Radiolysis via radioactivity is not responsible for rapid methane oxidation in subterranean air

Atmospheric methane is rapidly lost when it enters humid subterranean critical and vadose zones (e.g., air in soils and caves). Because methane is a source of carbon and energy, it can be consumed by methanotrophic methane-oxidizing bacteria. As an additional subterranean sink, it has been hypothesized that methane is oxidized by natural radioactivity-induced radiolysis that produces energetic ions and radicals, which then trigger abiotic oxidation and consumption of methane within a few hours. Using controlled laboratory experiments, we tested whether radiolysis could rapidly oxidize methane in sealed air with different relative humidities while being exposed to elevated levels of radiation (more than 535 kBq m-3) from radon isotopes 222Rn and 220Rn (i.e., thoron). We found no evidence that radiolysis contributed to methane oxidation. In contrast, we observed the rapid loss of methane when moist soil was added to the same apparatus in the absence of elevated radon abundance. Together, our findings are consistent with the view that methane oxidizing bacteria are responsible for the widespread observations of methane depletion in subterranean environments. Further studies are needed on the ability of microbes to consume trace amounts of methane in poorly ventilated caves, even though the trophic and energetic benefits become marginal at very low partial pressures of methane.


Introduction
Energetic radiation generates ions and radicals in fluids via radiolysis that can trigger subsequent chemical reactions [1], including the oxidation of organics. Radiolysis has likely affected the evolution of early microbial metabolisms and is crucial for powering the deep microbial biosphere [2,3]. However, few studies have addressed the quantitative importance of radiolysis for contemporary fluxes in the atmosphere and the critical zone, especially in comparison to processes that compete with biologically mediated transformations. PLOS  The concentration of methane (CH 4 ) in the atmosphere has more than doubled since 1850 to~1.85 ppmv (i.e., parts per million by volume) and now contributes~15% of anthropogenic forcing of climate change [4]. The Intergovernmental Panel on Climate Change (IPCC) report [5] includes secondary greenhouse warming effects of CH 4 and arrives at 1 W m -2 for CH 4 relative to 1.7 W m -2 for CO 2 , making CH 4 the second most important anthropogenic climate forcing agent. In the atmosphere, the removal of CH 4 is due primarily to oxidation via photochemically generated tropospheric OH• radicals ( [6], and refs. therein). In spite of intense radiation in the atmosphere from sun and space, the residence time of atmospheric CH 4 is~12 years. The second largest sink for atmospheric CH 4 is shallow subterranean environments containing aerated soils that are inhabited by CH 4 oxidizing bacteria, MOB [4,7]. MOB are also found in deeper aerated subterranean environments, such as caves in the vadose zone, although their contribution to global CH 4 cycling has not been quantified or incorporated into earth system models [4].
A growing number of studies have reported that, throughout the world, concentrations of CH 4 are often depleted in the air of caves suggesting that subterranean environments may represent an overlooked sink for atmospheric CH 4 (e.g., [8][9][10][11][12][13]). Based on ventilation rates and CH 4 pools, it is estimated CH 4 is rapidly consumed in caves on time scales ranging from hours to days [14,15]. Depletion of CH 4 in caves is often attributed to MOB. However, a study from Spanish caves proposed that rapid CH 4 oxidation may be attributed to non-biological processes via radiolysis and ionization of subterranean air by natural radioactivity that could lead to the oxidation of CH 4 at a sufficiently fast rate to account for appreciable consumption of CH 4 [10]. It has been proposed that α-radiation (e.g., from 222 Rn) can radiolytically ionize, or generate radicals from, atmospheric components (e.g., H 2 O) including CH 4 [16,17]. The study by Haynes and Kebarle [16] determined that α-radiation has a slow effect on pure CH 4 and mixed hydrocarbon gas in the absence of air, making it difficult to extrapolate results to CH 4 in air in the presence of ions and radicals from heteromolecules.
