First In-Situ Measurements of Plume Chemistry at Mount Garet Volcano, Island of Gaua (Vanuatu)

: Recentvolcanicgascompilations haveurged theneedtoexpandin-situplumemeasurements to poorly studied, remote volcanic regions. Despite being recognized as one of the main volcanic epicenters on the planet, the Vanuatu arc remains poorly characterized for its subaerial emissions and their chemical imprints. Here, we report on the ﬁrst plume chemistry data for Mount Garet, on the island of Gaua, one of the few persistent volatile emitters along the Vanuatu arc. Data were collected with a multi-component gas analyzer system (multi-GAS) during a ﬁeld campaign in December 2018. The average volcanic gas chemistry is characterized by mean molar CO 2 / SO 2 , H 2 O / SO 2 , H 2 S / SO 2 and H 2 / SO 2 ratios of 0.87, 47.2, 0.13 and 0.01, respectively. Molar proportions in the gas plume are estimated at 95.9 ± 11.6, 1.8 ± 0.5, 2.0 ± 0.01, 0.26 ± 0.02 and 0.06 ± 0.01, for H 2 O, CO 2 , SO 2 , H 2 S and H 2 . Using the satellite-based 10-year (2005–2015) averaged SO 2 ﬂux of ~434 t d − 1 for Mt. Garet, we estimate a total volatile output of about 6482 t d − 1 (CO 2 ~259 t d − 1 ; H 2 O ~5758 t d − 1 ; H 2 S ~30 t d − 1 ; H 2 ~0.5 t d − 1 ). This may be representative of a quiescent, yet persistent degassing period at Mt. Garet; whilst, as indicated by SO 2 ﬂux reports for the 2009–2010 unrest, emissions can be much higher during eruptive episodes. Our estimated emission rates and gas composition for Mount Garet provide insightful information on volcanic gas signatures in the northernmost part of the Vanuatu Arc Segment. The apparent CO 2 -poor signature of high-temperature plume degassing at Mount Garet raises questions on the nature of sediments being subducted in this region of the arc and the possible role of the slab as the source of subaerial CO 2 . In order to better address the dynamics of along-arc volatile recycling, more volcanic gas surveys are needed focusing on northern Vanuatu volcanoes. is given as an analog for plume density during our measurement period.


Introduction
Monitoring the composition and mass flux of subaerial volcanic emissions is key to understanding how volcanism impacts our planet, on both local and global scales.
Therefore, constraining major volatile budgets from subaerial volcanic arc emitters provides a tool to partially quantify the recycling efficiency of H 2 O, CO 2 , SO 2 and other volatiles via the subduction process. The contrasting volatile ratios in the depleted mantle (CO 2 /S T of~0.3-0.8) [13], subducted sediments (CO 2 /S T~1 -100) [14] and crustal fluids from limestone de-carbonatation (CO 2 /S T >>100) [15] imply that measuring volcanic gas composition can bear crucial information on volatile sources at depth. At the arc-scale, ever-increasing volcanic gas inventories are key to understanding the volatile cycle at subduction.
For instance, the Southwestern Pacific region is home to some of the most persistently degassing volcanoes on Earth [23], but Melanesian volcanoes have systematically been underrepresented in previous literature reviews. Recent advances in satellite remote sensing techniques, such as the ozone monitoring instrument (OMI), have greatly improved our knowledge on SO 2 flux emissions globally [23], revealing Vanuatu as a major volcanic SO 2 source globally, and urging the volcanic gas community to target volcanic gas studies in the poorly studied, yet strongly emitting volcanoes along this arc segment.
In-situ measurements of volcanic gas compositions can be attained by using relatively inexpensive multi-component gas analyzer systems (such as the multi-GAS) [24,25]. In addition to their great potential in eruption forecasting, these portable units have revolutionized our understanding of along-arc chemistry of volcanic gases at different regions of the world. Moreover, when combined with remotely sensed SO 2 fluxes, these measurements provide a means to constrain volcanic gas fluxes, especially CO 2 [17][18][19][20][21][26][27][28][29].
