The emission of methane from wetlands is spatially heterogeneous, as concurrently measured surface fluxes can vary by orders of magnitude within the span of a few meters. Despite extensive study and the climatic significance of these greenhouse gas emissions, it remains unclear what drives these large within-site variations, creating a knowledge-gap that impedes a mechanistic understanding of wetland fluxes. While geophysical variables including water table depth (WTD) and soil temperature are known to correlate with CH₄ flux, measurable variance in these parameters declines as spatial and temporal scales become finer. Here, we leveraged depth-stratified gene abundance and gene expression measurements of methanogenesis and methanotrophy to investigate CH₄ flux variance at an ombrotrophic peat bog. Our results show that the flux variance was strongly correlated to methanogen abundance and that peat depth also exerted significant control over CH₄ flux, methanogen abundance, and the relationship between the two. Correlations between CH₄ flux and either WTD or soil temperature were absent or minimal. These findings suggest that microbial factors likely underlie localized variance in wetland CH₄ flux, and that a greater reliance on biological predictors could improve our ability to understand wetland methane fluxes at finer scales than is currently possible.

The majority of the data consists of RNA and DNA abundances of specific genes (mcrA,  pmoA, mmoX) quantified via the digital droplet polymerase chain reaction (ddPCR) method. Relatedly, dsDNA abundance for corresponding samples was quantified using fluorometric methods on a Qubit. Additionally, there is accompanying environmental data from the study site, including soil temperature (measured via in-ground probe), water table depth (measured from surface using constructed wells), and refusal depth (measured as a distance with an in-ground depth probe). Lastly, there is methane flux data for our study site, quantified via cavity ring-down spectroscopy using a closed chamber method. All data has been processed in R Studio, using the packages specified in the accompanying manuscript. 

All datafiles can be opened in Excel or an equivalent CSV editor. The main script is in the form of an R markdown document.