Temperature sensitivity of anaerobic methane oxidation versus methanogenesis in paddy soil: Implications for the CH4 balance under global warming

The global methane (CH4) budget is based on a sensitive balance between methanogenesis and CH4 oxidation (aerobic and anaerobic). The response of these processes to climate warming, however, is not quantified. This largely reflects our lack of knowledge about the temperature sensitivity (Q10) of the anaerobic oxidation of CH4 (AOM)—a ubiquitous process in soils. Based on a 13CH4 labeling experiment, we determined the rate, Q10 and activation energy of AOM and of methanogenesis in a paddy soil at three temperatures (5, 20, 35°C). The rates of AOM and of methanogenesis increased exponentially with temperature, whereby the AOM rate was significantly lower than methanogenesis. Both the activation energy and Q10 of AOM dropped significantly from 5–20 to 20–35°C, indicating that AOM is a highly temperature‐dependent microbial process. Nonetheless, the Q10 of AOM and of methanogenesis were similar at 5–35°C, implying a comparable temperature dependence of AOM and methanogenesis in paddy soil. The continuous increase of AOM Q10 over the 28‐day experiment reflects the successive utilization of electron acceptors according to their thermodynamic efficiency. The basic constant for Q10 of AOM was calculated to be 0.1 units for each 3.2 kJ mol−1 increase of activation energy. We estimate the AOM in paddy soils to consume 2.2~5.5 Tg CH4 per year on a global scale. Considering these results in conjunction with literature data, the terrestrial AOM in total consumes ~30% of overall CH4 production. Our data corroborate a similar Q10 of AOM and methanogenesis. As the rate of AOM in paddy soils is lower than methanogenesis, however, it will not fully compensate for an increased methane production under climate warming.

AOM was estimated to consume ~300 Tg CH 4 year −1 in global marine environments (Hinrichs & Boetius, 2002), and ~200 Tg CH 4 year −1 in global freshwater wetlands (Segarra et al., 2015). This makes AOM a significant component of the global carbon cycle and represents a potential constraint on climate warming. Despite its clear significance, AOM in terrestrial ecosystems is not yet included in most modern process-based biogeochemical models due to the lack of information about the underlying mechanisms and controls.
The potential of AOM in marine and terrestrial habitats has been studied under controlled conditions in relation to CH 4 concentrations, electron acceptor availability, and environmental parameters such as temperature, salinity, alkalinity, and pressure (Bhattarai et al., 2019;Fan et al., 2019). Among these factors, temperature appears to be a decisive selection parameter for the distribution of ANMEs (Bhattarai et al., 2019) and NC10 (He et al., 2015). Moreover, AOM metabolic processes are, as all chemical and biochemical reactions, temperature-dependent (He et al., 2015), but the temperature sensitivity of AOM in soils has not yet been described.
The temperature sensitivity of metabolic reactions is commonly expressed as the Q 10 value, that is, the factor by which the reaction rate increases with a 10°C rise in temperature. The Q 10 value can be used to predict the potential feedback of metabolic reactions to climate change  and thus plays a pivotal role in global biogeochemical models. In the global carbon cycle, the CH 4 flux to the atmosphere is a result of three processes: methanogenesis, aerobic CH 4 oxidation, and AOM ( Figure 1). The rates of these processes, therefore, determine the amount of CH 4 emitted to the atmosphere, whereas the Q 10 describes the proportional increase of these rates in response to global warming. In nature, Q 10 commonly decreases with increasing temperature Karhu et al., 2014;Mahecha et al., 2010). For example, Q 10 values of methanogenesis decrease with temperature and range between 1.4-15.4 (Huang et al., 2016;Wei et al., 2021). In comparison, the Q 10 of aerobic CH 4 oxidation in paddy soil ranges from 2.2 to 9.8 (Cai &Yan, 1999). The Q 10 is directly related to the activation energy of a given biochemical reaction . The activation energy is described by the Arrhenius equation, which links the reaction rate with temperature. Higher activation energy should correspond to a higher Q 10 (Arrhenius, 1889). Perkins et al. (2012) demonstrated that the temperature dependence (Q 10 ) of respiration across ecosystems was equivalent to the activation energy. Nonetheless, the real-world values will additionally depend on the biochemical and biophysical conditions in the soil, for example, methanogenesis requires a certain low redox potential (<−150 mV;Wang et al., 1993), which is below that for the reduc-  (Keppler et al., 2006). Submerged paddy soils provide an ideal habitat for AOM-related microorganisms (Hu et al., 2014;Vaksmaa et al., 2017) due to the abundance of AEAs from organic and mineral fertilization (NO − 3 /NO − 2 , SO 2− 4 and dissolved organic matter) and from (sub)tropical soil development (Mn 4+ , Fe 3+ ).
Several metabolic pathways of AOM are active in paddy soils (Fan et al., 2021). The available data indicate that AOM could offset the CH 4 efflux from paddy soils by 10%-20% (Fan et al., 2019(Fan et al., , 2020. Here, we experimentally tested the response of methanogenesis and AOM to warming within a range of temperatures common in paddy soils from temperate to tropical climates (5-35°C). Based on the obtained temperature sensitivity of these processes, and using available literature data, we compared the rates and Q 10 values of AOM and methanogenesis, and predicted the significance of AOM as a CH 4 sink in response to climate change.

