This article was corrected on 11 AUG 2014. See the end of the full text for details.
Water Conservation Area 2A in the Florida Everglades is characterized by a nutrient gradient with high levels in the north from agricultural runoff and more oligotrophic conditions in the southern interior. Based on laboratory incubations and field studies, we found that the relative importance of methane (CH4) production mechanisms shifted along this gradient, with a greater contribution due to hydrogenotrophic methanogenesis at higher nutrient levels. The relative contributions of hydrogenotrophic and acetoclastic methanogenesis were determined from laboratory experiments and verified with field results. In the lab the relative contributions of the two pathways were determined from the differences in CH4 production rates in soil collected from sites along the nutrient gradient that was incubated with and without an inhibitor of acetoclastic methanogenesis (methyl fluoride, CH3F). In the nutrient-poor soil, most of the CH4 was formed via acetate fermentation and only 25% came from hydrogenotrophic methanogenesis. At the nutrient-impacted site CH4 was produced at fourfold higher rates and the proportion of CH4 produced via hydrogenotrophic methanogenesis increased to 50%. Isotopic fractionation factors for hydrogenotrophic and acetoclastic methanogenesis were calculated from the soil incubations and applied to δ13C-CO2 and δ13C-CH4 measured in pore water from the same transect. The trend of increased hydrogenotrophic relative to acetoclastic CH4 production along the nutrient-impacted gradient was mirrored in the field data, which produced similar results to the lab incubation work, with up to 23% of the CH4 produced from hydrogenotrophic methanogenesis at the nutrient-poor site and nearly half at the nutrient-impacted site.
CH4, a potent greenhouse gas, has been increasing in concentration in the atmosphere during the past century. Wetlands are the largest natural source of atmospheric CH4, contributing 150–180 Tg or > 20% of the global CH4 produced annually [Anderson et al., 2010; Bergamaschi et al., 2007; Bousquet et al., 2006]; hence, there is a need to better understand the mechanisms of CH4 production in these systems. The Florida Everglades are subtropical wetlands that cover a large area of the southern part of the Florida peninsula. The Everglades were historically oligotrophic, but in the past century nutrient influx has altered the ecosystem. High nutrient levels are correlated with elevated microbial activity, CH4 production, and changes in microbial population structures in the Everglades [Bachoon and Jones, 1992; Drake et al., 1996; Wright and Reddy, 2001]. There are two main pathways for CH4 production in terrestrial freshwater wetlands: fermentation of acetate (acetoclastic methanogenesis) and reduction of carbon dioxide (CO2) (hydrogenotrophic methanogenesis). Generally, more labile organic matter promotes the acetoclastic pathway [Hornibrook et al., 1997; Hornibrook et al., 2000b; Sugimoto and Wada, 1993]. Approximately 70% of the CH4 produced in terrestrial environments has been attributed to acetoclastic methanogenesis (Conrad  and review by Whiticar ), while 30% comes from the reduction of CO2 with H2. Hydrogenotrophic methanogenesis, however, dominates CH4 production in northern oligotrophic Sphagnum-dominated bogs as confirmed via incubation and microbial studies and isotopic evidence [Hines et al., 2001, 2008; Horn et al., 2003; Rooney-Varga et al., 2007]. Microbial community analysis has indicated the greater relative importance of hydrogenotrophic CH4 production in the peat soils of the Florida Everglades. In eutrophic Everglades soils, most probable numbers of hydrogenotrophs were >1000 times higher than acetotrophs and around 100 times higher in a nutrient transition zone and oligotrophic soils [Chauhan et al., 2004]. Thus, it appears that although hydrogenotrophic microbes dominate all Everglades soils along this nutrient gradient, the relative importance of the acetoclastic pathway depends upon the nutrient status of the soil. In high-latitude oligotrophic bogs, dominance of hydrogenotrophic over acetotrophic methanogenesis has also been observed [Kotsyurbenko et al., 2007; Hines et al., 2008] and attributed to low pH, among other factors. In contrast to the northern bogs, pH levels in Everglades peat are high (7 to 8).
The isotopic composition of CH4 is related to its origin because δ13C-CH4 is affected by the pathway from which the CH4 is formed, by temperature, and by the δ13C of the source carbon. Discrimination against the heavier isotope, 13C, during CH4 formation leads to lower δ13C in CH4 compared to the precursor substrate and subsequent isotopic enrichment in dissolved inorganic carbon [Corbett et al., 2013]. The apparent isotope fractionation factor, αapp, for methanogenesis is defined as αapp where acetoclastic methanogenesis occurs is typically lower (1.04 to 1.06) than αapp where hydrogenotrophic methanogenesis dominates (1.05 to 1.09) [Hornibrook et al., 2000a; Whiticar, 1999; Whiticar et al., 1986]. This difference is commonly used to determine the relative contribution of each pathway to the CH4 produced in different environments. More specific information about the methanogenic pathway can be obtained when the fractionation factor between the precursor CO2 or acetate and the resulting CH4 is known. Here we refer to the fractionation factor between CO2 and CH4 when CO2 is used as the substrate for methanogenesis as αCO2. Many values for αCO2 exist in the literature, ranging from 1.04 to 1.08 for pure cultures (review by Valentine et al. ) and 1.05 to > 1.09 in environmental samples (reviews by Conrad ; Conrad et al. ; Whiticar ).
