Figures
Abstract
Subsurface habitats harbor novel diversity that has received little attention until recently. Accessible subsurface habitats include lava caves around the world that often support extensive microbial mats on ceilings and walls in a range of colors. Little is known about lava cave microbial diversity and how these subsurface mats differ from microbial communities in overlying surface soils. To investigate these differences, we analyzed bacterial 16S rDNA from 454 pyrosequencing from three colors of microbial mats (tan, white, and yellow) from seven lava caves in Lava Beds National Monument, CA, USA, and compared them with surface soil overlying each cave. The same phyla were represented in both surface soils and cave microbial mats, but the overlap in shared OTUs (operational taxonomic unit) was only 11.2%. Number of entrances per cave and temperature contributed to observed differences in diversity. In terms of species richness, diversity by mat color differed, but not significantly. Actinobacteria dominated in all cave samples, with 39% from caves and 21% from surface soils. Proteobacteria made up 30% of phyla from caves and 36% from surface soil. Other major phyla in caves were Nitrospirae (7%) followed by minor phyla (7%), compared to surface soils with Bacteroidetes (8%) and minor phyla (8%). Many of the most abundant sequences could not be identified to genus, indicating a high degree of novelty. Surface soil samples had more OTUs and greater diversity indices than cave samples. Although surface soil microbes immigrate into underlying caves, the environment selects for microbes able to live in the cave habitats, resulting in very different cave microbial communities. This study is the first comprehensive comparison of bacterial communities in lava caves with the overlying soil community.
Citation: Lavoie KH, Winter AS, Read KJH, Hughes EM, Spilde MN, Northup DE (2017) Comparison of bacterial communities from lava cave microbial mats to overlying surface soils from Lava Beds National Monument, USA. PLoS ONE 12(2): e0169339. https://doi.org/10.1371/journal.pone.0169339
Editor: Lorenzo Brusetti, Free University of Bozen/Bolzano, ITALY
Received: July 23, 2016; Accepted: December 15, 2016; Published: February 15, 2017
Copyright: © 2017 Lavoie et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All sequences, taxonomic assignments, and R script files are available from the zenodo database at: https://zenodo.org/record/58072.
Funding: Funding was provided by Lava Beds National Monument (LABE) through the Colorado Plateau Cooperative Ecosystem Studies Unit (http://nau.edu/CEFNS/Forestry/cpcesu/), Award number J8410100028-UNM-56. Funder's role: LABE personnel Shane Fryer and Shawn Thomas helped in the selection of study sites, supplied metadata for all sites, and assisted with field work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Most life on Earth in the aphotic subsurface is microbial [1], but there is much that we do not yet know about subsurface life. Caves can provide a natural way to access subsurface environments ranging from very deep limestone caves (Krubera Cave in the Western Caucasus is more than 2,190 m deep [2]), to shallow caves, such as lava caves that have an overburden of up to 10 m [3]. Discovery of extensive lava flows and lava caves on Mars [4] supports the concept that Earth’s lava caves may serve as a model for the study of life on other planets (astrobiology) [5,6,7].
Lava caves, formed during active lava flows, contain diverse microbial mats that range from extensive mats covering walls and ceilings to small, scattered colonies (Fig 1 and Fig 2). Mat colors include white, yellow, tan, gold, orange, and pink, with shades in between [7–11]. Despite their extensive nature, little is known about microbial mat diversity (reviewed in [11,12]). Studies of microbial diversity [7–15] in lava caves lag behind such studies in karst caves. Stoner and Howarth [16] first described the mats or “slimes” in Hawaiian lava caves using culture-dependent methods for isolation of chemoheterotrophic microorganisms and reported on the presence of fungi and aerobic bacteria. They suggested that white and brown slimes might be important sites for nutrient cycling in caves, particularly nitrogen.
(A) Entrance to Valentine Cave. (B) Surface samples taken above Hopkins Chocolate Cave. (C) Yellow microbial mat sampling site in Valentine Cave. (D) Extensive yellow microbial mats on walls of Hopkins Chocolate Cave. (E) Passage in Valentine Cave showing less microbial mat coverage near the entrance. (F) Tan microbial mat sample taken in L-V460 Cave. Photos copyright Kenneth Ingham (A, D, E) and Diana Northup (B, C, F).
(A) Overview of predominantly yellow and white microbial mats, some separate and some intermixed. (B) Overview of tan and white microbial mats. (C) Intermixed tan, white, and yellow microbial mats. Close ups of (D) tan and white colony morphology, (E) tan colonies, and (F) yellow colonies.
Lava caves are extreme environments, simplified by the lack of photosynthesis in the deep or dark zone of the cave, resulting in extremely oligotrophic conditions. The simplified nature of caves makes them a model natural laboratory to study factors controlling biological diversification [17,18]. The isolation of most caves limits the ability of organisms to migrate, resulting in high levels of endemism among troglobionts and stygobionts as the norm [19]. Culver et al. [20] found that about 30% of cave-adapted invertebrate species in U.S. caves are found in only a single cave. The results of Hathaway et al. [21] show that the trend can be extended to bacterial diversity in Azorean and Hawaiian lava cave microbial mats. The authors were far from sampling total diversity, but less than 5% of the OTUs (operational taxonomic unit) found in lava caves occur in other caves or in other volcanic environments. Sequences were more likely to be related to samples from the same cave or the same island than between islands. If microbial distribution is ubiquitous, then they would expect a higher percentage of shared OTUs between the two island archipelagos.
Biospeleologists originally thought that cave microbes were simply a subset of surface microbes washed into underlying caves [22,23]. Ortiz et al. [24,25] recently published what we believe to be the first comparisons of cave microbial diversity with the overlying soil microbial community. Their studies focused on bacterial diversity across carbonate speleothem surfaces sampled by swabbing from a karst cave, Kartchner Cavern in Arizona. Comparison of bacterial taxonomic profiles to surface soil samples revealed major differences and only a 16% overlap between cave speleothem and surface soil OTUs [24].
Our study is the first and most comprehensive to compare lava cave bacterial mat communities to bacterial communities from the overlying surface soil of each cave. We examined a range of environmental, geographical, and chemical factors that may contribute to bacterial diversity in microbial mats of different colors (tan, white, and yellow) from lava caves in Lava Beds National Monument, California, USA.
Materials and methods
Field studies ethics statement
All sampling was done under Permit LABE-2011-SCI-0007 issued to Northup by the National Park Service (Lava Beds National Monument, 1 Indian Well Headquarters, PO Box 1240, Tulelake, CA 96134). Lava Beds National Monument is a federally-protected area under the National Park Service, Department of the Interior. No protected species were sampled. Shawn Thomas, Shane Fryer, and Joke Vansweevelt have given their consent to have photos of themselves, taken during sampling trips, used in this manuscript.
Sampling sites
Lava Beds National Monument (LABE) is located in northern California close to the borders of Oregon and Nevada [26]. The Monument covers 190 km2 on the NE flank of the Medicine Lake Volcano. Two-thirds of the lava came from the Mammoth and Modoc craters over the last two million plus years and as recently as 1,100 years ago. Flows are largely of basalt with smaller amounts of silica-rich basaltic andesite [27].
LABE has the largest number of lava caves in North America, with 778 known [26]. Twenty-five of the lava caves have signed entrances and developed trails for ease of visitation. The area is a high-elevation (1219–1737 m above sea level), semi-arid desert with average yearly precipitation of 375 mm. Temperature ranges from an average low of -5.4°C in January to an average high of 22.3°C in July and August. Some of the lava caves contain perennial ice [27].
We worked with LABE personnel to select seven lava caves to cover a range of parameters (Table 1): amount of human visitation, age of the lava flow, surface vegetation, elevation, number of entrances, branching complexity, and mapped length of the caveIn addition to the cave samples, a sample of soil overlying each cave entrance (e.g. Fig 1B) was collected for comparison and photo-documented at the collection site (Fig 1F).
Cave samples (Table 2) were characterized by color of the microbial mats (tan, white, and yellow e.g. Fig 2), distance from the nearest entrance, temperature, humidity, and RH, and pH when suitable water pools or dripping water were available near mats. All samples were photo-documented at the collection site
Temperature, RH, and pH measurements
Temperature (web bulb and dry bulb in order to obtain an approximate RH) was taken in April, 2011 and in September, 2012 with an IMC temperature probe (http://www.imcinstruments.com/), which was calibrated at frequent intervals in the cave to improve accuracy. Wet bulb readings were obtained with the IMC probe sheathed with wicking soaked in deionized water before each reading. For some RH samples a portable Kestrel 3000 wind meter (https://kestrelmeters.com/products/kestrel-3000-wind-meter) was used, which was calibrated at the beginning of each cave. A Javascript program (http://home.fuse.net/clymer/water/wet.html) used dry and wet bulb temperatures to approximate relative humidity. Readings for pH were taken with a Twin Cardy pH meter (Spectrum Technologies, Inc., http://www.specmeters.com/nutrient-management/ph-and-ec-meters/ph/cardy-twin-ph-meter/), calibrated with pH 7 buffer.
