Long time series of fumarolic compositions at Campi Flegrei (Italy), Mammoth Mountain (California), Panarea (Italy) and Nisyros (Greece) show rapid increases, up to orders of magnitude, of the CO2/CH4 ratio systematically with the occurrence of volcanic unrest periods. These easily detected anomalies originate with the arrival of CH4-poor magmatic fluids in the shallower levels of the volcanoes. The data suggest that volcanoes are characterized by magmatic activity at depth also in periods of apparent quiescence. The activity is constituted by the pulsing release of large amount of fluids which either cause unrest periods (seismicity and ground deformation) or possibly could precede volcanic eruption. This type of volcanic activity can be monitored trough the classical geophysical techniques together with the systematic sampling and analysis of fumaroles.
 Methane is a gas species which differentiates hydrothermal gases, where it is present in relatively high concentrations, from high temperature volcanic-magmatic fluids where it is normally absent or present in very low concentrations. Measured CO2/CH4 ratios in fumaroles from 23 hydrothermal systems on the world range from 10 to 104 (Figure 1) roughly in agreement with the theoretical values expected for a gas phase in chemical equilibrium at temperatures from 200°C to 400°C and redox conditions fixed by hydrothermal buffers [Chiodini and Marini, 1998]. Much higher equilibrium CO2/CH4 ratios, in the order of 106 to 1012, can be instead computed for volcanic fluids at high temperature and redox conditions fixed either by the magmatic gases redox buffer (H2S-SO2) [Giggenbach, 1987] or by the typical magmatic redox buffer (i.e., Ni-NiO, Figure 1). Methane in high temperature fumaroles, when detected, is thus ascribed at the mixing of magmatic fluids with hydrothermal fluids or to interaction of magmatic gases in the hydrothermal environment during the transfer toward the surface.
 The aim of this work is to show how CO2/CH4 ratio in fumaroles from hydrothermal systems located in active volcanoes shows important anomalies that correspond with unrest periods, indicating the arrival at the surface of fluids degassed at depth by magma bodies. The case of Campi Flegrei (Italy), where the largest time series of fumarolic compositions is available, is discussed firstly. Subsequently we show compositional time series of fumaroles located in other three volcanoes. All the fumarolic data used in this work, including those taken from the literature, are reported in the auxiliary material.
2. The 1983–2008 Time Series of Fumarolic Compositions at Campi Flegrei (Italy)
 Campi Flegrei caldera is one of the most dangerous volcanic areas in the world, including part of the city of Napoli, the town of Pozzuoli and several densely inhabited villages. After the last eruption in 1538, the caldera subsided until 1968 when an uplift phase started and reached a maximum value of about 3.5 m in 1985. The process occurred during two major uplift periods (bradyseisms) the first in 1968–71, the second in 1983–1985. Since that time the caldera subsided until 2002–2003 when an apparent stabile period starts at a maximum uplift of ∼2.5 m with respect to the beginning of the crises. This general behavior was interrupted by four minor uplifts that occurred systematically every 5–6 years (1989–1994–2000–2006). All the uplift periods were accompanied by seismicity and in particular by many thousands of earthquakes during the two major crises in 1969–72 and in 1983–84. Maximum ground deformation and the location of the earthquakes [Orsi et al., 1999] point to a restricted area in the centre of the caldera, near the fumarolic field of Solfatara.
 The main component of the fumaroles is H2O followed by CO2 and H2S, and the maximum temperature is 160°C that remained constant during the observation period. The absence of acidic gases (SO2, HCl, HF) and relatively high CH4 contents were interpreted as indicators of the stationing of the fluids in a hydrothermal system fed by a mixture of meteoric and magmatic components [Caliro et al., 2007]. Physical numerical simulations of the hydrothermal system [Todesco et al., 2003], suggested that magma degassing episodes have a relevant role in the triggering of the volcanic unrest periods that periodically affect the area [Chiodini et al., 2003]. This important, probably primary role of magmatic fluids released at depth in triggering volcanic unrest periods at Campi Flegrei is not surprising. According to Chiodini et al. , the thermal energy release associated with the gas emission at Solfatara (∼100 MW in 1998) is the most important term in the energetic balance of the whole Campi Flegrei caldera.
 The strict relation among CO2 and CH4 concentrations, and ground elevation is illustrated in the chronograms of Figure 2 referring respectively to the biggest crises of 1983–85 (Figure 2a) and to the last minor bradyseismic event of 2006 (Figure 2b). During both crises, the curve of ground deformation is followed by a negative peak of CH4 and a positive peak of CO2 with a delay of ∼4 months and ∼1 year respectively. In our interpretation the structured sequence of ground uplift episodes and changes in the chemical composition of the fluids expelled by the system is caused by the sudden input of CO2 rich, CH4 poor magmatic gases into the hydrothermal system.