In the Vanuatu arc, volcanic gas surveys have mostly focused on Yasur [30][31][32] and Ambrym [33][34][35] due to the persistent high levels of degassing and volcanic activity. However, Reference [36] reported the first estimate of volcanic SO 2 budget for the Vanuatu Island Arc, in which the authors highlight other vigorously degassing volcanoes along the arc, for which volcanic gas compositions have remained unconstrained.
In this paper, we present the results of a field campaign to measure gas compositions and estimate total volatile fluxes emitted from Mount Garet volcano (or Mount Gharat; also occasionally referred to in the literature as Gaua volcano), located on the island of Gaua, Vanuatu, whose activity in 2009-2010 prompted the evacuation of more than 1500 inhabitants [36]. Data collected in December 2018 is here integrated with previous SO 2 flux reports from both ground-based measurements [36] and satellite records [23] in order to assess the volatile budget of Mount Garet. Data from this study represents the first in-situ plume composition measurements obtained for Mount Garet and thus provides information on volcanic gas signatures in the northern part of the arc segment. We also integrate our novel dataset in the context of previously reported gas compositions for other Vanuatu volcanic centers and discuss the overall implications of this and other recent studies of Melanesian volcanoes [37] to global volatile budgets.

The Vanuatu Island Arc
The New Hebrides Island Arc is located in the southwest Pacific ( Figure 1). It is composed of a narrow chain of Neogene to recent volcanic islands originating from the eastward subduction of the Australian plate beneath the Pacific plate, with associated back-arc extension of the North Fiji basin on the Pacific side [38,39]. The recent tectonic evolution of the arc is dominated by its collision with the D'Entrecasteaux Zone (DEZ) [40], which began 1.5-3 Ma ago near Epi island, and has been migrating northward as a result of the obliquity of the DEZ relative to the trench [40,41]. The New Hebrides Island Arc is subdivided in four different segments, the northernmost being essentially submarine and poorly studied. The other three include the central and south segments (CVS and SVS, respectively) and the high-Mg andesite volcanic segment (HMAVS) [42].
Appl. Sci. 2020, 10, x FOR PEER REVIEW  3 of 16 submarine and poorly studied. The other three include the central and south segments (CVS and SVS, respectively) and the high-Mg andesite volcanic segment (HMAVS) [42]. Vanuatu extends over the CVS and SVC, for about 700 km. It consists of a partly emergent ridge with an average width of 200 km, that overlies a steeply eastward-dipping subducted slab [42]. The DEZ collision triggered the detachment of the Australian plate lithosphere beneath the central part of Vanuatu. However, this detachment is less pronounced in the Northern part of the Vanuatu arc (due to a westward and upward invasion of the Indian-type MORB mantle) [42] where the island of Gaua is located.
Gaua, also named Santa Maria, represents the emerged part of a 3000 m-high and 40 km-wide composite volcano [43]. It consists of a little-dissected symmetrical cone, about 25 km-across, truncated at an elevation between 500 m and 690 m (a.s.l.) by an 8.5 × 6 km central caldera, filled by a large crater lake (Lake Letas; Figure 1). The active cone of Mt. Garet rises in the middle of Lake Letas.