Sample collections for DNA and Scanning Electron Microscopy (SEM)
Sampling took place in April 2011 with additional sampling (GEM2, HCC1, SC2) in August 2012 to increase the number of white mat samples. Yellow, white, and tan mats were sampled from each cave. Samples for DNA extraction were collected aseptically with a flame-sterilized cold chisel into a sterile 50 cc Falcon tube. Soil samples were collected from above each cave entrance by removing any surface plant detritus and scooping the top 2 cm of soil into a sterile 50 cc Falcon tube. All samples for DNA were covered within hours with sucrose lysis buffer [28] to release and stabilize the DNA. All samples were brought to the Northup Lab at the University of New Mexico for further processing and analysis within seven days. We collected surface soils from above seven caves, and had seven samples of white and tan mats and nine samples from yellow mats, with at least one of each color from each cave.
Samples for scanning electron microscopy (SEM) from rock chips were mounted directly onto SEM stubs with super glue and placed in a carrying case for transport. A microbial mat sample (L-V460-110425-6), consisting of chips of the wall rock with white to pale yellow colonies, was taken approximately 200 m into Cave L-V460 at the bottom of a pillar. One of the two yellow microbial mat samples (HC110423-5) analyzed with SEM was a rock chip with yellow colonies from the floor of Hopkins Chocolate Cave, approximately 36 m into the cave. The second yellow mat sample (S-L280-110427-3) was taken approximately 60 m into Cave S-L280 and 1 m above the floor and 1 m below the ceiling.
Water chemistry analysis
Samples for chemical analysis were preserved with 6N hydrochloric acid in the field, as described in [29]. Amounts of chloride, nitrite, nitrate, phosphate, and sulfate were analyzed using a Dionex Ion Chromatograph DX-100 (Dionex, Sunnyvale, CA, USA) as described [29].
Molecular phylogeny
DNA extraction, sequencing, and sequence analysis.
DNA was extracted from triplicate samples of rock chips with microbial mats from each cave by mat color and from the surface soil samples using the MoBio Power SoilTM DNA extraction kit following manufacturer’s protocol except we used bead beating rather than vortexing, which the Northup Lab finds to be more effective at releasing DNA from Gram positive cells.
Polymerase Chain Reaction (PCR).
PCR was performed to verify the quality and quantity of the DNA prior to sequencing. One hundred twenty five to three hundred ng of purified DNA was used to amplify the 16S rRNA gene from environmental DNA by PCR with universal primers, p46 forward (5’-GCYTAAYACATGCAAGTCG-3’) and p1409 reverse (5’-GTGACGGGRGTGGTGTRCAA-3’; [30] and AmpliTaq LD (Applied Biosystems) with an MJ thermal cycler using: 4 min denaturation at 94°C followed by 35 cycles of 45 sec annealing at 55°C, extension for 2 min at 72°C, denaturation for 30 sec at 94°C, with a final 45 sec 55°C and a 20 min 72°C extension step after cycling was complete.
Sequencing and phylogenetic analysis.
Samples were analyzed with next-generation sequencing of the 16S SSU gene bacterial V1-3 region (primer 27F) using Roche FLX and Titanium 454 technology conducted by MR DNA, Shallowater, TX (http://www.mrdnalab.com/).
All 454 data were processed in QIIME 1.9.1 [31]. Quality control and trimming of the 454 dataset were done using the split_libraries.py command with a lower length (-l) of 100 bp and an upper length (-L) of 500. A quality score (-s) of 30 was chosen. Removal of erroneous sequences (denoising) and OTU clustering were done using pick_de_novo_otus.py pipeline with the sumaclust option [32]. The sumaclust algorithm is mainly useful to detect the 'erroneous' sequences created during amplification and sequencing protocols. OTUs were clustered at the 97% similarity level using sumaclust. The pick_de_novo command also picks the representative set and assigns taxonomy using uclust [33] against the greengenes 13.8 database [34]. The pipeline also aligns and builds a phylogenetic tree using pynast [35] and fasttree [36] from the representative sequence set. Chimera checking was done using USEARCH to detect artifacts created during sequencing. Good’s coverage showed that we were successful in getting nearly all of the diversity from our samples. Values ranged from 99.11% to 87.13% with an average value of 94.98%. Library sizes ranged from 471 to 4317. Rarefaction curves (S1 Fig) were generated using a custom function written by Bela Hausmann (https://github.com/joey711/phyloseq/issues/143) and is also available in our zenodo archive. These plots compare estimated total richness per sample at a given depth. The rarefaction curves are calculated with observed (total richness), Shannon, Chao1 and their associated uncertainties.
Diversity by phyla with the Proteobacteria separated out by class was compared for all samples. The L2 phyla data were reduced to eighteen groups, including unassigned phyla and a group we called minor phyla. The minor phyla are entries that had less than 1000 OTUs across all samples. Of a total of 140,848 OTUs, 26,609 were from the surface samples, and 38,214 from tan, 36,014 from white, and 40,011 from yellow mats. The process was repeated at the L6 genus level resulting in nine groups, an unassigned group, and minor genera with 471 taxa.
Sequences submitted to the NCBI GenBank database (www.ncbi.nlm.nih.gov/genbank/) were assigned Accession Numbers JX694094-JX702544, and the three additional white samples KP705489-KP706447.
Statistical analysis
Community dissimilarity was visualized using the phyloseq package [37] and ggplot2 [38] in R [39]. Alpha diversity was analyzed using observed OTUs, Shannon and Chao1 indices in the phyloseq package. Alpha diversity measures were carried out on the raw data as recommended [37]. Observed richness will scale with increasing library size; however both Chao1 and Shannon are robust measures of richness and diversity, which account for differences in library size (rarefaction curves in S1 Fig). Observed OTUs is the raw number of OTUs present in each sample of quality controlled and clustered sequences as described above. Beta diversity was analyzed using non-metric dimensional scaling (NMDS) with the Brays-Curtis distance using the vegan package [40] in R. The Brays-Curtis distance was picked because it is invariant to changes in units, unaffected by additions and removal of samples, and NMDS recognizes differences in total abundances when relative abundances are similar. The ordination of the samples was done using custom R scripts by Umer Zeeshan Ijaz available at: http://userweb.eng.gla.ac.uk/umer.ijaz/bioinformatics/ecological.html.
Differential abundance of taxa was characterized using the DESeq2 package [41] with parameters fitType = “local”. An adjusted p-value threshold of 0.1 was used to calculate log2 fold changes between surface soils and cave microbial mats.
Phylogenetic tree analysis was carried out in the phyloseq package. The data were subset by phyla that were differentially abundant as determined by DeSeq2. Tree files with tips label were written out using write.tree(phy_tree(phyloseq_obj),file = "phylum_name.newick"). The tree, tip labels, and traits (cave or surface) were loaded into Interactive Tree of Life v3 (http://itol.embl.de/) [42] for visualization. Traits were assigned to the tree tips as relative abundance of the OTUs present in the cave.
Total bacterial richness for cave samples was tested using a hierarchal generalized mixed effects model in R with the stanarm [43] package. Richness data were normalized in QIIME before modeling using normalize_table.py -i lavabeds.biom -a DESeq2 -o DESeq2_normalized_otu_table.biom–DESeq_negatives_to_zero. Cave, color of mat, visitation, and cave geometry were assigned as random effects (levels of hierarchy). The following options were used with stan_glmer: family = gaussian, dapt_delta = 0.95, prior = normal(location = 0, scale = 8), seed = SEED, iter = 50000, cores = 6.
Results
Water chemistry
Chemical analysis of water samples collected from each cave with available standing water is shown in Table 3. Chlorine levels averaged (5.086 ppm), slightly higher than the EPA Maximum Contaminant Level Goals for drinking water of 4 ppm [44]. EPA levels for nitrate (1 ppm) and nitrate (10 ppm) were not exceeded in most cave water samples. There are no phosphate level standards set by the EPA, but our cave samples were all low. Sulfate water standards are 250 ppm, and ours were all very low.