 The process would cause a generalized pore pressure increase registered at the surface by the observed ground uplifts. The positive CO2 anomaly is due to the higher CO2 content of the magmatic gas with respect to the hydrothermal one [Caliro et al., 2007]. In agreement with the observation, the time delay of the arrival of the CO2 peak at the surface was previously estimated to be ∼20 months by physical numerical simulations [Chiodini et al., 2003]. It is note to worth that the physical simulations do not predict any temperature anomaly at the surface in agreement with the data. According to [Caliro et al., 2007] CH4 is formed in the deeper and hottest part of the system by the conversion of CO2 into CH4 in the hydrothermal environment. The arrival of magmatic fluids into this zone can explain the negative CH4 anomaly. In fact magmatic fluids, poor in CH4 and probably characterized by significant contents of SO2, would dilute the hydrothermal component and would cause more oxidizing, transient conditions impeding the CH4 formation. Further effects on the CH4 could be due to variations on the temperature-pressure conditions in the CH4-forming zone. However these are probably minor effects considering the relatively rapid mixing events and the relatively slow kinetic of the CO2-CH4 reactions [Giggenbach, 1997]. The shorter time delay in the negative peak of CH4 with respect to that of CO2 (Figure 2) could reflect an higher mobility of this gas species, possibly indicating a chromatographic effect during the transfer of the gases within the hydrothermal system.
 This correlation between ground elevation and chemical signs deserves further considerations. The variation in the ground elevation at Campi Flegrei (Figure 3a) can be considered as the result of at least two processes acting in the subsoil. On one hand it may derive from ‘volcanic’ sources, i.e. from the variation in the pressure of the hydrothermal system due to the input of magmatic gases and/or from magma movement at depth. On the other, the Campi Flegrei caldera is affected by a secular subsidence. The forces causing subsidence should act also in periods of uplift, most probably deriving from the compaction of the thick layer of recent volcanic pyroclastic products [Dvorak and Mastrolorenzo, 1991]. In order to remove this effect and to highlight the effects due to the ‘volcanic’ sources in the last period (from the last big bradyseismic event until now) we added 0.016 m/a to each elevation measure (i.e. the secular subsidence [Troise et al., 2007]) obtaining the “corrected” curve reported in Figure 3a. The “corrected” curve shows an exponential decay in the time since 1985. This behavior is connected to the decline in the time of the effects of the large bradyseismic event occurred in 1982–1984. According to the best fit equation (Figure 3a) the decay process is characterized by a time constant of 1517 days (∼4.2 years). The residuals of the best fit curve with respect to the measured data (Figure 3b) should give a signal of the ground elevation filtered with both the secular subsidence and the effects of the large 1982–1984 crisis. Potentially this signal is most representative of the minor bradyseismic events that occurred after 1985 and is the most suitable parameter to be compared with our geochemical signals collected in the same period. Figure 3b compares the elevation residuals with the fumarolic CO2/CH4 ratio, reported on a logarithmic scale. Because the opposite behaviour during the crisis of CO2 and CH4, their ratio results in fact as a very good tracer of the input of magmatic fluids into the hydrothermal system of Solfatara. The similarity between the geochemical signal and the elevation residuals is very good and difficult to ascribe as fortuitous. Systematically every ground inflation corresponds to an increase of CO2/CH4, and systematically a decrease of the ratio accompanies any deflation for each of the four minor bradyseisms in the last 25 years. The geochemical parameter correlates well also with seismicity: CO2/CH4 peaks occur some months after the occurrence of earthquakes at Campi Flegrei (Figure 3c). This suggest that pressure pulses, caused by the arrival of magmatic fluids at shallow levels, possibly trigger seismicity by reducing the effective normal stress acting on slip planes. This possibility seems highly probable as similar fluid-triggered earthquakes are observed also in non-volcanic environment [Miller et al., 2003].
3. Time Series of Fumarolic Compositions at Mammoth Mountain (California), Panarea (Italy) and Nisyros (Greece)
 The sequence of seismicity and ground uplift followed by a peak of the CO2/CH4 fumarolic ratio, is not limited to Campi Flegrei, but has been observed at other volcanoes as illustrated in Figure 4.
