Our study took place during the growing season in June and July 2019 on the Albemarle Pamlico Peninsula (APP), in eastern North Carolina. We selected five areas: Palmetto-Peartree Preserve (PPP), Point Peter Rd within the Alligator River National Wildlife Refuge (PP), Swanquarter National Wildlife Refuge (SQ), Pocosin Lakes National Wildlife Refuge (PC), and Gull Rock State Game Lands (GR) (Martinez and Ardón, 2021). The coastal wetlands along the APP include non-riverine swamp forests, pocosins, and freshwater and brackish marshes. The APP has generally low salinity (one to seven parts per thousand) and is subject to frequent flooding from wind tides (Manda et al., 2014). More than 50% of the peninsula is within 1.5 m of mean sea level, making it highly vulnerable to sea level rise and saltwater intrusion (Moorhead and Brinson, 1995). The APP has been drained via extensive artificial canals and ditches to improve the agricultural value of the land (Manda et al., 2014), but these canals also allow saltwater to propagate further inland especially during storms, imperiling nearby vegetation communities (Bhattachan et al., 2019). Saltwater intrusion and increasing flooding events have increased the amount of forest die back in the region, creating extensive ghost forests (Smart et al., 2020).

Five snags were selected at each site (n = 25) based on stability (i.e., intact and not at risk of falling). There are different stages of decay for standing dead trees, which can be classified from 1 to 5, with stage 1 being intact and recently deceased while stage 5 being close to falling over (Odion et al., 2011). The snags selected for all sites, except PC, were all pine trees with most of the bark removed classified as stages ranging from 1 to 3. The snags selected at PC were a mixture of pine and cypress trees with most of the bark removed, also ranging from 1 to 3 stages of decay. Snag and soil CH₄ concentrations were measured using clear semi-rigid chambers at various heights ranging from 60–70 cm from ground, depending on optimal sealing conditions (Figure 1). The snag-stem chambers were designed after Siegenthaler et al. (2016), which is a flexible type of chamber made from a clear, polycarbonate sheet. Soil chambers (30 cm diameter) were installed 1–2 m a way from snags. The soil chamber tops included a vent tube, internal fan (0.003 m³/s; Jameco Electronics, Belmont, California, United States), and three sampling ports, two hose barbed connectors for portable gas analyzer tubing (Input/Output air sample) and air-tight tubing (12 inches) with a three-way stopcock for extra gas samples. The soil chambers were inserted 10 cm into the soil and left to stabilize for at least 1 week before taking measurements.

The CH₄ in snag and soils chambers was measured using two different methods: Gasmet portable gas analyzer (Gasmet Technologies, Finland) and the syringe method (Figure 1). The Gasmet is capable of analyzing a variety of greenhouse gases, but this study is focused only on CH₄. Chambers were attached for 10–20 min (10 min for snags and 20 for soils), and measuring gas concentrations every 20 s using the Gasmet, or 30–40 min measuring every 5 min using the syringe method. The syringe method was used at times when the Gasmet was not optimal due to either weather or battery failures. Gas samples using syringe method were stored in Exetainer vials (Labco–Lampeter, Wales, United Kingdom). Prior to field use, Exetainer vials were evacuated using a vacuum pump connected to a manifold and flushed repeatedly with ultra-pure N₂ gas. Gas samples were stored in vials and run within a 6-week period, which has been shown to maintain sample concentrations (Sturm et al., 2015). CH₄ gases were analyzed using an Agilent gas analyzer (7890 A GC system - Santa Clara, CA) equipped with a methanizer and Flame ionization Detector for CH₄ analysis. Detailed descriptions of the region, site, and methods can be found in Martinez and Ardón (2021).

CH₄ concentration inside snag heartwood were assessed following a similar protocol to Covey et al. (2012). A 1.27 cm diameter drill bit was used to drill to the center of the tree at three heights (30, 60, and 120 cm). After drilling, each hole was plugged using a rubber septum and pierced with a sampling wand connected to the Gasmet to sample for ∼3 min. The air immediately outside the drilled hole was also sampled for comparison and to allow for the higher gas concentration to flow out of the tubing.

