Learning from soil gas change and isotopic signatures during 2012 Emilia seismic sequence

Soil surveys were performed in Medolla (Italy), a peculiar area characterized by spotty high soil temperature, gas vent, and lack of vegetation, to determine the migration mechanisms and spatial behavior of gas species. Hereby we present soil gas measurements and their isotopic ratios measured between 2008 and 2015, including the 2012 Emilia-Romagna seismic sequence. We found that soil gas concentrations markedly changed during the main shocks of May 20 and 29, 2012 (Mw 6.1 and 6.0, respectively), highlighting the presence of a buried fault intersecting the gas vents. We suggest that crustal dilation associated with seismic activity favored the uprising of geogas towards the surface. Changes in the isotopic signature highlight the contribution of two distinct sources, one deeper, thermogenic and another superficial related to organic-rich layer, whose relative contribution varied before, during and after the earthquake. We suppose an increase of microbial component likely due to the ground shaking of shallower layers linked to seismic sequence, which masks the thermogenic contribution. Although the changes we detect are specific for an alluvial plain, we deduce that analogous processes may be active elsewhere, and that soil gas geochemistry represents an useful tool to discriminate the gas migration related to seismic activity.

Scientific REPORtS | 7: 14187 | DOI: 10.1038/s41598-017-14500-y Quaternary fluvio-lacustrine deposits of the Po plain 12,13 . This sector of the buried Apennine front is tectonically active in response to the general compressive stress field 14 dating from middle Pleistocene.
In the Medolla area, natural gas (CH 4 -dominated) reservoirs have been detected at depths between 2000 and 2700 m, in correspondence of reverse fault planes 15 from which fluids migrate upward along minor fault planes. At these depths, the main stratigraphic unit consists of upper Miocene marls and organic-rich clays 15 . Moreover, some other CH 4 -dominated gas occurrences are recognized at 200 m and between 650 and 900 m of depth in the Plio-Pleistocene formations 16 . According to Lindquist 16 and Mattavelli and Novelli 17 most of the gaseous hydrocarbons in the Po Plain have a biogenic origin (80%), while the remaining is equally distributed between thermogenic (10%) and mixed origin (10%).
In May-June 2012, a seismic sequence struck the Emilia-Romagna Region, with more than 2,400 aftershocks. The epicenters of the two main shocks, Mw 6.1 and 6.0 18 were located 15 and 2 km from the study area, respectively (Fig. 1a).

Results
The soil gas surveys were carried out in October 2008, December 2008, soon after the first main-shock, in September 2012 and in the following years (2013, 2014 e 2015), aiming at monitoring any variations in the gas chemistry, as well as in the seepage spatial distribution. Several gas species (CO 2 , CH 4 , H 2 , He, Ne, C 2 H 6 , 222 Rn) were determined, but in the following only CO 2 , H 2 and CH 4 measurements are reported, since they are the only gases where isotopic ratios were also analyzed.
Collected data were processed with a standard statistical approach (Supplementary Table S1) and used to create three-dimensional surface maps (Fig. 2).
Normal Probability Plot (NPP) were used to select background, anomalous values, and extra outliers 19 . In particular values above 2.00% v/v for CO 2   Scientific REPORtS | 7: 14187 | DOI:10.1038/s41598-017-14500-y more than one order of magnitude lower (CO 2 , 2.31% v/v; H 2 , 0.44 ppm v/v; CH 4 , 6.01 × 10 −4 % v/v). The highest CH 4 concentrations were measured in two areas with the absence of vegetation (0.048% v/v at M20 and 0.024% v/v at M3). It is likely that these macroseeps represent preferential migration pathways for deep gas hosted in the Mesozoic formations (>3000 m depth; Camurana 2 well log), as suggested by the δ 13 C-CH 4 measured on these points showing thermogenic values of −25.88 and −29.68‰.
Soon after the main shock on 20 th of May 2012, CH 4 average concentrations increased by more than three orders of magnitude (6.46% v/v), whereas H 2 and CO 2 showed an increment to 9.36 ppm v/v and 5.43% v/v respectively. The highest values of all gas species were observed at M20 and M3 (up to 39.00% v/v and 89.42 ppm v/v for CH 4 and H 2 at M20, and up to 13.50% v/v for CO 2 at M3) and in a newly formed area with lack of vegetation (M14, 11.83% v/v, 86.56 ppm v/v and 40.34% v/v for CO 2 , H 2 and CH 4 , respectively).