See Table 1 for the cave sample abbreviations. Bromide (not shown) all below detection limits.
Surface soil versus cave alpha diversity
Measures of alpha diversity among surface soil samples and cave samples by color of mat (Fig 3) show species richness (Observed and Chao1 [45]) and relative abundance (Shannon and Simpson’s). Simpson is less influenced by singletons (i.e. rare taxa) than Shannon’s Index, but Shannon’s in less sensitive to difference in library size. The means are very different, with much greater bacterial diversity in soil than cave. Soil bacterial diversity is more evenly distributed among OTUs. Cave samples are less evenly distributed, with a wide distribution of Simpson’s values with many outliers in contrast to soil samples. In-cave variation is much higher than surface soil variation. In terms of species richness in differently colored mats, tan samples have higher diversity, followed by yellow and then white. In terms of Shannon’s and Simpson’s indices, the three colors of microbial mats are not that different. Fig 3 shows the difference between cave microbial mats and surface soils in respect to total richness of each sample (observed OTUs) and Shannon Index. In all cases surface soils have more OTUs and higher Shannon indices indicating higher diversity. Rarefaction curves compare estimated total richness per sample at a given depth and are included in the supplemental information (S1 Fig). The rarefaction curves show a linear relationship between sampling depth and species richness. The Chao1 estimated richness and Shannon level off after 1000 and 500, respectively.
Box plots of surface soils and cave microbial mats by color. Surface soil samples are in blue and cave samples in red.
Taxonomy by phylum and proteobacteria class
Composition by the top five phyla and four proteobacterial classes, plus minor phyla based on OTUs, are presented in Fig 4. Comparisons of surface soil samples with the underlying cave samples show some differences. Most notable is the lower percentage of Actinobacteria in surface soil samples (21%) versus cave samples (39%), the reduction in the Nitrospirae in surface (3%) vs. cave samples (7%), and the increase in Alphaproteobacteria in the surface soil samples (17%) compared to the cave samples (10%). Smaller, but significant differences that decrease from surface to cave are seen with the Bacteroidetes (surface 8% and cave 2%); Gemmatimonadetes (surface 3% and cave <1%); and Planctomycetes (surface 2% and cave 1%). The Gammaproteobacteria increase between surface (18%) and cave (20%), while Betaproteobacteria (surface 4% and cave 3%) and Deltaproteobacteria (surface 4% and cave 1%) both decline in cave samples. None of the phyla differed significantly by mat color. Some differences were observed by mat color in the Actinobacteria (tan: 37%, white: 39%, yellow: 44%), Gammaproteobacteria (tan: 19%, white: 24%, yellow: 20%), Nitrospirae (tan: 11%, white: 4%, yellow: 7%), and Betaproteobacteria (tan: 3%, white: 6%, yellow: 2%).
Major and minor phyla by all cave samples and all surface samples. Cave samples have a greater relative abundance of Actinobacteria and Nitrospirae.
Molecular phylogeny by family or genus
The percentage of OTUs by genus or family is presented in Fig 5 for surface soil and cave mat samples. Results from the analysis of diversity by genera are interesting in several respects. First of all, a large percentage of OTUs are novel. Of the nine major genera, only four are identified to genus level, four to family, and one to class. There is also a group called unassigned, which could not even be classified to bacteria. These unassigned OTUs may represent archaeal DNA that amplified with our primers or may be organisms that are particularly difficult to classify through 454 sequencing, like members of the Verrucomicrobia. Bergmann et al. [46] used barcoded pyrosequencing with surface and subsurface soils to reduce primer bias, and found that Verrucomicrobia are ubiquitous and were often the dominant phylum in their samples. In our study, no individual Verrucomicrobia genus met our criteria for inclusion in Fig 5.
(A) mean surface soils and (B)mean cave microbial mats.
A major difference between surface and cave samples is the proportion of the family Pseudonocardiacae present with 14% of surface soil OTUs vs. 30% of cave OTUs. Other major genus level OTUs fell within the Acidobacteria, Nitrospiraceae, and Bacteroidetes. Within the Acidobacteria there are three groups, all belonging to the candidate class Chloroacidobacteria. There are two orders PK29 and PK10 that are not identified further. The Nitrospiraceae is classified to the genus Nitrospira. The Bacteroidetes group is identified to the family Chitinophagaceae, which are elevated in soils where insect chitin is higher than in the cave, a defining characteristic of this family [47], although new isolates from oligotrophic lake waters lacked this ability [48].
The remaining major bacterial OTU groups are all members of the Proteobacteria, with three from the Alphaproteobacteria and three Gammaproteobacteria. The Alphaproteobacteria OTUs are two members of the order Rhizospiriales, one family Hyphomicrobiaceae and one Methylobacteriaceae, and one Rhodospirillaes identified to the genus Magnetospirillum. Together the Rhizospiriales account for only 2% of cave species, but 5% of surface species, probably associated with the rhizosphere. The Gammaproteobacteria include the class Gammaproteobacteria and two members of the Sinobacteraceae with one family, Sinobacteraceae, and the genus Steroidobacteria. The type species is Steroidobacteria denitricans, a nitrate oxidizer [49], and the family Sinobacteraceae contains only two described species.
Dissimilarity among surface soils and cave microbial mats
Non-metric dimensional scaling (NMDS) in Fig 6 shows a separation of the cave microbial mat samples from the surface soil samples, indicating that the two sets of samples are not similar to each other. The surface soil samples cluster tightly in the far right, while the cave samples spread out to the left. To fill out the dataset with one white sample per cave, we had LABE personnel obtain additional samples from GE-L350, Hopkins Chocolate, and S-L280 Caves in 2012. These three samples (GEM2, HC1, SC2) have very low numbers of OTUs and diversity compared to the other samples collected in 2011, although they have comparable overall sequence numbers to some of the original samples (e.g. GE2). Tan and yellow mats cluster together, while white mat samples are widely spread out and group into three clusters. Within cave similarity is greater than between cave similarity, probably due to the three mat colors sampled in each cave, and suggests the necessity for multiple samples to cover the diversity comprehensively.
NMDS separates out lava cave mat communities at the phylum level, with Proteobacteria split out by class, from the overlying surface soils. Circles show the 95% confidence interval.
Differential abundance of OTUs between surface and cave
DESeq2 (differential analysis of count data) between all surface samples and all cave samples showed strong evidence for differential abundance of OTUs between surface soils and cave microbial mats (Fig 7). Eighteen phyla stood out as being significantly differential over or under represented between cave microbial mats and surface soils. The major phyla and Proteobacteria class that are significantly higher in the cave than in surface soil are the Actinobacteria (p>0.00000) and Nitrospirae (p = 0.00232). Major groups with greater representation in surface soils than the cave are Gammaproteobacteria (p>0.00000), Verrucomicrobia (p = 0.00001), Bacteroidetes (p = 0.00021), and Alphaproteobacteria (p = 0.02373). Of the minor phyla, GAL15 (p = 0.00273), WS3 (p = 0.03444), and SBR1093 (p = 0.04227) have higher numbers in the cave samples compared to surface soils. The groups with higher amounts in surface soils than cave samples are TM7 (p>0.00000), OD1 (p>0.00000), Armatimonadetes (p>0.00000), FBP (p = 1e−05), Cyanobacteria p = 0.00438), Fibrobacteres (p = 0.00698), and Elusimicrobia (p = 0.04176). The remaining two minor phyla are approaching significant differences; Planctomycetes (p = 0.05177) and NC10 (p = 0.05655).
The band is the median, and the box delineates the upper and lower quartile. The whiskers show the maximum and minimum values. All data points are shown.
Scanning electron microscopy
The three samples (one white, two yellow) examined with scanning electron microscopy (SEM) revealed extensive microbial morphologies present with similarities and differences observed across the three samples. The white sample from Cave L-V460 contained the most unusual morphologies, which were filaments covered in curly putative pili/fimbrae with spheroids emerging from the tips (Fig 8A–8C). The spheroids range in size from 0.8 to 1.2 μm in diameter, and in Fig 8B, one can observe what appears to be a “neck” on one of the spheroids. Some of the filaments with numerous hair-like extensions appeared to be segmented (Fig 8B). Some parts of the sample also contained smooth, long filaments, while other areas had small spheroids (0.8 μm in diameter) emerging from the biofilm (not pictured). A somewhat similar morphology (filaments with spheroids protruding from the ends) was observed in a yellow sample from Hopkins Chocolate Cave (Fig 8D). This sample also had many colonies covered by, or partially emerging from, a lawn of biofilm (not pictured). The second yellow sample, from Cave S-L280, had extensive biofilm that appeared to bury filaments, and was dotted with large colony masses. We interpret the fuzzy areas along the margins of the colonies (Fig 8E) as being biofilm. Many strands of beads-on-a-string morphologies are seen on the colonies and biofilm (Fig 8F).