| Soil collection and samples
The sampling site is located near Jinjing town of Hunan province in China (28°33′04″N, 113°19′52″E). Soils were sampled from an ongoing long-term field experiment with different fertilization treatments for rice cultivation . The typical paddy field has a tillage history of more than 1000 years of rice production. In a previous study, we found that soils from 20-30 cm depth under pig manure fertilization showed the highest AOM potential (Fan et al., 2020). We therefore chose soils from this depth and fertilization treatment for the experiment. The fertilization treatment comprised 60 Mg pig manure ha −1 year −1 (containing 250 g C kg −1 , 16.8 g N kg −1 , 5.3 g P kg −1 , 2.5 g K kg −1 ; pH 8.0), plus conventional background fertilization (60 kg N ha −1 year −1 as urea, 18 kg P ha −1 as Ca(H 2 PO 4 ) 2 , and 83 kg K ha −1 ). Samples from four soil cores were mixed to form one composite sample per plot. The samples were not sieved to minimize exposure to air because both methanogenesis and AOM are highly sensitive to oxygen contamination.

| Experimental design
Twenty grams of field-moist soil (30% soil weight-based water content) and 15 ml deionized water were loaded into 100-ml Kimble KIMAX borosilicate laboratory glass jars to prepare the soil slurry.
Before use, all jars and septa for sealing were autoclaved twice at 121°C for 20 min. Anoxic conditions were created by cycles of headspace evacuation and N 2 refilling (Fan et al., 2019). The slurry was pre-incubated in the dark at either 5 ± 0.5, 20 ± 0.5, or 35 ± 0.5°C for 2 weeks to establish a new equilibrium and consume any O 2 remaining in the microcosms. To raise the representativeness of soil samples, two jars were set up for every plot. To exclude contamination with atmospheric O 2 , all manipulations were conducted in an anaerobic glovebox (N 2 /H 2 , 97/3%).
Subsequently, 5 ml of CH 4 containing 10 atom% 13 CH 4 were injected to quantify the net anaerobic oxidation of 13 CH 4 to 13 CO 2 .

F I G U R E 1
Conceptual diagram illustrating the effects of temperature sensitivity (Q 10 ) and activation energy (E a ) on CH 4 oxidation including (i) methanogenesis, (ii) aerobic, and (iii) anaerobic oxidation of methane (AOM) in soils. Question mark: unknown effects of temperature sensitivity for each process. Color changes from blue through yellow to red reflect the increasing process rates. "Energy efficiency" arrow: the relative affinities of CH 4 -oxidizing archaea and bacteria for alternative electron acceptors (AEA, in blue) under anoxic conditions. The process rates under anoxic conditions (e.g., AOM) increase faster with warming compared to oxic conditions, thus leading to similar process rates under high temperature (two curved arrows) [Colour figure can be viewed at wileyonlinelibrary.com] This yielded an initial average headspace CH 4 concentration of 3.1%. A similar volume of N 2 instead of CH 4 was used to maintain the same gas pressure in the natural abundance control. Gas samples were collected at 3 and 7 days during the pre-incubation period, and at 3, 7, 14, 21, 28 days after 13 CH 4 injection. Gas sampling and all measurements, including stable C isotope analysis of CO 2 using isotope ratio mass spectrometry, were conducted according to (Fan et al., 2020). Isotope data are reported as δ 13 C-values relative to the Vienna Pee Dee Belemnite standard.