The fractionation between acetate methyl and CH4 (αac) has been measured in cultures and environmental samples and ranged from 1.010–1.032 [Blair and Carter, 1992; Conrad, 2005; Conrad and Klose, 2011; Goevert and Conrad, 2009; Penning et al., 2006]. Although we did not measure the δ13C of acetate, αapp is meaningful due to intramolecular differences in δ13C between the methyl group and the carboxyl group which have been measured to be 20‰ [Blair et al., 1987]. Cleaving such a molecule to produce CH4 and CO2 would yield αapp near 1.020 because it is the methyl group that is incorporated into CH4.
We hypothesize, based on the microbial data of Chauhan et al. , that hydrogenotrophic CH4 production is the dominant production pathway in Everglades soils. Based on these microbial results, we further hypothesize that the relative importance of each methanogenic pathway changes depending on the extent to which the soil is impacted by eutrophication. One might expect acetoclastic methanogenesis to increase relative to CO2 reduction when the soil is enriched with nutrients because of greater production of more labile organic matter which supports increased acetate fermentation [Hornibrook et al., 1997; Hornibrook et al., 2000b; Sugimoto and Wada, 1993]. Based on microbial data [Chauhan et al., 2004], this does not appear to be the case in the northern Everglades. Following these microbial results, we propose that the proportion of CH4 produced via hydrogenotrophic methanogenesis increases with nutrient concentration.
We will elucidate the CH4 production pathways in the Everglades in two related ways. First, we will present data from incubations of soil collected at three sites along the nutrient gradient in the northern Everglades. These soils were incubated with and without an inhibitor of acetoclastic methanogenesis in order to determine in vitro rates of CH4 production and rates of hydrogenotrophic CH4 production alone. From these incubations, we were able to establish α for hydrogenotrophic methanogenesis. From this α and the amount of CH4 produced with and without inhibitor, we estimated α for acetoclastic methanogenesis in these soils. The fractionation factors were then used in combination with the apparent fractionation factors measured from field studies of δ13C-CH4 and δ13C-CO2 in pore water collected from the three sites to determine the proportion of CH4 produced via each pathway in situ.
Everglades Water Conservation Area 2A (WCA-2A) is a subtropical sawgrass wetland covering 544 km2 in southern Florida (Figure 1). A 1–2 m thick layer of peat overlies the limestone bedrock [Gleason et al., 1974] and seasonal flooding occurs within the Kissimmee River/Lake Okeechobee/Everglades watershed. The Kissimmee River discharges into Lake Okeechobee, which overflows during periods of high rainfall, and this water enters the Everglades. Some 1.5 metric tons of phosphorus is delivered to Lake Okeechobee per day from dairy farms and other agricultural operations located upstream. In addition, much of the water entering WCA-2A comes from the Everglades Agricultural Area to the north of WCA-2A. Thus, the water entering the historically oligotrophic Everglades is loaded with 10–20 times the pre-twentieth century amount of phosphorus and nitrogen, leading to high soil phosphorus levels and increased carbon deposition [Corstanje et al., 2007; Wu et al., 1997]. The nutrient gradient is accompanied by a change in vegetation from sawgrass (Caladium jamaicense) and open water with cyanobacterial diatom mats in the southern, oligotrophic part of the transect to dominance of cattail (Typha latifolia and Typha domingensis) in the north [Craft and Richardson, 1997; Lagerwall et al., 2012; Noe et al., 2001]. The transition site is intermediate between the pristine site U3 and the nutrient-impacted F1 site, with patches of mixed cattail and sawgrass. Stormwater Treatment Area 2, located northwest of WCA-2A, comprises 63 km2 of wetlands that were constructed to remove nutrients from agricultural runoff before it is discharged into the Everglades (http://www.sfwmd.gov).
Pore water samples were collected at three sites along a transect from the northern, nutrient-rich part of WCA-2A (site F1), across a transitional area (site F4) to the relatively unimpacted interior (site U3) in October 2009 and April 2010. Nutrients, productivity, sulfate, and vegetation at these sites are well studied [Debusk et al., 1994; Koch and Reddy, 1992; Reddy et al., 1993; Wright and Reddy, 2007].