Mt. Garet (797 m, a.s.l.) has two distinct craters. Since 1991, the activity has clustered on the southeast crater, which was also the site of the 2009-2010 explosions [44] and the survey location for the in-situ volcanic gas measurements reported here. A previous study [45] on the geology of the volcano identified a thick pile of basaltic and andesitic lava flows with associated rubble horizons and beds of red cinders or agglomerates. Mallick and Ash [45] also recognized an extensive cap of ash surrounding the caldera. On the flanks, small parasitic cones and fissure-fed lava flows are distributed along fractures concentric to the caldera rim. The morphology of vents and lava surfaces mantling the lower slopes suggest that these peripheral extrusions are very young. Analysis of 2009-2010 lavas, products of the last major eruption, revealed identical compositions to those previously reported in literature [43,45,46]. The occurrence of basaltic andesite and basaltic trachyandesite volcanites with highly calcic plagioclase suggest that the 2009-2010 eruption of Gaua was fed by an alkaline (K-rich) magmatic source, typical of the northern part of the Vanuatu arc [44].  Vanuatu extends over the CVS and SVC, for about 700 km. It consists of a partly emergent ridge with an average width of 200 km, that overlies a steeply eastward-dipping subducted slab [42]. The DEZ collision triggered the detachment of the Australian plate lithosphere beneath the central part of Vanuatu. However, this detachment is less pronounced in the Northern part of the Vanuatu arc (due to a westward and upward invasion of the Indian-type MORB mantle) [42] where the island of Gaua is located.
Gaua, also named Santa Maria, represents the emerged part of a 3000 m-high and 40 km-wide composite volcano [43]. It consists of a little-dissected symmetrical cone, about 25 km-across, truncated at an elevation between 500 m and 690 m (a.s.l.) by an 8.5 × 6 km central caldera, filled by a large crater lake (Lake Letas; Figure 1). The active cone of Mt. Garet rises in the middle of Lake Letas.
Mt. Garet (797 m, a.s.l.) has two distinct craters. Since 1991, the activity has clustered on the southeast crater, which was also the site of the 2009-2010 explosions [44] and the survey location for the in-situ volcanic gas measurements reported here. A previous study [45] on the geology of the volcano identified a thick pile of basaltic and andesitic lava flows with associated rubble horizons and beds of red cinders or agglomerates. Mallick and Ash [45] also recognized an extensive cap of ash surrounding the caldera. On the flanks, small parasitic cones and fissure-fed lava flows are distributed along fractures concentric to the caldera rim. The morphology of vents and lava surfaces mantling the lower slopes suggest that these peripheral extrusions are very young. Analysis of 2009-2010 lavas, products of the last major eruption, revealed identical compositions to those previously reported in literature [43,45,46]. The occurrence of basaltic andesite and basaltic trachyandesite volcanites with highly calcic plagioclase suggest that the 2009-2010 eruption of Gaua was fed by an alkaline (K-rich) magmatic source, typical of the northern part of the Vanuatu arc [44].

The 2009-2010 Eruption of Mount Garet (Gaua)
According to local inhabitants, the 2009-2010 eruption of Mount Garet was the strongest witnessed since 1963. Bani et al. [44] assessed magma composition, lithology of volcanites, ash mineralogy, SO 2 mass discharge, halogen release and lake water chemistry, and provided a comprehensive understanding of the 2009-2010 sequence of eruptive events described below.
A series of mild explosions on 29 September 2009, marked the beginning of a year-long eruption at Mount Garet. Shortly after, on October 3 and 14, airborne differential optical absorption spectroscopy (DOAS) measurements revealed SO 2 fluxes of 3024 ± 1063 and 2765 ± 956 t d −1 , respectively [36]. Despite high degassing rates, the first two months of volcanic activity were largely characterized by low explosivity, with the exception of four large explosive events that took place between 18-27 November 2009 and that displaced 250 inhabitants from the western coast of Gaua, the area most affected by ash fallout. By 17 December, SO 2 discharge rates had remained high (3084 ± 1080 t d −1 ) [36], although the low proportions of juvenile materials and the abundance of hydrothermally altered rock and lava fragments deposited until then suggested an initial eruptive phase fueled by the subsurface interaction between the rising magma and a hydrothermal system [44].