| Calculations
The rate of methanogenesis was calculated using the difference of headspace CH 4 concentration between two sampling time points during the pre-incubation period (i.e., before 13 CH 4 injection).
AOM, as expressed by the amount of 13 CO 2 (i.e., the end-product of AOM) generated, was calculated using the isotope mixing-model: where C OX (μg) is the amount of 13 CH 4 oxidized based on the released 13 CO 2 , C Total is the total amount of C in the corresponding pool (i.e., CO 2 ), δ 13 C Total is the delta value of 13 CO 2 in the samples treated with 13 CH 4 , δ 13 C Control is the delta value of 13 CO 2 in the reference (no 13 CH 4 addition), and δ 13 C OX is the delta value of the added tracer 13 CH 4 with 10 atom% enrichment.
The Q 10 of methanogenesis and AOM were calculated by the transformed Arrhenius equation (Arrhenius, 1889): where R is the methanogenesis or AOM rate (ng C g −1 DW (dry weight) h −1 ), T is the incubation temperature (°C), a and b are fitted coefficients.
The activation energy was calculated by the Arrhenius equation (Arrhenius, 1889): Taking the natural logarithms of both sides of Equation (4) and separating the exponential and pre-exponential terms, yielded: where k is the rate (ng C g −1 soil h −1 ), A is the frequency of molecular collisions, R is the universal gas constant (8.314 J K −1 mol −1 ), and T is the temperature in Kelvin (273.15 K). The activation energy (E a , J mol −1 ) was calculated as the slope of ln(k) versus (−1/RT).

| Data collection from the literature
We extracted data on methanogenesis and AOM rates from peerreviewed articles published until April 2021 using Web of Science (http://apps.webof knowl edge.com/) and Google Scholar (http:// schol ar.google.com/) databases. The following key words were used for the search of (a) methanogenesis rate: "methane production" or "CH 4 production" or "methanogenesis"; (b) AOM rate: "anaerobic oxidation of methane" or "anaerobic oxidation of CH 4 " or "anaerobic methane oxidation" or "anaerobic CH 4 oxidation" or "AOM". Both parameters were further refined by the key words "soil" or "wetland" or "peatland" or "river sediment" or "lake sediment" or "freshwater sediment." Experiments to be included in the dataset had to meet the following criteria: (a) laboratory incubation; (b) declared incubation temperature; (c) the unit can be transferred to ng C g −1 h −1 ; (d) data were collected only from natural, untreated samples. The latter are often reported as control/reference for various treatments.
For example, amendments with electron acceptors for AOM, or added substrates for methanogenesis, were not considered. If a paper reported multiple temperature treatments, then each treatment was included separately in the dataset. If a paper reported rates at different depths of soils/sediments, then their average values were included. Likewise, if a paper reported rates at different incubation times, average values were included. The data obtained in our current study were incorporated into the dataset.

| Statistical analyses
To exclude outliers in the prepared literature datasets (Tables S1 and S2), the rates of methanogenesis and AOM in each dataset were normalized by temperature (i.e., rate/temperature); the normalized rates were thus transferred to z-scores. A given rate was considered as an outlier when its z-score was >3 or <−3.
The correlations between rates (methanogenesis and AOM) and temperature were fitted by six observations in the Equation (2). Two-way ANOVA of variance was used to determine differences in CH 4 and CO 2 production, δ 13 C-CO 2 , and cumulative AOM against incubation time and temperatures. The normality and homogeneity of the residuals of the variances were tested before applying ANOVA. t-tests were used to characterize the differences between methanogenesis and AOM. We used an exponential model (1) C OX = (δ 13 C Total − δ 13 C Control ) (δ 13 C OX − δ 13 C Control ) × C Total , (2) R = aexp (b * T) , (3) Q 10 = exp (10 * b) ,

| CH 4 and CO 2 production and δ 13 C values of CO 2
CH 4 and CO 2 production increased with temperature during preincubation (Figure 2a,b). After the 13 CH 4 injection, both incubation duration and temperature affected the CO 2 concentration [CO 2 ].