Perforated tubes attached to syringes were used to collect pore water from near the soil surface down to 40 cm soil depth at 3–10 cm intervals. Five pore water profiles were taken from each site in October 2009 and three in April 2010. Profile locations were chosen at random and were spaced approximately 5 m apart. Pore water was stored in the syringes on ice for up to 8 h then filtered with 0.2 µm filters into evacuated 10, 20, or 30 ml glass vials that had been sealed with butyl rubber stoppers and aluminum crimp seals. Most October samples were immediately acidified with 0.3 ml of 21% phosphoric acid. At the pH levels in Everglades pore water (~8), most of the inorganic carbon is in the form of carbonate (CO32−) and bicarbonate (HCO3-). Acid was added to convert these forms of carbon to CO2 for measurement of δ13C-ΣCO2 and concentrations of ΣCO2 (all forms of aqueous CO2: H2CO3, HCO3-, CO32−, and CO2). April pore water samples were not acidified immediately so that dissolved free CO2 (aq) could be measured. After the δ13C-CO2 and concentrations of dissolved CO2 (aq) were measured later in the lab, April samples were acidified to convert HCO3- to CO2 so that δ13C-ΣCO2 and concentrations of ΣCO2 could also be measured. After transfer to glass vials, samples were stored upside down in ice for up to 24 h during transport to the lab. Helium was added to the vials to atmospheric pressure and the vials were then frozen, upside down, until analysis. The δ13C values and concentrations of the CO2 (aq), total CO2 (ΣCO2), and CH4 were measured using a Finnigan Mat Delta V isotope ratio mass spectrometer coupled to a gas chromatograph [Merritt et al., 1995]. Each sample was measured twice, and analytical precision based on repeated measurements of a standard was <0.15‰.
Soil cores were collected from each site in April 2010 5–10 m distant from each other. Three cores from each site were transported to the lab where they were combined following homogenization by hand. Approximately 20 ml of soil slurry was placed in glass vials which were then sealed with butyl rubber stoppers and aluminum crimp seals. The headspace was flushed with N2 gas for 2–3 min and vials were shaken in the dark for 2 days. Vials were flushed with N2 gas again, and CH3F (methyl fluoride 99.99%, Specialty Gases of America, Toledo, Ohio) was added to some vials to inhibit acetoclastic methanogenesis. CH3F was added in initial amounts between 0.02% and 1.4% of the headspace volume to test the effect of varying concentrations of the inhibitor and to determine the optimal amount to inhibit acetoclastic methanogenesis, but not affect hydrogenotrophic methanogenesis. To account for the solubility of CH3F in water in calculating the initial CH3F concentration in the headspace, a Bunsen coefficient of 0.99 was used [Frenzel and Bosse, 1996]. Three replicates for each CH3F concentration and four replicates of the control were prepared. Samples were kept in the dark at room temperature (22–24°C). Water temperature data is limited to 1 day per month at WCA-2A (http://sfwmd.gov/dbhydroplsql/). Water temperature on 10 November 2009 (the closest date to our sampling time) was 23°C at all three sites. Average water temperature was 22.6–23.6°C on 14 April 2010. These temperatures were similar to field temperatures in nearby Water Conservation Area 3A (WCA-3A), for which multiple daily temperatures are available. In October 2009 the water temperature in WCA-3A ranged from 23.1 to 26.3°C and in April 2010 from 20.0 to 23.5°C.
The headspace was analyzed for CH4 concentrations several times over the course of the incubations using a Shimadzu Mini 2 gas chromatograph equipped with a 2 m Hayesep Q column (temperature: 30°C; flow rate: 30 ml min−1) beginning within 24 h of the start of the incubation to determine the rate of CH4 production. The δ13C values and concentrations of the CO2 (aq) and CH4 in the headspace of the incubation vials were measured using the same Finnigan Mat Delta V isotope ratio mass spectrometer that was used to measure pore water samples.
3.1 Pore Water
Pore water CH4 concentrations increased with increasing nutrient levels. CH4 concentrations in pore water were lowest at site U3, the most oligotrophic site, where they ranged in October from near zero to 0.3 mM over the upper 40 cm (mean 0.2 mM) and in April the range was near zero to 0.2 mM (mean 0.1 mM ) (Figure 2). In October, the most nutrient-impacted site (F1) exhibited relatively high average CH4 levels, 0.2 to 0.9 mM (mean 0.5 mM), while at the transition site, F4, CH4 was 0.1 to 0.5 mM (mean 0.3 mM) in the upper 40 cm of the soil. In April, CH4 concentrations at the transition and eutrophic sites were similar (mean 0.3 mM). At the transition site, the concentration range was 0.2–0.4 mM, and at the eutrophic site, concentrations ranged from 0.1 to 0.3 mM.