Some studies, however, have raised questions about the relative importance of abiotic CH 4 oxidation based on theoretical considerations of kinetics, the inability of α-radiation from metallic uranium and radon to trigger fast oxidation of CH 4 [15,18]. Laboratory and field experiments implicated MOB with the rapid decline in cave CH 4 concentrations [18], while isotopically uncharacterized radon was unable to remove CH 4 from air in an Australian cave [15]. Studies on radon typically focus on 222 Rn because its longer half-life of 3.83 days facilitates quantification. No study has yet examined the radiolytic effect on CH 4 oxidation of the relatively more energetic decay of 220 Rn (called thoron, with a half-life 55.6 s), particularly in the air close to cave walls and floors where 220 Rn is relatively more abundant. Also, direct experiments linking the constraints of air humidity and natural radiation from specific radon isotopes to CH 4 oxidation in air are lacking. The current study fills these gaps with detailed independent experiments in two laboratories using energetically distinct radiation levels from isotopes of radon ( 222 Rn and 220 Rn) at different humidities and contrasting the results with CH 4 -depletion by MOB.

Materials and methods
The authors of this study belong to two teams that had no knowledge of each others' experiments at Indiana University (IU) and Royal Holloway University of London (RHUL). After completion of all experiments, the two groups decided to jointly report their complementary results. Work at IU afforded superior analytical control on radon isotopes and could accurately measure higher dose rates, whereas the more gas-tight experimental setup at RHUL provided more straightforward evidence for the inability of natural radiation levels to rapidly oxidize atmospheric CH 4 at its natural atmospheric abundance.
Details of materials and methods are available from protocols.io under http://dx.doi.org/10. 17504/protocols.io.s7aehie. We employed two separate, complementary experimental approaches at IU and RHUL. The following two sections offer brief overviews.

Apparatus at IU for active, time-resolved measurements of gas concentrations with circular flow
At IU, we constructed an experimental apparatus to assess the loss of CH 4 in an active (i.e., with pumping of air) and time-resolved manner with or without added radiation from radon isotopes and their progeny (Fig 1). The use of pumping qualifies this method as active and time-resolved in contrast to passive measurements of radon that integrate over time [19]. Approximately 6 L of air was recirculated in the sealed apparatus that included (i) a glass tube with optional thorium carbonate to generate 220 Rn (also called thoron), and (ii) a glass tube containing uranium ore to generate 222 Rn, with an overlying layer of coconut charcoal to limit the escape of co-produced, short-lived 220 Rn. Blank experiments without elevated radiation identified a reproducible loss of CH 4 (likely by diffusion through polymer tubing within the sealed analytical SARAD RTM 2200 instrument) that was subtracted from all other experiments at IU to arrive at net CH 4 losses that are due to other factors, such as radiolysis or microbial methanotrophy.
We conducted a number of experiments at IU to assess the importance of α-radiation intensity, relative humidity, and the presence or absence of soil on CH 4 dynamics. Moisture is critical for the emanation efficiency of radon isotopes from solid sources (i.e., the escape of noble gas radon atoms from the interior of minerals into H 2 O-containing pore space via recoil subsequent to radioactive decay of parental nuclides; e.g., [20,21]) and for stabilizing ions and radicals in air. Individual experiments differed in terms of their optional use of elevated humidity, thorium carbonate, and gas flowing through the tube containing uranium ore. The trapped~6-L volume of air was initially spiked with CH 4 from natural gas to~70 ppmv and with CO 2 to~5,000 ppmv (except for experiments with soils) to distinguish it from room air and to increase the analytical precision during the time-series of measurements that lasted over a few days to weeks. Elevated CO 2 concentrations are typical for many cave environments [10].
