Fig. 3: Simulations with the full range of microbial traits led to CH₄ emission predictions consistent with observations.

Modeled and observed daily CH₄ emissions in the A fen and B bog sites. For the palsa site, the model accurately captured the observed NEE and CH₄ emissions (Supplementary Fig. 7). Blue-shaded areas show modeled daily mean emissions (2004–2009) due to the full range of microbial traits and solid lines show the best fit of the modeled emissions. Orange circles represent daily averaged CH₄ emissions observed for each day of the year (DOY) across the simulation period when at least two years of data are available³⁷. Given the absence of Autumn data in 2005, 2006, and 2008, DOY average Autumn CH₄ emissions may not be representative of true inter-annual variability. Full-time series of results are provided in Supplementary Fig. 5. Simulated daily CH₄ emissions over the simulation period are evaluated by calculating Pearson correlation coefficient (R) and root mean square error (RMSE) when quality control measurements are available (Fen: R = 0.51, RMSE = 80 mgC m⁻² d⁻¹, n = 359; bog: R = 0.44, RMSE = 22 mgC m⁻² d⁻¹, n = 392). Modeled daily CH₄ emissions averaged over each season in C fen and D bog sites. The boxes represent the median and the first and third quartiles of daily emissions for each season (n = 1300) during 2004–2009. The whiskers represent the minimum and maximum values within 1.5 times the interquartile range. E Sensitivity analysis of annual CH₄ emissions attributed to each microbial trait measured by Mean Elementary Effect (“Methods”) for dominant microbial functional groups. Results for all microbial traits including traits related to nitrogen cycling are provided in Supplementary Fig. 8, which shows a much lower sensitivity of CH₄ emissions to these traits compared with the five dominant microbial functional groups. Each solid square bar depicts the interannual mean of 2004–2009 (n = 6), and the error bars represent the standard deviation around the mean. R_max maximum specific respiration rate, HM hydrogenotrophic methanogen, AM acetoclastic methanogen, Aero_Heter aerobic heterotroph, DOC dissolved organic carbon. The substrate is shown as a superscript for some functional groups. Note that a positive Mean Elementary Effect of a microbial trait for CH₄ emissions implies more CH₄ is emitted when increasing that microbial trait. K_m also contributes to the variation of CH₄ emissions through its relationship with R_max (Supplementary Fig. 3).