The carbon isotope composition (δ13C) of pore water CH4 (Figure 2) was the most 13C enriched at the oligotrophic site (range: −62 to −53‰; mean: −55‰ in October and range: −52 to −44‰; mean: −47‰ in April), while F1 (eutrophic) contained the most 13C-depleted CH4 and the δ13C values there varied the least between fall and spring (range: −65 to −62‰; mean: −64‰ in October and range: −68 to −62‰; mean: −65‰ in April). The δ13C values of methane at the transition site were similar to the oligotrophic area in October (range: −60 to −51‰; mean: −57‰) and were more negative in April (range: −62 to −59‰; mean: -61‰).
In acidic peatland systems most of the dissolved inorganic carbon is in the form of CO2 and so there is little difference between the ΣCO2 and CO2 (aq) following acidification. But because of the high pH of Everglades water (between 7 and >8), a large proportion of the inorganic carbon in pore water is present as HCO3-. When these samples are acidified, ΣCO2 is significantly larger than CO2 (aq) and the calculated fractionation factors, α, are affected. Because CO2 (aq) rather than HCO3- is used during methanogenesis [Bott and Thauer, 1989; Vorholt and Thauer, 1997], we used δ13C-CO2 (aq) to calculate α values in the soil incubations (they were not acidified). For the most part, pore water sampled during October was acidified, and therefore, only total dissolved inorganic carbon (ΣCO2) isotopes and concentrations are available as direct measurements (except for one profile that was measured before and after acidification). Pore water collected in April was measured both before and after acidification, yielding CO2 (aq) as well as ΣCO2 concentrations and isotope values (Figure 2). δ13C-CO2 (aq) and concentrations for each site from April were averaged and compared with the ΣCO2 so we could estimate δ13C-CO2 (aq) and concentrations in the pore water profiles which were collected in October and only measured following acidification.
The δ13C-CO2 (aq) was more negative than δ13C-ΣCO2, reflecting the isotopic fractionation between CO2 (aq) and HCO3- [Deuser and Degens, 1967; Mook et al., 1974]. The δ13C-CO2 (aq) was offset from δ13C-ΣCO2 by −6.5‰ (±1.0) at U3, −7.5‰ (±1.0) at F4, and −6.5‰ (±0.9) at F1 in April, and this shift was used to estimate δ13C-CO2 (aq) for October pore water. The δ13C-CO2 (aq) in pore water was between −18 and −12‰ at all three sites during both seasons and did not vary consistently with soil depth (Figure 2). The CO2 (aq) concentrations in April in oligotrophic pore water were 26% (±8%) of ΣCO2. At the transition and eutrophic sites CO2 (aq) concentrations were 29% (±6%) and 39% (±12%), respectively, of ΣCO2 (Figure 2). ΣCO2 and CO2 (aq) concentrations were higher in the transition and eutrophic pore water than in the oligotrophic pore water and increased at all sites with depth in the soil. We have confidence in our calculated values for October CO2 (aq) because pH, which controls the CO2 speciation, was similar in October 2009 to April 2010, although South Florida Water Management District data shows that monthly average pH was slightly higher in surface water at the nutrient-poor site (7.5 to 7.7) than the transition and nutrient-impacted sites (7.0 to 7.5) (http://sfwmd.gov/dbhydroplsql/).
3.2 Soil Incubations
The two main reactions forming CH4 in flooded freshwater soils are acetoclastic and hydrogenotrophic methanogenesis (equations (1) and (2a)).
Note that reaction (2a) does not result in net CO2 consumption. The hydrogen is produced from organic matter as shown below:
so that both pathways (1) and (2a) + (2b) have the same net result (equation (2c)) assuming that most of the organic matter is derived from cellulose, a valid assumption for most aquatic sediments, peats, and other wetlands [Conrad, 1999].