Volcanic activity intensified again in January 2010. Eruption columns regularly reached up to 3000 m a.s.l. over the following five months (Volcanic Ash Advisory Center, Darwin) (GVP: 05/2011 (BGVN 36:05)) and ash fallout extended at least 10 km away from the source [44]. Following this, magma composition became more basic and more homogeneous, as evidenced by the lithological characteristics of ash fragments and further by Scanning Electron Microscopy (SEM) analysis [44]. The prominent role of more juvenile magma in this eruptive phase is attested by larger proportions of vesiculated fragments, and by a concomitant decrease in hydrothermally altered materials emitted. By the end of January, up to 75% of the erupted products were juvenile, and composed of 50% of vesiculated fragments [44].
Finally, in July 2010, volcanic activity at Mount Garet started to decline and the eruption ended in September 2010. The Vanuatu Geohazards Observatory (VGO) maintained the alert level at two throughout the eruption, and no further evacuation was carried out. During the 2009-2010 Gaua eruption, Lake Letas also went through a partial color change from blue to brown [44], which could still be seen in December 2018 in the area adjacent to Mt. Garet ( Figure 1).

Materials and Methods
We used a multi-GAS unit [24,25] to measure in-plume concentrations of major volatile species at Mount Garet's southeast crater (~797 m a.s.l.). The instrument was composed of a Gascard nondispersive infrared (NDIR) CO 2 spectrometer from Edinburgh Sensors and City Technology SO 2 , H 2 S and H 2 electrochemical sensors (see supplementary material S1 for details on sensor's manufacture models, sensor's accuracy and repeatability, calibration ranges and gas standards). Gas sensors were calibrated before and after the field campaign at the Earth and Marine Sciences Department (DiSTeM) of the University of Palermo. The system also included temperature (T) and relative humidity (Rh) KVM3/5 Galltec-Mela sensors. All components and respective power sources were contained inside a weather-resistant plastic case, with inlet/outlet ports to provide access to ambient air.
In the field, volcanic gas was pumped through the in-series connected sensors using a small pump with a flow rate of 1.8 l/min. In-plume gas concentrations were measured non-continuously for approximately 4 h (see Figure 2 for the whole concentration time series of CO 2 and SO 2 ) at a frequency of 1 Hz. Each measurement routine was preceded by instrument warm-up (3 min) and two minutes of ambient air flow to flush residual volcanic gas that remained trapped within the circuit.
Multi-GAS data were processed using the Ratiocalc software [47]. We selected specific acquisition time windows showing good temporal correlation (R 2 ≥ 0.70) between volatile concentrations measured simultaneously [48][49][50] (see Figure 3a for different gas species). Molar ratios and errors reported in Table 1 were calculated from the slope and the R 2 value of the best-fit line (linear regression, such as the example of Figure 3b, in black), respectively. Following an identical procedure, bulk compositions and associated errors reported in Tables 1 and 2 were obtained from all data points for a given x/SO 2 pair. The x/SO 2 dot-to-dot ratio versus in-plume SO 2 concentration [47,51] (Figure 3a) is also shown to demonstrate that ratios, widely scattered at low SO 2 concentrations, converge to a less-varying ratio value (statistically indistinguishable from bulk x/SO 2 ratios reported in Table 1) at high SO 2 concentrations (30-70 ppmv). Table 1. Individual x/SO 2 molar ratios measured by the multi-component gas analyzer system (multi-GAS) in the Mount Garet's plume. These correspond to time windows showing good temporal correlation between gas species (R 2 ≥ 0.70) and strong volcanic gas signal (≥30 ppm of SO 2 ). Ratios and associated errors correspond to the slope and the R 2 value of the best-fit regression line, respectively. concentrations, converge to a less-varying ratio value (statistically indistinguishable from bulk x/SO2 ratios reported in Table 1) at high SO2 concentrations (30-70 ppmv). Table 1. Individual x/SO2 molar ratios measured by the multi-component gas analyzer system (multi-GAS) in the Mount Garet's plume. These correspond to time windows showing good temporal correlation between gas species (R 2 ≥ 0.70) and strong volcanic gas signal (≥30 ppm of SO2). Ratios and associated errors correspond to the slope and the R 2 value of the best-fit regression line, respectively.  (3) is then shown in Figure 3 for detailed calculation procedure.  (Table 1) combined. SO 2 and CO 2 concentration (in ppmv) time series at the top illustrates plume conditions at which individual ratios were calculated. Note that CO 2 /SO 2 ratio (3) is then shown in Figure 3 for detailed calculation procedure.   Table 1. Cross-sensitivity SO 2 effects on the H 2 S sensor were estimated at 14% during calibration and corrections were applied during data processing to calculate interference-corrected H 2 S concentrations in the gas plume.