(A) White microbial mat from Cave L-V460 showing filaments covered with putative pili/fimbrae with spheroid shapes emerging from the ends. (B) Close-up of these morphologies from A. (C) Overview of a field of these morphologies in the same white microbial mat from Cave L-V460. (D) Yellow microbial mat from Hopkins Chocolate Cave showing similar morphologies to those seen in images A-C, and including some biofilm and smooth filaments. (E) Overview of a yellow colony from Cave S-L280, showing extensive biofilm in the background and on colony edges. Beads-on-a-string morphologies are observed lying on the background biofilm. (F) Close-up of beads-on-a-string morphology on biofilm from Cave S-L280 yellow microbial mat.
Energy dispersive X-ray spectroscopy revealed the presence of high silica and oxygen peaks, along with a carbon peak from the biofilm and microorganisms. The silica is precipitated as coatings on the basalt samples, inter-grown with biofilm and microorganisms.
Cave environmental and geometry effects on richness
Several variables (Tables 1 and 2) collected from the field were tested to determine how much environmental conditions and cave geometry contribute to overall bacterial richness in the cave microbial mats. Temperature in the fall of 2012 was a positive predictor (41.1) of richness, while relative humidity in the fall was positive, but a small predictor of richness (6.2). Microbial mat color contribution to richness varied depending on color. Tan mats were a positive predictor (47.9). Yellow and white mats were negative predictors of richness (-11.5 and -35.1, respectively). Cave geometry variables were both positive and negative predictors. The number of entrances was a positive predictor (69.7). The complexity (nodes) of the cave was a negative predictor (-6.2). Caves with split (V-shaped) geometry were negative predictors (-20.2), while tube shaped was positive (21.9). With the exception of two caves (GD 20.6 and HC -19.8), caves were not predictive of overall richness. Neutral predictors (~ -3 to 3) were elevation of cave, overall length, distance from entrance, age of lava flow, and human impact.
Discussion
The general belief for decades has been that cave microorganisms are a subset of the microbial community found in surface environments overlying the cave [50], eking out a minimal chemoheterotrophic existence in the cave. Our study is the first to test the hypothesis that bacterial communities in the soil overlying lava caves are substantially different from the bacterial communities found in the cave microbial mats. Our study is the most robust to date of lava cave microbial diversity, with surface soil and three colors of microbial mats (tan, white, and yellow) from each of seven different caves in Lava Beds National Monument, CA, USA.
Cave and surface soil sample bacterial communities differ in major ways
Soil overlying caves can be a source of bacteria for colonization of underlying caves; however, the actual overlap in OTUs in our study is only 11.2%, suggesting that soil microorganisms either do not make it into the cave or die off once there. However, there are some similarities at bacterial higher taxonomic levels, i.e. phylum, between cave and surface, and between our soils and those of other regions. Both surface soils and cave microbial mats contain the phyla Actinobacteria, Alpha-, Beta-, Delta-, and Gammaproteobacteria, Acidobacteria, Chloroflexi, Nitrospirae, Verrucomicrobia, Gemmatimonadetes, Planctomycetes, and Bacteroidetes. These findings overlap with Janssen’s review [51] of studies of the microbial diversity of soils from many environments found in clone libraries. Across many habitats, these studies found that soils are dominated by Proteobacteria, averaging 39% of soil bacteria, and by Acidobacteria, Actinobacteria, Chloroflexi, Verrucomicrobiota, Planctomycetes, Gemmatimonadetes, Bacteroidetes, Firmicutes, and other/unknown. Our soils are fairly similar, but lack a significant amount of Firmicutes.
While not significantly different between cave and surface or among mat colors, Acidobacteria are a major phylum across all samples. Jones et al. [52] studied Acidobacteria across 87 soils and found them to be both ubiquitous and abundant. Their abundance relative to other taxa was variable, but correlated strongly with pH (R = -0.80, p<0.001). They suggest that pH is an effective habitat filter for Acidobacteria, with the greatest abundance below pH 5.5. Our pH values are from dripwater and pools and show no effect on bacterial diversity in LABE. Future studies will analyze pH of the rock substrate to determine if pH is a relevant controlling factor for diversity.
A study of Icelandic lava caves by Northup et al. [53] compared surface soil microbial communities with microbial communities of different types (mats, slimes, snottites, organic oozes, and soil) from four lava caves. Surface soils were the most diverse at two sites. As in our study, the mats were similar at the phylum level with Actinobacteria dominant followed by Acidobacteria, and Alpha-, Beta-, and Gamma- Proteobacteria. Unlike our study, there was a lack of Nitrospirae in the Icelandic caves. They found sample type to be the most important factor in bacterial diversity.
Ortiz et al. [25] collected surface soils from three locations above Kartchner Cavern. Analysis of their three soil samples was limited to qPCR for comparison of domain distributions within their karst cave speleothem swab samples. They found bacterial abundance in the cave was comparable to their soil samples, but Archaea were significantly higher in the cave, and fungi were below the detection limit in cave samples. Comparison of cave samples with one of the soil samples from above the cave showed an overlap of 16% of OTUs between cave and surface [24], which is comparable to our study of seven cave and surface soil samples, with an overlap of 11.12% OTUs. Our study and the Ortiz et al. [24,25] studies support the hypothesis that cave bacterial communities are significantly different from surface soil bacterial communities.
Actinobacteria are more abundant in cave samples
A major difference between surface and cave bacterial communities in our study, and in comparison to the Ortiz et al. [24] study in a carbonate cave, is the abundance of Actinobacteria. While Actinobacteria occurred in moderate numbers in surface soils in our study (21%), they occurred at much higher levels in cave samples (39%). Actinobacteria, a dominant group in the 454 sequence libraries, are also probably some of the dominant forms seen in the scanning electron microscopy (e.g. Fig 8).
Barka et al. [54] reviewed Actinobacteria taxonomy, physiology, and secondary metabolite production. Actinobacteria are a large phylum among the Bacteria. Most Actinobacteria are heterotrophic, feeding on organic carbon, and some are known to fix nitrogen both as symbionts and free-living. New studies have also established that they can be chemolithoautotrophic, such as in the case of the uncultured TM3 subdivision of Actinobacteria that have been shown to exhibit nitrate-dependent iron oxidation [55]. Actinobacteria are key members of the microbial community in caves (e.g. [56]). Actinobacteria include organisms that give caves their musty odor due to the chemical geosmin [57]. Snider et al. [58] showed that Actinobacteria within two lava caves with roots penetrating the ceiling tended to be where more moisture is available, and their numbers fall off substantially in drier areas of the caves.
The color of the microbial mats may be produced by the bacteria present, in particular Actinobacteria. Production of melanoid pigments varies by strain, nutrients, and age [54]. Pigment colors range from red, yellow, orange, blue, green, and black [54]. Porca et al. [56] also reported that several Pseudonocardiaceae (relatives of their OTU group III) were shown to produce yellow pigments [59,60]. Lee [61] identified two new members of the Pseudonocardiaceae from soil and dry bat guano from a cave in the Republic of Korea that produced yellow and grey-white colored colonies.
Within the Actinobacteria recovered from our cave samples, the dominant organisms belonged to the family Pseudonocardiacea. Barton et al. [62,63] hypothesized that the Pseudonocardiaceae-related phylotypes found at one of their sites are degraders of plant matter coming into the caves. The Pseudonocardiacea are also known to produce a variety of secondary metabolites [64]. One note of caution is that universal bacterial primers are unable to amplify the 16S rRNA gene in many Actinobacteria despite 100% homology to the primers [65]. Thus, it is entirely likely that both our study and others are missing a portion of the Actinobacteria present in the environments sampled.
Broader implications of actinobacterial diversity in caves: The search for bioactive compounds.
Caves are an extreme habitat and the cave microbiome has great potential as a novel resource for drug discovery [54,66,67]. Antimicrobials are one example of secondary metabolites produced by microorganisms, but may have broader roles as toxins, ionophores, bioregulators, and signal molecules produced by “metabolically talented” microbes [68]. Actinobacteria produce 2/3 of antibacterial agents in use, but are also important in biotechnology, medicine, and agriculture. A review of their extensive secondary metabolites by Barka et al. [54] includes antibacterials, anifungals, antivirals, antihelminths, antitumor, immunosuppressive and immunostimulatory agents, biopesticides, herbicides, and plant growth promoters. Natural products are the most promising source of novel antibiotics and cave environments have great potential for the development of new bioactive microbial metabolites especially from Actinobacteria, which are dominant in lava caves [69].