Soils were incubated with CH3F, a selective inhibitor of acetoclastic methanogenesis [Frenzel and Bosse, 1996; Janssen and Frenzel, 1997] and compared with controls to which no inhibitor was added (Figure 3). The slope of the regression line represents the CH4 production rate. More CH4 was produced in vials with no CH3F than in those in which CH3F was added to the headspace. At high concentrations of CH3F, hydrogenotrophic methanogenesis may be inhibited as well as acetoclastic methanogenesis and the ideal CH3F concentration to inhibit mostly acetoclastic methanogenesis appears to vary among different systems [Chan and Parkin, 2000; Conrad, 1999]. To determine the optimal amount of inhibitor to use in Everglades soils, different amounts of CH3F (from 0.02 to 1.4%) were added to the headspace of the incubation vials containing the top 40 cm of soil collected in April 2010 from all three sites. Increasing CH3F concentration led to decreased CH4 production in soils from all sites (Figure 4a). Between the additions of 0.7 and 1.4% CH3F, there was little to no change in CH4 production in oligotrophic and transition soils, but there was a further decrease from 69 to 59 µmol g−1 in eutrophic soils. Coincident with the leveling off of CH4 production rates, the apparent fractionation factor between CO2 and CH4 (αapp) increased rapidly with CH3F up to around 0.3% and then began to stabilize at 0.7 to 1.4% (Figure 4b). Addition of CH3F above 0.7% to the headspace caused no change in αapp in oligotrophic soils (1.073) and only a small increase in transition and nutrient-impacted soil incubations (αapp increased from 1.079 to 1.082 in transition soil incubations going from 0.7% to 1.4% CH3F and from 1.078 to 1.079 in eutrophic soil incubations). CH3F in our incubations degraded over time and after 61 days; concentrations had decreased by approximately one half, and the δ13C of the CH3F became more positive (data not shown). The inhibitor concentrations were apparently low enough after 61 days to allow acetate fermentation to proceed, as δ13C of the accumulated CH4 became more enriched in 13C. In Everglades soils, it appears that at least 0.3% to 0.7% CH3F is necessary to inhibit acetoclastic CH4 production and since CH3F is degraded over time, the addition of around 1 to 2% CH3F is likely optimal.
CH4 production increased along the nutrient gradient from the oligotrophic U3 site (0.8 µmol g−1d−1) to F4, the transition site (3.1 µmol g−1d−1) to the eutrophic F1 site (3.7 µmol g−1d−1) (Table 1). Soil from all three sites produced less CH4 in the presence of CH3F since the acetoclastic CH4 production pathway was suppressed. By subtracting the CH4 production rate in CH3F-amended incubations (= hydrogenotrophic methanogenesis) from the rates measured in the controls, we found that one quarter of the CH4 produced came from hydrogenotrophic methanogenesis in oligotrophic and transition soils, while in the nutrient-impacted soil, around one half was from this pathway (Table 1).
Table 1. CH4 Production in Soil Collected in April 2010
Production rates were determined from the slopes of the lines in Figure 3.When no inhibitor was added, the CH4 produced was from a combination of hydrogenotrophic and acetoclastic methanogenesis. With 1.4% CH3F in the headspace, we assume that all of the CH4 was produced via hydrogenotrophic methanogenesis. The difference between the total methane produced and that from hydrogenotrophic methanogenesis is the amount produced via acetoclastic methanogenesis. Percentages that each production pathway contributed to total CH4 production are given in parentheses.
In the absence of CH3F, δ13C-CH4 ranged from −65 to −57‰ at the different sites (Table 2), similar to the −65‰ to 60‰ reported by Chanton et al.  for Everglades CH4. When the production of CH4 via acetate fermentation was inhibited (i.e., in the presence of CH3F), δ13C of the CH4 produced ranged from −89 to −81‰. Fractionation factors without CH3F were 1.044–1.055 (representing the combination of both CH4 production pathways), and when inhibitor was added, α ranged from 1.073 and 1.081. The latter values should represent αCO2 because CH4 produced by acetoclastic methanogenesis was inhibited. To determine the fractionation factor for acetoclastic methanogenesis, mass balance (equations (3) and (4)) was used, in which CH4total represents the total production rate of CH4; CH4CO2 is the rate of CH4 production via hydrogenotrophic methanogenesis, and CH4acet is the rate at which CH4 was produced from acetoclastic methanogenesis. CH4total and CH4CO2 were measured in soil incubations with and without CH3F. CH4acet was calculated from the difference between CH4total and CH4CO2. Similarly, αapp was determined from incubations with no CH3F-inhibiting acetoclastic methanogenesis, and αCO2 was measured in incubations with CH3F.
Table 2. Stable Carbon Isotopes in CH4 and CO2 (aq) and Fractionation Factors (α) From Incubations With and Without CH3F
Acetoclastic + Hydrogenotrophic (No Inhibitor)
Hydrogenotrophic (+ Inhibitor)
The measured apparent fractionation factor between CO2 (aq) and CH4 in unamended soil incubations is αapp. The measured fractionation factor in soils with CH3F is αCO2. The αacet was calculated from the amount of CH4 produced with and without CH3F, αapp, and αCO2.
Values for oligotrophic site with inhibitor are from day 28 of the incubation. All other values are from 7 to 8 days of incubation.
The apparent fractionation factors for acetate fermentation were estimated from equation (4) to be 1.034, 1.035, and 1.030 for the top 40 cm of soil from oligotrophic, transition, and eutrophic sites, respectively (Table 2). It is assumed that CH4 that is not produced via CO2 reduction came from acetoclastic methanogenesis and that CH3F completely inhibited acetoclastic CH4 production without affecting hydrogenotrophic methanogenesis. The mean fractionation factor for hydrogenotrophic CH4 production in our soil incubations from the three sites was 1.076 (± 0.004) and for acetoclastic methanogenesis was 1.033 (± 0.002).
