Does Microbial Diversity of Cave Ecosystems Differ from Outside? The Case of the Azé Show Cave (France)

Badr Alaoui-Sosse UMR 6249 CNRS/Université Bourgogne Franche Shinji Ozaki UMR 6249 CNRS/Université Bourgogne Franche Lionel Barriquand Université Savoie-Mont-Blanc Daniele De Luca University of Naples Federico II Paola Cennamo University of Naples Suor Orsola Benincasa Benoit Vallot UMR 6249 CNRS/Université Bourgogne Franche Laurence Alaoui-Sosse UMR 6249 CNRS/Université Bourgogne Franche Pascale Bourgeade UMR 6249 CNRS/Université Bourgogne Franche Faisl Bousta Laboratoire de Recherche des Monuments Historiques (LRMH) Lot Aleya UMR 6249 CNRS/Université Bourgogne Franche-Comté UsC INRA Stéphane Pfendler (  stephane.pfendler@univ-fcomte.fr ) Laboratoire Chrono Environnement https://orcid.org/0000-0002-1571-2559

Does Microbial Diversity of Cave Ecosystems Differ from Outside? The Case of the Azé Show Cave (France) 1

. Introduction
Caves are oligotrophic ecosystems characterized by the absence or low light level, stable temperature and high humidity [1]. Despite low amounts of organic matter [2], Barton and Jurado (2007) [3] have estimated that 10 6 microorganisms per gram of rock inhabit in these ecosystems. Microorganisms grow inside self-made anionic EPSs (Extracellular Polymeric Substances) [4], called bio lm, that has a threedimensional architecture [5]. Two types of bio lms may be easily distinguished in the caves. First, the photosynthetic bio lm mainly composed by cyanobacteria, micro-algae, diatoms and bacteria [6]. Then, the non-photosynthetic bio lms that are less studied and inhabited by bacteria such as Actinobacteria [7].
For their expansion on cave rock, all microorganisms require ve growth conditions [8]. First, the bioreceptivity that has dissimilar levels depending on the type of substrata [9]. On this regard, Jones et al. (2017) [10] have reported that surface type signi cantly controlled 70 to 90% of the variance in phylogenetic diversity. Then, moisture, temperature and nutrients are essential for microorganism survival, growth, and expansion [11,12,13]. Finally, photosynthetic bio lms need a natural (usually at the cave entrance) or an arti cial lighting source (mounted for the tourists) depending on their location in the cave [1].
Since several decades, the cave microbiome has been studied for several reasons [14] such as medical or pharmaceutical interests (e.g., research of antibio lm compounds) [15] and cave preservation (e.g., Paleolithic drawings and rock paintings curation) [16, 17,18]. These studies have been carried out using a large set of microorganism identi cation technics such as microscopic methods [19,20], isolation and culture in laboratory [21], cloning followed by Sanger sequencing [22,23,24], ampli ed ribosomal DNA restriction analysis [25] or high throughput sequencing [6,26,27,28]. Results of these studies have demonstrated that cave microorganisms developing inside bio lms may be involved in speleogenesis processes. Indeed, it has been reported that the complex microbial communities are capable of re ecting in substantial rock alterations, involved in calci cation processes [21,29,30,31], formation of new structures [32], corrosion of the mineral surfaces [4] and cave expansion [1,33]. Thus, the biodeterioration process is of great concern to cave owners and managers who nd themselves faced with a dual problem: the esthetic impact of bio lms [34,35,36] and the biodeterioration [31,37].
Despite phototrophic bio lms have been studied for several decades, studies comparing the microbial communities occurring inside and outside cave environments are very limited [38, 39,40]. The differences between communities would provide a better understanding of the microorganism sources and thus about bio lm genesis inside the cave. Moreover, it will permit to know how much the communities differ.
In the present work, we have analyzed the bio lm communities growing inside the Azé show cave through high-throughput sequencing and compared them with those growing outside the cave.

Site description
This study was carried out in one of the caves of Azé, which are located in the Monts du Mâconnais  [41]. The entrance of the prehistoric cave, the one we have sampled in this study, is located at 275 m above sea level. At 180 to 200 m of the entrance, a large area was unobstructed from the sediment during the winter 1976-1977, and many bear bones (160,000-year-old Ursus deningeroides) were discovered [42]. Several vestiges of the Lower and Upper Paleolithic, Bronze Age, Gallo-Roman period and Middle Age (including a wall), also remain in the cavity, and the caves are nowadays visited by 28,000 tourists each year [6].

