Assessing the effect of water table level on carbon dioxide and methane exchange from a tropical peatland mesocosm experiment using automated soil flux chambers

The majority of tropical peatlands located in Southeast Asia are being affected by artificial drainage, which lowers the water table level (WTL) to provide conditions suitable for establishing profitable plantation crops, such as Acacia crassicarpa. Nevertheless, drawdown of the WTL also exposes peat organic matter to belowground heterotrophic microbial activity, subsequently generating greenhouse gas (GHG) emissions. Over the recent years, a consensus has emerged in the scientific community suggesting that reversing drainage effects through rewetting peatlands is the key to mitigating GHG emissions: higher WTLs lead to lower GHG emissions overall. In this study, we tested this emerging paradigm by constructing a mesocosm experiment comprising duplicates of three different WTLs in a degraded tropical peatland near Jambi, South Sumatra. Instead of directly controlling the WTL, peat surfaces were excavated to different depths (i.e., no removal as control, -25cm and -50cm of surface peat removed) with an area of 100 m² for each plot. All plots were subject to similar seasonal WTL fluctuations but the respective WTL distance from the surface varied depending on how much peat was removed at each plot. A total of eight automated chambers monitored the carbon dioxide (CO₂) and methane (CH₄) soil fluxes at hourly intervals for the period August 1^st 2022 to December 8^th 2022. A Random Forest model was developed for each of the eight chambers to identify the driving environmental variables (including WTL) and gap-fill the missing data when technical problems prevented data collection. Our results show that the WTL-CO₂ emissions relationship at our study site was close to linear with greater emissions (mean ± stdev; 3.85 ± 1.87 µmol m^-2 s^-1) at the plots where distance of WTL from the surface was lowest, and lower emission (1.44 ± 1.21 µmol m^-2 s^-1) at the control plots. The effect of WTL on CH₄ fluxes was, however, more complex and ranged between -18.01 and 1692.36 nmol m^-2 s^-1, with the greatest mean emission rate and variability (0.88 ± 1.25 nmol m^-2 s^-1) in one of our control sites with standing Acacia. Differences in vegetation and root-mediated gas release thus seem to influence CH₄ emissions at a greater rate than WTL at our study site. Additionally, our Random Forest model determined that soil surface temperature was a better explanatory variable than WTL in predicting both CO₂ and CH₄ spatial and temporal variability. Our results suggest that effectively managing GHGs from degraded peatlands may be more complicated than looking at WTL alone. Consequently, we recommend running long-term continuous time series and adding other environmental variables, such as soil temperature, in models to better understand, predict and manage GHGs exchange from tropical peatlands.