A few months later, in September 2012, average H 2 concentrations increased further (12.28 ppm v/v), whereas CH 4 and CO 2 concentrations remained substantially stables, with the highest values (39.80% v/v and 12.25% v/v, respectively) in correspondence of M3 site. Some CO 2 anomalous spots (with values ranging from 8.41% v/v to 8.94% v/v) were measured outside the macroseeps zone, in the northern and southern part of the study area. In absence of hints for a deep origin (as the association with other geogas) these anomalies can be attributed to organic material oxidation, microorganism or plant respiration 21 .
Between 2013 and 2014 the CH 4 mean concentrations increased up to 7.53% v/v (highest value of 84.20% v/v at M20), whereas H 2 decreased to 3.64 ppm v/v with the highest value of 37.2 ppm v/v always at M20. Mean CO 2 concentrations decreased to 2.72% v/v (maximum value of 11.01% v/v on M3).
Finally, in 2015 the average concentrations of all gases dropped to 2.03% v/v, 4.61 ppm v/v and 0.70% v/v for CO 2 , H 2 and CH 4 , respectively, remaining overall higher than the values collected in 2008. CH 4 highest values were found on macroseeps (maximum value of 8.97% v/v on M14), whereas the highest CO 2 and H 2 values (6.32% v/v and 30.7 ppm v/v) were located in the northern and eastern part of the studied area.
The origin of soil gas was investigated by isotopic ratios of CH 4 and CO 2 measured on the points with highest concentrations (Table S2). As already mentioned δ 13 C-CH 4 , δD-CH 4 measured on M3 and M20 in October ranged from −29.86‰ to −25.88‰ vs VPDB and −92.26‰ to −106.44‰ vs VSMOW, respectively. A few months later (December 2008) these points sensibly changed their CH 4

Discussion
The anomalous concentration of CO 2 , H 2 and CH 4 , highlighted a general E-W trend connecting the main macroseeps (Fig. 2). The association of more than one gas species along a linear trend suggests the presence of a previously unknown blind fault, similarly to what observed in other seismically active areas 4 . In soils, the average H 2 concentration is about 0.5 ppm v/v, almost the same as that found in the atmosphere. High H 2 concentrations, up to several thousand ppm, are restricted to active faults 22 . CH 4 and H 2 concentrations are positively correlated over time, with a Pearson coefficient ranging from 0.66 to 0.86. This relationship suggest that CH 4 can act as a carrier for H 2 which would be otherwise unable to reach the surface due to its low concentration 2,22 .
No evidence for this linear trend were observed before 2012, thus it can be inferred that seismic activity has altered stress field enhancing a pre-existent phenomenon, such as the gas seepage through the soil. Earthquakes and crustal deformation 4 can indeed alter the hydraulic properties of soils, such as permeability and porosity, favoring advective migration of deep gases toward the surface caused by variations in pressure and temperature following preferential pathways.
In a low permeability soil, such as the Plio-Pleistocene sediments of the area, the overpressure generated by an earthquake can remain active for a long period and the effects on deformations and on fluid flow could be visible after several months 23 . The significant decrease of soil gas concentrations in 2015 may be due to both a reduction of permeability and porosity of rocks and soils in the rupture zone and a closure of the pathways opened by the seismic activity after the overpressure generated by the earthquake had reduced.
Concerning the isotopic ratios (δ 13 C-CH 4 , δD-CH 4 and δ 13 C-CO 2 ) of soil gas, the traditional Shoell's plot 24 (Fig. 3a) highlights two different groups of data: one with prevailing microbial and mixed origin, and a few points falling in the thermogenic (TD) field. Thermogenic signature is positively correlated with low CH 4 concentrations, whereas for increasing CH 4 concentrations microbial and mixed sources prevail.
