Contribution of Microorganisms with the Clade II Nitrous Oxide Reductase to Suppression of Surface Emissions of Nitrous Oxide.

The sources and sinks of nitrous oxide, as control emissions to the atmosphere, are generally poorly constrained for most environmental systems. Initial depth-resolved analysis of nitrous oxide flux from observation wells and the proximal surface within a nitrate contaminated aquifer system revealed high subsurface production but little escape from the surface. To better understand the environmental controls of production and emission at this site, we used a combination of isotopic, geochemical, and molecular analyses to show that chemodenitrification and bacterial denitrification are major sources of nitrous oxide in this subsurface, where low DO, low pH, and high nitrate are correlated with significant nitrous oxide production. Depth-resolved metagenomes showed that consumption of nitrous oxide near the surface was correlated with an enrichment of Clade II nitrous oxide reducers, consistent with a growing appreciation of their importance in controlling release of nitrous oxide to the atmosphere. Our work also provides evidence for the reduction of nitrous oxide at a pH of 4, well below the generally accepted limit of pH 5.

Artificial Neural Network Language Models Predict Human Brain Responses to Language Even After a Developmentally Realistic Amount of Training.

Artificial neural networks have emerged as computationally plausible models of human language processing. A major criticism of these models is that the amount of training data they receive far exceeds that of humans during language learning. Here, we use two complementary approaches to ask how the models' ability to capture human fMRI responses to sentences is affected by the amount of training data. First, we evaluate GPT-2 models trained on 1 million, 10 million, 100 million, or 1 billion words against an fMRI benchmark. We consider the 100-million-word model to be developmentally plausible in terms of the amount of training data given that this amount is similar to what children are estimated to be exposed to during the first 10 years of life. Second, we test the performance of a GPT-2 model trained on a 9-billion-token dataset to reach state-of-the-art next-word prediction performance on the human benchmark at different stages during training. Across both approaches, we find that (i) the models trained on a developmentally plausible amount of data already achieve near-maximal performance in capturing fMRI responses to sentences. Further, (ii) lower perplexity-a measure of next-word prediction performance-is associated with stronger alignment with human data, suggesting that models that have received enough training to achieve sufficiently high next-word prediction performance also acquire representations of sentences that are predictive of human fMRI responses. In tandem, these findings establish that although some training is necessary for the models' predictive ability, a developmentally realistic amount of training (∼100 million words) may suffice.

GNOME: A Gaseous Nitrogen Oxide Measuring front-End for aqueous environmental materials.

GNOME is a straightforward and easy to build sample handling apparatus, designed to prepare dissolved nitrogen oxide species in aqueous environmental samples. GNOME is designed to serve as a sample preparation device for downstream chemiluminescent analysis. It is based on the familiar chemistry ring stand; the major advantage scalability designed to accommodate the needs of the user. Additionally, since GNOME is constructed of discrete, snap-to-fit components, the modular design allows users to easily substitute or replace parts. Given that there are few to zero commercial equivalents, construction plans to fill this hardware gap are offered herein. The inlet can resolve down to a lower limit of at least 0.05 nmoles NOX, and is instrumentally linear to a ≥10 nmoles NOX. The approach increases sample throughput, data quality, and overall user experience, making it more efficient than the off-the-shelf commercial equivalent.

Hydrogen and dark oxygen drive microbial productivity in diverse groundwater ecosystems.

Around 50% of humankind relies on groundwater as a source of drinking water. Here we investigate the age, geochemistry, and microbiology of 138 groundwater samples from 95 monitoring wells (<250 m depth) located in 14 aquifers in Canada. The geochemistry and microbiology show consistent trends suggesting large-scale aerobic and anaerobic hydrogen, methane, nitrogen, and sulfur cycling carried out by diverse microbial communities. Older groundwaters, especially in aquifers with organic carbon-rich strata, contain on average more cells (up to 1.4 × 107 mL-1) than younger groundwaters, challenging current estimates of subsurface cell abundances. We observe substantial concentrations of dissolved oxygen (0.52 ± 0.12 mg L-1 [mean ± SE]; n = 57) in older groundwaters that seem to support aerobic metabolisms in subsurface ecosystems at an unprecedented scale. Metagenomics, oxygen isotope analyses and mixing models indicate that dark oxygen is produced in situ via microbial dismutation. We show that ancient groundwaters sustain productive communities and highlight an overlooked oxygen source in present and past subsurface ecosystems of Earth.