Microbial Community Characterizing Vermiculations from Karst Caves and Its Role in Their Formation

The microbiota associated with vermiculations from karst caves is largely unknown. Vermiculations are enigmatic deposits forming worm-like patterns on cave walls all over the world. They represent a precious focus for geomicrobiological studies aimed at exploring both the microbial life of these ecosystems and the vermiculation genesis. This study comprises the first approach on the microbial communities thriving in Pertosa-Auletta Cave (southern Italy) vermiculations by next-generation sequencing. The most abundant phylum in vermiculations was Proteobacteria, followed by Acidobacteria > Actinobacteria > Nitrospirae > Firmicutes > Planctomycetes > Chloroflexi > Gemmatimonadetes > Bacteroidetes > Latescibacteria. Numerous less-represented taxonomic groups (< 1%), as well as unclassified ones, were also detected. From an ecological point of view, all the groups co-participate in the biogeochemical cycles in these underground environments, mediating oxidation-reduction reactions, promoting host rock dissolution and secondary mineral precipitation, and enriching the matrix in organic matter. Confocal laser scanning microscopy and field emission scanning electron microscopy brought evidence of a strong interaction between the biotic community and the abiotic matrix, supporting the role of microbial communities in the formation process of vermiculations. Supplementary Information The online version contains supplementary material available at 10.1007/s00248-020-01623-5.


Introduction
The hypogean environments are the least known and studied on Earth [1]. Despite the prohibitive abiotic factors (e.g., oligotrophy, total darkness, and high mineral concentrations) for life development, they represent interesting ecological niches, hosting extremophile microorganisms, highly specialized and perfectly adapted to this peculiar ecosystem, showing an unexpected biodiversity within the Bacteria domain and countless novel species [2]. To overcome the limiting factors, microorganisms create mutualistic networks, cooperating in communities and favoring each other's survival. The autotrophic microorganisms generally draw energy by chemosynthesis, using chemical elements (such as Ca, Mg, Fe, Mn, and S) and organic and inorganic compounds abundant in the host rocks, cave sediments, groundwater, and atmosphere. Concurrently, several microbial groups rely on mixed metabolic pathways (mixotrophy) [3]. In any case, such microbial communities may contribute to the formation of caves, influencing several biogeochemical processes [1,[4][5][6][7]. In particular, they act inducing the precipitation [8,9] or dissolution of minerals of speleothems and other structures occurring in underground environments, like moonmilk and vermiculation deposits [10,11]. The genesis of all these examples is, indeed, difficult to be explained only by pure abiotic physicochemical processes [2].
Among cave structures, vermiculations are enigmatic deposits recurring on rock surfaces in caves all over the world [12][13][14], characterized by variable morphologies, colors, and dimensions [15,16], and generally composed of calcite, associated with quartz, and traces of clay minerals [17]. Recent studies highlighted microbial evidences supporting their biological origin [10,[17][18][19]. Vermiculations can be indeed considered "life hotspots" and a precious support for the studies on cave geomicrobiology. To our knowledge, there are still few studies on their microbial characterization and most of these concern vermiculations from sulfuric acid speleogenetic systems [10,20,21].
Aimed at shedding light on the microbial community of vermiculations from the Pertosa-Auletta Cave (Campania, southern Italy) and on its role in their formation, this work represents one of the first microbiological studies of vermicular deposits from a normal epigenic karst system. To this end, molecular biology approaches have been employed. In addition, giving an important contribution to the knowledge of the hidden biological aspects of vermiculations, it represents a key step toward the protection and conservation of these peculiar biosignatures and of the whole cave ecosystem.

Vermiculation Samplings
Eleven different points were sampled in the four main branches of the Pertosa-Auletta Cave (Fig. 1), a limestone show cave in southern Italy. Approximately, 2 g of vermiculation deposits was collected. The four branches of the studied karst cave are characterized by various degrees of frequentation, namely (i) Active (A), (ii) Fossil (F), (iii) Paradise (P), and (iv) Tourist (T), where Active indicates the branch still influenced by an active water flow, Fossil identifies inactive conditions of water flow, Paradise is a short piece of the active branch, lit and frequented by humans, and Tourist is the illuminated trail opened to the public for regular visits.
