Differential depth distribution of microbial function and putative symbionts through sediment-hosted aquifers in the deep terrestrial subsurface

Abstract

An enormous diversity of previously unknown bacteria and archaea has been discovered recently, yet their functional capacities and distributions in the terrestrial subsurface remain uncertain. Here, we continually sampled a CO₂-driven geyser (Colorado Plateau, Utah, USA) over its 5-day eruption cycle to test the hypothesis that stratified, sandstone-hosted aquifers sampled over three phases of the eruption cycle have microbial communities that differ both in membership and function. Genome-resolved metagenomics, single-cell genomics and geochemical analyses confirmed this hypothesis and linked microorganisms to groundwater compositions from different depths. Autotrophic Candidatus "Altiarchaeum sp." and phylogenetically deep-branching nanoarchaea dominate the deepest groundwater. A nanoarchaeon with limited metabolic capacity is inferred to be a potential symbiont of the Ca. "Altiarchaeum". Candidate Phyla Radiation bacteria are also present in the deepest groundwater and they are relatively abundant in water from intermediate depths. During the recovery phase of the geyser, microaerophilic Fe- and S-oxidizers have high in situ genome replication rates. Autotrophic Sulfurimonas sustained by aerobic sulfide oxidation and with the capacity for N₂ fixation dominate the shallow aquifer. Overall, 104 different phylum-level lineages are present in water from these subsurface environments, with uncultivated archaea and bacteria partitioned to the deeper subsurface.

Fig. 1. Crystal Geyser’s 5-day eruption cycle measured during the 2015 sampling period exhibited variations in downhole water pressure and electrical conductivity that define three phases.

In each phase, electrical conductivity (EC) and geochemical measurements (6,710 measurements each, no technical replicates; Supplementary Fig. 1) are used to identify relative depths of source water compositions: intermediate for the recovery phase (2,330 measurements), deep for the minor eruptions (2,820 measurements) and shallow for the major eruptions (1,560 measurements). The numbered horizontal grey bars indicate the time periods for each metagenomic sample (25 samples in total) and coloured numbers indicate the grouping of samples from each phase.

Fig. 2. Diversity of recovered genomes based on 16 concatenated ribosomal proteins.

Genomes were reconstructed for organisms from 104 different phylum-level lineages; 503 different lineages are shown (two lineages did not exhibit >50% alignment coverage and are thus not displayed). Phyla in bold were assigned names in this study. The scale corresponds to per cent average amino acid substitution over the alignment. Asterisks mark yet-to-be-cultivated phyla, which thus have a Candidatus status. OD1, Parcubacteria. A full tree with reference sequences can be found in Supplementary File 1.

Fig. 3. Hydrogeology and community composition of subsurface fluids sourced from Crystal Geyser throughout an entire eruption cycle.

a, The Crystal Geyser site lies within one of the several natural CO₂ reservoirs within the Paradox Basin. The CO₂ was probably generated from thermal decomposition of Pennsylvanian-aged carbonate rocks^{, ,}. CO₂ gas and brine formed by groundwater dissolution of Paradox evaporites migrate via faults and fractures^,. b, The community profile of 505 organisms strongly followed the succession of the geyser eruptions (blue lines, NMDS). One data point corresponds to one metagenomic sample. The samples show a clear pattern following the succession of the geyser cycle. c, Entire community profile of 505 organisms tracked across the 5-day cycle of the geyser. Each colour corresponds to one genome. d,e, Profiles of the CPR and DPANN community, respectively, show an increase in the overall abundance during the minor eruptions when groundwater has the deepest source composition. f, Downhole electrical conductivity time series during the sampling of the cycle illustrating the individual phases of the geyser (6,710 samples were measured, see Fig. 1 and Supplementary Fig. 1). Number of biological replicates in panels b–e was 24. EC, electrical conductivity; GW, groundwater.

Fig. 4. Microbial source tracking and changes of metabolic potential.

a–c, The abundances of organisms that are significantly enriched in groundwater from the different depths (for details on organisms please see Supplementary Table 5, number of biological replicates are given in parenthesis of panels a–c). The pie charts indicate the diversity of CPR, DPANN, other bacteria, and other archaea associated with each relative depth. d, Different carbon fixation pathways predominate in groundwater from the three different depths. Nitrogen fixation and the reverse TCA cycle occur in one organism, Sulfurimonas (a). e, Metabolic pathway analysis shows distinct metabolic profiles associated with the groundwater from the different depths (individual metabolic capacities of each organism are listed in Supplementary Table 7). Each circle displays the cumulative relative abundance of genomes contributing to this single metabolic process. Arrows display if an increase or decrease is significant (P < 0.05). CBB, Calvin-Benson-Bassham; disprop., disproportionation.

Fig. 5. Putative symbiotic interaction of Ca. “Altiarchaeum” and Ca. “H. crystalense”

a, Linear correlation analysis of relative abundance of the two archaea across 25 metagenome samples (full cycle of the geyser). b, Scanning electron micrograph of what are inferred to be Ca. “Altiarchaeum” cells (“SM1”) taken during the minor eruptions of the geyser. Tiny cell-like structures appear to be attached to the surface (“?N”). This structure was observed in two out of five samples taken for scanning electron microscopy analysis from the geyser fluids. cor., correlation coefficient. More images are available under Supplementary Fig. 9.

Fig. 6. Conceptual representation of a relatively stable microbiome in deeper sandstone aquifer sources.

The microbiome is dominated by Ca. “Altiarchaeum” (SM1) and their putative DPANN symbionts and populated by many CPR and other bacteria, some of which are probably symbiotic partners for CPR. We envision facile distribution of the very small CPR and DPANN cells through the sandstone pore spaces, providing periodic opportunities for establishment of the symbiont–host interactions that are probably required for CPR and DPANN cell replication. This figure provides a conceptual diagram of generalized microbial habitats in the aquifer based on an approximate pore size of sandstone. However, we note the subsurface is a heterogeneous three-dimensional system and physical properties will vary substantially. The Carmel and Kayenta formations are expected to act as aquitards (confining barriers) that separate the high permeability sandstone aquifers (Fig. 3a), with each aquifer largely confined, both hydrologically and microbiologically, from other aquifers by these low-permeability shale/mudstone units. This physical separation by low-permeability units probably contributes to the distinctive microbial communities associated with the three relative groundwater source depths as documented in the study.