Other bacteria of interest in lava cave microbial mats
In addition to the Actinobacteria, the gammaproteobacterial orders Xanthomonadales and unclassified Gammaproteobacteria were slightly elevated in cave samples (20%) in comparison to surface soil communities (18%). The order Xanthomonadales shows up in other cave studies on NCBI (i.e. accession numbers: DQ066611, FJ347998, HM592533). A poorly characterized taxon, 0319-6A21, shows up with much greater abundance in the lava caves than in the surface soils. 0319-6A21 was first isolated in 2004 from Australian desert soil [70], and is classified in the class Nitrospirales and may provide clues into nitrogen cycling in caves. These results, together with the elevated Pseudonocardiacea, parallel the findings of Porca et al. [56] who found these groups to be major constituents of microbial communities in their studies of yellow colonies in carbonate caves in Spain, Slovenia, and the Czech Republic. Their study revealed three major core groups that included the Pseudonocardiacea, the Chromatiales, and the Xanthomonadales, which Porca et al. [56] suggest may be “true cave dwellers.”
An observation from the comparison of soil and cave samples at the genus level is the occurrence of many OTUs that were present in smaller numbers (genera that were present in three or fewer samples with less than 100 sequences), which we grouped together into the category minor genera. The genus level diversity in the minor genera in surface samples represents 35% of the diversity in surface soils samples and 20% of the diversity in cave samples.
One of the few lava cave microbial mat studies that extends the discussion of bacterial diversity to the level of genus is the Hawaiian study from 16 caves by Spilde et al. [71]. The samples included white and yellow microbial mats, along with pink-orange mats. At the genus level they reported an abundance of Bacillus, which were very rare in our LABE samples, Nitrospira, and two new genera of Actinobacteria: Crossiella and Euzebya.
One bacterial community is different
Valentine 13 is not typical of either cave or surface samples. Betaproteobacteria (36.92%) dominate, while Actinobacteria make up only 8.56% of the total phyla. At the L6 genus level, only 1.44% of sequences are Pseudonocardia. Several variables may account for the differences. The site is located at a junction where the passage splits into two tubes. Very near the sample site is a large pile of breakdown, the only occurrence of breakdown of this size in the entire cave. The breakdown includes packed, wet sediment that has fallen through cracks in the ceiling above the pile, and the site is wetter than other locations in the cave. LABE personnel have observed millipedes and springtails around the edges of the breakdown, indicating a small connection with the surface. We speculate that the ceiling connection may be a source of microbes or nutrients unique to Valentine 13. The shape of the passage, with a higher ceiling at the collection site, may also direct more visitors to pass by the sample site, possibly also affecting the community structure or nutrient inputs. The floor also changes from pahoehoe to a’a lava at this point, perhaps indicating a change in rock chemistry. Regardless, we kept the sample in for analysis, and point out the need for multiple samples to mitigate such variations.
Phylogenetic analysis of surface soils and cave microbial mats
The Nitrospirae phylum illustrates differences between OTUs found in cave samples versus surface soil samples. The phylogenetic tree of the Nitrospirae (Fig 9) shows the proportion of OTUs between cave and surface samples colored by family. The Nitrospirae vary by family in terms of whether they are primarily found in the cave vs. surface samples. The major Nitrospirae in our samples are found in three families, the Nitrospiraceae, the Leptospirillaceae, and a candidate family, 0319-6A21. The majority of the 0319-6A21 family is found in the cave microbial mats and the remaining families both in the surface soil and the cave microbial mats. A similar tree was constructed for Actinobacteria (S2 Fig); however the proportion of cave taxa to surface taxa was nearly 50:50.
Approximate maximum likelihood, mid-point rooted tree with pynast aligned 454 sequences.
Bacterial diversity among different colored mats
Bacterial composition was not substantially different among the mat colors, with some interesting differences. Members of the Actinobacteria dominated lava cave microbial mats of all colors. The Actinobacteria vary somewhat by mat color (S3 Fig), with 37% in tan mats, 39% in white mats, and 44% in yellow mats. Gammaproteobacteria varied the most by mat color with 19% in tan, 37% in white, and 21% in yellow.
Northup et al. [14] compared white and yellow microbial mats and selected secondary minerals based on color from four lava caves in each of three different locations; tropical and semi-arid lava caves in Hawaiˋi, temperate lava caves in the Azores, and semi-arid lava caves in New Mexico. Mats are more extensive in areas with greater rainfall. They found 13 phyla across all white and yellow mats from all three locations. The number of phyla per cave ranged from 5–11, with slightly greater diversity at the phylum level in yellow over white mats, but not in all phyla.
Hathaway et al. [21] was the first in-depth comparison of white and yellow microbial mat communities from lava caves on different archipelagos, but did not conclude that mat color was significant among differently colored mats. The authors concluded that geographic location is important in determining the composition of microbial communities, with Hawaiˋi showing greater diversity than Terceira. Principle Coordinate Analysis (PCoA) by island showed that only 13.7% of the variability was explained by geographic location and levels of nitrogen, organic carbon, and copper. Rainfall, especially in Hawaiˋi, accounted for another 10.8% of the variability. PCoA by cave showed a strong influence of geography. Actinobacteria (16%) dominated the clones recovered from Hawaiian samples, while on Terceira Acidobacteria dominated with 21% of the clones recovered. Alphaproteobacteria made up 13% of the diversity in Hawaiˋi and 15% in Terceira. Bacterial sequences recovered with no known phyla were 14% of the total in Hawaiˋi and 12% in Terceira. A nearest neighbor comparison among caves in Hawaiˋi and Terceira with other caves in the ARB database showed only thee OTUs (0.22%) overlap. This high level of alpha diversity means that bacterial communities found in each cave are different. These differences may be partially due to the greater range of habitats in Hawaiˋi, ranging from semi-arid to tropical rainforest conditions. Novel bacteria were found in all sites showing the need for conservation of caves as sources of novel bacteria and as simplified natural ecosystems for study of larger ecological questions.
Riquelme et al. [72] studied white, yellow, and tan mats from 14 lava caves on two islands in the Azores using 16S rDNA clone libraries. Environment and chemistry showed no relation to OTU diversity and composition of the microbial mats. Similar to other studies, including ours, there was a dominance of cosmopolitan OTUs. There was a greatest influence on β diversity (composition) between islands and caves than α diversity (community). The absence of clear differences across mat colors could be due to insufficient geologic time on these islands for microbial communities to diversify, and/or convergent evolution due to the selective pressure of extreme environments in caves.
We predict that a comparison of LABE microbial mats with those of the Azores, New Mexico, and Hawai`i, as new sequencing technologies become available for these sites, will show lower diversity in LABE than in Hawai`i in particular. Hathaway et al. [21] found that yellow and white color of the mat was not a predictive factor in diversity. Our findings in LABE were that yellow and white mat color was a negative predictor of diversity, but tan color was a strong positive predictor.
Nitrogen cycling bacteria in LABE caves
Nitrogen is often a limiting nutrient in oligotrophic environments. Our cave sequences document the presence of several key organisms in the nitrogen cycle, some of which occur in elevated numbers in comparison to surface soil samples. We found Nitrospirae, which contain nitrite oxidizers and a Nitrospira spp. that can carry out complete nitrification, to be higher in cave samples (7%) in comparison to surface soil samples (3%). Nitrospirae OTUs varied from nine to 1064 across different caves and colored mats, with the highest number of OTUs occurring in tan mats from Golden Dome and GE-L350 Caves. The Nitrospirae contain nitrite oxidizers, the Alphaproteobacteria include nitrogen fixers, and Betaproteobacteria contain ammonia oxidizers, all key players in the nitrogen cycle. A new genus of Nitrospira has been described by Daims et al. [73] that can complete the entire nitrification cycle, taking ammonia to nitrate. Differences from cave to cave could be further investigated with a study of nitrogen available in the basaltic matrix and from infiltrating waters. Nitrite levels in our limited samples (Table 3) were largely low, but three caves had slightly elevated nitrate levels above the limit for drinking water of 1.0 ppm, and may indicate either surface inputs with nitrate or in situ production in these caves.