The sampling was performed using disposable and sterile scalpel blades and Eppendorf tubes, carefully avoiding damage to the walls. Stored at 4°C, the samples were immediately sent to the Instituto de Recursos Naturales y Agrobiologia of Sevilla (IRNAS-CSIC, Spain) and maintained at − 80°C, until processing.

Molecular Analyses
Total DNA was extracted using FastDNA TM Spin Kit for Soil, according to the producer's protocol (MP Biomedical). The DNA quality was determined by a Nanodrop ND-1000 Spectrophotometer, whereas the amount by a Qubit 2.0 Fluorometer (Invitrogen).
The extracted DNA (with a minimum concentration of~5 ng/μL), after purification by Genomed and Genomic DNA Clean & Concentrator™-10 (Zymo Research), was analyzed by via next-generation sequencing (NGS) targeting the V3-V4 hypervariable region of Prokaryotes 16S rRNA, using Illumina MiSeq 2 × 250 paired end, according to Macrogen (Seoul, Korea) library preparation protocol. Chimeras were identified and removed by means of USEARCH [29]. Resulting reads were processed in Qiime [30], whereas UCLUST [29] was used for the similar sequences assignment to operational taxonomic units (OTUs) by clustering with a 97% similarity threshold. Paired-end reads were merged using FLASH [31]. RDP Release 11 was used as against reference database for taxonomic identification of query sequences. Alpha diversity analysis, including estimation of Chao1, Shannon, Simpson, and Good's Coverage indices, and rarefaction curves, based on the observed species metric, were performed through Qiime.
The graphs relative to molecular analysis data were elaborated in the R 3.6.0 programming environment [32]. The barplots, showing the relative abundances at phylum, class, and order levels for each sample, with associated dendrograms explaining the similarities among the samples, were created using "ggplots2", "dendextend", and "RColorBrewer" packages. Pearson's correlation coefficients (r values) were obtained using cor function to evaluate associations (for α = 0.05) between geochemical characteristics and microbial phyla as well as among biological properties of the analyzed vermiculations. Non-metric multidimensional scaling (NMDS) analysis, with superimposition of confidence ellipses for branches (α = 0.05), and principal component analysis (PCA) were performed using meta.mds function, based on Euclidean distance metric, and prcomp function, respectively, both from "vegan" package.

Microscopy
The nucleic acids of the whole cells were visualized using the specific SYBR Green fluorescent dye (1:100 dilution), on samples not handled further, under an Olympus FluoView FV1000 confocal laser scanning microscope, and the 488nm excitation laser line with emission signal being collected at 510-530 nm. Images were analyzed with the FluoView 2.1 software (Olympus). FESEM images were acquired using FEI Teneo (Thermo Fisher, MA, USA). To this end, samples were prepared as reported in Addesso et al. [17]. In particular, they were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) at 4°C for 2 h and washed thrice in cacodylate buffer. Subsequently, they were treated with 1% osmium tetroxide for 1 h at 4°C and dehydrated by subsequent dilution series in ethanol and acetone finishing with 100% acetone before drying. The samples were dried in a EM CPD 300 (Leica Microsystem, Wetzlar, Germany) critical point drying device at 34.5°C. Finally, samples were mounted on SEM stubs and sputter-coated with gold (5-10 nm).

Results
All the 11 studied vermiculations, developing on limestone substratum (except A1 and A3, in the Active branch, which were growing at the interface between limestone host rock and bat guano crusts), showed a considerable biological diversity.
