Northern wetlands and bogs are a large source of biogenic methane (CH₄) to the atmosphere and shape climate. There is then substantial interest in quantifying the effect sizes of controls on bog CH₄ emissions. Yet quantification is complicated because wetland emissions are, like the controls that regulate them, spatially and temporarily heterogeneous within individual wetland systems. To account for this heterogeneity, we established 55 measurement plots in a sampling design that had extensive spatial, and repeated temporal, coverage over one ombrotrophic bog. This coverage allowed us to robustly estimate the effect sizes of known environmental controls on CH₄ emissions such as temperature, water table depth (WTD), and plant functional composition. As expected, 89% of the cumulative measured flux came from 16% of the plots, with consistently high CH₄ emissions over two consecutive growing seasons. Soil temperature exhibited the strongest control on emissions, while WTD and vegetation composition had much smaller relative effects compared to temperature. Interestingly, at high temperatures, WTD was a strong determinant of whether a plot had high vs. lower emissions. Despite temperature’s large effect size, the controls we measured only explained ~29% of the variance in CH₄ fluxes. Overall, our data confirm observations from other wetlands that CH₄ emissions are highly variable at local scales, and yet our data also reveal that effect sizes can be quantified robustly within this heterogeneity. Highly-replicated, local designs may therefore be useful for generating data that can serve to help refine wetland CH₄ emissions uncertainties.

This dataset is comprised of three tabs in an Excel file. The “metadata tab contains information about the variables measured. The “GHG flux data” tab contains daily CH₄ and CO₂ flux information and environmental variables measured simultaneously to flux measurements. The “Plot vegetation data” tab has percent cover data for each subplot in the experiment.

The code file is named ‘TayloretalJGRBCode.txt’ and contains all R (the statistical freeware package) code used for data analysis in the manuscript. Analytical steps are explained ahead of each section of code within the code file.