ABSTRACT: The microbial ecology of the deep biosphere is difficult to characterize, owing in part to sampling challenges and poorly understood response mechanisms to environmental change. Pre-drilled wells, including oil wells or boreholes, offer convenient access, but sampling is frequently limited to the water alone, which may provide only a partial view of the native diversity. Mineral heterogeneity demonstrably affects colonization by deep biosphere microorganisms, but the connections between the mineral-associated and planktonic communities remain unclear. To understand the substrate effects on microbial colonization and the community response to changes in organic carbon, we conducted an 18-month series of in situ experiments in a warm (57°C), anoxic, fractured carbonate aquifer at 752 m depth using replicate open, screened cartridges containing different solid substrates, with a proteinaceous organic matter perturbation halfway through this series. Samples from these cartridges were analyzed microscopically and by Illumina (iTag) 16S rRNA gene libraries to characterize changes in mineralogy and the diversity of the colonizing microbial community. The substrate-attached and planktonic communities were significantly different in our data, with some taxa (e.g., Candidate Division KB-1) rare or undetectable in the first fraction and abundant in the other. The substrate-attached community composition also varied significantly with mineralogy, such as with two Rhodocyclaceae OTUs, one of which was abundant on carbonate minerals and the other on silicic substrates. Secondary sulfide mineral formation, including iron sulfide framboids, was observed on two sets of incubated carbonates. Notably, microorganisms were attached to the framboids, which were correlated with abundant Sulfurovum and Desulfotomaculum sp. sequences in our analysis. Upon organic matter perturbation, mineral-associated microbial diversity differences were temporarily masked by the dominance of putative heterotrophic taxa in all samples, including OTUs identified as Caulobacter, Methyloversatilis, and Pseudomonas. Subsequent experimental deployments included a methanogen-dominated stage (Methanobacteriales and Methanomicrobiales) 6 months after the perturbation and a return to an assemblage similar to the pre-perturbation community after 9 months. Substrate-associated community differences were again significant within these subsequent phases, however, demonstrating the value of in situ time course experiments to capture a fraction of the microbial assemblage that is frequently difficult to observe in pre-drilled wells.