Figure 4a refers to Mammoth Mountain (California) in the period from 1980 to 2000. Seismicity beneath the mountain was low until 1989 when a 6-month-long seismic crisis of low-magnitude earthquakes occurred. The swarm was considered noteworthy because of its duration, evidence that it was accompanied by dike emplacement beneath the mountain, and the onset of long-period (LP) earthquakes [Sorey et al., 1998, and references therein]. Important observed variations included an increase in the temperature, flow and 3He/4He ratio of the main fumarole (MMF) and the development of large areas of tree kill induced by enormous diffuse flux of cold CO2 on the flanks of the mountain. Using the data reported by Sorey et al. , Figure 3a shows a strong increase (from ∼10000 to ∼100000) in CO2/CH4 ratio of MMF fumarole after the seismic swarm of 1989 and in concomitance with the tree kill process induced by the arrival at the surface of the deep gases.
Figure 4b shows the CO2/CH4 variations observed at Panarea (Eolian Islands, Italy) in the period 2002–2005. The sea floor near Panarea is affected by numerous submarine gas emissions that were studied during the 1990's and interpreted as discharges of a hydrothermal system [Italiano and Nuccio, 1991]. On November 3rd 2002, an anomalous degassing event affected the area, probably in response to a submarine explosion [Chiodini et al., 2006]. The process was evident at the sea surface where a large 8–10 m gas bubble formed. Microearthquakes with Md generally less than 1 (Md max = 1.8) were recorded between November 3rd and 13th. In that period, monitoring of the main gas emissions commenced. Figure 3b shows the variation in the time of the CO2/CH4 ratio measured at emission #2 [Chiodini et al., 2006]. After the November 3rd event and the correlated seismic activity, the CO2/CH4 ratio increased in one month to the highest values (∼500000) and subsequently declined to ∼15000.
 Finally Figure 4c shows the increase of the CO2/CH4 ratio, from 250 to 1500, observed at the fumaroles of Phlegeton crater (Nisyros caldera, Greece) in the 1999–2002 period [Chiodini et al., 2002]. Interestingly a pair of years before the observed geochemical change, during 1996–1998, the volcano showed signs of renewed activity, which included intense seismicity and ground deformation [Sachpazi et al., 2002]. The increase in local seismicity began by late 1995, and culminated on August, 1997, with the occurrence of the two largest earthquakes of the sequence (MS = 5.3 and 5.2). During two separate field surveys held in March and July 1997, Sachpazi et al.  recorded an approximate daily rate of 35 and 50 events/day of Magnitudes in the 0.7–2.1-range. In the same period a vertical ground uplift of about 140 mm was revealed by SAR interferometry [Sachpazi et al., 2002].
4. Discussion and Conclusion
 The CO2/CH4 ratio is order of magnitudes higher in magmatic gases than in hydrothermal fluids, in principle constituting a powerful tracer of magma degassing episodes occurring beneath volcanoes with ongoing hydrothermal activity. This theoretical consideration is supported by the analysis of long time series of fumarolic CO2/CH4 ratio measured at Campi Flegrei (Italy), Mammoth Mountain (California), Panarea (Italy) and Nisyros (Greece). In each of these volcanoes, the CO2/CH4 ratio shows quick and large increases concurrently with the occurrence of volcanic unrest.
 The most complete and longest time series of fumarolic composition, ground deformation and seismicity are available at Campi Flegrei. The ground elevation data of the last 23 years were filtered from the effects of a secular subsidence affecting the area, and from the effects of the decay of the last large bradyseism (1983–1985) in order to obtain a data set suitable to be compared with the geochemical signals. The CO2/CH4 data strictly overlaps in time with the filtered ground movements.
 I speculate that this correlation is due to a common parameter, fluid pressure within the hydrothermal system. Fluid pressure is affected by the amount of magmatic fluids in the system which also controls CO2/CH4 ratio. The fluid pressure of the hydrothermal system drives ground movement. Since 1985, four pulses of the magmatic source have caused the generalized increase in the pressure of the hydrothermal system and the observed 4 periods of ground uplift and seismicity (minor bradydeisms).
 Finally the case of Campi Flegrei together with those of Mammoth, Panarea and Nisyros suggest that volcanoes are characterized by magmatic activity at depth also in periods of apparent quiescence. The activity is essentially constituted by the pulsed release of large amount of fluids which can cause unrest (seismicity and ground deformation) and may precede volcanic eruption. A such volcanic activity can be monitored trough classical geophysical techniques together with systematic sampling and analysis of fumaroles.
 The author would like to thank two anonymous reviewers for their useful suggestions and the Centro di Monitoraggio team of Osservatorio Vesuviano for the seismological data. This work was financially supported by INGV, ‘Dipartimento di Protezione Civile’ of Italy (UNREST project) and by EU (VOLUME project).