were additionally filtered for a minimum length of 400 bp and a maximum length of 520 bp. The obtained ASVs were taxonomically classified using a pre-trained naive Bayes classifier, which was trained on the PR2 database v. 5.0.0 [38]. Taxa with low support were additionally manually checked against the NCBI nucleotide database using the BLAST algorithm. Non-eukaryotic ASVs and those belonging to Embryophyta, chloroplasts, mitochondria, and Metazoa were removed before further analysis.

All statistical analyses and visualizations were performed within the R environment v. 4.4.0 using phyloseq v. 1.48.0 [39], ggplot2 v. 3.5.1 [40], MicrobiotaProcess v. 1.16.0 [41], and microbiome v.1.26.0 [42] packages. Indices of alpha diversity were calculated using the alpha function from the microbiome R package [42]. The Bray–Curtis dissimilarities and weighted UniFrac distances [43] were used for β-diversity comparison of protist communities based on the ASV distribution and phylogenetic differences, respectively. Permutational multivariate analyses of variance (PERMANOVA) of Bray–Curtis dissimilarities and weighted UniFrac distances were conducted using the “mp_adonis” function in the MicrobiotaProcess [41] package with 999 permutations. Principal coordinate analysis (PCoA) plots of Bray–Curtis dissimilarity or weighted UniFrac distances (based on Hellinger-transformed ASV relative abundances) were constructed using the “mp_plot_ord” function in the MicrobiotaProcess [41] package.

Marker taxa were identified using LEfSe (Linear discriminant analysis Effect Size) analysis [44] on CPM (counts per million)-normalized ASV abundance, with multiple testing correction options (“multigrp_strat”), as implemented in the microbiomeMarker v. 1.6.0 [45] package. This process was set with the Kruskal (p = 0.05) test based on linear discriminant analysis (LDA) effect size (LEfSe) and Wilcox (p = 0.05), corrected with False Discovery Rate (FDR).

2.4. Functional Annotation of ASVs

In this study, we used previously published works [46,47] to assign ASVs into three major functional groups: consumers (Ancyromonadida, Ciliophora, Rhizaria, Obazoa non-Ichthyosporea, CRuMs (Collodictyonida, Rigifilida, and Mantamonadida), Amoebozoa, non-Xanthophyceae and non-Peronosporomycetes Stramenopiles, Centroplasthelida, Telonemia, Malawimonadida, and Diplonemidae), phototrophs (Archaeplastida, Cryptophyceae, and Xanthophyceae), and parasites (Apicomplexa, Peronosporomycetes, Phytomyxea, Perkinsida, and Rozellomycota). Besides, we considered as parasites exclusively predatory protists such as Orciraptor, Viridiraptor, Allantion, and Colpodellida. All other groups of protists, for which the function could not be clearly defined, were classified as “unknown”.

3. Results

In this study, we conducted a comparative evaluation of the diversity, taxonomic composition, and functional composition of protist communities in Distant Lake inside Shulgan-Tash Cave in contrast to Blue Lake located outside the cave. Two types of biotopes, plankton and biofilm (mats and pool fingers), were investigated in Distant Lake. The rarefaction curves formed based on the Chao1 index calculated with ASVs reached a plateau except for BL communities, indicating that the sequencing depth of each sample inside the cave was sufficient (Figure S1).

3.1. Diversity and Taxonomic Composition of Protist Communities

3.1.1. Alpha Diversity

Diversity estimation based on sequencing the V4 region of the 18S rRNA gene revealed fairly rich and diverse communities of protists in water sites of Shulgan-Tash Cave: 1621 ASVs in total, with 450–474 ASVs in the richest community of Blue Lake and 41–60 ASVs in the least rich community of pool fingers (Table S1). The taxonomic richness of communities, assessed using the Chao1 index, in Blue Lake, located in the illuminated site of the cave entrance, was about twice as high as in the inner cave sites (Wilcoxon rank-sum test, p ≤ 0.05; Figure 3). Similar ratios were observed using the Shannon index (diversity assessed with a focus on rare taxa, which depends more on richness than evenness). The Gini–Simpson index (diversity assessed with a focus on common taxa, which depends more on evenness than richness) and Pielou’s evenness (evenness of relative abundances for all ASVs) demonstrated similar patterns of alpha diversity in the protist communities, which were substantially lower even in relatively richer biofilm communities (MDL and PF of Distant Lake) than in more even water communities (Figure 3). A pairwise comparison of the alpha diversity indices for protist communities of the cave biotopes did not reveal significant differences, except for the Chao1 richness (Wilcoxon rank-sum test, p ≤ 0.05) between the communities of MDL and PF sites. Nevertheless, the Chao1 richness values of the communities in CEW, MDL, and WPF sites were higher than those in PF and WDL sites.