Other studies have suggested the importance of nitrogen-based systems in caves. Hathaway et al. [74] investigated the diversity of ammonia oxidation (amoA) and nitrogen fixation (nifH) genes in lava caves of Terceira, Azores, Portugal. They found that Nitrosospirae related sequences dominated the ammonia-oxidizing bacteria and that a key nitrogen fixation gene, nifH, was found among Klebsiella pneumoniae-like sequences (Gammaproteobacteria). Tetu et al. [75] found evidence that microbial slime curtain communities in the submerged Weebubbie Cave under the Nullarbor Plain in Australia had primary productivity based on the combined activity of ammonia-oxidizing Archaea and bacterial nitrite oxidizers, especially Nitrospirae. The ability of microorganisms to cycle nitrogen is key to supplying community needs in an oligotrophic environment. Future studies of LABE caves will employ metagenomics and transcriptomics to investigate the range of capabilities that LABE microorganisms possess and employ in the caves.
Impact of human visitors on bacterial diversity
We divided our LABE study caves into high visitation (about 30,000 people a year) and low visitation, which have controlled access and are rarely visited by researchers and park personnel (up to 10 visitors some years). High and low visitation caves (Table 1), are comparable in terms of alpha diversity and show no significant differences in microbial community structure. When you make the jump to Carlsbad Cavern with over 400,000 visitors per year, Griffin et al. [76] reported significant human impacts of visitors on the microbiota. Using cultivation, they reported that Staphylococcus spp. were the dominant bacteria in the air along tourist trails compared to Knoellia spp. off trail. Knoellia is a new genus of Actinobacteria first isolated from a cave in China [77]. Fungal spores of Penicillium and Aspergillus showed a general decrease with distance from the entrance, but with a peak in the Lunchroom where visitors rest, eat, and wait for the elevator to return to the surface. A second study in Griffin et al. [76] using molecular techniques showed Enterobacteriaceae dominating along the descent rail and in the Lunchroom. They concluded that humans were important sources of non-indigenous microorganisms into Carlsbad Cavern, and recommended mitigation steps. There was only one Staphylococcaeae identified, and only 25 OTUs of Enterbacteriacae identified to the level of family detected at any LABE cave or surface sample. What we may be seeing is a threshold of visitors before we see human impacts. Right now, that threshold may be somewhere between 30,000 and 500,000 visitors per year, and merits further study.
Environmental variables impact bacterial diversity differences
Levels of PO4-, SO42, NO3- in LABE lava caves were compared to those reported in lava caves from Terceira Island, Azores [74]. Phosphate levels in LABE samples are low, but much higher than in Terceira (ranges LABE 0.4697–0.6715 ppm vs. Terceira 0.003–0.0696 ppm), much lower in sulfate (ranges LABE 0.3251–1.1231 ppm vs. Terceira 1.42–37.29 ppm), and higher in nitrate (ranges LABE 0.5483–8.0036 ppm vs. Terceira 0.08–6.92 ppm). The nitrate results are unexpected given the large number of cattle roaming the surface above Terceira lava caves. Pools in Lechuguilla Cave [78], a carbonate cave in NM, had a more narrow range of nitrate values (ranges LABE 0.5483–8.0036 ppm vs. Lechuguilla 0.740–1.90 mg/L (ppm)) and a much greater range of sulfate values (ranges LABE 0.3251–1.1231 ppm vs Lechuguilla 28 to 112 mg/L (ppm)).
The strongest positive predictors of LABE bacterial diversity were the number of entrances to the cave and tan color. Tan mats also had the highest overall diversity. White and yellow colors were both negative predictors of diversity. Cave geometry varied as a predictor, with complex branching patterns (nodes) negative while a simple tube shape was a positive predictor of diversity. The cave itself had little effect on predicting diversity. All other variables were weak or neutral predictors of diversity, although temperature in Fall 2102 was a moderate predictor.
Cave geomicrobiological studies show that caves are nutrient-limited with redox interfaces in microniches where we see the interaction of microbial activities and minerals [79,80]. Interactions with minerals may be the dominant force driving microbial diversity in lava caves [81], but further study of basalt geochemistry at LABE is needed to test this hypothesis in the caves, but it seems likely that mineralogical factors do influence within cave diversity differences.
Conclusions
Overlap in OTUs between surface and cave communities at LABE is only 11.12%, revealing that the cave bacterial communities differ substantially from surface soil communities and are not a subset of surface soils as previously assumed. Of particular significance are the differences in Actinobacteria, Alphaproteobacteria, Nitrospirae, and Gammaproteobacteria composition between surface soil and cave samples. Surface soil diversity is higher than that observed in cave samples and considerable novel diversity exists in both surface soil and cave mat samples. However, because many of the earlier studies are based on clone sequences that are limited in number, this conclusion may change as newer sequencing is applied to these sites. Communities in different mat colors do not differ substantially in composition, similar to the findings of Hathaway et al. [21]. A combination of cave geometry and environmental parameters influence microbial richness in lava caves.
LABE microbial cave community diversity at the phylum and Proteobacteria class level is comparable to diversity found in other major lava cave areas in New Mexico [8,10], Hawaiˋi [21,71], and the Azores [21,56,69]. Communities in different mat colors show significant overlap, but with many unique members that may contribute to the color of the mats in our study. So what does account for the differences in mat colors in lava caves? Caves are not homogeneous environments. There are zones related to distance from the entrance; seasonal variations; three-dimensional geometry; differential cooling and deposition of minerals during lava flows; and microhabitats that may vary at the level of the individual mineral grain. We think species composition and mineral availability at the microhabitat scale accounts for the color differences seen in the microbial mats.
Our study is the most extensive bacterial diversity study of lava caves to date that compares the bacterial diversity in three colors of microbial mats across seven caves with surface soil overlying each cave. Variability inherent in sampling supports the need for replicated study of microbial community structure. Further studies should examine diversity in other lava cave areas around the world using newer sequencing technologies. There are preliminary indications of world-wide microbial biogeography and we should work to fill in the gaps.
Supporting information
S1 Fig. Rarefaction curves by alpha diversity for microbial mats and surface soils.
https://doi.org/10.1371/journal.pone.0169339.s001
(PDF)
S2 Fig. Phylogenetic tree of Actinobacteria by LABE lava cave and surface soil.
Approximate maximum likelihood, mid-point rooted tree with pynast aligned 454 sequences. Outer bars show relative proportion of taxa found in cave microbial mats and surface soils. Inner band is colored by genus.
https://doi.org/10.1371/journal.pone.0169339.s002
(PDF)
S3 Fig. Cleveland dot plot of relative abundance of OTUs.
Showing major and minor phyla for all cave and surface samples.
https://doi.org/10.1371/journal.pone.0169339.s003
(PDF)
Acknowledgments
We thank personnel at Lava Beds National Monument, especially Shane Fryer, Shawn Thomas, Katrina J. Smith, Emma Brinley Buckley, and David A. Riggs for assistance with field work, collection of mat samples and environmental data, and their continuing enthusiasm for the project. Shane helped us secure funding for the project. Bruce Rogers, Joke Vansweevelt, and Pat Helton did sample collection, TDS, and climate monitoring in the caves, and surface support during the five days of the expedition. We’re especially thankful for Bruce Rogers’ help in drafting maps of the climate measuring stations and getting us interested in investigating LABE bacterial diversity. Kenneth Ingham was extremely helpful in photographing the caves, microbial mats, and science in action. We thank Abdul-Mehdi S. Ali for his analysis of our water samples.
Author Contributions
- Conceptualization: DEN.
- Data curation: DEN.
- Formal analysis: ASW KJHR KHL DEN.
- Funding acquisition: DEN.
- Investigation: KJHR EMH MNS DEN.
- Methodology: DEN KJHR ASW.
- Project administration: DEN.
- Resources: KJHR DEN.
- Software: ASW.
- Supervision: DEN.
- Validation: DEN ASW KHL.
- Visualization: KHL DEN ASW.
- Writing – original draft: KHL DEN ASW.
- Writing – review & editing: KHL DEN ASW.
References
- 1. Edwards KJ, Becker K, Colwell F. The deep, hot energy biosphere: Intraterrestrial life on earth. Ann Rev Earth Planet Sci. 2012;40: 551–568.
- 2.
Depth records. Available: www.karstworlds.com/2012/08/krubera-voronya-record-depth-record
- 3.
Palmer AN. Cave Geology. Dayton, OH: Cave Books; 2007.
- 4. Léveillé RJ, Datta S. Lava caves and basaltic caves as astrobiological targets on Earth and Mars: A review. Planet Space Sci. 2010;58: 592–598.
