Islands Within Islands: Bacterial Phylogenetic Structure and Consortia in Hawaiian Lava Caves and Fumaroles

Abstract

Lava caves, tubes, and fumaroles in Hawai'i present a range of volcanic, oligotrophic environments from different lava flows and host unexpectedly high levels of bacterial diversity. These features provide an opportunity to study the ecological drivers that structure bacterial community diversity and assemblies in volcanic ecosystems and compare the older, more stable environments of lava tubes, to the more variable and extreme conditions of younger, geothermally active caves and fumaroles. Using 16S rRNA amplicon-based sequencing methods, we investigated the phylogenetic distinctness and diversity and identified microbial interactions and consortia through co-occurrence networks in 70 samples from lava tubes, geothermal lava caves, and fumaroles on the island of Hawai'i. Our data illustrate that lava caves and geothermal sites harbor unique microbial communities, with very little overlap between caves or sites. We also found that older lava tubes (500-800 yrs old) hosted greater phylogenetic diversity (Faith's PD) than sites that were either geothermally active or younger (<400 yrs old). Geothermally active sites had a greater number of interactions and complexity than lava tubes. Average phylogenetic distinctness, a measure of the phylogenetic relatedness of a community, was higher than would be expected if communities were structured at random. This suggests that bacterial communities of Hawaiian volcanic environments are phylogenetically over-dispersed and that competitive exclusion is the main driver in structuring these communities. This was supported by network analyses that found that taxa (Class level) co-occurred with more distantly related organisms than close relatives, particularly in geothermal sites. Network "hubs" (taxa of potentially higher ecological importance) were not the most abundant taxa in either geothermal sites or lava tubes and were identified as unknown families or genera of the phyla, Chloroflexi and Acidobacteria. These results highlight the need for further study on the ecological role of microbes in caves through targeted culturing methods, metagenomics, and long-read sequence technologies.

Keywords: cave microbiology; fumaroles; lava caves; microbial consortia; networks; taxonomic distinctness; volcanic environments.

Figure 1

A principal coordinates analysis (PCoA) of the weighted Unifrac distance measured between samples, capturing 49% of the variation in multivariate space. Samples are displayed as the cave or site they were collected from. This illustrates that samples from geothermal sites and lava caves are two distinct populations in terms of phylogenetic community composition and abundance.

Figure 2

Faith's Phylogenetic Diversity (PD) vs. estimated site age. The approximate age of a particular site was defined by the latest flow recorded for the location. Steam vents (sites P & S) are features in two possible flow events and were thus given the range of 65–400 years, representing the earliest and latest potential time periods for those flows. Phylogenetic diversity increases with the increasing age of the lava flow at that site. PD was higher among sites that were 500–820 year old, when compared to younger sites. This includes Kaumana cave, a lava tube, which is only ~130 years old.

Figure 3

(A) Relative abundance by the site of class-level taxa with >1% abundance in geothermally active sites. Geothermal caves are dominated by Cyanobacteria (Oxyphotobacteria; 40.2% ± 25.3) and Chloroflexi including classes Ktedonobacteria (10.6% ± 15.9) and Chloroflexia (5.8 ± 9.8). Steam vent sites hosted the highest abundance of Oxyphotobacteria. Ktedonobacteria was higher in geothermal caves. Deinococci were most abundant in Kīlauea Caldera's Big Mouth cave. (B) Upset plot illustrating overlapping ASVs among geothermal sites. Steam vents hosted many unique ASVs (2,518), followed by Big Ell cave. This shows that geothermal sites are largely unique environments, with few overlapping ASVs; the highest number of overlapping ASVs was in Big Ell and Pahoa Steam vents, which may in part be due to the larger number of samples from these locations. Only one ASV occurred in all sites (Class Oxyphotobacteria, Chlorogloeopsis sp.).

Figure 4

(A) Relative abundance by the site of class-level taxa with >1% abundance in lava tubes. Dominant taxa in these features were Actinobacteria (16.0% ± 24.4), Gammaproteobacteria (15.4% ± 11.0), Nitriliruptoria (13.7% ± 23.3), and Alphaproteobacteria (9.8% ± 10.7). Longimicrobia and Acidimicrobiia ASVs were considerably more abundant in the Kipuka Kanohina cave system in the south of the island of Hawai‘i. (B) Upset plot of lava tube sites shows that individual lava tubes are moderately unique, with Maelstrom cave hosting 1,277 unique ASVs, followed by Kaumana cave (1,022 unique ASVs). Only seven ASVs occurred in all lava tubes.

Figure 5

Phylogenetic distinctness (Delta+) on the left (A) and variation of phylogenetic distinctness (Lambda+) on the right (B) for all samples based on patristic distances between identified taxa (ASVs). Samples are labeled by location. Most samples are either within expected values of Delta+ (null = randomly distributed) or higher than expected, meaning samples are phylogenetically over-dispersed, possibly due to competitive exclusion (A). Variation in phylogenetic distinctness (Lambda+) is within expected values or lower than expected. Most samples had lower than expected variation in distances between branches on the phylogenetic tree, when compared to the null model of random assemblage, and represent that more distinctly related organisms are assembled into the community (B).

Figure 6

Hub network for geothermal sites, illustrating a high level of complexity within the network. There are 100 nodes that have an above-average hub score (represented by larger circles in the network). Hubs are taxa that most likely have important ecological roles, such as keystone species, and help maintain the overall network structure.

Figure 7

Geothermal neighbor interaction frequency plots at the class level of the three classes with the greatest number of interactions (edges) within the geothermal network: (A) Oxyphotobacteria, (B) Alphaproteobacteria, and (C) Ktedonobacteria. Oxyphotobacteria have the largest number of interactions or co-occurrence with Alphaproteobacteria. However, Alphaproteobacteria have the most interactions with Acidobacteriia, followed by Oxyproteobacteria and then other Alphaproteobacteria. Ktedonobacteria, a common Chloroflexi group in geothermal sites, also has the greatest number of interactions with Alphaproteobacteria, followed by interactions with other Ktedonobacteria.

Figure 8

Hub network for lava tubes, suggesting lower complexity than the geothermal network. There are 69 ASVs with an above-average hub score, represented by larger circles on the network graph. Hubs are taxa that most likely have important ecological roles, such as keystone species, and help maintain the overall network structure.

Figure 9

Lava tube neighbor interaction frequency plots at the class level of the two classes with the greatest number of interactions (edges) within the network: (A) Gammaproteobacteria and (B) Alphaproteobacteria. Gammaproteobacteria have the highest frequencies of neighbor interactions with other Gammaproteobacteria, as well as Alphaproteobacteria. Gammaproteobacteria also interact or co-occur with Acidimicrobiia (phylum Actinobacteria), and Deltaproteobacteria. Alphaproteobacteria co-occur with Gammaproteobacteria at the greatest frequency and have much fewer interactions with other Alphaproteobacteria. Deltaproteobacteria and Chloroflexi class Anaerolineae also had a high frequency of interactions with Alphaproteobacteria within lava tubes.