| Rates of methanogenesis and AOM
The cumulative AOM increased with incubation temperature (p < 0.01; Table S4) and peaked at 370 ng C per gram dry soil after 28 days at 35°C. This was 17 and 2.8 times more than at 5 and 20°C, respectively (Figure 3b). The methanogenesis rate increased exponentially with temperature (Figure 3c), as did the AOM rate ( Figure 3d), the latter being 0.07, 0.34, and 0.85 ng C g −1 soil h −1 at 5, 20, and 35°C, respectively. The methanogenesis rates over the 5-35°C temperature range were significantly higher than the AOM rates in the studied paddy soil.
Note, however, that the Q 10 of AOM and of methanogenesis were similar at both 5-20°C and at 20-35°C, or for the overall 5-35°C range in paddy soil (Figure 4a-c).
The activation energy (E a ) of methanogenesis was similar at the 5-20 and 20-35°C intervals (Figure 5a), whereas the E a of AOM was higher at 5-20 than at 20-35°C (1.8) (Figure 5b). The overall (5-35°C) E a of methanogenesis was higher than that of AOM (Figure 5c).
The E a of AOM increased linearly at a rate of 3.2 kJ mol −1 by every 0.1 unit of Q 10 (Figure 5d).

| Literature data on methanogenesis and AOM rates
To elucidate the temperature dependency of AOM versus methanogenesis, we collected literature data reporting the rates of these processes in natural samples at various temperatures (after outliers scanning: 364 methanogenesis rates from 82 studies; 91 AOM rates from 28 studies including our own data; see chapter 2.4., Tables S1 and S2). The literature dataset corroborated the exponential increase of both rates with temperature (Figure 6a,b). The mean methanogenesis rate across terrestrial ecosystems was about 3.6 times higher than that of AOM (Figure 6c), whereas the mean Q 10 and E a of both processes were similar (Figure 6d,e).

| DISCUSS ION
The increase in δ 13 C values of CO 2 after injecting 13 C-labeled CH 4 , as compared to the unlabeled control, is clear evidence of AOM in the anoxic paddy soil (Figure 3a). The 13 C enrichment of CO 2 was higher at 20 and 35°C than at 5°C, reflecting an increase of the AOM rate with temperature, as expected. Based on our results and literature data, the AOM and methanogenesis rates exponentially increase with soil temperature (Figures 3c,d and 6a,b), and thus follow the general pattern of the temperature effect on microbial respiration (Karhu et al., 2014).