Most experiments at IU discriminated between α-radiation from radon 222 Rn versus thoron 220 Rn. Whereas radon 222 Rn with a half-life of 3.83 days is relatively homogeneously distributed in cave air (also in our apparatus), the much shorter lived thoron 220 Rn with a half-life of only 55.6 s [22] cannot travel far from its parent nuclei residing in minerals [19], thus thoron's highest concentrations in cave air are near cave walls and the floor. The higher α-decay energy of 220 Rn (6.3 MeV) relative to 222 Rn (5.49 MeV) prompted us to design experiments for separate examinations of the ability of both radon isotopes to trigger the oxidation of CH 4 . The more energetic α-decay of thoron 220 Rn should ionize air more efficiently than 222 Rn. Thoron was generated from thorium carbonate that was optionally loaded into a glass tube attached to the round-bottom flask. In other experiments, 222 Rn decay measuring up to 327 kBq m -3 was produced in-situ in the glass apparatus by uranium ore chips (Fig 1). Escape of co-produced thoron from ore was reduced by using a layer of coconut charcoal in the upper part of the glass tube as a filter [23]. The resulting adsorption of 220 Rn on charcoal increased the residence time in the glass tube and let 220 Rn decay before it could enter the 5-L glass flask.
We quantified the concentrations of 222 Rn, 220 Rn, CH 4 and CO 2 during experiments at IU at an air flow rate of~0.2 L min -1 once every hour while operating the diffusion pump in the SARAD RTM 2200. 220 Rn radiation intensity was either measured via α-spectroscopy at a faster flow rate of 1 L min -1 in 10-min increments (n � 10) while temporarily operating the more powerful membrane pump, or values from flow rates �0.2 L min -1 with the diffusion pump were doubled to adjust for fast 220 Rn decay (see S1 File for detailed control experiments and graphed data). Elevated relative humidity fosters the stabilization of ions in air via attachment to clusters of water molecules and may enhance the ability of ions to trigger oxidative degradation of CH 4 (discussed in [10]). Therefore, at IU we recorded humidity in the apparatus along with temperature, air pressure, flow, and battery voltage on an hourly basis. The Approximately 6 L of air was recirculated in a sealed apparatus to assess the loss of CH 4 with or without added radiation from radon isotopes and their radioactive progeny. At the beginning of each experiment, the trapped air was slightly enriched in CH 4 (and CO 2 , except for experiments with soils), followed by hourly measurements of gas concentrations over a few days to weeks. Radon 222 Rn was generated by uranium ore while charcoal retained 220 Rn (a). The air intake of the 3-neck 5-L glass flask was directed to the bottom of the flask with a plastic insert to facilitate the mixing of air (b); thorium carbonate is not shown. Depicted components of the apparatus (c) are not drawn to scale. https://doi.org/10.1371/journal.pone.0206506.g001 Radiolysis via radioactivity is not responsible for rapid methane oxidation in subterranean air PLOS ONE | https://doi.org/10.1371/journal.pone.0206506 November 1, 2018 accuracy of data from the SARAD RTM 2200 was independently evaluated via direct comparison with a newly manufactured and factory-calibrated Thoron Scout instrument (SARAD GmbH, Dresden, Germany; details available in S1 File).
We conducted a number of experiments at IU to test for the effects of radiation and microbial activity on CH 4 dynamics in our experimental apparatus. Multi-day time-series of data were collected in closed-circuit air reflux mode (i) as duplicated blank experiments without added radon or thoron, (ii) with enhanced 220 Rn concentration in dry or moist air, (iii) with enhanced 222 Rn concentration in dry or moist air, and (iv) with jointly enhanced 220 Rn and 222 Rn concentrations in moist air to depict an extreme scenario where cave air had a highly elevated α-radiation level. Furthermore, (v) we tested for CH 4 oxidation after placing moist soils, which we assumed contained methanotrophic bacteria (MOB), into the 5-L glass flask, without elevated radioactivity. Certain impurities in industrially conditioned natural gas may act as MOB inhibitors, for example acetylene and carbon monoxide (p. 335 in [24]). As a precaution, the CH 4 spikes in experiments employing two different soils were derived from gas that was collected from a natural seepage of shale gas in New York State [25]. Natural shale gas is not known to contain acetylene or carbon monoxide.