Overall, during the survey record, the isotopic signatures of most samples remained substantially consistent. However the most active M3 and M20 macroseeps switched their signature from thermogenic (October 2008) to both microbial and mixed origin (December 2008-2015; Fig. 3b). We suggest that in this area two different sources of methane coexist, one thermogenic, deriving from the deeper Mesozoic reservoirs 17 (>3000 m), and another microbial 25 produced at a shallower depth (between 200 and 900 m). It is likely that the relative increase of microbial component masks the thermogenic contribution. The lack of an evolutionary trend between thermogenic and mixed-microbial origin suggest that a fractioning process does not occur. On the contrary, the shift between mixed and microbial origin suggests the presence of a variable contribution of shallow input over time. A vertical profile, performed in 2015 at the M20 site ( Fig. 3b; Supplementary Table S3) to verify this hypothesis, showed increasing thermogenic contribution with depth, shifting from microbial (0.20 m), to mixed (1 m), to thermogenic (2.5 m).
Samples with thermogenic origin showed δ 13 C-CO 2 values (from −10.96‰ to −37.04‰) typical of organic and/or soil-derived origin 26 . On the contrary, extremely negative values, between −42.15 and −70.01‰, were recorded on samples with CH 4 microbial and mixed origin. These negative values have the same 13 C/ 12 C ratio of CH 4 , suggesting that CO 2 directly derives from microbial CH 4 .
This hypothesis is consistent with Galand et al. 27 , Valentine et al. 28 and Whiticar 29 which showed that processes as bacterial methanogenesis may produce a carbon isotopic fractionation between CH 4 and the coexisting CO 2 ranging from 41‰ to 72‰, 22‰ to 58‰, and 49‰ to 95‰, respectively. These ranges are higher than the isotopic difference between CH 4 and CO 2 detected at Medolla area (from 28‰ to −2‰; Table S2). In fact, the depletion of 13 C in CO 2 is more compatible with a 13 C/ 12 C kinetic fractionation due to a partial CH 4 to CO 2 conversion which occur in the presence of free oxygen and methanotrophic bacteria [9][10][11] . Indeed, according to geochemical and biological data from Capaccioni et al. 9 and Cappelletti et al. 11 , anomalous ground heating of Medolla area is due to oxidative conditions and bacterial activity (genus Methylocaldum), promoting the exothermic oxidation of methane within the most aerated soil layer at 0.6 and 0.7m depths 9 . This exothermic reaction (800 kJ × mol −1 ) 30 could also explain the anomalous soil heating measured on macroseeps (from 30 to 48 °C) in the 2012-2015 period. This process had always been active over the time, although with different intensity, and could explain both the observed ground heating up, and the diffuse emission from the soil of CO 2 characterized by an extremely negative isotopic ( 13 C/ 12 C) signature.
In this study, soil temperatures measured in other parts of the investigated area, showed average values of 24 °C consistent with air temperatures and regional geothermal gradient (1 °C/100 m) 31 , supporting the theory of CH 4 oxidation as the main process for macroseeps soil heating.
Although the soil heating mechanism is independent from tectonic activity we suggest that 2012 seismic sequence might have enhanced this phenomenon increasing the preexisting CH 4 emissions. Indeed the highest soil temperature (51.8 °C) was measured in May 2012 together with the highest CO 2 and CH 4 concentrations (11.7% v/v and 38.8% v/v, respectively). On the other hand when a gas uprising is high, CH 4 is not completely consumed by bacteria 32 . According to the prevailing local conditions, some fraction of methane may escape oxidation reaching the surface.

Conclusions
Soil gas distribution and their isotopic signature were investigated in the Medolla farming area, between 2008 and 2015. After the 2012 seismic sequence the soil gas concentrations of CO 2 , H 2 and CH 4 markedly increased along an E-W preferential direction, suggesting the presence of a buried tectonic lineament linking CH 4 macroseeps. Seismic crustal deformation favored the fluid migration towards the surface by increasing the pore pressure and/or enhancing permeability of soils following preferential pathways. In 2015, these concentrations gradually decrease towards the initial values, although they remain still higher than those observed in 2008. This decreasing is likely due to a natural lowering of permeability and porosity in the rupture zone after the overpressure generated by the earthquake had reduced.
Isotopic ratios of CH 4 and CO 2 highlight two different gas sources, one deeper, thermogenic and another shallower, microbial. These sources coexist producing a variable isotopic 13 C/ 12 C ratio from microbial to mixed, depending on the contributions of each source. The lack of an evolutionary trend between thermogenic and mixed-microbial origin suggest that a fractioning process does not occur.