3.1.2. Beta Diversity

Principal coordinate analysis based on the Bray–Curtis and weighted UniFrac distances grouped the cave protist communities in three clusters according to the main cave biotopes: inner and outer water sites, cave aquatic biofilms, and pool fingers (Figure 4). Protist communities were confirmed to be associated with specific biotopes by both Bray–Curtis dissimilarity (PERMANOVA; F = 5.03, R² = 0.71, p ≤ 0.001) and the weighted Unifrac algorithm (PERMANOVA; F = 9.46, R² = 0.82, p ≤ 0.001). In general, all protist communities inside Shulgan-Tash Cave differed significantly from the communities of Blue Lake (weighted UniFrac; F = 4.93, R² = 0.26, p ≤ 0.01). Communities of different biotopes in Distant Lake such as plankton (CEW, WDL, WPF), mats (MDL), and pool fingers (PF) also differed significantly from each other (weighted UniFrac; F = 8.97, R² = 0.62, p ≤ 0.001).

Blue Lake protist community featured ASV11 and ASV22, which belonged to an unclassified phototrophic chrysophyte alga Ochromonadales. The cave mat community was characterized by specific ASV2 (unclassified amoeboid photosynthetic or heterotrophic stramenopile of the class Synchromophyceae), ASV7, and ASV18 (both belonged to a heterotrophic cercomonad flagellate Sandona). The latter two ASVs have homologous sequences in GenBank assigned to Sandona hexamutans HQ918174.1 [48] and Sandona pentamutans EU709147.1 [49] with 99% and 100% identity, respectively. ASV3 was a feature of the pool finger protist community and was identified as an unclassified heterotrophic flagellate from the family Bicoecaceae, whereas ASV4 characterized the communities of both pool fingers and mats and belonged to an unclassified heterotrophic flagellate Choanoflagellatea.

Although the protist communities of the water samples from Distant Lake and Blue Lake formed isolated clusters based on the Bray–Curtis dissimilarity, PCoA ordination using phylogenetically weighted UniFrac distances placed all water samples in a common cluster (Figure 4), which confirmed, at the supraspecies level, the close phylogenetic relatedness of the protists living in the water inside and outside the cave.

3.1.3. Taxonomic Composition

A total of 9 supergroups of eukaryotes, 16 divisions, 23 subdivisions, 61 classes, 133 orders, 186 families, and 272 genera were identified, except for unidentified taxa (Figures S2–S8). However, the following discussion focuses only on protist taxa (ASVs) represented by more than 1% of the reads, namely 5 supergroups, 8 divisions, 12 subdivisions, 24 classes, 31 orders, 33 families, and 32 genera (Table S2). The taxonomic structure of the protist communities at the genus level and at higher taxonomic levels demonstrated similar patterns. In all biotopes, the most abundant supergroups were Obazoa, represented primarily by Opisthokonta, and TSAR, mainly represented by Stramenopiles, Alveolata, and Rhizaria (Figures S2 and S3; Table S2). Blue Lake (BL) had a significantly higher proportion of Archaeplastida compared to the other sites (PF, MDL, CEW, WPF, WDL) (Figure S2; Table S2).

Supergroups Haptista, Cryptista, and CRuMs were present in small numbers, each representing less than 1% of the total abundance in all communities (Figure S2; Table S2).

Stramenopiles dominated in all biotopes studied, occupying 26–49% of the total abundance in the cave lake communities and up to 78–81% in Blue Lake (Figure S3; Table S2). Blue Lake was dominated by the mixotrophic flagellates unclassified Ochromonadales (28–32%) and heterotrophic flagellates Spumella (22–23%), whereas in pool fingers, unclassified Bicoecales were the most abundant (24–37%) (Figure 5 and Figures S6 and S8; Table S2). In MDL, Stramenopiles were represented by unclassified Bicoecales, Synchromophyceae, and unclassified Gyrista (Figure 5 and Figures S5 and S6). Diatoms, mainly Melosirales, were found in different samples (Figure 5 and Figure S6; Table S2).