Gas-tight terrarium experiments at RHUL
Experiments at RHUL at atmospheric CH 4 abundance used a gas-tight glass terrarium (i.e., an aquarium without water holding a volume of 13.45 L; Fig 2) with a hermetically sealing glass lid. Two air-tight gas ports allowed the withdrawal of 1-L air samples into Tedlar bags without changes in atmospheric pressure. Fragments of uraninite-bearing pitchblende served as a source of radioactivity. An AlphaLab Air Ion Counter with an integrated fan was placed in the terrarium to measure the abundance of ions in air in 30-s intervals. The α-radiation was quantified on 1-h intervals with a Canary Pro monitor (Airthings, Oslo, Norway) via α-spectrometry. Gas samples in Tedlar bags were analyzed for CH 4 mole fractions with a Picarro G1301 CRDS (Cavity Ring-Down Spectrometer, Picarro Inc., Santa Clara, California, USA).
The initial RHUL experiment #1 (Fig 2A) assessed the production of negative ions and the abundance of 222 Rn over~6 h (i.e., stage 1) without either pitchblende or a beaker with water in the terrarium that had been flushed initially with laboratory air, and subsequently for~15 h in the presence of pitchblende and a beaker with 130 mL of 38˚C warm water in the terrarium (stage 2).
The subsequent RHUL experiment #2 (Fig 2B) in the same terrarium included monitoring of the CH 4 mole fraction of laboratory air sealed in the terrarium where pitchblende and a beaker with 130 mL of water (initially at 38˚C) had been placed to provide for elevated radioactivity and relative humidity. Elevated relative humidity was needed to simulate cave conditions. The AlphaLab Air Ion Counter failed to provide useful data due to static interference with Tedlar bags. The second experiment lasted for 76 h and 50 min and reached a 222 Rn-based radiation level in excess of 50 kBq m -3 after 5 h. Approximate 1-L aliquots of air sampled from the terrarium were analytically compared with aliquots of exterior laboratory air on four occasions.

Active time-series measurements with circular flow at IU
Our controlled experiments with and without 220 Rn and/or 222 Rn were designed to directly test whether or not radiation can oxidize CH 4 in cave air on ecologically relevant time scales (i.e., hours to days). We relied on comparisons of CH 4 inventories in experiments with (i) high radiation intensity from in-situ generated 220 Rn and/or 222 Rn with those from (ii) duplicate (a) A hermetically sealed glass terrarium was filled with laboratory air containing atmospheric CH 4 . A radon monitor provided data on 222 Rn abundance, while an AlphaLab Air Ion Counter measured the concentration of negative ions. After~6 h into the RHUL experiment #1, the placement of a beaker filled with deionized, warm water elevated the relative humidity to > 85%. At the same time, two fragments of pitchblende (containing uraninite as a radiation source) were placed into the terrarium blank experiments with no artificially enhanced radiation to demonstrate the sensitivity of our setup to detect CH 4 -losses. In addition, we conducted (iii) two experiments with moist soils in the absence of added radon isotopes to assess the potential for environmental microorganisms (i.e., MOB) to remove CH 4 as has been demonstrated elsewhere by members of our research team [14,18]. The comparisons among experiments covered a common range of CH 4 concentration and thus only differed in the lengths of their time windows needed to lower the CH 4 concentration from the upper to the lower threshold (i.e. yellow rectangle in Fig 3A). The 'common window' of CH 4 decline for all 11 experiments maximized the data available for comparison.