Gas samples for stable isotope analysis were sampled once at each site in June 2019 for GR and PP and July 2019 for PC, SQ, and PPP from inside the snag (n = 25) and soil (n = 25) chambers after chamber incubation using a gas tight syringe. Both snag and soil chambers were modified to sample with a syringe either through a rubber septum (snag chambers) or tubing using a three-way stopcock (soil chambers). Heartwood gas samples were also collected for stable isotope analysis with a syringe through a rubber septum 60 cm from ground (n = 25).

Belowground porewater sippers were installed at 30 cm near three snags at each site. Porewater was analyzed to help provide insights into whether CH₄ emitted by stems was produced in soils or groundwater (Barba et al., 2018). Porewater was drawn after snag and soil gas sample collections and measured for δ ¹³C-CH₄, δ ²H-CH₄, and CH₄ concentration. CH₄ concentrations were measured using ultra-high purity N₂ using headspace equilibration methods (Sturm et al., 2015). Approximately 6 ml of porewater was injected into a pre-evacuated 12 ml exetainer. Liquid exetainer samples were stored upside down to prevent gas leakage and placed immediately in a cooler in the field. At the end of each field day collection, 6–7 ml of ultra-pure N₂ gas was added to provide over-pressure preventing access of atmospheric air during transport. Samples were wrapped in aluminum foil, stored upside down, and kept in cool refrigerator (4°C) until lab analysis. Samples were taken out of fridge 24 h prior to GC analysis to reach room temperature to allow equilibration of gases in headspace and water column. The volume of the porewater sample and headspace was determined by weighing the exetainer vial. The concentration of gas in the original water sample was determined by calculating the aqueous concentration in water and headspace after equilibration (Helton et al., 2014).

Porewater samples collected for isotopes δ ²H-CH₄ and δ ¹³C-CH₄ were acidified with HCl acid in the field to a pH < 2 and capped with no headspace bubbles. Two sets of porewater samples were collected per sipper for isotopes δ ²H-CH₄ and δ ¹³C-CH₄, which are analyzed separately. All gas and porewater samples for stable isotope measurements were shipped to UC Davis Stable Isotope Facility. There the stable isotope ratios of carbon (δ¹³C) and hydrogen (δ²H) in CH₄ were measured using a ThermoScientific Precon concentration unit interfaced to a ThermoScientific Delta V Plus isotope ratio mass spectrometer (ThermoScientific, Bremen, Germany).

Snag tree cores were taken (June and July 2019) to provide estimates of CH₄ emissions by incubation to test the potential of wood to produce CH₄ (n = 100; PMP and PMO). Tree cores (5 mm diameter × 10–30 cm length) included the bark (if present), sapwood, and heartwood. Heartwood rot/wet wood conditions provide a conducive habitat for microbial CH₄ production, therefore potential CH₄ production (PMP) from incubations could suggest tree decomposition as an important source of CH₄ (Covey et al., 2012; Barba et al., 2018). Tree cores were taken at two heights (120 and 60 cm) from the five sites, to perform wood incubations to measure potential CH₄ production (n = 50, 25 per each height). Cores were placed in 120 ml serum vials and flushed with N₂ for 8–10 min. Gas samples were taken from the vial at 0, 12, 24, 48, and 72 h after incubating. A separate empty vial was flushed with N₂ as a field blank to measure potential background concentrations. Snag tree cores from GR and PP were run in June 2019 measuring at time points 0, 12, 24, and 48 h, while snag tree cores from PPP, SQ, and PC sites were run in July 2019 measuring at time points 0, 24, 48, and 72 h. The 12 h time point for July incubation was deemed unnecessary, and we extended the incubation to 72 h to determine if a longer incubation was needed to measure change in CH₄ concentration during treatments.

