Longitudinal analysis of the Five Sisters hot springs in Yellowstone National Park reveals a dynamic thermoalkaline environment

Abstract

Research focused on microbial populations of thermoalkaline springs has been driven in a large part by the lure of discovering functional enzymes with industrial applications in high-pH and high temperature environments. While several studies have focused on understanding the fundamental ecology of these springs, the small molecule profiles of thermoalkaline springs have largely been overlooked. To better understand how geochemistry, small molecule composition, and microbial communities are connected, we conducted a three-year study of the Five Sisters (FS) springs that included high-resolution geochemical measurements, 16S rRNA sequencing of the bacterial and archaeal community, and mass spectrometry-based metabolite and extracellular small molecule characterization. Integration of the four datasets facilitated a comprehensive analysis of the interwoven thermoalkaline spring system. Over the course of the study, the microbial population responded to changing environmental conditions, with archaeal populations decreasing in both relative abundance and diversity compared to bacterial populations. Decreases in the relative abundance of Archaea were associated with environmental changes that included decreased availability of specific nitrogen- and sulfur-containing extracellular small molecules and fluctuations in metabolic pathways associated with nitrogen cycling. This multi-factorial analysis demonstrates that the microbial community composition is more closely correlated with pools of extracellular small molecules than with the geochemistry of the thermal springs. This is a novel finding and suggests that a previously overlooked component of thermal springs may have a significant impact on microbial community composition.

Figure 1

Sampling occurred at the Five Sisters spring system in the White Creek Drainage in YNP. (a) Map of YNP indicating the location of the Five Sisters springs north of Old Faithful generated using Adobe Illustrator vCC, adobe.com . (b) Picture of the Five Sisters springs with all five pools labeled.

Figure 2

Relative microbial abundances were determined for each spring from 2017 to 2019. (a) Phylum-level relative abundance of each spring on a yearly basis. Microbial population composition changes over the three-year period. Each panel represents a year from 2017 to 2019 with springs FS1-FS5 relative abundance levels. (b) Phylum-level abundance across years for each of the Five Sisters springs.

Figure 3

Microbial alpha diversity was calculated using a Shannon Index for all springs each year. (a) Bacterial alpha diversity from 2017 to 2019. (b) Archaeal alpha diversity from 2017 to 2019. Shannon Index analysis accounts for both the evenness and richness in a system.

Figure 4

A geochemical analysis was completed at each sampling event. (a) 2D-PCA scores plot of all five springs by year. Shaded regions indicate 95% confidence intervals. (b) One-way parametric ANOVA analysis indicating geochemical features whose levels are significantly different (p-value < 0.05) between the three years.

Figure 5

Intracellular small molecules were isolated and metabolomic profiles were analyzed. (a) PCA of the intracellular small molecules. Metabolites from 2017 group separately from 2018 and 2019. Metabolites from 2018 and 2019 group independently but overlap in the 95% confidence intervals for several springs, most strongly in FS4 and FS5. (b) Heatmap of the significant identified metabolites between years determined using an ANOVA (p < 0.05). Sampling year is indicated at the top of the figure with each column representing one sample and each row representing an identified metabolite. The dendrogram at the top groups 2018 and 2019 separately from 2017 based on the relative abundance of the identified metabolites. The figure was generated using MetaboAnalyst v4.0 (c) Metabolic pathways derived from the identified metabolites with the y-axis showing the p-value from an ANOVA of the groups and the x-axis showing the metabolic impact based on the metabolites identified and their role in the specific metabolic pathway.

Figure 6

Extracellular small molecules were isolated and small molecule profiles were analyzed. (a) PCA for extracellular small molecules in the sediment. 2018 and 2019 group very closely together relative to 2017. FS2-4 from 2017 group together and are closer to 2018 and 2019 than FS1 and FS5 from 2017. (b) Heatmap of the top 50 discriminating small molecules between 2017, 2018 and 2019. The top of the figure showing the year of each sample with a dendrogram indicating that 2017 groups independently from 2018 and 2019. The figure was generated using MetaboAnalyst v4.0 (c) Plot showing unique sulfur, nitrogen and sulfur and nitrogen containing molecules. Element specific molecules are shown in each year from 2017 to 2019. Formulas were determined from the mass spectrometry datasets using an error of 15 ppm.

Figure 7

Data from all available datasets were examined to determine correlative features. (a) Correlogram as described in the methods section investigating archaeal ZOTUs and sulfur containing extracellular small molecules. (b) Correlogram of archaeal ZOTUs and nitrogen containing extracellular small molecules. The intensity of the color indicates the strength of the correlation, with blue demonstrating positive and red demonstrating negative correlations. An asterisk indicates a p-value of < 0.05 for the correlation. Correlograms were created using the mixOmics package in R.

Figure 8

A phylogenetic tree of archaeal sequences was generated and ZOTUs that correlated strongly to the extracellular small molecule datasets were identified in the tree. The Inner ring is populated by Archaea found in the Five Sisters site using metagenomic data. The colored shapes in the inner ring indicate phyla of previously cultured and characterized archaeal species corresponding to the taxonomic classifications shown. Archaeal sequences isolated in the correlogram analysis are indicated by red dashes in the outer ring. The figure was created using Interactive Tree of Life v6.