Multiple trials in our experimental apparatus revealed that CH 4 dynamics were unaffected by radiation within the precision of measurements. Repeat blank experiments with dry (experiments #1 and #2) or moist air (experiments #3 and #4) without artificially elevated radon or thoron concentrations resulted in reproducible and systematic small losses of both CH 4 and CO 2 over time (Fig 3A and 3B; Table 1A). Although radon isotopes, CH 4 and CO 2 could not diffuse through glass and metal in our apparatus, the SARAD RTM 2200 and its Axetris laser OEM Module LGC F200 methane detector were internally and externally connected to glass and metal components with short segments of various types of clear polymer tubing (Fig 1) that resulted in slow losses via gas diffusion through polymers. The rate of diffusion across a layer of polymer is dependent on the difference in partial pressures between the interior and exterior air, and hence the rates of CH 4 and CO 2 losses via diffusion over time follow curves that asymptotically approach equilibria (Fig 3F). At a CH 4 concentration of~59 ppmv in the apparatus (i.e., the midpoint of the common CH 4 range; Fig 3A) and outside air with 1.85 ppmv, the mean CH 4 diffusive loss from air in the apparatus during blank experiments #1 to #4 consistently amounted to 0.39 ppmv h -1 regardless of humidity and small variations in room temperature and air pressure (Table 1A; S1 File). Such a loss of gas over time could theoretically result from a small internal leak in the system. However, the non-parallel pattern of CO 2 losses in blank experiments (Fig 3B) is inconsistent with a leak and instead argues for varying diffusivity of the polar molecule CO 2 through permeable material at different humidities. The observed degree of CH 4 loss from the system was unavoidable and had to be subtracted from the observed bulk CH 4 losses in experiments with enhanced radiation and soils to arrive at any specific losses that are due to radioactivity or presumed microbial methanotrophy.
There was a comparable loss of CH 4 in recirculating air for all experiments without soil, regardless of the absence or presence of radiation from 220 Rn, 222 Rn, or both 220 Rn and 222 Rn, in dry or moist air (Fig 3A). The time needed to cross the 'common window' of CH 4 decline from 67.2 to 50.9 ppmv was not shorter when radiation from 220 Rn and/or 222 Rn was added (Table 1A, 1B). The slopes of lines representing CH 4 decline within the common window in Fig 3A were not higher for experiments with elevated radiation (mean~0.38 ppmv h -1 ) than for blank experiments without added radon isotopes (mean~0.39 ppmv h -1 ; Table 1A and 1B). The mean levels of added radiation from 220 Rn, and especially the cumulative radiation in experiment #9 from simultaneously added 220 Rn and 222 Rn, ranged between~50 and 535 kBq m -3 after doubling of experimental 220 Rn values that were measured at �0.2 L min -1 (Table 1B) and thus always exceeded the radiation levels reported in cave air [26], including the air in all Spanish caves where abiotically driven CH 4 oxidation due to radiolysis has been to generate 222 Rn. Tedlar bags in the terrarium are not shown in the photograph. (b) Diagram of the sampling procedure to collect <1-L aliquots of air from the terrarium in RHUL experiment #2. This experiment lasted for 76 h and 50 min and reached a 222 Rn-based radiation level in excess of 50 kBq m -3 after 5 h. https://doi.org/10.1371/journal.pone.0206506.g002 reported [10]. For example, the average rate of CH 4 consumption in Spanish Altamira Cave air of -0.03 ppmv h -1 occurred at a maximum 222 Rn radiation level of~6 kBq m -3 , which is roughly one to three orders of magnitude less than the radiation in any of our experiments  4 concentrations in experiments #1 to #9 (without soil) followed similar trajectories depending on original concentrations, despite major differences in radiation intensity (see the yellow rectangle that identifies a window of CH 4 concentrations that is common to all experiments). Declining CH 4 concentrations are independent of the intensity of α-radiation. Blank experiments #1 to #4 without elevated radiation identify a reproducible loss of CH 4 by diffusion that was subtracted from all other experiments to arrive at net losses that are due to other factors, such as radiolysis or microbial methanotrophy. (b) The decline of CO 2 concentrations in a range of experiments without soil followed similar patterns. In addition to loss due to diffusion through plastic, it was likely influenced by adsorption, solution in water, or possible chemical uptake. In addition, CO 2 was generated from moist soils in experiments #10 and #11. (c) Radon 222 Rn and (d) thoron 220 Rn concentrations partially depended on relative humidity; soil no. 2 in experiment #11 generated low levels of 222 Rn over time presumably due to traces of uranium in minerals; 220 Rn concentrations are original data from low flow rates at <0.2 L min -1 when laminar flow conditions in the 5-L glass flask caused heterogeneity and occasional spikes. (e) Noise in relative humidity data partially derived from the automatic battery recharge cycle that influenced the internal temperature of the SARAD RTM 2200 and the algorithm to calculate humidity. (f) Experiments #10 and #11 with moist soils without added 222 Rn or 220 Rn resulted in a long-term exponential decline of CH 4 concentrations while CO 2 was generated biologically.
