Advances in nucleic acid sequencing technology have enabled expansion of our ability to profile microbial diversity. These large datasets of taxonomic and functional diversity are key to better understanding microbial ecology. Machine learning has proven to be a useful approach for analyzing microbial community data and making predictions about outcomes including human and environmental health. Machine learning applied to microbial community profiles has been used to predict disease states in human health, environmental quality and presence of contamination in the environment, and as trace evidence in forensics. Machine learning has appeal as a powerful tool that can provide deep insights into microbial communities and identify patterns in microbial community data. However, often machine learning models can be used as black boxes to predict a specific outcome, with little understanding of how the models arrived at predictions. Complex machine learning algorithms often may value higher accuracy and performance at the sacrifice of interpretability. In order to leverage machine learning into more translational research related to the microbiome and strengthen our ability to extract meaningful biological information, it is important for models to be interpretable. Here we review current trends in machine learning applications in microbial ecology as well as some of the important challenges and opportunities for more broad application of machine learning to understanding microbial communities.

Biological sensors can be engineered to measure a wide range of environmental conditions. Here we show that statistical analysis of DNA from natural microbial communities can be used to accurately identify environmental contaminants, including uranium and nitrate at a nuclear waste site. In addition to contamination, sequence data from the 16S rRNA gene alone can quantitatively predict a rich catalogue of 26 geochemical features collected from 93 wells with highly differing geochemistry characteristics. We extend this approach to identify sites contaminated with hydrocarbons from the Deepwater Horizon oil spill, finding that altered bacterial communities encode a memory of prior contamination, even after the contaminants themselves have been fully degraded. We show that the bacterial strains that are most useful for detecting oil and uranium are known to interact with these substrates, indicating that this statistical approach uncovers ecologically meaningful interactions consistent with previous experimental observations. Future efforts should focus on evaluating the geographical generalizability of these associations. Taken as a whole, these results indicate that ubiquitous, natural bacterial communities can be used as in situ environmental sensors that respond to and capture perturbations caused by human impacts. These in situ biosensors rely on environmental selection rather than directed engineering, and so this approach could be rapidly deployed and scaled as sequencing technology continues to become faster, simpler, and less expensive. IMPORTANCE Here we show that DNA from natural bacterial communities can be used as a quantitative biosensor to accurately distinguish unpolluted sites from those contaminated with uranium, nitrate, or oil. These results indicate that bacterial communities can be used as environmental sensors that respond to and capture perturbations caused by human impacts.

With the rapid advances in sequencing technology, the cost of sequencing has dramatically dropped and the scale of sequencing projects has increased accordingly. This has provided the opportunity for the routine use of sequencing techniques in the monitoring of environmental microbes. While metagenomic applications have been routinely applied to better understand the ecology and diversity of microbes, their use in environmental monitoring and bioremediation is increasingly common. In this review we seek to provide an overview of some of the metagenomic techniques used in environmental systems biology, addressing their application and limitation. We will also provide several recent examples of the application of metagenomics to bioremediation. We discuss examples where microbial communities have been used to predict the presence and extent of contamination, examples of how metagenomics can be used to characterize the process of natural attenuation by unculturable microbes, as well as examples detailing the use of metagenomics to understand the impact of biostimulation on microbial communities.

To better understand the impacts of Corexit 9500 on the structure and activity levels of hydrocarbon-degrading microbial communities, we analyzed next-generation 16S rRNA gene sequencing libraries of hydrocarbon enrichments grown at 5 and 25°C using both DNA and RNA extracts as the sequencing templates. Oil biodegradation patterns in both 5 and 25°C enrichments were consistent with those reported in the literature (i.e., aliphatics were degraded faster than aromatics). Slight increases in biodegradation were observed in the presence of Corexit at both temperatures. Differences in community structure were observed between treatment conditions in the DNA-based libraries. The 25°C consortia were dominated by Vibrio , Idiomarina , Marinobacter , Alcanivorax , and Thalassospira species, while the 5°C consortia were dominated by several species of the genera Flavobacterium , Alcanivorax , and Oleispira . Most of these genera have been linked to hydrocarbon degradation and have been observed after oil spills. Colwellia and Cycloclasticus , known aromatic degraders, were also found in these enrichments. The addition of Corexit did not have an effect on the active bacterial community structure of the 5°C consortia, while at 25°C, a decrease in the relative abundance of Marinobacter was observed. At 25°C, Thalassospira , Marinobacter , and Idiomarina were present at higher relative abundances in the RNA than DNA libraries, suggesting that they were active in degradation. Similarly, Oleispira was greatly stimulated by the addition of oil at 5°C. IMPORTANCE While dispersants such as Corexit 9500 can be used to treat oil spills, there is still debate on the effectiveness on enhancing oil biodegradation and its potential toxic effect on oil-degrading microbial communities. The results of this study provide some insights on the microbial dynamics of hydrocarbon-degrading bacterial populations in the presence of Corexit 9500. Operational taxonomic unit (OTU) analyses indicated that several OTUs were inhibited by the addition of Corexit. Conversely, a number of OTUs were stimulated by the addition of the dispersant, many of which were identified as known hydrocarbon-degrading bacteria. The results highlight the value of using RNA-based methods to further understand the impact of dispersant on the overall activity of different hydrocarbon-degrading bacterial groups.