Environmental parameters
Environmental conditions in the Azé cave have been monitored for several years (Table 1) showing two different seasonal regimes were measured. First, during the winter a convection loop is established and the CO 2 level is then normal. Then, during the summer, the warm outside air blocks the cold air in the cave and the CO 2 level can exceed 3.5% in the deepest parts of the cave. The entrance zone undergoes strong temperature variations while in the terminal zone of the cave the temperature varies approximately between 11.5 and 12.0°C. The entrance is highly in uenced by the outside humidity, while the air is always saturated with humidity beyond 80 m from the entrance. As well as natural light, arti cial light intensities were collected inside the show cave. The light meter detector was placed (i) on the phototrophic bio lms and (ii) at 2 cm from light source.

Colorimetric measurement of bio lm
The sampled bio lms differed according to their colors. For this reason, the bio lm colorimetric parameters were measured with a spectrophotometer (CM-600d KONICA MINOLTA, illuminant D65, SCI mode and 8 mm diameter target mask). Color measurements were analyzed according to the CIELAB color system. The a* scale is associated with changes in redness-greenness (positive a* is red and negative a* is green) [43].

Quantum yield measurements
In order to distinguish photosynthetic bio lms from non-photosynthetic bio lms, quantum yield measurements, corresponding to phototrophs metabolism, were carried out. Before each sampling, ve measurements were taken on each bio lm, which were previously placed for 30 min in the dark. Quantum yield parameter (Fv/Fm) was monitored using the photosynthesis yield analyzer mini-PAM (WALZ, Germany) as follows: Fv/Fm = ϕPSII / qP 2.6. Bio lm sampling The phototrophic bio lms were collected outside, at the entrance ( rst 27 meters) and inside the cave (from the 27th meter to the bottom of the cave). Five samples were collected from the limestone rock (hereinafter referred as "Out_rock") and other ve samples from the soil (called "Out_soil") outside the cave. Then, eight samples were taken at the entrance area ("Ent"), and twelve other samples from the end of the entrance to the bottom of the farthermost cave area ("Ins"): Rotonde ("rot"), Gour, North Gallery, "14 Juillet" room ("July"), and on a bear bone (bear room).
In accordance with the sequencing platform (Microsynth AG, Balgach, Switzerland), 100 to 200 mg of fresh matter was taken from each bio lm sample. To avoid unwanted contamination, bio lms were directly scraped using 2-ml sterile tubes containing a buffer and balls for mechanical lysis, and subsequently kept on dry ice (-78°C). Samples were then conserved by MicroSynth AG at -20°C until total DNA extraction.

Molecular methods and sequencing
For DNA extraction, PowerBio lm DNA Isolation Kit was used by MicroSynth AG following the manufacturer's instructions (MoBio Laboratories, Inc., Carlsbad, CA, USA). The polymerase chain reaction (PCR) ampli cation followed a two-step PCR protocol using a state-of the-art high delity polymerase. This two-step PCR was applied in order to increase reproducibility and to improve the production of highquality multiplex amplicon libraries. PCRs were performed with the primers p23SrV_f1, p23SrV_r1 [44,45] and 16S 799 f (5' -AACMGGATTAGATACCCKG-3') and 16S 1115 r (5' -AGGGTTGCGCTCGTTG-3'

Sequence data analysis
Reads were assigned to each sample according to a unique barcode. Paired reads were assembled into contigs using the Mothur pipeline [48]. An in-silico PCR kept only reads containing the used forward and reverse primer. For 16S generated sequences, the contigs were pre-clustered at 99% similarity, while 23S sequences were dereplicated as unique sequences. Rare sequences, represented by less than 10 reads for 16S and 23S primers, were removed from the analysis. The 16S and 23S taxonomic assignments were performed using SSU and LSU SILVA database (v132), respectively. Operational taxonomic units (OTUs) were then constructed using the Needleman-Whunch distance and average neighbor clustering (UPGMA) at a distance of 0.03, and 0.05 for 16S and 23S primer sequences, respectively. The number of reads per sample was calculated after a random sub-sampling of 10,000 reads. Nucleotide data were deposited in the GenBank database under the BioProject ID: PRJNA723481.

Statistical analysis
The relationship between the solar light intensity and the distance from the entrance was analyzed by the linear regression. The coe cient of determination (R 2 ) was calculated if the relationship was signi cant (p-value < 0.05) under ANOVA. OTU Richness, Simpson's diversity index, and Simpson's evenness at outside, entrance and inside were compared for both phototroph and bacterial communities, using the nonparametric Kruskal-Wallis test because of skewed distribution of OTUs. Community composition was separately analyzed for phototrophic and bacterial communities. An analysis of similarity (ANOSIM) test was performed on the non-metric multidimensional scaling (NMDs) analysis.
All statistical analyses were performed using R software version 4.0.2 (R Development Core Team, 2016).
Kruskal-Wallis tests and multiple comparisons of samples were performed by the "kruskal" function from the "agricolae" package.