https://doi.org/10.1371/journal.pone.0206506.g003  Fig 3A) was interpolated from hourly spaced data. with added radon isotopes (#5 through #9). Thus, in terms of radiation intensity, our experiments represent an extreme test of the radiolysis hypothesis. Only the air in shafts of underground uranium mines has been observed to reach even higher radiation levels of one million or more Bq m -3 [27]. The consistent pattern of CH 4 decline in our experiments without soils can be better appreciated in light of the observed CO 2 dynamics (Fig 3B). CO 2 is more polar than CH 4 , can be more easily adsorbed on surfaces, and is more water-soluble and reactive than CH 4 . Therefore, it is possible that changes in room temperature (21.1 to 27.5˚C) and atmospheric pressure (96.7 to 99.3 kPa) may have affected adsorption and solubility of CO 2 during our experiments. Moreover, after one week of measurements with a wet paper tissue in the 4-L glass flask without soil, fungi had discolored the paper tissue and metabolically generated CO 2 , thus partially stabilizing the CO 2 partial pressure (experiment #6, Fig 3B), apparently without affecting the CH 4 decline (Fig 3A). The paper tissue had been hung by a thread from the central glass stopcock to maximize surface area and to avoid any anoxic microenvironments that could facilitate biological methanogenesis (Fig 1). Subsequent experiments in moist air without soil replaced the wet paper tissue with added deionized water at the bottom of the 4-L glass flask. Experiments with soils initially generated CO 2 via microbial and fungal remineralization of soil organic matter, followed after several days by a decline due to diffusive loss of CO 2 .
In the two experiments with moist soils, we documented a CH 4 loss of of~0.09 ppmv h -1 within the common window of CH 4 concentration decline (Fig 3A), as determined by subtracting the diffusive CH 4 loss in blank experiments from the bulk CH 4 loss in experiments #10 and #11 with soils (Table 1). It is well established that heterogeneously distributed methanotrophic biofilms in the subsurface [28] are capable of scavenging CH 4 from the atmosphere (e.g., [29,30]). Soil gas can often reach 222 Rn radiation levels of many thousand Bq m -3 , depending on local geology [31,32]. If radiolysis would indeed be able to trigger fast oxidative decay of CH 4 in soil gas, such an important CH 4 sink in dry soils without abundant methanotrophic activity would likely have been documented. Also, radiolysis would compete with methanotrophs in moist soils for CH 4 and would have been identified as a factor in soil CH 4 studies.

Experiments in gas-tight terrarium at RHUL
The first stage of experiment #1 at RHUL (Figs 2A and 4) established background conditions for the abundance of negative ions (~3800 ions cm -3 ) and the concentration of 222 Rn (17 to 51 Bq m -3 ) in laboratory air at temperatures from 21.4 to 21.7˚C and relative humidities from 26.6 to 29.0%. After the onset of stage 2, the placement of pitchblende and a beaker with 130 mL, 38˚C warm water into the sealed terrarium strongly increased the abundance of negative ions in air (up to~200,000 ions cm -3 ) and the concentration of 222 Rn (~118 kBq m -3 ). The relative humidity exceeded 85%, and the air temperature intermittently rose by 5˚C. The measurement uncertainty of the Canary Pro radon monitor increased with the 222 Rn radiation level (Fig 4). However, the factory-documented uncertainty at the highest measured radiation level and the steadily increasing abundance of negative ions in air suggested that after a run time of 17 h, the 222 Rn-based radiation level exceeded 100 kBq m -3 (Fig 4; data shown in S1 File). Experiment #2 at RHUL (Fig 2B) used the same sealed terrarium with pitchblende and high humidity to monitor and compare the CH 4 mole fractions in the air of both the terrarium and the outside laboratory air over~77 h. The Canary Pro radon monitor in the terrarium indicated an increase in 222 Rn over time parallel to RHUL experiment #1. After 5 h into RHUL experiment #2, the 222 Rn-based radiation in the terrarium was consistently > 50 kBq m -3 . Despite high levels of ionization and 222 Rn-based radiation in the terrarium, the CH 4 mole fraction of 1.9941 ± 0.0036 ppm in terrarium air after being sealed for~77 h was indistinguishable from the starting value of 1.9971 ± 0.0122 ppm within the uncertainty of measurements ( Table 2).