Although Opisthokonta were found in all samples, their proportion was the largest in the communities of the oligotrophic Distant Lake, ranging from 10 to 39% (Figure S3). Moreover, the biofilm communities of mats and pool fingers were characterized by the predominance of sessile heterotrophic flagellates, which belonged to unclassified Choanoflagellatea (7–18%) (Figure 5 and Figures S5 and S6; Table S2).

Fungi were represented by Basidiomycota (Ustilaginomycotina, Pucciniomycotina, and Agaricomycotina), Chytridiomycota, and Ascomycota (Saccharomycotina), occupying a significant proportion in all aquatic communities (11–15% in WDL; 15–20% in CEW; 32–35% in WPF). Surprisingly, fungi-like saprotrophic and parasitic protists Peronosporomycetes (4–14%) and fungi Ustilaginomycotina (6–23% in CEW and WPF only) were the abundant orders in aquatic communities (Figures S5 and S6; Table S2).

The pool fingers community was the only one in which representatives of Filasterea (heterotrophic amoeboid protists) were present as the dominant taxon (Figure S4). Interestingly, Ancyromonadida also constituted a large proportion (4–13%) in the pool finger community (Figure S3; Table S2).

Rhizaria, another representative of TSAR, was predominant in the inner cave communities and differed from the community of Blue Lake in both diversity and genera representation. The presence of Sandona (up to 45%) was a feature of the mat community. Rhizaria were also represented by the genera Neoheteromita, Paracercomonas, and Teretomonas (Figure 5 and Figure S3).

3.2. Protist Taxa with Differential Abundance between Main Groups of Samples

LEfSe analysis was used to identify biomarkers (enriched or differentially abundant taxa) associated with the cave biotopes and an outside Blue Lake (Table S3). In Blue Lake communities, 71 genera had significantly larger proportions in contrast to the cave sites (p < 0.01, FDR, LDA > 2). Obviously, the most represented genera belonged to the photosynthetic members of classes Chrysophyceae (Hydrurus, heterotrophic flagellates Paraphysomonas), Xanthophyceae (Vaucheria), Chlorophyceae (Planophila), and Bacillariophyceae (Planothidium, Cocconeis, Diatoma) (Table S3).

Among cave communities, as expected, planktonic and biofilm (mat and pool finger) communities differed in numerous markers. In the aquatic samples CEW, WDL, and WPF, 70 differentially abundant taxa were identified, among which Alveolata, Cryptophyta, and Discoba were significantly enriched (p < 0.05, FDR, LDA > 4) (Table S3). Ancyromonadida and Evosia were enriched in the mat and pool finger communities. The enriched genera in the cave planktonic communities included Spumella, Segregatospumella, Oikomonas, an unidentified genus of Saprolegniales, Melosira, Wickerhamomyces, Rhodotorula, Colpidium, Malassezia (Figure S9; Table S3). At the same time, taxa enriched in the MDL and PF together included mainly taxa of suprageneric levels, namely divisions Ancyromonadida and Stramenopiles, subdivision Filasterea, classes Choanoflagellatea and Variosea, and family Bicoecaceae (Figure S9; Table S3).

In the communities of the inner cave biotopes (CEW, WDL, WPF, PF, and MDL), 88 taxa with significant differential abundance (p < 0.05, FDR, LDA > 2) were found (Figure S10; Table S3). Generally, in every cave biotope, the specifically enriched division was distinguished: Rhizaria (LDA = 5.1) in MDL, Ancyromonadida (LDA = 4.9) in PF, and Alveolata (LDA = 5.1) in plankton, respectively. At the genus level, the communities of mats were enriched with genera Sandona, Flectomonas, Cercomonas (belonging to the Rhizaria clade), Bracteacoccus, and Archaeorhizomyces. Pool finger communities were enriched with only one taxon of the genus level, Schizoplasmodium; 21 enriched taxa of higher taxonomic levels were found. Segregatospumella, Spumella, Paracercomonas, Cryptococcus, and Pseudocyrtolophosis genera were enriched in the communities of the cave plankton.