4.1 Laboratory Studies to Partition Pathways and to Determine Fractionation Factors
The effectiveness of CH3F in inhibiting acetoclasts without affecting hydrogenotrophic methanogenesis appears to vary with different soil or sediment types. Chan and Parkin  reported that 1% CH3F inhibited 39% of the total methanogenesis and 10% resulted in 93% inhibition of CH4 production. They did not differentiate between acetoclastic and hydrogenotrophic methanogenesis. Several other studies have shown variable methanogenic inhibition at different CH3F concentrations in landfill and compost soils, salt marsh sediments, plant rhizospheres, and peat soil [Epp and Chanton, 1993; King, 1996; Oremland and Culbertson, 1992]. CH3F at 5% was required to inhibit acetoclastic methanogenesis in granular soils where the CH3F might have been adsorbed to the grain surfaces, limiting its effectiveness [Hao et al., 2011]. Conrad et al.  found that only 2% CH3F was sufficient to completely inhibit acetoclastic methanogenesis in sediments from a lake in Germany and 0.5 to 2% CH3F was used in experiments on sediments from Amazonia [Conrad et al., 2010]. Based on incubations of Everglades soil with 0.02 to 1.4% v/v CH3F, our data indicate that ~1.4% CH3F inhibited acetoclastic methanogenesis without affecting hydrogenotrophic methanogenesis and that oligotrophic (U3) soil appears to be more sensitive to CH3F (Figure 4). We are confident that hydrogenotrophic methanogenesis was not inhibited in our soils since the rate of CH4 production did not decrease with added CH3F up to 1.4%, which would be expected if hydrogenotrophic methanogenesis was being increasingly inhibited. Also, the contribution of hydrogenotrophic methanogenesis to total CH4 produced is greater in the incubated soils than estimates based on stable carbon isotope values of pore water CH4 and CO2 (discussed later). If the CH3F was inhibiting hydrogenotrophic CH4 production, soil incubations would yield lower estimates of the relative importance of this pathway than field data with no inhibitor.
Fractionation factors from incubations amended with 1.4% CH3F are interpreted to represent αCO2 (the fractionation factor for hydrogenotrophic CH4 production). Values of αCO2 obtained from soil incubations ranged from 1.073 to 1.081 (Table 2) and are within the range reported for αCO2 in a variety of environments including high-latitude bogs and fens, rice field soils, and pure cultures (1.04 to 1.09). Specifically, αCO2 determined in this study are similar to the average αCO2 values of 1.075 and 1.085 found for the more comparable tropical lake sediments [Conrad et al., 2010, 2011] and 1.082 from a temperate fresh water estuary [Avery and Martens, 1999]. The fractionation factors for acetate fermentation determined from soil incubations were between 1.030 and 1.035 and are also within the range of previously published values (1.01 to 1.055) [Blair and Carter, 1992; Penning et al., 2006; Whiticar et al., 1986].
4.2 Field Studies Corroborate Lab Results
The soil incubations from each of the three sites along the nutrient gradient supported the hypothesis that hydrogenotrophic CH4 production becomes a more important methanogenic pathway at nutrient-impacted sites. To determine whether this trend exists in situ based on pore water determined values, we used the αCO2 and αacet derived from the soil incubations and mass balance (equation (5)) to calculate the relative amount of CH4 produced from hydrogenotrophic methanogenesis (f) in the field from pore water samples.
The αapp is known from the isotopic composition of pore water CH4 and CO2, and the fractionation factors for hydrogenotrophic methanogenesis (αCO2 = 1.076) and acetate fermentation (αapp = 1.033) were derived from mean values from the soil incubations. We assume that these fractionation factors were similar to in situ values. Using equation (5), if αapp is 1.076, the average fractionation factor for hydrogenotrophic methanogenesis, it would signal that 100% (f = 1) of the CH4 came from CO2 reduction. Likewise, if αapp = 1.033, the amount of CH4 produced via acetate fermentation would be estimated at 100%.