The dendrograms (Fig. 2), showing similarities and divergences between specimens based on taxon relative abundances, highlighted three groups, keeping enough in the graph representations of all three taxonomic levels. The clustering analysis showed a clear separation of P1, the only sample located in Paradise branch, from the other two groups, closer to each other ( Fig. 2a-c). At the phylum level (Fig. 2a), A2, F1, F3, and T2 clustered together from the rest (A1, A3, F2, F4, T1, T3). At the class level (Fig. 2b), F2 grouped with A2, F1, F3, and T2, splitting up from A1, A3, F4, T1, and T3. Lastly, at the order level (Fig. 2c), F1, F3, F4, and T2 assembled a new cluster divided from the remaining samples. Figure 2 also shows the corresponding PCAs based on the total bacterial communities at the phylum (Fig. 2d), class (Fig. 2e), and order ( Fig. 2f) levels. Analogous clusters of the dendrograms were also observed in PCA plots. The first (PC1) and the second (PC2) principal components accounted together for 86.04%, 83.81% and 84.73% of the data variance, respectively for phylum, class, and order taxonomic levels.

Microbial Community Richness and Diversity
The rarefaction curve plots, built based on the number of observed microbial groups vs. the number of sequences per sample, for both the four branches and the 11 individual samples, are reported in Online Resource 2 (a and b, respectively). Most curves tended to approach the saturation plateau, reinforcing the sufficiency of sequencing analysis, adequately representative of the investigated communities.
Alpha diversity estimation, using several metrics, is reported in Table 2. The total OTUs generated for each sample ranged from a maximum of 2127 to a minimum of 1323, whereas the average value of Good's Coverage was 99.78%, indicating that the analysis well covers the microbial diversity in vermiculation samples. Chao1 richness estimator resulted between 1444.8 and 2313.3. Shannon and Simpson diversity indices presented similar estimates among the samples (around 7 and close to 1, respectively), except for P1, which presented the lowest values (5.78 and 0.87, respectively).
The NMDS biplot (Fig. 3), based on the microbiological and geochemical-mineralogical [17] characteristics of the analyzed vermicular deposits, showed a clear separation of the confidence ellipses grouping the Tourist and Fossil branches. The vermicular deposits from the active trail revealed intermediate characteristics, as highlighted by the partial overlapping of its confidence ellipse with the other two. Between the two most abundant minerals (calcite and quartz), calcite characterized the vermiculations from the four trails, whereas quartz mainly those of the Tourist and Active trails. Among the 24 elements (total Al, Ba, C, Ca, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, N, Na, Ni, P, S, Si, Sr, Ti, V, and Zn and organic C) analyzed, N, S, and organic C, mostly abundant in the vermiculations from the lightened trails (Paradise and Tourist), together with C, and to a lesser extent Mo, P, and Sr, showed a strong relationship with bacterial communities.
Confocal microscopic observations performed on samples A4 (Fig. 4a, b), F1 (Fig. 4c), and T1 (Fig. 4d) provided interesting information about the distribution and density of microbial communities (green-colored zones) on the mineral surface. As revealed by FESEM images (Online Resource 3), microbial structures were found mainly associated with clay minerals.

Discussion
Although vermiculations represent a perfect substratum suitable for microbes, probably participating also to their formation as mediators of geochemical processes [10,20], very little is known about the microbiota of such enigmatic deposits. In this context, our study provides, for the first time, an overview  Table 3 Pearson correlation matrices between microbial phyla and geochemical characteristics of the studied vermiculations (a) and among their biological properties (b). Significant (***p < 0.001; **p <  Table 3 (continued) and the geochemical characteristics (black labels) of the same vermiculations, as reported in Addesso et al. [17] on the microbial life associated with vermiculations from nonsulfidic karst systems. The NGS approach revealed a biodiversity comparable to those observed in several matrices from different caves [10,19,20,[33][34][35]. Proteobacteria (41.3-54.8%), represented (in decreasing order) by Gamma-, Beta-, Alpha-, and Deltaproteobacteria classes, was the dominant phylum, likely in relation to the wide ranges in metabolism and phenotype, offering the capability to degrade a broad spectrum of organic substrates and to adapt to and thrive in the hostile cave environment [2]. The presence of Proteobacteria is often associated with Fe-Mn deposits [36,37], both chemical elements were observed in vermiculations from Pertosa-Auletta Cave [17], and mainly related to the geochemical characteristics of the substratum, as highlighted by the NMDS. The carotenoidproducing gammaproteobacterial Xanthomonadales order was also detected, typical of yellow-colored colonies found in caves [38,39]. Among Alphaproteobacteria, the Rhizobiales order, represented by members able to fix nitrogen and to oxidize iron and manganese, and the Rhodospirillales order, equally participating to the nitrogen cycle, were observed. They are typical surficial microorganisms [2], but, as suggested by Lavoie et al. [40], their presence in caves can be related to the migration of microorganisms from above lying soils, and once in the cave they start an adaptation process to the new surrounding environmental conditions. Similar to vermiculations from the Pertosa-Auletta Cave, those in the sulfuric acid Fetida Cave (Apulia, Italy) showed a great abundance of Proteobacteria (44-46%), but with copious microbial communities belonging to Deltaproteobacteria (25%) and Epsilonproteobacteria (16%), respectively, dominated by Desulfobacterales and Campylobacterales, involved in the sulfur cycle [19,20].