- 5. Boston PJ, Ivanov MV, McKay CP. On the possibility of chemosynthetic ecosystems in subsurface habitats on Mars. Icarus. 1992;95: 300–308. pmid:11539823
- 6. Boston PJ, Spilde MN, Northup DE, Melim LA, Soroka DS, Kleina LG, et al. Cave biosignature suites: Microbes, minerals, and Mars. Astrobiol. 2001;1: 25–55.
- 7. Northup DE, Welbourn WC. Life in the twilight zone-Lava-tube ecology. NM Bur Mines Mineral Res Bull. 1997;155: 69–81
- 8. Northup DE, Connolly CA, Trent A, Peck VM, Spilde MN, Welbourn WC, et al. The nature of bacterial communities in Four Windows Cave, El Malpais National Monument, New Mexico, USA. Assoc Mex Cave Stud Bull. 2008;19: 119–125.
- 9.
Garcia MG, Moya M, Spilde MN, Stone FD, Northup DE. 2009. Discovering new diversity in Hawaiian lava tube microbial mats. In: White WB (ed). Proc 15th Int Congr Speleol, Kerrvile, TX, USA, July 19–26, 2009. 2009; 1, Symposia Part 1: 364‒369.
- 10. Moya M, Garcia MG, Spilde MN, Northup DE. Composition of bacterial mats in El Malpais National Monument, New Mexico, USA: Comparisons and contrasts with bacterial communities in Hawaiˋi lava tubes. Proc 15th Int Congr Speleol. 2009;2: 709–713.
- 11.
Northup DE, Hathaway JJM, Snider JR, Moya M, Garcia MG, Stone FD. Life in Earths lava caves: Implications for life detection on other planets. In: Hanslmeier A, Kempe S, and Seckbach J (eds.) Life on Earth and Other Planetary Bodies. Cellular Origin, Life in Extreme Habitats and Astrobiology Series. Berlin: Springer. 2012; p 459–484.
- 12.
Northup D, Lavoie KH. 2015. Microbial diversity and ecology of lava tubes. In: Life in Extreme Environments: Microbial Life of Cave Systems. Series Ed: Wagner R. Volume Ed. Engel AS. De Gruyter. 2015;3: 161‒192.
- 13. Forti P. Genetic processes of cave minerals in volcanic environments: an overview. J Cave Karst Stud. 2005;76: 3–13.
- 14. Northup DE, Melim LA, Spilde MN, Hathaway JJM, Garcia MG, Moya M, et al. Lava cave microbial communities within mats and secondary mineral deposits: Implications for life detection on other planets. Astrobiol. 2011;2: 1–18.
- 15. White WB. Secondary minerals in volcanic caves: data from Hawaiˋi. J Cave Karst Stud. 2010;72: 75–85.
- 16.
Stoner MF, Howarth FG. Community structure and niche differentiation in Hawaiian lava tubes. In: Island Ecosystems: Biological Organization in Selected Hawaiian Communities. Mueller Dombois D, Bridges KW, Carson HL (eds) Hutchinson Ross Publ. Co., Stroudsburg PA, 1981; pp 318–336.
- 17. Poulson TL, White WB. The cave environment. Science. 1969;165: 971–981. pmid:17791021
- 18.
Culver DC, Pipan T. The Biology of Caves and Other Subterranean Habitats. Oxford UK. Oxford University Press; 2009.
- 19. Christman MC, Culver DC, Madden MK, White D. Patterns of endemism of the eastern North American cave fauna. J Biogeogr. 2005;32: 1441–1452.
- 20. Culver DC, Master LL, Christman MC, Hobbs HH III. Obligate cave fauna of the 48 contiguous United States. Conserv Biol. 2000;14: 386–401.
- 21. Hathaway JJM, Garcia MG, Moya Balasch M, Spilde MN, Stone FD, Dapkevicius MdLNE, et al. Comparison of bacterial diversity in Azorean and Hawaiian lava cave microbial mats. Geomicrobiol J. 2014a;31(3): 205–220.
- 22. Laiz L, Groth I, Gonzalez I, Saiz-Jimenez C. Microbiological study of the dripping waters in Altamira Cave (Santillana del Mar, Spain). J Microbiol Meth. 1999 36: 129–138.
- 23. Simon KS, Benfield EF, Macko SA. Food web structure and the role of epilithic biofilms in cave streams. Ecology. 2003;84: 2395–2406.
- 24. Ortiz M, Neilson JW, Nelson WM, Legatzki A, Byrne A, Yu Y, et al. Profiling bacterial diversity and taxonomic composition on speleothem surfaces in Kartchner Caverns, AZ. Microb Ecol. 2013;65: 371–383. pmid:23224253
- 25. Ortiz M, Legatzki A, Neilson JW, Fryslie B, Nelson WM, Wing RA, et al. Making a living while starving in the dark: metagenomic insights into the energy dynamics of a carbonate cave. ISME J. 2014;8(4): 478–491,
- 26.
Lava Beds National Monument, CA. http://www.nps.gov/labe/
- 27.
Donnelly-Nola JM, Natheson M, Champion DE, Ramsey DW, Lowenstern JB, Ewert JW. Volcano hazards assessment of Medicine Lake Volcano, Northern CA. US Geolog Surv Sci Invest Rep. 2007-5174-A. 26 p.
- 28. Giovannoni SJ, DeLong EF, Schmidt TM, Pace NR. Tangential flow filtration and preliminary phylogenetic analysis of marine picoplankton. Appl Environ Microbiol. 1990;56: 2572–2575. pmid:2403262
- 29.
Pfaff JD, Hautman DP, Munch DJ. 1997. Method 300.1 Determination of inorganic anions in drinking water by ion chromatography. Cincinnati, OH: National Exposure Research Laboratory, Office of Research and Development, U.S. EPA.
- 30. Northup DE, Snider JR, Spilde MN, Porter MN, Van deCamp JL, Boston PJ, et al. Diversity of rock varnish bacterial communities from Black Canyon, New Mexico. J Geophys Res. 2010;115
- 31. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows analysis of high-throughput community sequencing data. Nature Meth. 2007;7(5): 335–336.
- 32. Mercier C., Boyer F, Bonin A., Coissac E. SUMATRA and SUMACLUST: fast and exact comparison and clustering of sequences. SeqBio Invited Talks. 2013:27–29 Available: http://www.gdr-bim.cnrs.fr/seqbio2013/wp-content/uploads/2013/12/seqbio2013-actes.pdf#page=28
- 33. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010;26: 2460–2461. pmid:20709691
- 34. McDonald D, Price MN, Goodrich J, Nawrocki EP, DeSantis TZ, Probst A, et al. An improved greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J. 2012;6(3): 610–18. pmid:22134646
- 35. Caporaso JG, Bittinger K, Bushman FD, DeSantis TZ, Andersen GL, Knight R. PyNAST: a flexible tool for aligning sequences to a template alignment. Bioinformatics. 2010; 26(2): 266–7. pmid:19914921
- 36. Price MN, Dehal PS, Arkin AP. FastTree 2–approximately maximum-likelihood trees for large alignments. PLoS One. 2010 Mar 10;5(3):e9490. pmid:20224823
- 37. McMurdie PJ, Holmes S. phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One. 2013 Apr 22.
- 38.
Wickham H. ggplot2: elegant graphics for data analysis. New York: Springer; 2009.
- 39.
R development core team. R: A language and environment for statistical computing. Vienna: R foundation for statistical computing. 2012. Available: http://www.R-project.org.
- 40.
Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O'Hara RB, et al. Package ‘vegan’. Community ecology package, version. 2013 Dec 12; 2(9).
- 41. Love M, Anders S, Huber W. Differential analysis of count data–the DESeq2 package. Genome Biol. 2014 May 13; 15: 550. pmid:25516281
- 42. Letunic I, Bork P. Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucl Acids Res. 2016 Apr 19:gkw290.
- 43.
Gabry A, Goodrich B. rstanarm: Bayesian Applied Regression Modeling via Standard R package version 2.12.1. 2016. https://CRAN.R-project.org/package=rstanarm
- 44.
EPA regulated water contaminants. Available: https://www.epa.gov/ground-water-and-drinking-water/table-regulated-drinking-water-contaminants
- 45. Chao A. Non-parametric estimation of the number of classes in a population. Scand J Stat. 1984;11: 265–27.
- 46. Bergmann GT, Bates ST, Eilers KG, Lauber CL, Caporaso JG, Walters WA, et al. The under-recognized dominance of Verrucomicrobia in soil bacterial communities. Soil Biol. Biochem. 2011;43: 1450–1455. pmid:22267877
- 47.