The relative importance of each pathway for CH4 production that we calculated using pore water apparent fractionation factors is shown in Figure 5. Our hypothesis that higher nutrient concentrations lead to increased relative importance of hydrogenotrophic methanogenesis is supported by pore water δ13C-CH4 and δ13C-CO2 values, which is in agreement with the incubation data. Although our field results depend on fractionation factors determined in the incubation experiment, the fractionation factors we calculated in this work were similar to values reported in the literature [Hornibrook et al., 2000a; Whiticar, 1999; Whiticar et al., 1986]. At the nutrient-impacted site (F1), 48% of the CH4 in October and 43% in April was produced from hydrogenotrophic methanogenesis (Table 3). These percentages are averages calculated from the eight to ten sampled depths in the top 40 cm of each core. The percent CH4 contributed by hydrogenotrophic methanogenesis at each depth is based on one to five replicates for each depth. In contrast, fractionation factors in pore water at the low-nutrient site (U3) suggest that only 23% of the CH4 was produced via hydrogenotrophic methanogenesis in October and that nearly all CH4 was produced via acetoclastic methanogenesis in April. At the transition site (F4), 24% of CH4 was produced from the hydrogenotrophic pathway in October, similar to the low-nutrient site, while in April, the production pathway was more similar to the nutrient-impacted site (39% from hydrogenotrophic methanogenesis). The trend of increasing relative importance of hydrogenotrophic CH4 production along the nutrient gradient from oligotrophic U3 to eutrophic F1 is supported by both laboratory and field data.
Table 3. Comparison of Stable Isotope Ratios and Fractionation Factors Between Incubations and Field Data
Hydrogenotrophic Methanogenesis (%)
After 7–8 days of incubation.
Average values of samples taken in upper 40 cm of soil.
Our estimates of the contribution of hydrogenotrophic methanogenesis were consistent with the trends expected based on the microbial observations of Chauhan et al. , who showed that there were tenfold more hydrogenotrophs in soil from the nutrient-impacted site than in oligotrophic soil. However, Chauhan et al.  found that hydrogenotrophic bacteria outnumbered acetotrophs by 2 to 3 orders of magnitude at all three sites. Our results indicate that hydrogenotrophic methanogenesis is of less relative importance than acetate fermentation, contrary to these microbial results. Chauhan et al.  obtained their numbers from most probable number analyses, which may mean that although a much higher number of hydrogenotrophs grew on the culture material than acetotrophs, there may not have actually been more hydrogenotroph activity in WCA-2A soil. Thus, while the trends in the different data sets agree (microbial rates and isotopes), they disagree in the relative importance of these processes in WCA-2A.
More recent microbial observations based on quantitative polymerase chain reaction (qPCR) yield better agreement with our isotopic studies. The qPCR results indicate that hydrogenotrophic methanogen numbers were approximately twice as high as acetoclastic methanogen numbers in soil from the eutrophic and transition sites (ratios 2.2 and 1.9, respectively), and in the oligotrophic soil the proportion of hydrogenotrophs to methanotrophs was more similar (ratio 1.4) [Bae et al., 2011; Ogram et al., 2013]. Our CH4 δ13C analyses also indicate that hydrogenotrophic methanogenesis contributed relatively more to CH4 production at the eutrophic site than at the transition or oligotrophic sites, but we found that overall, more CH4 was produced via acetoclastic than hydrogenotrophic methanogenesis at the transition and oligotrophic sites.
It is possible that sulfate reduction could have led to higher relative amounts of CO2 reduction in nutrient-impacted soil. While sulfate levels are low throughout our study site (Koch-Rose et al.  reported that sulfate concentrations in WCA-2A soils were < 0.6 mM), they were slightly higher in April 2010 pore water from the oligotrophic site (0.19 mM) than at the transition and nutrient-impacted sites (0.09 and 0.08 mM, respectively) averaged over the upper 35 cm (H.-S. Bae, unpublished data, 2013). Results of Keller and Bridgham  suggest that sulfate is rapidly cycled in peatlands and even at low concentrations may contribute to carbon mineralization. We suggest that at the lower sulfate concentrations found in the eutrophic soil of F1, methanogens were able to better compete with sulfate reducers for H2, while the slightly higher sulfate concentrations at the oligotrophic site enabled sulfate reducers to consume more of the H2.
4.4 Other Concerns
Several possible issues with our interpretation of the results will be examined below. First, excess CH3F can suppress hydrogenotrophic methanogenesis, leading to underestimates of this pathway of CH4 production by decreasing the amount of CH4 produced in incubations. However, our experiments containing different amounts of CH3F suggest that this is not the case. The leveling out of CH4 production rate along with the stabilizing of δ13C-CH4 and αCO2 values (Figure 4) indicate that at and below concentrations of 1.4% CH3F, CH4 production via hydrogenotrophic methanogenesis was not affected. Also, partial inhibition of hydrogenotrophic CH4 production in the soil incubations with CH3F would have resulted in lower estimates for this pathway from our incubation results relative to what we determined from pore water data (Table 3), but this was not the case.