Natural and arti cial light pro les
Outside the cave, the light intensity reached 779 m − 2 s − 1 (data not shown), while inside the cave, the solar light decreased exponentially (R 2 = 0.99) from the entrance (0th meter: 9.5 mol m − 2 s − 1 ) to the 37th meter (0 mol m − 2 s − 1 ) (R 2 = 0.99; Fig. 1a). The arti cial light intensities (Fig. S1 a) have showed a disparity of radiations (137 to 5060 mol m − 2 s − 1 ) along the cave. Available light for the bio lms (Fig. 1b), that is presenting on Fig. S1b, showed 0.2, 7.4, and 27.5 mol m − 2 s − 1 as the minimum, the average, and the maximum light intensity, respectively. Finally, the distance between the lamps and bio lms varied between 0.23 to 5.06 meters (data not shown).

Quantum yield and colorimetric measurements
The data (Fig. S2) have showed that, whatever the light intensity or light sources (natural or arti cial), the quantum yield e ciency of the sampled bio lms (Fig. 1b) was greater than 0.6 (average: Fv/Fm = 0.67).
Our results showed that bio lms located in the rst 60 meters of the cave (i.e., Rotonde and Entrance samples) have the lower green-red scale values (Fig. S3). In fact, the bio lms have exhibited negative values (average = -4.17) corresponding to green color and thus higher microorganism density. All the Gour, '14 Juillet' and Bear bone samples, located in the second half of the cave, have positive values (from 1.68 to 3.37), corresponding to lower green intensity and higher red intensity (limestone wall colour).

Taxonomic composition of bio lms
The Fig. 2 shows the distribution of the photosynthetic organisms according to their sampling localization. Our results showed that eukaryotes were always the dominant taxon in comparison to cyanobacteria. Photosynthetic eukaryote (74%) and cyanobacteria (26%) proportions were similar between both inside and entrance bio lms, while outside bio lms were represented for 31% by cyanobacteria.
The Chlorophyta phylum was prevailing in bio lms regardless the localization (57.6% inside, 90.4% at the entrance and 66.7% outside). The less represented phylum was the Bacillariophyta that was poorly recorded inside and at the entrance of the cave (< 0.7%). On the contrary, outside the cave they were represented by 6.5% of the total eukaryotes sequenced. Our results have also showed high proportion of unclassi ed eukaryotes (32.7%) inside the cave. Finally, Streptophyta were represented in all bio lm samples. However, it is necessary to consider that bryophytes are multicellular organisms and thus can deeply prevaricate semi-quantitative information provided by the sequencing.
The proportion of cyanobacteria populating the bottom of the cave highlighted an important proportion of Nostocales (98%). At the entrance and outside the cave, high proportion of unde ned cyanobacteria (84.2% and 47.8%, respectively) were recorded. Outside the cave, excluding the unde ned cyanobacteria, the most represented cyanobacteria were Synechococcales (28.6%), Nostocales (14.3%) and Chroococcidiopsidales (6.6%).

In uence of bio lm localization
Richness of phototrophic organisms signi cantly decreased from the outside to the inside of the cave (pvalue < 0.001) (Fig. 3a). Bacterial richness showed the same decreasing trend, but no signi cant difference was obtained between the outside and the entrance of the cave (Fig. 3b). The Simpson indices for phototrophs and bacteria also showed a similar trend. However, the index for phototrophs showed signi cant difference (p-value < 0.001) only for the outside samples, whereas the index for bacteria showed no signi cant difference between the three areas (Fig. 3c, 3d). On the contrary, Simpson evenness for phototrophic organisms was signi cantly lower at the entrance than the others (p-value < 0.01) (Fig. 3e). No signi cant difference of Simpson evenness for bacteria was observed (Fig. 3f).

Composition of the bacterial and phototrophic communities
Phototrophic communities were different among the outside, the entrance and the inside of the cave, although the communities at the entrance and the outside were more similar in comparison with these at the inside (Fig. 4a). On the other hand, a greater variability of the phototrophic organisms was observed at the inside of the cave. In the same way, the samples from entrance have a sparser distribution in comparison to outside communities. These samples were scraped on blue bio lms (Fig. 1a), while the others on green spots.
As well as the eukaryotic phototroph communities (Chlorophyta, Bacillariophyta and Streptophyta), the prokaryotes showed the same trend (Fig. 4b). On this regard, three types of prokaryote communities were more distinguished than phototrophic communities. Moreover, bio lms at outside could be divided into two groups. The rst group corresponded to bio lms scraped from the soil, while the second group was sampled on limestone rocks.