In addition to measuring potential CH₄ produced under anaerobic conditions, snag cores from similar heights (120 and 60 cm from ground) were incubated under higher concentrations of CH₄ to measure potential CH₄ oxidation (PMO) under aerobic conditions (n = 50). Vials were flushed with a gas standard mixture of CO₂, CH₄, and N₂O near ambient levels with a much higher CH₄ concentration. Snag cores from GR and PP were incubated at a CH₄ concentration of ∼45 ppm, while snag cores from PPP, SQ, and PC were incubated at a CH₄ concentration of ∼120 ppm. The difference in CH₄ concentrations was due to difficulties diluting Pure CH₄ (99.999%). CH₄ can potentially be oxidized by methanotrophs and/or nitrifying bacteria which have an affinity for CH₄ (Chan and Parkin, 2001). Gas samples were taken at similar intervals as the potential CH₄ production incubations. An empty vial flushed with the same gas standard mixture was measured as a field blank to assess potential gas leaks or air seepage inside the vial.

We calculated snag tree core CH₄ fluxes, represented as ng g⁻¹ hr⁻¹, using the volume of the vial (minus the volume of the snag tree core) and the weight of the snag tree core as a substitute for area. Criteria were used to filter data using a minimum detectable concentration difference (MDCD) between the initial and final measurements. The MDCD in this study is based on the detection limits of the GC, which is 20% (or 0.2 ppm) for CH₄, so |Final_{CH4 Concentration}–Initial_{CH4 Concentration}| > 0.2 (meets MDCD criteria). Observations that did not meet the criteria were not used (marked as NA). The incubation fluxes were calculated using the HMR approach (HMR R package), which uses both linear and non-linear approaches (Pedersen et al., 2010). Only data with significant correlations (p < 0.1) were kept.

Due to the lack of normality in the data, Kruskal-Wallis was used to test differences in stable isotopes (δ¹³C-CH₄ and δ²H-CH₄) among source types, and heartwood CH₄ concentrations among sites, and between heights (30, 60, and 120 cm). Dunn’s multiple pairwise comparisons were then used for post-hoc analysis when Kruskal-Wallis showed significant differences, with a Bonferroni p-value adjustment (p < 0.05). Linear regression models were used to assess the relationships between stable isotopes and CH₄ concentrations with all source types (snag chambers, soil chambers, heartwood, and porewater), and also between CH₄ heartwood concentrations and stem height. CH₄ concentration data was log transformed due to lack of normality. All statistical analyses were performed in R program (R Core Team, 2020).

Heartwood CH₄ concentrations differed among sites and between heights (p < 0.01, Figure 2). Concentrations from PC (mean ± SE: 904 ± 415 ppm) were an order of magnitude higher than all other sites, which were on average 13 ppm (p < 0.01, Figure 2). For all sites, CH₄ concentrations decreased with increasing height, with measurements from 30 cm being 2–4 times higher than measurements from 120 cm (p = 0.01). The linear regression model for CH₄ concentration (log transformed) by height for all sites combined (Eq. 1) was significant (p < 0.01) although explained a small amount of variance (r²=0.04). f ( C H 4   ( p p m ) ) = − 0.01 × H e i g h t   ( c m ) + 3.34 (1)

The δ¹³C-CH₄ value of CH₄ ranged from −74 to −39‰, while δ²H-CH₄ ranged from -327 to -34‰. Both δ¹³C-CH₄ and δ²H-CH₄ stable isotopic values became more enriched (more positive) from porewater < heartwood < soil chamber < snag chamber (Figures 3, 4). Porewater δ¹³C-CH₄ (mean: −60 ± 1.7‰) and δ²H-CH₄ (−327 ± 4.8‰) were more depleted than soil chamber, heartwood, and snag chamber values (p < 0.1) (Figures 3, 4). Soil chamber δ¹³C-CH₄ values (−48 ± 1.5‰) were not significantly different from snag chamber (−47.7 ± 0.9‰; p = 1.0) or heartwood values (-52.2 ± 1.5‰; p = 0.13), but snag chamber values were more enriched than heartwood (p = 0.06). Similarly, soil chamber δ²H-CH₄ values (−168 ± 14‰) were also within the range of snag chambers (−132 ± 12‰; p = 0.70), and heartwood (−209 ± 14‰; p = 0.53) values, but heartwood values were more depleted than snag chamber values (p = 0.006). Soil chamber δ¹³C-CH₄ ranged from −73 to −39‰ while δ²H-CH₄ ranged from -304 to -34‰. Heartwood δ¹³C-CH₄ ranged from −73 to −44‰, while δ²H-CH₄ ranged from −327 to −105‰. Overall, there was enrichment in isotopic signature in both δ¹³C and δ²H for all source types with decreasing CH₄ concentrations across all sites (Figures 3, 4), except for δ¹³C from SQ.