The waters of the Eastern Mediterranean are characterized by unique physical and chemical properties within separate water masses occupying different depths. Distinct water masses are present throughout the oceans, which drive thermohaline circulation. These water masses may contain specific microbial assemblages. The goal of this study was to examine the effect of physical and geological phenomena on the microbial community of the Eastern Mediterranean water column. Chemical measurements were combined with phospholipid fatty acid (PLFA) analysis and high-throughput 16S rRNA sequencing to characterize the microbial community in the water column at five sites. We demonstrate that the chemistry and microbial community of the water column were stratified into three distinct water masses. The salinity and nutrient concentrations vary between these water masses. Nutrient concentrations increased with depth, and salinity was highest in the intermediate water mass. Our PLFA analysis indicated different lipid classes were abundant in each water mass, suggesting that distinct groups of microbes inhabit these water masses. 16S rRNA gene sequencing confirmed the presence of distinct microbial communities in each water mass. Taxa involved in autotrophic nitrogen cycling were enriched in the intermediate water mass suggesting that microbes in this water mass may be important to the nitrogen cycle of the Eastern Mediterranean. The Eastern Mediterranean also contains numerous active hydrocarbon seeps. We sampled above the North Alex Mud Volcano, in order to test the effect of these geological features on the microbial community in the adjacent water column. The community in the waters overlaying the mud volcano was distinct from other communities collected at similar depths and was enriched in known hydrocarbon degrading taxa. Our results demonstrate that physical phenomena such stratification as well as geological phenomena such as mud volcanoes strongly affect microbial community structure in the Eastern Mediterranean water column.

Microbial adaptation to extreme conditions takes many forms, including specialized metabolism which may be crucial to survival in adverse conditions. Here, we analyze the diversity and environmental importance of systems allowing microbial carbon monoxide (CO) metabolism. CO is a toxic gas that can poison most organisms because of its tight binding to metalloproteins. Microbial CO uptake was first noted by Kluyver and Schnellen in 1947, and since then many microbes using CO via oxidation have emerged. Many strains use molecular oxygen as the electron acceptor for aerobic oxidation of CO using Mo-containing CO oxidoreductase enzymes named CO dehydrogenase. Anaerobic carboxydotrophs oxidize CO using CooS enzymes that contain Ni/Fe catalytic centers and are unrelated to CO dehydrogenase. Though rare on Earth in free form, CO is an important intermediate compound in anaerobic carbon cycling, as it can be coupled to acetogenesis, methanogenesis, hydrogenogenesis, and metal reduction. Many microbial species—both bacteria and archaea—have been shown to use CO to conserve energy or fix cell carbon or both. Microbial CO formation is also very common. Carboxydotrophs thus glean energy and fix carbon from a “metabolic leftover” that is not consumed by, and is toxic to, most microorganisms. Surprisingly, many species are able to thrive under culture headspaces sometimes exceeding 1 atmosphere of CO. It appears that carboxydotrophs are adapted to provide a metabolic “currency exchange” system in microbial communities in which CO arising either abiotically or biogenically is converted to CO2and H2that feed major metabolic pathways for energy conservation or carbon fixation. Solventogenic CO metabolism has been exploited to construct very large gas fermentation plants converting CO-rich industrial flue emissions into biofuels and chemical feedstocks, creating renewable energy while mitigating global warming. The use of thermostable CO dehydrogenase enzymes to construct sensitive CO gas sensors is also in progress.