Comparison of organism communities
Phototroph organisms encompassed 154 OTUs, while 3,872 bacterial OTUs were recorded. The obtained phototroph OTUs, which were recorded only on one speci c area, represented < 4% of the total DNA reads, while OTUs shared at least in two cave sections represented 3.6%. Moreover, 23 OTUs are shared by samples from outside the cave, at the entrance and inside. They were represented by 92.5% of the total reads. These OTUs were composed by 10 chlorophytes, 9 cyanobacteria, two unclassi ed eukaryotes, one bacillariophyte and one streptophyta (Fig. 5a).
The Venn diagram, obtained from bacteria data (Fig. 5b), has shown that 419 OTUs (corresponding to 14.6 % of all the OTUs) were shared by the three sampled cave areas. However, they corresponded to 56.2% of the total bacterial reads. Thus, our results have indicated that 1,597 OTUs (55.6% of all the OTUs) were speci c to one of the three sampled sites. Nevertheless, these OTUs were only representing 9.7% of total reads.

Discussion
In the present study, we have compared both bacterial and phototrophic communities inhibiting (i) the cave area illuminated by arti cial light, (ii) the entrance, characterized by low natural light intensity, and (iii) the outside of the cavity, irradiated by high intensity of natural light.

Bio lm community diversity
Both the studied Eukaryote and Prokaryote microorganisms have showed a richness decrease from the outside to the bottom of the Azé cave. The obtained results in Azé are not surprising considering that the outside part of the cave is bordered by the forest and that the 60th rst meters of the cave are strongly in uenced by the outdoor parameters [41]. From the 60th meter to the bottom of the cave, environmental conditions are relatively stable and thus less in uenced by the outside, which explain the lowest richness of bacteria and photosynthetic organisms. The microorganism richness decrease was already reported by Roldán et al. (2004) [47], Piano et al. (2015) [48] and Abris et al. (2020) [49] for the phototrophs and fungal propagules, respectively. The authors have described a negative correlation between distance from the cave entrance and propagule deposition rate. This is explained by the gradual reduction of the air currents that spread microorganisms from the entrance to the bottom of the cave [50]. In the case of Azé, the touristic activity factor also plays a role, but it is probably limited to the arti cial lighting that allows the growth of phototrophs and the resulting development of bacteria. Moreover, insects and animals may probably play a more important role in the microorganism cave spreading in comparison to the touristic activity [28]. However, further studies should be implemented for a better understanding of the in uence of insect and animal in the spread of cave microorganisms. In fact, our data have shown that only four OTUs, corresponding to rare OTUs and < 0.1% of the total reads, are speci c to the cave inside. Moreover, the entrance data have showed the same trend. These results indicate that there is no speci c phototrophic community inside the cave. However, the communities inhabiting the cavities may adapt to speci c conditions such as low nutrient, temperature and light intensity [51].

Communities change depending environmental factors
In the Azé cave, the bio lm communities gathered several phototrophic phyla such as the well-known Chlorophyta, Cyanobacteria, Bacillariophyta and Streptophyta [4]. These phyla are frequently described by authors studying the cave microorganism proliferation [17,52,53,54,55]. However, the community distribution and abundance obtained in this study are not comparable to those previously published by other authors. More generally, there is often variation from one study to another. This observation may be due to many environmental factors (i.e., temperature, moisture, nature of the substrata, light intensity  [4] and illuminated surfaces [20], such as in the outside the Azé cave, are the main factors for Bacillariophyta development.

Conclusion
Our results have highlighted that caves are colonized by a high diversity of bacterial community depending on the substrata and the environmental conditions. Comparing to the outside of the cave, these communities seem speci c to the inside. Since arti cial lights were used for touristic purpose, the Azé cave ecosystem has been modi ed leading to new development of phototrophs. In the present study, we have demonstrated that these communities are not signi cantly different from the cave outside phototrophs. The main differences consist on a richness decrease from the outside to the bottom of the cave, explained by the decreasing in uence of the cave outside environment. It is noteworthily to highlight the difference of communities from one bio lm to another inside the cave; these differences may be linked to the community functional traits, especially the limestone biodeterioration processes. Finally, further studies should examine how the substratum physico-chemical parameters in uence the bio lm community proliferation.

Declarations
Acknowledgement First of all, we are grateful to the curators of the Azé Cave, who kindly gave us permission to access the cave and to carry out all our eld experiments. We also thank the

Figure 2
Distribution of the photosynthetic organisms according to their sampling localization (outside, entrance or inside the cave).

Figure 3
The richness, and both Simpson index and evenness indices of bacteria and phototrophic organisms were calculated (Out = outside, Ent = entrance, Ins = inside; results of the statistical analyses are represented by the letters a, b, c, ns and *, ** or ***).