Acidobacteria represented the second most abundant phylum, whose genetic and metabolic diversity is comparable to the highly diverse Proteobacteria [41][42][43]. Acidobacteria often occur together with chemolithoautotrophic Gammaproteobacteria, suggesting a mutualistic association between them: Acidobacteria gain energy oxidizing the reduced organic compounds (chemoorganotrophy) obtained from Proteobacteria autotrophic metabolism, an ecological advantage in cave oligotrophic environments [44]. Only in the green P1 vermiculation, in the lightened Paradise branch, the most represented phylum after Proteobacteria was Actinobacteria (33.8%), with Actinomycetales order, clearly different from the other vermiculations (1.9-10.3%), confirmed also by PCAs. The abundance of Actinobacteria in this vermiculation is justified by their association with Cyanobacteria, a well-known relationship in lightened subterranean environments [45].
Commonly found in soil systems, Actinobacteria may have an important ecological role in biogeochemical cycles of cave ecosystems, mediating mineralization processes [34] and producing bioactive compounds, such as antimicrobials that allow the biotic control on other populations [46]. Cuezva et al. [7] demonstrated they are able to capture CO 2 from the atmosphere and precipitate CaCO 3 polymorphs, as shown in FESEM images of the same samples reported in Addesso et al. [17]. In particular, Actinomycetales are able to degrade recalcitrant organic compounds [47]. The relative humidity and availability of endo-and exogenous organic matter in the Paradise branch can explain their colonization success. In fact, here, the moisture reaches approximately 100%, due to the presence of an underground river nearby, promoting the proliferation of Actinomycetes [48]. Moreover, the Paradise trail is lit and frequented by tourists who, together with photoautotrophic communities growing close to artificial light systems, bring an important input in terms of organic compounds, facilitating heterotrophic populations, including Actinomycetales [49].
The aerobic chemolithoautotrophic nitrite-oxidizing Nitrospirae and the anaerobic ammonium-oxidizing Planctomycetes, together with Firmicutes, able to reduce/ oxidize sulfur, as well as chemo-or phototrophic Chloroflexi, were also found elsewhere in small amounts. Moreover, numerous less-represented taxonomic groups (with relative abundance < 1%) were observed in the 11 vermiculations investigated and their ecological role in this kind of ecosystem is still debated [2]. Among them, Archaea were also present, with the Thaumarchaeota, Euryarchaeota, and Woesearchaeota phyla, despite the archaeal DGGE profile highlighted a major number of bands in terms of core species richness. The same were found in considerable amount (< 4.3%) in Fetida Cave [20], where the relative abundances change (Proteobacteria > Planctomycetes > Acidobacteria > Chloroflexi > Bacteroidetes > Actinobacteria > Nitrospirae), likely due to the more extreme acidophilic environment, promoting the development of some bacterial groups rather than others. Despite the scarcity of knowledge about the archaeal group in cave ecosystems, it is well known that they give a relevant contribution to the global carbon nitrogen and sulfur cycles [22,50,51]. This may explain both the strong association between Euryarchaeota and N, and the relation of unclassified Archaea phylum groups, Thaumarchaeota, and Woesearchaeota with C and organic C highlighted by NMDS.