CH₄ concentrations in snag tree core incubations changed very little over time (Figure 5). The overall number of snag tree incubation cores that had a measurable flux was 10 out of 100 cores taken. From the measurable fluxes, 5 were from the potential CH₄ oxidation (PMO) treatment, while the other five were from the potential CH₄ production (PMP) treatment (Figure 5). The highest CH₄ flux within the PMP treatment was 2.12 ng g⁻¹ hr⁻¹ for cores 60 cm from ground, while the smallest increase was 0.35 ng g⁻¹ hr⁻¹ for cores 120 cm from ground. One core from 120 cm height produced a small amount of CH₄ under PMO conditions, 0.5 ng g⁻¹ hr⁻¹. The highest oxidized CH₄ within the PMO treatment was -2.5 ng g⁻¹ hr⁻¹ from 120 cm height core. The sample size was too low to perform any statistical analysis on the incubation flux calculations.

CH₄ concentrations in heartwood varied across the sites. The CH₄ concentrations at PC were higher (904 ± 415 ppm) because there was standing water year-round, unlike all other sites (Figure 2; Table 1). When drilling into the heartwood at PC, some snags had a small stream of dripping water from within the tree (personal observation). The high moisture (and likely low oxygen levels) within the snags were conducive to CH₄ production, which would explain why their concentrations were so much higher than all other sites. The mean heartwood CH₄ concentration for the other sites (∼13 ppm) were below those found in snags from a similar region within the APP (100 ppm) (Carmichael et al., 2017). The difference between snags from the Carmichael et al. (2017) study and ours could be due to site differences similar to PC within our own study. The snags from Carmichael et al. (2017) were in standing water year-round.

There was a gradual decrease in CH₄ concentration with increasing stem height (Figure 2), which has been found in other studies when measuring heartwood CH₄ concentrations in live and dead trees (Covey et al., 2016; Carmichael et al., 2017), but this decrease also affects tree stem CH₄ fluxes (Pangala et al., 2017; Jeffrey et al., 2019). The decrease in heartwood CH₄ is likely due to distance from source (CH₄ originating from within soils) and increasing oxidation with increasing stem height, which is also why tree-stem CH₄ flux would be affected (Pangala et al., 2017; Jeffrey et al., 2019). Using a linear regression model, we calculated that CH₄ concentrations inside the heartwood would decrease to 0 near 3.34 m (Eq. 1). Snags across the study sites ranged in height from 3 to 15 m. Our results suggest that snags can be important areas for CH₄ oxidation. It is important to note that measurements were taken during a relatively warm time of year (Table 1), so it is possible that during time periods when the air is cooler, there might be less CH₄ oxidation in the snags (King and Adamsen, 1992).

Stable isotope data from all sites had similar patterns (enrichment from porewater < heartwood < soil chamber < snag chamber) with the exception of δ¹³C from SQ (Figure 3). As we predicted most of the CH₄ produced from each source: snag chamber, heartwood, soil chamber, and porewater, was generated through the acetoclastic pathway which ranges from δ¹³C ∼ −65 to −50‰, and δ²H ∼ −400 to −250%, although a few samples (n = 4) were less than δ¹³C −65‰ from porewater, soil, and heartwood samples. One sample from heartwood in PC had really high CH₄ concentrations (∼40,000 ppm) and a very depleted δ ¹³C signature (−73‰), suggesting that hydrogenotrophic pathway might happen in stems standing in water. The δ¹³C-CH₄ from various freshwater wetlands have been determined to be within the range of −86‰ to −31‰ (Bréas et al., 2001). A previous study conducted in freshwater wetlands concluded that dissolved CH₄ in porewater at ca. 25−cm depth had δ ²H values of ∼ –340‰ or less (Hornibrook et al., 2000), which is also within the range of our study. Overall, the δ¹³C and δ²H values indicate that most of the CH₄ measured was produced through the acetate fermentation pathway which is common in most freshwater soils (Bréas et al., 2001).