Synopsis of combined results
The absence of any experimental evidence for accelerated loss of CH 4 in the presence of elevated radiation makes it highly unlikely that radiation from radon isotopes is important in nature where 220 Rn and 222 Rn concentrations are typically much lower. Our data indicate that natural radiation in cave air cannot be responsible for the rapid consumption of CH 4 in air on time-scales of days, even in caves with high relative humidity. The same conclusion had been reached from earlier laboratory experiments [18] and from observations in Australian cave air [15]. In the second stage, the addition of pitchblende and a beaker with hot water provided a source of 222 Rn and high humidity to simulate conditions characteristic of cave environments. Although the final 222 Rn concentration exceeded 100 kBq m -3 and was thus higher than in most caves, the elevated radiation and ionization of air in the terrarium was unable to lower the atmospheric abundance of CH 4 over 77 h in the subsequent RHUL experiment #2 ( Many caves experience seasonally different degrees of venting and even reversals of air flow, which results in differences in air temperature and humidity and is difficult to simulate in laboratory experiments. Still, most cave environments at sufficient distances from cave entrances and vent holes are thermally buffered by surrounding rock and therefore do not express the relatively high diurnal and seasonal temperature and humidity variations as outside environments. Our experiments in laboratories were conducted at relatively constant room temperatures similar to many cave environments. Room temperatures in air-conditioned laboratory buildings are similar to actual temperatures in sub-tropical and tropical caves [14]. The use of water and moist soil in many of our experiments simulated the range of humidity in natural cave air. One possible caveat in terms of dissimilarity between our laboratory settings and actual caves may be the fact that our experiments allowed daylight to reach our experimental setups. However, the amount and timing of indirect light (no direct sunshine) was insufficient to let any photoautotrophs (algae) observably grow in our experiments. A necessary difference between air in our experiments at IU and actual cave air was the presence of traces of CH 4 in our experiments. Some CH 4 was needed to test for possible radiolytic destruction of CH 4 . In contrast, most natural cave air is depleted in CH 4 relative to outside air. We conclude that the experimental conditions during experiments at IU and RHUL were reasonable approximations to simulate cave conditions. In the open atmosphere, solar radiation is mainly responsible for the generation of OH• radicals ( [6], and refs. therein) that are the longest-lived potential radical reactant with CH 4 in air. Subterranean radiolysis by radioactivity involves far more energy than photochemical dissociation of molecules by solar radiation, hence the speciation of resulting ions and radicals is different. A host of highly energetic, short-lived ions and radicals other than OH• is generated in subterranean air. The first abstraction of an H atom from CH 4 requires a far higher activation energy than those of H atoms from methyl CH 3 and methylene CH 2 moieties. We argue that cave environments with elevated radioactivity may host short-lived, yet highly energetic radicals and ions that can supply the needed activation energy for first H-abstraction from CH 4 more efficiently than OH• in the open atmosphere. Thus, the application of kinetic and energetic findings of photochemical CH 4 oxidation in the open atmosphere may not be warranted for subterranean environments.
The α-radiation level in cave air is typically higher than in the open atmosphere because cave air is relatively close to rock and sediment surfaces with minerals harboring radioactive nuclides. The ionization rate in air via 222 Rn radon decay is larger close to the ground, as reported for a Finnish forest [33], a spa [34], and in houses [35]. The effect is due to (i) strongly elevated radon concentrations in the air in porous, uranium-containing substrates and the rapid dilution of radon above surfaces upon mixing with the open atmosphere, especially during windy conditions. In contrast, cave air far from cave entrances is typically less turbulent and allows for a more even distribution of 222 Rn in cave air. (ii) Short-lived 220 Rn will always exhibit a greater abundance in air close to its parent nuclides in soil, rock, cave walls and floors [19]. Regardless, even exceptionally high combined radiation levels of 220 Rn and 222 Rn provided no evidence for accelerated CH 4 oxidation in our experiments.
