Comparison of Actinobacteria communities from human-impacted and pristine karst caves

Abstract

Actinobacteria are important cave inhabitants, but knowledge of how anthropization and anthropization-related visual marks affect this community on cave walls is lacking. We compared Actinobacteria communities among four French limestone caves (Mouflon, Reille, Rouffignac, and Lascaux) ranging from pristine to anthropized, and within Lascaux Cave between marked (wall visual marks) and unmarked areas in different rooms (Sas-1, Passage, Apse, and Diaclase). In addition to the 16S rRNA gene marker, 441 bp fragments of the hsp65 gene were used and an hsp65-related taxonomic database was constructed for the identification of Actinobacteria to the species level by Illumina-MiSeq analysis. The hsp65 marker revealed higher resolution for species and higher richness (99% operational taxonomic units cutoff) versus the 16S rRNA gene; however, more taxa were identified at higher taxonomic ranks. Actinobacteria communities varied between Mouflon and Reille caves (both pristine), and Rouffignac and Lascaux (both anthropized). Rouffignac displayed high diversity of Nocardia, suggesting human inputs, and Lascaux exhibited high Mycobacterium relative abundance, whereas Gaiellales were typical in pristine caves and the Diaclase (least affected area of Lascaux Cave). Within Lascaux, Pseudonocardiaceae dominated on unmarked walls and Streptomycetaceae (especially Streptomyces mirabilis) on marked walls, indicating a possible role in mark formation. A new taxonomic database was developed. Although not all Actinobacteria species were represented, the use of the hsp65 marker enabled species-level variations of the Actinobacteria community to be documented based on the extent of anthropogenic pressure. This approach proved effective when comparing different limestone caves or specific conditions within one cave.

Keywords: 16S rRNA gene sequencing; Actinobacteria; Paleolithic cave; cave anthropization; hsp65 sequencing; metabarcoding.

Figure 1

(a) Map of Dordogne area indicating the cave locations (Lascaux, Rouffignac, Reille, and Mouflon; white squares—pristine; gray squares—anthropized). (b) Map of Lascaux Cave (entrance Sas‐1, Passage, Apse, and Diaclase)

Figure 2

(a) Rarefaction curves for 37 common samples of the hsp65 and 16S rRNA genes at the 97% and 99% OTU cutoffs. The X‐axis denotes the number of samples, and the Y‐axis denotes the number of OTUs. (b) Pairwise molecular distances between sequences of the hsp65 (Y‐axes) and 16S rRNA gene (X‐axes) genes among Actinobacteria species from different taxonomical levels with the correlation coefficient (R ²) for each equation. OTU, operational taxonomic unit

Figure 3

Sammon projection of nonmetric multidimensional scaling (NMDS) based on the Bray–Curtis distance matrices of the hsp65 and 16S rRNA gene markers for different caves (a) and Lascaux Cave locations, and marked/unmarked areas (b). The F and p values of overall AMOVA (97% OTU cutoff) are indicated in Table A7. AMOVA, analysis of molecular variance; OTU, operational taxonomic unit

Figure 4

Significantly different OTUs between pairs of caves (Metastats, p < 0.005). For each cave, the number of OTUs that differed proportionally from other caves and taxonomically assigned OTUs proportionally the most abundant in the respective cave is indicated (99% OTU cutoff). OTU, operational taxonomic unit

Figure 5

Average proportions of Actinobacteria families based on hsp65 (a) and 16S rRNA gene (b) sequence libraries from different caves (Rouffignac, Lascaux, Mouflon, and Reille), Lascaux Cave locations (Sas‐1, Passage B, Passage IP, Apse, and Diaclase) and marked (S)/unmarked (U) areas within Lascaux

Figure 6

Co‐occurrence networks of Actinobacteria OTUs: (a) hsp65 marker and (b) 16S rRNA gene marker; OTUs differed significantly between marked (black) and unmarked (white) areas in Lascaux Cave, and those that did not differ between these areas (gray) using Metastats (p < 0.05). The letters indicate OTUs that were specific for the respective Lascaux Cave locations (S, Sas‐1; PB, Passage banks; PI, Passage inclined planes; A, Apse; and D, Diaclase) using LEfSe (p < 0.03). Strong significant connections (Spearman's correlation >0.8 and <0.5 for 16S rRNA gene and <0.35 for hsp65) are displayed (99% OTU cutoff). Red lines indicate negative correlations and black lines indicate positive correlations. OTU, operational taxonomic unit

Figure A1

Venn diagrams showing the number of different Actinobacteria taxa recovered with hsp65 (blue) or 16S rRNA gene (red) at different taxonomic ranks. (a) Class, (b) order, (c) family, and (e) genus

Figure A2

Indices of alpha diversity (richness, Chao‐1; evenness, Simpson evenness; diversity, inverse Simpson) for Actinobacteria communities among caves, different locations within Lascaux Cave, and marked/unmarked areas. Significant differences between groups are shown with letters (ANOVA and Tukey post hoc test; p < 0.05; 97% OTU cutoff). ANOVA, analysis of variance; OTU, operational taxonomic unit

Figure A3

Venn diagrams showing unique and shared OTUs between (a) different caves and (b) different locations within Lascaux Cave based on hsp65 and 16S rRNA gene markers (99% OTU cutoff). OTU, operational taxonomic unit

Figure A4

Taxonomic compositions of the core microbiome between (a) caves and (b) different locations within Lascaux Cave based on the hsp65 and 16S rRNA gene markers (99% OTU cutoff). OTU, operational taxonomic unit

Figure A5

Phylogenetic tree showing the taxonomic diversity and allocation of hsp65 sequences from caves retrieved by a metabarcoding approach. OTUs were used to build this figure. The evolutionary history was inferred using the neighbor‐joining method (Saitou & Nei, 1987) and reveals the relationship of sequences from caves obtained in this study with sequences belonging to type and reference strains classified within the Actinobacteria class. The tree is based on the Kimura two‐parameter method (Kimura, 1980) with the confidence values of the branches determined by bootstrap analyses (Felsenstein, 1985) based on 1000 replicates. Only values >80% are shown at the nodes. This analysis involved 116 nucleotide sequences. The scale bar represents 0.020 substitutions per nucleotide position. Evolutionary analyses were conducted in MEGA 6 (S. Kumar et al., 2018). OTU, operational taxonomic unit

Figure A6

Average proportions of species within the genus Streptomyces in the Actinobacteria hsp65 amplicon sequence libraries significantly different in marked (S) and unmarked areas (U) of Lascaux Cave (Metastats, p < 0.05).

Figure A7

Average proportions of genus Mycobacterium in the Actinobacteria hsp65 gene amplicon sequence libraries in different caves (Rouffignac, Lascaux, Mouflon, and Reille) and different locations within Lascaux Cave (Sas‐1, Passage B, Passage IP, Apse, and Diaclase).

Figure A8

Average proportions of genus Nocardia in the Actinobacteria hsp65 gene amplicon sequence libraries in different caves (Rouffignac, Lascaux, Mouflon, and Reille) and different locations within Lascaux Cave (Sas‐1, Passage B, Passage IP, Apse, and Diaclase).