3.3. Functional Composition of Protist Communities

We investigated the functional diversity of protists in Blue Lake and inner cave sites, based on the relative abundance of consumers, phototrophic organisms, and parasites (Figure 6 and Figure S11). Out of all, 1118 (69%) ASVs were classified as consumers, 250 (15.5%) as phototrophs, 139 (8.5%) as parasites (including some predatory protists), and 114 ASVs (7%) were assigned to unknown categories due to their unclear functional roles. Consumers were prevalent (84.8%) in the Distant Lake communities, whereas phototrophic and parasitic groups had minor proportions (5.3% and 4.5%, respectively) (Figure 6). Phototrophic protists were prevalent in the plankton community of Blue Lake (55%), although consumers were also widely represented (37.2%). Parasites were equally abundant in communities inside the cave (4.4%) and in Blue Lake (4.5%) and were mostly represented by parasitic Peronosporomycetes (Gyrista) and Rozellomycota (Fungi). Consumers were represented mainly by the supergroup TSAR and its division Stramenopiles. Phototrophic protists were mostly from the division Stramenopiles (TSAR) and from the division Chlorophyta (supergroup Archaeplastida) with a smaller proportion.

The functional diversity of protists in Distant Lake was analyzed in three biotopes, including plankton (CEW, DLL, WPF), pool fingers (PF), and mats (MDL). Consumers were dominant in all biotopes, accounting for 94.5%, 89%, and 72% in PF, MDL, and cave plankton, respectively (Figure S11). Parasitic (11.9%) and phototrophic (11.6%) protists comprised a fairly significant part of cave plankton communities, while in PF and MDL, they constituted only small proportions (0.08% and 0.58% for parasites; 2.2% and 1.1% for phototrophs, respectively) (Figure S11).

4. Discussion

DNA metabarcoding has revealed quite rich protist communities in the water sites of Shulgan-Tash Cave. Among the studied cave biotopes, the richest community was found in Blue Lake located at the cave entrance, whereas the least even communities were in the cave mats (MDL) and pool fingers (PF) (Figure 3). Distant Lake is located in a remote dead-end gallery of the cave, where, apart from the absence of photosynthesis, there is almost no direct influx of air and water from the surface, except for scant infiltration through the thickness of the epikarst and condensate from the upper gallery. Therefore, the higher alpha diversity metrics (richness and Shannon index) for protist communities of Blue Lake compared to the cave sites can be easily explained by the intensive growth of photosynthetic protists (Figure 6), many of which are absent in Distant Lake, and the intake of organic matter and nutrients, as well as other protists, into Blue Lake from surrounding terrestrial habitats, as noted generally for subsurface aquatic habitats [50]. At the same time, water from Distant Lake most likely flows into the underground Shulgan River and then into Blue Lake (Figure 1). This connection suggests the ingress of microorganisms from Distant Lake into Blue Lake, but a comparison of taxonomic lists for both locations revealed only 51 common ASVs, which may be attributed to both transfer through water flow and a poor discriminant power of the V4 region of the 18S rRNA for some closely related taxa of protists (Figure S12). This finding clearly indicates that in the depth of the cave, the specific communities of Distant Lake adapted to extreme conditions (cold, darkness, and substantially limited nutrients) are formed, as noted for other caves [6,19,51]. According to our preliminary data (unpublished), a unique microbial community represented by mats has formed in Distant Lake (Figure 2), with some members possibly acting as primary chemosynthetic producers.

The protist communities in pool fingers (PF) and mats (MDL) were characterized by the least evenness, with a high proportion of a few dominant taxa. Dominant taxa in MDL communities included Sandona, unclassified Synchromophyceae, and Choanoflagellatea, whereas, in PF, they were represented mostly by unclassified Choanoflagellatea and Filasterea, families Bicoecaceae and Sandonidae. All plankton communities inside the cave and outside (Blue Lake) were more even than the biofilm communities because they were richer, and their dominant taxa occupied lower proportions in the communities. In all sites and biotopes, three high-rank taxa of protists, Stramenopiles, Opisthokonta, and Rhizaria, dominated (Figure 4 and Figure S3). Stramenopiles is a rather diverse taxon, including heterotrophic flagellates, chrysophyte and diatom algae, fungi-like protists such as oomycetes and labyrinthulomycetes, etc. Stramenopiles have been revealed to prosper in other caves and groundwaters [3,20,50,52]. Opisthokonta, another megataxon of eukaryotes including heterotrophic organisms such as choanoflagellates, fungi, and multicellular animals (Metazoa), is also known to be dominant in caves and groundwaters [20,23,50]. However, in all mentioned ecosystems, the dominant Opisthokonta were represented mainly by different fungi, such as Ascomycota, Basidiomycota, Mucoromycota, and Chytridiomycota, but not Choanoflagellatea [20,23,50]. The predominance of Choanoflagellatea in communities of the lake mats and pool fingers seems to be a unique feature of Shulgan-Tash Cave. Choanoflagellates have been mentioned as dominants in the cave biotopes only once, namely in the pristine cave Allas in France [23]. However, a single finding of choanoflagellate Salpingoeca amphoridium is known from a plankton sample of the Echo River, a subterranean stream in Mammoth Cave (Kentucky, USA) [53].