The extremely negative values of δ 13 C-CO 2 , recorded on samples with CH 4 microbial and mixed origin, and the high soil temperatures are ascribed to the exothermic oxidative reactions of CH 4 in CO 2 which occur in the presence of free oxygen and methanotrophic bacteria. The macroseeps CH 4 emission and the high soil temperature have been known since the late nineteenth century and are therefore independent from tectonic activity of 2012. The earthquakes might have enhanced the gas seepage phenomenon, favoring the uprising of microbial and mixed CH 4 . According to isotopic results, the ground shaking linked to the 2012 seismic sequence enhanced the migration of soil gases from the shallower layers of Plio-Pleistocene deposits, increasing the microbial contribution of methane and covering the low amount of deeper thermogenic gases.
Soil gas geochemistry represents an useful tool to discriminate the gas migration related to seismic activity. The long term geochemical monitoring allowed to recognize that after an initial variation of soil gas distribution linked to seismic activity, the system in the Medolla area is slowly returning to its pre-seismic condition. Obtained results encourage the research about soil gas geochemistry on seismic active area highlight the importance to have a dataset before, during and after earthquakes.

Method
Sampling procedure. 37 soil gas samples were collected on a yearly basis from October 2008, except for two surveys in 2012. All surveys were conducted during a period of stable and dry weather conditions and in a short time to minimize any variations induced by different sampling periods. Samples were collected from the unsaturated or vadose zone using a steel probe driven into the ground to a depth of 0.8 m; this depth is considered to be below the major influence of meteorological variables 33,34 .
Chemical analysis. The soil-gas concentrations (N 2 , O 2 , CO 2 , CH 4 , He, C 2 H 6 , H 2 ) were analyzed in the Fluid Geochemistry Laboratory at INGV Rome, by a MicroGC Agilent 4900 CP, equipped with two Thermal Conductivity Detectors, responding to the difference in thermal conductivity between the carrier gas (Ar) and the sample components, with an error of ±3%.
Isotopic analysis on free gas (δ 13 C-CO 2 , δ 13 C-CH 4 , δD-CH 4 ,) were performed at ISO4 S.n.c. Laboratory. The results were obtained by preparing the sample according to ref. 35 . Results are expressed in VSMOW and VPDB, following the International Atomic Energy Agency protocol.
Normal probability plots (NPPs), were elaborated by Sinclair method to distinguish different populations and amore objective approach to statistical anomaly threshold estimation.
The normal probability plot of CO 2 shows a data distribution characterized by six populations: i) background values, ranging from 0 to 1% v/v; ii) threshold anomaly, ranging from 1 and 2% v/v; iii) weak local anomaly, ranging from 2 to 3% v/v; iv) moderate anomalous values, up to 6% v/v; v) high anomalous values, ranging from 6 to 10% v/v; vi) outliers, with values higher than 10% v/v. Scientific REPORtS | 7: 14187 | DOI:10.1038/s41598-017-14500-y The NPP of H 2 highlights a quite homogeneous distribution for values up to 10 ppm v/v, with background values from 0 to 1.8 ppm v/v, weak anomalies between 4 and 10 ppm v/v, moderate anomalies between 10 and 36 ppm v/v, high anomalies between 36 and 60 ppm v/v and outlier values over 60 ppm v/v. For CH 4 , six populations were identified. The threshold anomaly is comprised between 0 and 1000 ppm v/v. The other populations are characterized by values ranging from 1000 ppm v/v to 4% v/v (local anomaly), from 4 to 10% v/v (weak anomaly), from 10 to 18% v/v (anomalous values), from 18 to 30% v/v (high anomalous values) and values over 30% v/v (outliers).
Data were displayed as 3D Surface maps were used to create a three-dimensional shaded rendering from a grid file. The height of the surface corresponds to the Z value of the associated grid node. These maps use colours to indicate the local orientation of the surface relative to a user-defined light source direction. The program Surfer 12.0 (Golden Software) determines the orientation of each grid cell and calculates reflectance of a point light source on the grid surface. The light position for all the maps is 135° for the horizontal angle and 45° for the vertical angle.