Skerman VBD. Genus Chitinophaga Sangkhobol and Skerman, 285VP. In Bergey's Manual of Systematic Bacteriology, 1st edn,. Eds; Staley JT, Bryant MP, Pfennig N, Holt JG. Baltimore: Williams & Wilkins. 1989;3: 2074–2077.
- 48. Lim JH, Baek S-H, Lee S-T. Ferruginibacter alkalilentus gen. nov., sp. nov. and Ferruginibacter lapsinanis sp. nov., novel members of the family ‘Chitinophagaceae’ in the phylum Bacteroidetes, isolated from freshwater sediment. IJSEM. 2009;59: 2394–2399. pmid:19620358
- 49. Fahrbach M, Kuever J, Remesch M, Huber BE, Kämpfer P, Dott W, et al. Steroidobacter denitrificans gen. nov., sp. nov., a steroidal hormone-degrading Gammaproteobacterium. Int J Syst Evol Microbiol. 2008; 58(Pt 9):2215–2213. pmid:18768632
- 50.
Engel AS, Northup DE. Caves and karst as models for advancing microbial sciences. In: Martin JB, White WB. Frontiers of Karst Research: Proceedings and Recommendations of the Workshop Held May 3 through 5, 2007 in San Antonio, Texas, USA. Leesburg, VA: Karst Waters Institute, Inc. 2008; pp. 37‒48.
- 51. Janssen PH. Identifying the dominant soil bacterial taxa in libraries of 16S rRNA and 16S rRNA genes. Appl Environ Microbiol. 2006;72: 1719–1728. pmid:16517615
- 52. Jones RT, Robeson MS, Lauber CL, Hamady M, Knight R, Fierer N. A comprehensive survey of soil acidobacterial diversity using pyrosequencing and clone library analyses. ISME J. 2009;3: 442–453. pmid:19129864
- 53.
Northup DE, Stefánsson ÁB, Medina MJ, Caimi NA, Kooser AS. Microbial communities of Icelandic lava caves. Proc 16th Int Symp Vulcanospeleo, Ocean View, HI. 2016: http://www.cavepics.com/IVS17/NORTHUP.pdf
- 54. Barka EA, Vatsa P, Sanchez L, Gaveau-Vaillant N, Jacquard C, Klenk H-P, et al. Taxonomy, physiology, and natural products of Actinobacteria. Microbiol Mol Biol Rev. 2016;80: 1–43. pmid:26609051
- 55. Kanaparthi D, Pommerenke B, Casper P, Dumont MG. Chemolithotrophic nitrate-dependent Fe(II)-oxidizing nature of actinobacterial subdivision lineage TM3. ISME J.2013;7(8): 1582–1594. pmid:23514778
- 56. Porca E, Jurado V, ŽGur-Bertok D, Saiz-Jimenez C, Pašic L. Comparative analysis of yellow microbial communities growing on the walls of geographically distinct caves indicates a common core of microorganisms involved in their formation. FEMS Microbiol Ecol. 2012;81: 255–266. pmid:22486654
- 57. Gerber NN. Volatile substances from actinomycetes: Their role in the odor pollution of water. CRC Crit Rev Microbiol. 1979; 7(3):191–214. pmid:396107
- 58.
Snider JR. Comparison of microbial communities on roots, ceilings and floors of two lava caves in New Mexico. MSci Thesis. University of New Mexico; 2010. Available: http://repository.unm.edu/handle/1928/11135
- 59. Maskey RP, Kock I, Helmke E, Laatsch H. Isolation and structure determination of phenazostatin D, a new phanazine from a marine actinomycete isolate Pseudonocardia sp. B6273. Z Naturforsch. 2003;58: 692–694.
- 60. Qin S, Su YY, Zhang YQ, Jiang CL, Xu LH, Li WJ. Pseudonocardia ailaonensis sp. nov., isolated from soil in China. Int J Syst Evol Microbiol. 2008;58: 2086–2089. pmid:18768609
- 61. Lee SD. Amycolatopsis jejuensis sp. nov. and Amycolatopsis halotolerans sp. nov., novel actinomycetes isolated from a natural cave. Int J Syst Evol Microbiol. 2006;56: 549–553 pmid:16514025
- 62. Barton HA, Taylor NM, Kreate MP, Springer AC, Oehrle SA, Bertog JL. The impact of host rock geochemistry on bacterial community structure in oligotrophic cave environments. Int J Speleol. 2007;36(2): 93–104.
- 63. Barton HA, Jurado V. What’s up down there? Microbial diversity in caves. Microbe. 2007;2: 132–138.
- 64. Bérdy J. Bioactive microbial metabolites: A personal view. J Antibiot. 2005;58(1): 1–26. pmid:15813176
- 65. Farris MH, Olson JB. Detection of Actinobacteria cultivated from environmental samples reveals bias in universal primers. Lett Appl Microbiol. 2007;45(4): 376–381. pmid:17897379
- 66.
Gabriel CR, Northup DE. Microbial Ecology: Caves as an extreme habitat. In: Cave Microbiomes: A Novel Resource for Drug Discovery. Cheeptham N (ed). Springer. 2013. pp. 85–108.
- 67. Pelaez F. The historical delivery of antibiotics from microbial natural products–can history repeat? BioChem Parmacol. 2006;
- 68.
Zhang L. Integrated approach for discovery of novel drugs from microbial natural products. In Zhang L, Demain AL (eds). Natural Products. Drug Discovery and Therapeutic Medicine. Humana, Totowa, NJ. 2005; p. 30–55.
- 69. Riquelme C, Hathaway JJM, Dapkevicius M de LNE, Miller AZ, Kooser A, Northup DE, et al. Actinobacterial diversity in volcanic caves and associated geomicrobiological interactions. Front Microbiol. 2015a;: 1–16.
- 70. Holmes AJ, Bowyer J, Holley MP, O'Donoghue M, Montgomery M, Gillings MR. Diverse, yet-to-be-cultured members of the Rubrobacter subdivision of the Actinobacteria are widespread in Australian arid soils. FEMS Microbiology Ecology. 2000 Aug 1;33(2):111–20. pmid:10967210
- 71.
Spilde MN, Northup DE, Caimi NA, Boston PJ, Stone FD, Smith S. Microbial mats communities in Hawaiian lava caves. Proc 16th Int Symp Vucanospeleo. 2016; http://www.cavepics.com/IVS17/SPILDE.pdf
- 72. Riquelme C, Rigal F, Hathaway JJM, Northup DE, Spilde MN, Borges PA, et al. Cave microbial community composition in oceanic islands: disentangling the effect of different colored mats in diversity patterns of Azorean lava caves. FEMS Microbiol Ecol. 2015b;91: fiv 141.
- 73. Daims H, Lebedeva EV, Pjevac P, Han P, Herbold C, Albertsen M, et al. Complete nitrification by Nitrospira bacteria. Nature. 2015;528: 504–509. pmid:26610024
- 74. Hathaway JJM, Sinsabaugh RL, Dapkevicius MdLNE, Northup DE. Diversity of ammonia oxidation (amoA) and nitrogen fixation (nifH) genes in lava caves of Terceira, Azores, Portugal. Geomicrobiol J. 2014b;31: 221–235.
- 75. Tetu SG, Breakwell K, Elbourne LDH, Holmes AJ, Gillings MR, Paulsen IT. Life in the dark; metagenomic evidence that a microbial slime community is driven by inorganic nitrogen metabolism. ISME J. 2013;7: 1227–1236. pmid:23426011
- 76. Griffin DW, Gray MA, Lyles MB, Northup DE. The transport of nonindigenous microorganisms into caves by human visitation: a case study at Carlsbad Caverns National Park. Geomicrobiol J. 2014;31: 175–185
- 77. Groth I, Schumann P, Schütze B, Augsten K, Stackebrandt. Knoellia sinensis gen. nov., sp. nov. and Knoellia subterranea sp. nov., two novel actinobacteria isolated from a cave. Int J Sys Evol Microb. 2002; 52: 77–84.
- 78. Levy DB. Geochemical trends in selected Lechuguilla Cave pools. J Cave Karst Stud. 2007 69(3): 342–350.
- 79. Barton HA, Northup DE. 2007. Geomicrobiology in cave environments: Past, current, and future perspectives. J Cave Karst Stud. 2007; 69: 163–178.
- 80. Northup DE, Lavoie KH. Geomicrobiology of caves: a review. Geomicrobiol J. 2001;18: 199–222.
- 81. Jones AA, Bennett PC. Mineral microniches control the diversity of subsurface microbial populations. Geomicrobiol J. 2014;31: 246–261.