Gas Species
Water concentration ratios were calculated using measurements of T, P and Rh and the Arden Buck equation [47]. CO 2 and H 2 O concentration ratios were estimated after subtraction of background-air concentration ratios acquired at plume-free areas, and their relative abundance is given in Figure 4 following the above described procedure.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 16 Cross-sensitivity SO2 effects on the H2S sensor were estimated at 14% during calibration and corrections were applied during data processing to calculate interference-corrected H2S concentrations in the gas plume.
Water concentration ratios were calculated using measurements of T, P and Rh and the Arden Buck equation [47]. CO2 and H2O concentration ratios were estimated after subtraction of background-air concentration ratios acquired at plume-free areas, and their relative abundance is given in Figure 4 following the above described procedure.

Volcanic Gas Composition
Gas emissions at Mount Garet were measured at close proximity to the southeast crater. Therefore, over the non-continuous, 4-h measurement period, the multi-GAS consistently captured a strong plume signal (up to ~70 ppm SO2). Background air-corrected CO2 concentrations varied from <1 to ~60 ppm (Figures 3 and 4). Rare plume batches yielded H2S concentrations above the 14% crosssensitivity to SO2 but occasionally reached in-plume concentrations up to ~7 ppm ( Figure 3). We recorded H2 concentrations up to 3 ppm (Figure 3). The volcanic H2O signal (up to ~3500 ppm; Figures 3 and 4) was intermittently resolvable from background noise caused by high humidity conditions characteristic of tropical environments.
Derived volatile ratios (in the form of x/SO2 ratios; where x is either H2O, CO2, H2S, or H2) for Mount Garet were obtained in measurement windows where strong positive co-variations (R 2 ≥ 0.70) were observed between SO2 and the other volatile species measured (Figure 3). Interference-corrected H2S/SO2 ratio and H2/SO2 ratios (0.13 ± 0.01 and 0.03 ± 0.01, respectively) remained constant throughout the entire measurement period. Likewise, CO2/SO2 ratios varied narrowly between 0.65 and 1.07 (avg. 0.87 ± 0.24). Due to variable mixing between atmospheric H2O and the magmatic plume, H2O/SO2 measurements were more variable, ranging from ~40 to ~56 (avg. 47.2 ± 5.7; Tables  1 and 2). Similarly, H2O/CO2 ratios also showed wide variations across the entire 4-h window, reaching values as low as ~40 and as high as ~70. However, the best correlation (R 2 = 0.75) was found at the upper limit of that range, yielding a H2O/CO2 ratio of ~69 (Figure 4). The volcanic plume composition of Mount Garet was estimated at 95.9 mol% H2O, 1.8 mol% CO2, 2.0 mol% SO2, 0.3 mol% H2S, and 0.06 mol% H2 (uncertainties reported in Table 2 are expressed as the standard error of the regression analysis and subsequent error propagation).

Volcanic Gas Composition
Gas emissions at Mount Garet were measured at close proximity to the southeast crater. Therefore, over the non-continuous, 4-h measurement period, the multi-GAS consistently captured a strong plume signal (up to~70 ppm SO 2 ). Background air-corrected CO 2 concentrations varied from <1 to~60 ppm (Figures 3 and 4). Rare plume batches yielded H 2 S concentrations above the 14% cross-sensitivity to SO 2 but occasionally reached in-plume concentrations up to~7 ppm (Figure 3). We recorded H 2 concentrations up to 3 ppm (Figure 3). The volcanic H 2 O signal (up to~3500 ppm; Figures 3 and 4) was intermittently resolvable from background noise caused by high humidity conditions characteristic of tropical environments.