A plausible reason for slow radiolytic reaction kinetics is the mismatch between the large number of CH 4 molecules in 1 m 3 of atmosphere containing 1.85 ppmv CH 4 at standard conditions (i.e.,~4.55 � 10 19 molecules CH 4 ) relative to the small number of radon-related nuclear decay events in the same volume of air (e.g., 10 kBq m -3 from 222 Rn resulting from the decay of 10,000 atoms of 222 Rn per second). The following simplistic numerical example illustrates the lack of feasibility of radiation-induced rapid oxidation of CH 4 . If we assume that 1 m 3 of atmosphere entering a cave with 10 kBq m -3 , even if every decay of 222 Rn leads to the oxidation of one molecule CH 4 , it would require a geologic time period of~144 million years to oxidize all CH 4 . In reality, the nuclide-specific radiation from the decay of 222 Rn alone is dwarfed by the total radiation from radon, thoron, their radioactive progeny, and any other radioactive nuclides present in a given environment [19]. S1 File offers alternative calculations based on the assumptions that either (i) all energy from α-decay is exclusively invested in radiolytic dissociation of CH 4 and results in the oxidation of multiple molecules of CH 4 per decay event, or (ii) that only a fraction of the energy from α-decay is dissociating CH 4 in the overwhelming presence of other molecules and atoms. The calculated time periods needed to degrade 1.85 ppmv CH 4 at a 222 Rn radiation level of 10 kBq m -3 range from 45.1 to 153,000 years, respectively. Even the most optimistic assumptions cannot speed up the radiolytic reaction kinetics to consume atmospheric CH 4 within hours to days.
We can use the most optimistic scenario for consumption of 1.85 ppmv CH 4 during 45.1 years at 10 kBq m -3 and calculate a radiation level of~165 MBq m -3 that would be required to perform the same task in 24 h, which would be commensurate with kinetic CH 4 observations in caves. Natural radiation levels of a few MBq m -3 have been measured in air where 222 Rn emanates through geologic faults from underlying uranium minerals [36]. Radiation levels in the range of MBq m -3 have been observed in the air of uranium mines [27]. Still, no location is known to offer values close to the required~165 MBq m -3 . We conclude that there is no natural cave environment on earth where the α-radiation level is strong enough to rapidly degrade CH 4 . The same conclusion was recently described in a study that included arguments based on radiolytic kinetics of ion-induced reactions [15] that complement our calculations using αdecay and activation energy.
Subterranean radiation does not provide a mechanism for a fast-acting sink of atmospheric CH 4 that would extend to arid and hyperarid environments, unlike microbial methanotrophy. Our study does not invalidate the geochemical data from previous studies documenting CH 4 dynamics in subterranean ecosystems [10]. We do not call into question the fundamental importance of radiolysis of H 2 O (and other air components) and subsequent redox reactions that are documented in the geologic record (e.g., [37]) or the long-term subterranean radiolytic impact on sedimentary organic matter [1]. However, the exceedingly slow chemical rates of reaction caused by natural rates of radiolysis would likely take years to geologic time periods in cave environments to deplete trace amounts of atmospheric CH 4 in cave air. As long as no alternative mechanisms have been identified, microbial methanotrophy serves as the only known fast-acting sink for subterranean CH 4 in the critical and vadose zones.

Conclusions
Strong radiation from radon isotopes and subsequent radiolysis of air proved unable to rapidly oxidize methane in dry or moist air. In the absence of a feasible alternative methane oxidation mechanism other than microbial methanotrophy, further studies are needed on the ability of microbes to consume trace amounts of methane in poorly ventilated caves, even though the trophic and energetic benefits become marginal at very low partial pressures of methane.
Supporting information S1 File. An Excel file contains a first sheet "read me" with instructions and an overview on additional sheets offering analytical details and radiolysis calculations.