Dominant Rhizaria were represented by bacteriotrophic (Sandona, Flectomonas, etc.) and predatory (Allantion, Orciraptor) cercozoan flagellates, which agrees with other observations of Cercozoa predominance in caves and subsurface waters [3,20,23,50,52]. Representatives of the genus Sandona (subdivision Cercozoa) have previously been described as small, rounded, gliding cells that are found ubiquitously in soils [48,49], likely under conditions similar to those of biofilms attached to underground stream beds. Interestingly, Cercozoa dominated ice communities, reaching 72% of total sequences, which confirms their ability to grow under low temperatures in caves [54,55]. Cercozoa also occupied a significant proportion of sediments and biofilms in the hot springs [56,57].

Pool fingers in Shulgan-Tash Cave, unique stalactoid speleothems formed in the cave water, were inhabited by specific communities [30]. The biofilms on the terminal parts of pool fingers were dominated mainly by sessile and gliding heterotrophic flagellates, including Choanoflagellatea, Ancyromonadida, cercomonad Sandona, and heterotrophic amoeboid protists Filasterea. Currently, it is the first study of the protist community of pool fingers, and further investigations are necessary to fully characterize the unique and enigmatic communities of pool fingers and the role of protists in their formation and functioning. Another interesting finding, noted for the first time for caves, seems to be a specific feature of Shulgan-Tash Cave: the predominance of Synchromophyceae in the cave mats. Synchromophyceae, amoeboid stramenopiles, include only three genera inhabiting surface waters: heterotrophic Leukarachnion and algae Chlamydomyxa labyrinthuloides and Synchroma grande [58].

The presence of diatom alga Melosira and green alga Bracteacoccus is not a specific trait of Shulgan-Tash Cave. Previously, algae in the cave were recorded using microscopic methods [59]. Algoflora has been found in almost all available habitats, including caves [15,51]. A number of studies show the possibility of diatom mixotrophy. Particularly, Claus [60] and Abdullin [61] performed growth experiments with algae in a cave and found that some species are able to survive in mineral-rich environments, even in complete darkness. Alternatively, some algae may be found in the caves due to their anthropogenic invasion [23]. At the moment, we have limited suggestions about the algoflora in caves, but as information is accumulating and the most predominant taxa are being identified, it would be promising to search for algae, especially diatoms, in the underground world.

The parasite representation was quite similar to that of other freshwater bodies, where the percentage of parasites was approximately 5% [46,47]. Interestingly, the taxonomic composition of the parasites differed between the communities of Blue Lake and the communities inside the cave. Namely, members of the Rosellomycota class, which infect fungi and green algae [62], dominated among the parasites of Blue Lake. The parasites of Distant Lake communities were mainly presented by predatory members of Glissomonadida and fungi-like protists Peronosporomycetes. The high relative abundances of parasitic fungi Ustilaginomycotina and fungi-like protists Peronosporomycetes seem to be a specific trait of water communities inside Shulgan-Tash Cave, as previous observations have not found these protists in large proportions in other caves [3,20,23,63,64,65,66], although Peronosporomycetes were predominant in groundwaters [50].