Both δ ²H and δ ¹³C of porewater CH₄ were significantly depleted compared to all other sources (Figures 3, 4). This would indicate that the CH₄ concentrations from porewater samples were closer in composition to that of the anoxic zone, or anoxic ‘‘microsites’’, leading to lower δ ²H and δ ¹³C (Sugimoto and Fujita, 2006). Porewater CH₄ concentrations are thought to be the best representation of CH₄ production, before any processes (i.e., diffusion and/oxidation) changes the stable isotope signature (Bréas et al., 2001), and therefore those values were used in this study to assess degree of oxidation in heartwood, snag chamber, and soil chamber. The end product of complete CH₄ oxidation is water vapor, which is also why δ²H becomes enriched (heavier), although rates of hydrogen transfer due to organic matter degradation can also affect δ²H (Bréas et al., 2001).

Stable isotopes from heartwood and soil chambers were more similar to each other and had wider ranges than snag chambers (Figures 3, 4). Heartwood δ¹³C and δ²H were on average similar to the mean porewater δ¹³C and δ²H, indicating that the degree of oxidation was less than the pathway from porewater to soil chamber, although there was a wide range for both δ¹³C and δ²H (Figures 3, 4). CH₄ oxidation occurred more in samples from snag chambers and were closest to ambient air stable isotopes (−46‰ and −114‰, δ¹³C and δ²H respectively), providing further evidence that is consistent with CH₄ continuing to be oxidized as it moves through the snags (Figures 3, 4). A previous study also showed δ¹³C-CH₄ enrichment with increasing stem height in a flooded forested ecosystem, suggesting oxidation during tree stem transport (Jeffrey et al., 2021). This same study also found methane oxidizing bacteria (primarily Methylomonas) to be abundant within bark representing an important sink (Jeffrey et al., 2021). We have collected evidence of methanotrophs within snags in GR at ∼60 cm from ground, although the abundance is not yet known (unpublished data; Carmichael et al. In Prep). The degree of oxidation in our study likely depends on the stage of decay of the snags. Later stages of decay (stages 4–5) likely having more porous stems (lower density) therefore increasing oxidation, whereas earlier stages of decay (stages 1–3) may continue emitting CH₄ due to moist tree stems which are more advantageous to methanogens, similar to the snags measured in this study.

The overall positive correlation between δ²H and δ¹³C of CH₄ in freshwater wetlands has also been shown in other studies (Sugimoto and Wada, 1995; Sugimoto and Fujita, 2006). Isotopic fractionation, in both δ¹³C and δ²H, has been shown to decrease with decreasing CH₄ concentrations, which is what we observed in all sites, except SQ (Figures 3, 4) (Riveros-Iregui & King, 2008). It is possible that the higher salinities and thus higher SO₄ ^2- content in the sediments in SQ inhibited methanogenesis because SO₄ ^2- is a preferential electron acceptor in saturated sediments (Weston et al., 2006). If there is less CH₄ (substrate) in the sediment, then there will be less methanotrophy and less enrichment in δ¹³C signal (Edmonds et al., 2009).

Tree cores were incubated under two different simulated conditions to further investigate potential production and/or oxidation in snags internally (heartwood and sapwood). Very few (n = 10) cores exhibited a measurable difference in CH₄ from the beginning to the end of the incubation (Figure 5). This suggests that any amount of CH₄ that is present in the heartwood is largely soil derived. Under the PMP treatment, the highest measurable CH₄ flux occurred in cores taken from the base of the tree, while in the PMO treatment the highest measurable CH₄ oxidation flux occurred in cores from 120 cm from base of the tree. It is possible that there are more abundant methanogen communities closer to the base where conditions are more conductive for methanogenesis (i.e., higher moisture content, thus lower O₂ availability), but not as much as deep soil layers where majority of the CH₄ is derived. The fact that there was very little change in CH₄ concentrations throughout both treatments at both heights (PMP and PMO) further demonstrates that the majority of the CH₄ emitted from snag stems is largely soil derived (Figure 5). A study looking at CH₄ fluxes from stems in the Amazon basin, similarly found that very few of their tree core incubations produced CH₄, and interpreted those results as support for soil as the source of CH₄ (Pangala et al., 2017).