One potential reason for smaller values for αapp and lower estimates for the proportion of hydrogenotrophic methanogenesis in field samples from U3 in April is CH4 oxidation, which can complicate the interpretation of our field CH4 isotope data. This process removes the 13C-depleted CH4 preferentially, enriching the residual CH4 in 13C and the CO2 pool in 12C, resulting in lower field αapp values. Thus, δ13C-CH4 would be expected to increase (become less negative) as δ13C-CO2 decreased (became more negative) where CH4 oxidation is important. As seen in Table 3, there is no such correlation in our data. For example, in U3 in April, the field profile where the estimated contribution of hydrogenotrophic methanogenesis was 5%, the δ13C-CH4 was less negative (-47‰) compared to the soil and October field data (Table 3), but the corresponding δ13C-CO2 was not more negative (−14‰).
Also, aerobic CH4 oxidation would not have been a factor in our incubation-determined estimates of the relative importance of the two pathways of CH4 production. The measurements were conducted under strictly anaerobic conditions, and they agree with the field-determined values. Additionally, aerobic CH4 oxidation would be expected to be manifested most strongly at the surface of the sediment where oxygen penetrates and should be apparent in the depth profiles of δ13C-CH4, with 13C-enriched CH4 in surface soil and consequently lower αapp. However, in pore water from the nutrient-poor site in October and April, no increase was evident in near-surface δ13C-CH4 values (Figure 2). The αapp value was lower at the soil surface in about half of the profiles, but our 0% estimates of CO2 reduction were determined from pore water below 20 cm, not near the soil surface. However, oxygen may enter the soil at various depths from the rhizosphere of plants. Since the nutrient-impacted site was more heavily vegetated, aerobic CH4 oxidation should have been stronger, but δ13C-CH4 values at the site were the least 13C-enriched of all three sites.
Finally, as with δ13C-CH4, the hydrogen isotope composition, δ2H, of the residual CH4 increases (becomes less negative) when CH4 is oxidized [Chanton et al., 2005]. The absence of a positive correlation between δ2H and δ13C in either pore water or soil CH4 at WCA-2A suggests that oxidation of CH4 does not noticeably alter the isotopic composition of the CH4 (M. E. Holmes, unpublished data, 2010).
Although there is no evidence for aerobic CH4 oxidation, CH4 can be oxidized anaerobically by electron acceptors such as sulfate [Boetius et al., 2000]. In addition to the isotope evidence that oxidation of CH4 is unimportant in WCA-2A soils, sulfate concentrations appear to be too low to support CH4 oxidation. When sulfate concentrations are below approximately 1 mM, anaerobic oxidation of CH4 decreases rapidly [Beal et al., 2011]. Nitrate and nitrite may also be used in the anaerobic oxidation of CH4 [Raghoebarsing et al., 2006; Zhu et al., 2012], but nitrate was 0.01 mM or less at all three sites, and nitrite was undetectable at most depths, with the exception of F4 where it was measured at 2 depths at concentrations of 0.004 mM or less (H.-S. Bae, unpublished data, 2013).
Inhibition of acetoclastic methanogenesis with CH3F proved an effective way of estimating the contribution of hydrogenotrophic methanogenesis to total CH4 production in Everglades peat soils. In soil incubations with and without this inhibitor, we show that while total CH4 production rates were higher in soil impacted by high nutrient influx, the relative contribution of acetate fermentation was higher in soil from an oligotrophic site. From these incubations, we were able to estimate that the average carbon isotope fractionation factor associated with hydrogenotrophic methanogenesis was 1.076 and for acetoclastic methanogenesis was 1.033 in WCA-2A soils. Application of these fractionation factors to δ13C-CH4 and δ13C-CO2 measured in pore water revealed that field data mirrored the trend observed in incubated soil, with decreasing relative importance of hydrogenotrophic methanogenesis with increasing distance from nutrient input. The greater abundance of hydrogenotrophic methanogens reported previously is not reflected in the actual rates of hydrogenotrophic CH4 production we measured, although the spatial trends in both data sets agreed.
Eutrophication-induced changes in the methane production pathways of wetlands may have implications for the isotope signature of CH4 flux to the atmosphere. It has been suggested that the δ13C values used in CH4 mass balance budgets for CH4 flux from northern peatlands to the atmosphere should be adjusted to reflect the more negative values associated with nutrient-poor acidic bogs [Hornibrook, 2009; Hornibrook and Bowes, 2007] relative to the more 13C-enriched CH4 emitted from fens. Our findings show that human-induced nutrient enrichment can lead to increased hydrogenotrophic methanogenesis in the Everglades and thus a decrease in the δ13C value of emitted CH4.
The authors thank Claire Langford, Todd Osborne, Tyler Mauney, and Linda Renner for their help with the laboratory and field work and Rachel Wilson for insightful comments. This research was funded by National Science Foundation grant DEB 0841158.
For the originally published version of this article, the author has requested that the colors in Figure 1 be revised to improve the accuracy of the information. The colors in the figure have now been revised and this version of the article may be considered the authoritative version of record.