Fungi Malassezia and Mortierella (Tables S1 and S2) found in Shulgan-Tash Cave have been recorded previously in other caves [64,65,66]. Cold-resistant Malassezia is associated with animals and is capable of growing at low temperatures [65,67]. It has been shown that fungal outbreaks in caves may be associated with the development of Mucor circinelloides and Cladosporium microsporum in Castañar Cave, Spain [66], and Fusarium solani in Lascaux Cave, France. [68]. These fungal outbreaks in caves clearly demonstrate the fragility of cave ecosystems. In this regard, caves as specific ecosystems require a more reverent attitude. No potentially dangerous fungi were found in Shulgan-Tash Cave in our study, indicating that it can serve as a reference point for future monitoring of the cave.

In conclusion, it is worth emphasizing that the specific composition of the protist communities is obviously determined by different environmental conditions in the water sites inside and outside Shulgan-Tash Cave. Unfortunately, environmental physicochemical parameters were not measured during sampling, but some features of the biotopes studied might be identified easily. The main difference of Distant Lake from Blue Lake is full darkness, constant cold, and substantial limitation of nutrients. However, a high concentration of CO₂ may contribute to the growth of autotrophic prokaryotes capable of chemosynthesis, especially in mats (unpublished data). These prokaryotes, together with others brought by karst water from a vadose zone, are the first component of the simplest food chain in the cave community and serve as food for abundant bacterivorous protists, mainly flagellates and, more rarely, amoebae. Bacteriotrophic protists with different types of swimming behavior and movement were revealed in the cave protist communities, as shown in a microscopic study of protists in aquifers [69]. In mats and pool fingers, they were represented by numerous creeping or gliding flagellates (Sandona, Cercomonas, Flectomonas), sessile flagellates (Bicoecaceae, Choanoflagellatea), and amoeboid protists (Copromyxa, Filamoeba, Synchromophyceae). In the water communities, there was a large abundance of actively swimming flagellates (Spumella elongata) and ciliates (Colpoda, Halteria), as well as flagellates with ambiguous behavior (Segregatospumella, some Bicoecaceae). The bacterivorous protists are in turn consumed by parasitic (Apicomplexa, Peronosporomycetes, Phytomyxea, Perkinsida, Rozellomycota) and predatory (Orciraptor, Viridiraptor, Allantion, Colpodellida) protists. Finally, the remains of organic matter are consumed by fungi (Basidiomycota, Chytridiomycota, Ascomycota) and fungi-like protists (Labyrinthulomycetes). Similar food chains have been found in groundwater and aquifers [50,69]. The food chain of Blue Lake, located in the illuminated area, differed from the food chain of Distant Lake, with a large proportion of phototrophic protists, mainly Ochromonadales and green algae.

We believe that this study enriches our knowledge of protist communities in karst caves. However, it has some limitations that should be noted here. First of all, the method of DNA metabarcoding based on the high-throughput sequencing amplicons of SSU rRNA genes is the best choice for the evaluation of total richness, phylogenetic diversity, and community dynamics of protists [70]. However, this approach provides information on short target regions of the gene that have insufficient discrimination power for the identification of protists at the species and genus level [70]. The identification of protist species and genera using the method of alignment against gene databases and the search for the closest homolog is often complicated due to blurry limits between genetic intra- and interspecific variation for most taxa [71,72]. The deposition of sequences with unreliable and often even unpublished taxonomic descriptions significantly complicates the systematic classification of protist sequences. For instance, some genera with high diversity and complex taxonomic structure, such as Spumella, cannot be identified reliably using current databases due to numerous deposited organisms formerly identified as Spumella, but after a thorough examination, they were described as new genera, e.g., Segregatospumella, Poteriospumella, etc., and species [73,74]. Similar issues occur with certain genera, e.g., Oikomonas, which have never been characterized according to the current taxonomic standards in protistology.

Another limitation of DNA metabarcoding is insufficient compliance of its results with estimates of the number and biomass of living protists [70]. Metabarcoding provides data on the relative abundance of organisms only, whereas for ecological research, the number and biomass of protists are usually the most crucial. These parameters may currently be estimated with direct microscopic observation and size measurement. The inability to distinguish active vegetative cells from dormant cysts, which are especially widespread in soils and aquifers [69], is an issue of metabarcoding that cannot be resolved with genetic methods. However, the most obvious way to resolve the main limitations of DNA metabarcoding involves a combination of this method with RNA metabarcoding, which provides information on metabolically active, rather than dormant, parts of the community, as well as the use of classical microscopic examination and cultivation methods to reveal the number, biomass, and morphological characteristics of protist species [70].