F I G U R E 2
Headspace CH 4 (a) and CO 2 (b) concentrations before and after 13 CH 4 injection in paddy soil (means ± SEs, n = 6) depending on temperature. Red arrow: 13 CH 4 injection [Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 3 Methanogenesis and anaerobic oxidation of methane (AOM) in paddy soil. (a) Dynamics of δ 13 C value of CO 2 over 28 days of incubation with 13 CH 4 versus soil without 13 CH 4 injection (=natural abundance unlabeled control; means ± SEs, n = 6). (b) Cumulative AOM at three temperature levels (5, 20, 35°C) during the 28-day incubation (means ± SEs, n = 6). (c) Average methanogenesis rate for each temperature, and the exponential relationship between the rate and temperature. (d) Average AOM rate for each temperature level, and the exponential relationship between AOM rate and temperature. Blue The Q 10 of AOM decreased from 2.7 (5-20°C) to 1.8 (20-35°C) in paddy soil (Figure 4b, p = 0.024), consistent with theoretical predictions . Therefore, AOM is a strongly temperature-dependent microbial process, and the AOM-driven CH 4 sink is more effective in a low-versus the high-temperature environment expected as a consequence of global warming. The physiological temperature optimum of rice growth, however, is typically above the low-temperature range (Peng et al., 2004), and most CH 4 is produced during the rice growth season (Huang et al., 1997;Tokida et al., 2011).
Notably, while the Q 10 of methanogenesis and AOM were similar, the methanogenesis rate was higher in both low-and high-temperature ranges (Figures 3c,d, 4c, and 6). Thus, the counterplay of CH 4 production versus AOM still yields net CH 4 emission from anoxic soils.
Compared to methanogens, anaerobic methanotrophs respond faster to warming at low than at high temperatures ( Figure 4b).
Consequently, AOM activity is more temperature-dependent than methanogenesis. The activation energy (E a ) of AOM decreased from 5-20 to 20-35°C, but the E a of methanogenesis remained stable ( Figure 5a,b). This additionally supports the conclusion that methanogenesis is less temperature-dependent but relied more on the redox potential and/or substrate availability. The higher E a of methanogenesis than that of AOM in paddy soil (Figure 5c) demonstrated that methanogens needed more energy to overcome the reaction threshold of anaerobic organic matter fermentation (acetoclastic pathway) and/or CO 2 reduction with hydrogen (hydrogenotrophic pathway) to form CH 4 (Parkin, 1993;Whiticar et al., 1986). In contrast, anaerobic methanotrophs more easily reduce NO − 3 /NO − 2 , humic acids, Mn 4+ , Fe 3+ , and SO 2− 4 to oxidize CH 4 anaerobically (Conrad, 2009;Fan et al., 2021;Smemo & Yavitt, 2011). The E a of AOM shows a linear relationship with Q 10 (Figure 5d). This agrees with previous findings that the temperature dependence of respiration across ecosystems is consistent with the activation energy (Perkins et al., 2012). Based on this linear relationship, we calculated an increase of E a by 3.2 kJ mol −1 for each 0.1 unit of Q 10 . This fundamental constant for the E a and Q 10 changes of AOM with temperature is crucial for process-based modeling of CH 4 oxidation in paddy soils under climate warming.
The Q 10 of AOM falls quite well into the Q 10 range of CO 2 efflux from soils (1.3-3.6; Lenton & Huntingford, 2003), which is commonly used in the global-scale models (Friedlingstein et al., 2006;Mahecha et al., 2010). Considering environmental conditions, the Q 10 of AOM could be affected by abiotic (e.g., electron acceptors availability, substrate diffusion coefficient) and biotic factors (e.g., AOM-related microbiota, involved enzymes). The Q 10 values increase with the incubation period (Fuchs et al., 2016;Wei et al., 2021), that was also observed in our 28-day experiment ( Figure 4d). We explain this Q 10 increase by the initial preferential consumption of more energetically efficient AEAs (e.g., NO − 3 /NO − 2 ), leaving AEAs with a lower redox potential (e.g., SO 2− 4 ) for use in the later stages of the incubation. First, decreasing the availability of suitable AEAs with incubation duration should increase Q 10 values of AOM because the microorganisms coping with the AEAs also have different affinities to substrates. For instance, the microbial affinity constant of NO − 3 /NO − 2 -dependent AOM (<0.6 μM) is four orders of magnitude higher than that of SO 2− 4 -dependent AOM (>16 mM; Raghoebarsing et al., 2006). Second, the Q 10 of AOM should increase in the following order depending on the thermodynamic energy yield of electron acceptors: NO − 3 /NO − 2 < humic substances < Mn 4+ < Fe 3+ < SO 2− 4 (Figure 1; see detailed results of the effects of AEAs on AOM potential for the same paddy soil in Fan et al., 2020;Smemo & Yavitt, 2011). Third, the restricted diffusion of CH 4 and electron acceptors in soils is a limiting factor for AOM; accordingly, the Q 10 of AOM should be higher in soils than in aquatic environments. Nevertheless, substrate diffusion/ supply co-varies with temperature, and this confounding effect could induce a relatively higher Q 10 value, that overestimates each individual process (e.g., AOM, methanogenesis; . A consistent parameterization of the temperature sensitivity beyond Q 10 (e.g., E a ) is therefore urgently needed Lane & Martin, 2010;Perkins et al., 2012).
Anaerobic oxidation of methane is ubiquitous in soils but is still an underappreciated CH 4 sink (Gauthier et al., 2015).

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the Supporting Information of this article.