The Simpson index displays values close to 1 for all the samples, considering the dominant groups in the community and excluding the rare ones, indicating a low biodiversity and a high dominance. From the NGS results, it emerges that the dominant groups are unclassified already at the phylum level and this increases with the taxonomic level specificity. Values close to 7 were, instead, obtained for Shannon index, sensible also to the rare species, abundantly present in all the samples and certainly important from an ecological point of view.
Overall, geochemical and microbiological characteristics of the studied vermiculations differed among branches of the Pertosa-Auletta Cave, with the greatest differences observed between those from tourist and unvisited branches. Anyway, macroelements (C, N, S, and P), as well as the organic matter, were mostly abundant in the vermiculations from the Paradise and Tourist branches, highlighting the presence of more abundant biomass in lightened trails, where the photoautotrophs proliferate. In these samples, also Mo and Sr were more abundant, indicating that a specialized microbial community could have resulted from some microbial lineages able to oxidize minerals containing such elements [52]. However, F4 sample showed a high abundance of Nitrospirae phylum compared to the other vermicular deposits that displayed also a high correlation with molybdenum, probably due to its content in the membrane-associated enzyme of the nitrite-oxidizing system [53]. Furthermore, the higher content of organic C in vermiculations from Fossil and Tourist trails [17] may explain the major abundance of Nitrospirae in such locations, where the availability of ammonia by ammonificators can increase the presence of nitrites, in turn usable by nitrite-oxidizing Nitrospirae group bacteria [33]. From Pearson correlation analysis, several associations emerge between biological and geochemical properties, as well as among the taxonomic groups, especially the rarest, but they are not at all easy to explain, due to the lack of information about their biogeochemical role in the cave ecosystem [33].
Confocal microscopy images showed a localization of DNA only in specific sites, recognizable in the green zones. This was confirmed also by FESEM images, as reported in Addesso et al. [17], showing the clayey deposits always associated with biogenic filamentous material, not ruling out the possibility that the microbes can interact or influence their behavior and evolution in the environment [54,55].
The findings of the present study support the theory formulated by Jones et al. [10], suggesting that microorganisms play an active role in vermiculation genesis, producing organic matter and secondary minerals, enriching the calcite matrix, trapping and binding sediment particles and dissolving, through etching or pitting, the rock. This may happen in different environments, from sulfuric acid to normal karst caves. However, beyond the biological evidences, the possibility of coexistence of several processes remains. For example, decalcification of rock walls, due to the dissolution processes caused by the acidity of seeping or condensation waters, can contribute to create the primordial calcite matrix [56][57][58][59]; thereafter, neutralization of electrical charges in the small particles, associated to wet-dry phenomena, can determine the different morphologies [16,60]. Nevertheless, further studies are required to clarify to what extent some processes prevail over others, determining the variety of vermiculations described.
The present study, describing the microbiota present in the vermicular deposits of the Pertosa-Auletta Cave and its relationships with geochemistry of vermiculations, fills the gap characterizing these topics in karst caves. The analyses carried out indicate a certain diversity of biological communities living in vermicular deposits, with a considerable percentage of unclassified lineages, already at the phylum level, demonstrating once more that the underground ecosystem hosts still a high number of unknown taxa. Proteobacteria and Acidobacteria were the predominant phyla, as generally observed in such environments, whereas Actinobacteria showed an increased growth due to the high humidity conditions and the input of organic matter from the considerable presence of tourists in the show cave. The involvement of such communities in the biogeochemical cycles is indisputable and the highlighted biological evidences confirm a tight interaction between biotic and abiotic factors in the formation of vermiculations. The obtained findings represent a crucial step for the protection and conservation of such unique ecological niches, making still more intriguing the knowledge and comprehension pathway of vermiculations.