Phenology is the dominant control of methane emissions in a tropical non-forested wetland

Abstract

Tropical wetlands are a significant source of atmospheric methane (CH₄), but their importance to the global CH₄ budget is uncertain due to a paucity of direct observations. Net wetland emissions result from complex interactions and co-variation between microbial production and oxidation in the soil, and transport to the atmosphere. Here we show that phenology is the overarching control of net CH₄ emissions to the atmosphere from a permanent, vegetated tropical swamp in the Okavango Delta, Botswana, and we find that vegetative processes modulate net CH₄ emissions at sub-daily to inter-annual timescales. Without considering the role played by papyrus on regulating the efflux of CH₄ to the atmosphere, the annual budget for the entire Okavango Delta, would be under- or over-estimated by a factor of two. Our measurements demonstrate the importance of including vegetative processes such as phenological cycles into wetlands emission budgets of CH₄.

Fig. 1. Mean monthly flux of CH₄ measured by eddy-covariance over Cyperus papyrus.

Measurements taken at Guma Lagoon (18°57′53.01″S; 22°22′16.20″E) in the permanently flooded part of the Okavango Delta from August 2017 to August 2020. Fluxes (FCH₄) are presented as function of (a) maximum gross primary productivity (GPP_MAX), and (b) enhanced vegetation index (EVI). The solid lines represent linear regressions (equations and t-test two-sided p-value given in the panels): standard error of (a) ±0.19 and 0.12, and (b) ±0.51 and 0.17 for slope and intercept, respectively.

Fig. 2. Diel cycles of methane fluxes measured by eddy-covariance over Cyperus papyrus.

Measurements taken at Guma Lagoon (18°57′53.01″S; 22°22′16.20″E) in the permanently flooded part of the Okavango Delta from August 2017 to August 2020. Half-hourly methane flux data points (FCH₄) were averaged to hourly values on a monthly basis using all available data for the period August 2017–August 2020 were used. The coloured ribbon represents the standard deviation of the mean and the grey rectangles symbolise night-time.

Fig. 3. Local and delta-scale methane emission estimates.

Monthly emission estimates of CH₄ net fluxes (g m⁻² d⁻¹) from eddy-covariance (EC) measurements (round symbols) over Cyperus papyrus at Guma Lagoon (18°57′53.01″S; 22°22′16.20″E), in the permanently flooded part of the Okavango Delta, and inferred from satellite observations of column CH₄ over the entire Okavango Delta (Tg yr⁻¹; square symbols). Data are presented as mean values ±2 standard deviations from the mean. The monthly EC budgets were constructed by summing diel cycles of hourly means; total uncertainty was obtained by propagating hourly standard deviation of the mean in quadrature (n = 24 independent hourly data points, see Eq. (1)). Due to the stochastic nature of the inversions used to derive emission estimates from satellite information, it is not possible to report a deterministic number of samples.

Fig. 4. Relationship between monthly methane fluxes, water level and air temperature.

Monthly median CH₄ fluxes (FCH₄, g m⁻² day⁻¹) ± inter-quartile range (IQR) from August 2017 to August 2020 as a function of (a) monthly water level, and (b) mean air temperature at the seasonal floodplain measurement site (19°32′53″S; 23°10′45″E). The number of half-hourly flux data points (n) from which median and IQR were calculated changed from month to month because of the variability of points filtered out by the micrometeorological quality control filter (see Methods). Consequently, n ranged from 33 (August 2020) to 570 (September 2019). The solid lines represent linear regressions on a temporal subset of the data (austral winter to summer 2017 and 2018) excluding the 2019 drought period (equations and t-test two-sided p-value given in the panels): standard error of (a) ±0.10 and 0.09, and (b) ±0.01 and 0.16 for slope and intercept, respectively.

Fig. 5. Annual CH₄ emission budgets by ecohydrological zones and for the entire Okavango Delta.

The budgets were obtained from upscaled eddy-covariance (EC) measurements and satellite observations. Total EC budgets are broken down into annual emissions from two hydrological zones (perennial and seasonal wetlands). The budgets for the occasionally flooded areas were negligible and were therefore left out. Individual budgets were constructed by summing the monthly emission estimates (n = 12) for each year. The error bars represent the uncertainty range for the respective emissions budgets; these were calculated by summing monthly uncertainties in quadrature (n = 12, see Methods and Eq. (1)).

Fig. 6. Ecohydrological zones of the Okavango Delta.

The map illustrates the spatial distribution and extent of the three main zones in 2019, based on a 25-year flood record and frequency-determined floodplain vegetation communities^, .

Fig. 7. RGB Sentinel-2 imagery of the area surrounding the eddy-covariance (EC) tower at Guma Lagoon (18°57′53.01″S; 22°22′16.20″E).

The region of interest, from which pixels were sampled to calculate the enhanced vegetation index (EVI) of the floating papyrus stand, is shown as a blue polygon. The approximate location of the water level measurement sensor is also indicated.