Fine-Scale Spatial Variability of Greenhouse Gas Emissions From a Subantarctic Peatland Bog

Peatlands are recognized as crucial greenhouse gas sources and sinks and have been extensively studied. Their emissions exhibit high spatial heterogeneity when measured on site using flux chambers. However, the mechanism by which this spatial variability behaves on a very fine scale remains unclear. This study investigates the fine-scale spatial variability of greenhouse gas emissions from a subantarctic Sphagnum peatland bog. Using a recently developed skirt chamber, methane emissions and ecosystem respiration (as carbon dioxide) were measured at a submeter scale resolution, at five specific 3 × 3 m plots, which were examined across the site throughout a single campaign during the Austral summer season. The results indicated that methane fluxes were significantly less homogeneously distributed compared with ecosystem respiration. Furthermore, we established that the spatial variation scale, i.e., the minimum spatial domain over which notable changes in methane emissions and ecosystem respiration occur, was <0.56 m2. Factors such as ground height relative to the water table and vegetation coverage were analyzed. It was observed that Tetroncium magellanicum exhibited a notable correlation with higher methane fluxes, likely because of the aerenchymatous nature of this species, facilitating gas transport. This study advances understanding of gas exchange patterns in peatlands but also emphasizes the need for further efforts for characterizing spatial dynamics at a very fine scale for precise greenhouse gas budget assessment.


S1. Chamber design
As previously described by Thalasso et al. (2023) and depicted in Figure S1, the chamber used in this study was a pyramidal trunk basket (Model 47970, Spectrum, Mexico) with a base (opening) measuring 0.32 × 0.29 m and a height of 0.22 m.A low-density polyethylene film (1.4 × 1.4 m; 0.025 mm thick; Frost King, Mexico) was positioned above the chamber and securely attached to the bottom.A battery-operated fan (Portable Fan, Cazokasi, Mexico) was affixed to one of the lateral faces of the chamber, opposite the side facing the sun, and operated at an airflow speed of about 1.2 m s -1 .Two units of flexible polyurethane tubing (external diameter: 6 mm, internal diameter: 4 mm; PUN-6X1-DUO-BS, Festo, Mexico) were attached inside the chamber, to opposite sides of the basket, approximately two-thirds of the chamber's height, passing from below the chamber's edge and connecting to the UGGA.During flux measurements, the chamber was positioned with the opening facing downwards, the plastic skirt was expanded around the chamber, and the steel chain (0.27 kg m -1 ) was placed above the plastic film, wrapping the base of the chamber three times (Figure S1).

S2. Flux measurements
The flux determination protocol involved four steps: step 1 -the ground air concentration (C L ) of CH 4 (C L,CH4 ) and CO 2 (C L,CO2 ) was measured for 5 min, just above the vegetation cover (where the chamber was placed); step 2 -the chamber was positioned and the gas concentration inside the chamber was measured.Once steady state was reached, C B of CH 4 (C B,CH4 ) was measured over a 5-minute period; step 3 -a pulse of about 1 mL of standard CH 4 (99.99 %, Linde, Chile) was injected with a plastic syringe through a septum connected on the waste line of the UGGA (returning to the chamber).It's important to note that the exact amount of CH 4 injected does not require precision, as outlined in Thalasso et al. (2023).Thus, we used small, easily transportable vials as the CH 4 source.This injection caused an abrupt and artificial increase of the CH 4 concentration, which then decreased asymptotically due to gas exchange between the chamber and the environment, i.e. the leaks that are measured with this method.The decreasing CH 4 concentration was used to determine  C (Ström et al. 2003).This step was maintained for 5 to 7 minutes, until a stable CH 4 concentration was observed; step 4 -a dark screen was placed on top of the chamber for 5 minutes to measure the CO 2 concentration (C B,CO2 ), which correspond to the CO 2 flux in absence of light (respiration).In the present work two fluxes were measured, the CH 4 flux (F CH4 ), and the CO 2 flux under dark conditions, thus corresponding to the CO 2 respiration rate of the ecosystem (R CO2 ).

S3. Ebullitive versus diffusive fluxes
It is worth noting that, for its design, the skirt-chamber captures and measures the total flux reaching the chamber, including ebullitive events, well described in peatlands (Baird et al., 2019;Strack et al., 2005).In this context, the open dynamic nature of the skirt-chamber should allow distinguishing both emission modes; i.e., ebullitive and diffusive, as previously observed in a similar chamber concept deployed in freshwater ecosystems (Gerardo-Nieto et al., 2019).However, with the skirt-chamber in peatlands, we have never observed peak increases in CH 4 or CO 2 concentration that would be the result of a bubble reaching the chamber.Our hypothesis is that the skirt-chamber, deployed under non-submerged conditions with the water table depth typically ranging from 0.1 to 0.6 meters, allowed bubbles released at the water surface to undergo progressive dilution until they reached the peat surface and the chamber.This process minimizes the abrupt increase in concentration observed in chambers deployed under submerged conditions when bubbles directly reach the chamber volume.4); the purple area corresponds to the area that would be observed under a perfectly homogeneous distribution of F CH4 , the orange area represents the area corresponding to the deviation to that perfect case, which define the ht parameter (Eq.5); A <0 represents the percentage of area initial where negative fluxes are observed; A 5 indicates the percentage or area where an emission superior to five time the mean emission (hotspots) and A 90 represents the percentage of area responsible for 90% of the total emission (see main document for details).Panel A, shows an example, based on F CH4 measured at Plot 4, with large heterogeneities, hotspots and negative fluxes; Panel B shows a theoretical example with small heterogeneities, no hotspots and no negative fluxes (M' j never exceed 1.0); Panel C shows a theoretical example with large heterogeneities, hotspots and no negative fluxes.
Figure S3: Heatmaps of the measured ground height relative to the water table (H) at the five 3 × 3 plots.Crosses indicate location where F CH4 were measured.Diversity Index (-)

Figure S2 :
Figure S2: Graphical representation of the numerical homogeneity model (NHM).The red continuous line depicts the pair formed by the normalized cumulated and ordered mass of CH 4 emitted by the plot (M') and the corresponding normalized cumulated area (A'; Eqs. 3, 4); the purple area corresponds to the area that would be observed under a perfectly homogeneous distribution of F CH4 , the orange area represents the area corresponding to the deviation to that perfect case, which define the ht parameter (Eq.5); A <0 represents the percentage of area initial where negative fluxes are observed; A 5 indicates the percentage or area where an emission superior to five time the mean emission (hotspots) and A 90 represents the percentage of area responsible for 90% of the total emission (see main document for details).Panel A, shows an example, based on F CH4 measured at Plot 4, with large heterogeneities, hotspots and negative fluxes; Panel B shows a theoretical example with small heterogeneities, no hotspots and no negative fluxes (M' j never exceed 1.0); Panel C shows a theoretical example with large heterogeneities, hotspots and no negative fluxes.

Figure S6 :
Figure S6: Shade-plot of the relative abundance of the seven vegetation classes across the 80 locations.