Derived volatile ratios (in the form of x/SO 2 ratios; where x is either H 2 O, CO 2 , H 2 S, or H 2 ) for Mount Garet were obtained in measurement windows where strong positive co-variations (R 2 ≥ 0.70) were observed between SO 2 and the other volatile species measured (Figure 3). Interference-corrected H 2 S/SO 2 ratio and H 2 /SO 2 ratios (0.13 ± 0.01 and 0.03 ± 0.01, respectively) remained constant throughout the entire measurement period. Likewise, CO 2 /SO 2 ratios varied narrowly between 0.65 and 1.07 (avg. 0.87 ± 0.24). Due to variable mixing between atmospheric H 2 O and the magmatic plume, H 2 O/SO 2 measurements were more variable, ranging from~40 to~56 (avg. 47.2 ± 5.7; Tables 1 and 2). Similarly, H 2 O/CO 2 ratios also showed wide variations across the entire 4-h window, reaching values as low as~40 and as high as~70. However, the best correlation (R 2 = 0.75) was found at the upper limit of that range, yielding a H 2 O/CO 2 ratio of~69 (Figure 4). The volcanic plume composition of Mount Garet was estimated at 95.9 mol% H 2 O, 1.8 mol% CO 2 , 2.0 mol% SO 2 , 0.3 mol% H 2 S, and 0.06 mol% H 2 (uncertainties reported in Table 2 are expressed as the standard error of the regression analysis and subsequent error propagation).

Discussion
Our observations constrain the gas plume composition at Gaua as hydrous (~95.9 mol% H 2 O), with SO 2 dominating over H 2 S (H 2 S/SO 2 of 0.13 ± 0.01), and with a H 2 /H 2 O ratio of~0.001.
Therefore, by using gas-phase H 2 /H 2 O (~0.001) and H 2 S/SO 2 (~8.4) redox couples and using published thermodynamic constants [52,53], we can estimate the oxygen fugacity (fO 2 ) of Mount Garet's volcanic gases (e.g., Moussallam et al., 2019). Equations (1) and (2) also provide the means to estimate the gas-melt equilibrium temperature: The value of fH 2 O used in (2) is 0.959 given that, at~1 bar the (Ptot × nH 2 O)/ntot = [(1 bar) (0.959ntot)]/(ntot) = 0.959 bar (where fugacity of a gas is equal to its partial pressure and that P(H 2 O) = P is the pressure in bar and ni the amount of species i in mol%) [54].
Solution yields an equilibrium temperature of~641 • C and a logfO 2 equivalent to ∆QFM = +2.0 where QFM refers to the quartz-fayalite-magnetite buffer [55], and where ∆QFM = logfO 2 −logfO 2 of QFM at the corresponding temperature range estimated above. This solution plots well within the global trend highlighted in [22].
As such, an equilibrium temperature of~641 • C is higher than the temperature at which sulfur partitioning into the liquid hydrothermal system or gas scrubbing reactions in a shallow subsurface environment may occur [56,57]. Therefore, the absence of contamination of Gaua's magmatic gas allows for comparison with other high-temperature arc gases worldwide ( Figure 5) [17,22,28]. Being the first for Gaua, our results contribute to our understating of volcanic gas signature and fluxes along the Vanuatu arc, as discussed below.  ). Ambrym [33,34] and Yasur [31] are also plotted for comparison, as well as Manam (Papua New Guinea) [37]. For within-plot field delimitations (e.g., hydrothermal gases and arc volcanism) see [28].

Along-Arc and Inter-Arc Volcanic Gas CO2/ST Variations
Recent global volcanic gas compilations have revealed systematic inter-arc and along-arc variations in gas CO2/ST ratios [15,21] which are thought to be mainly governed by variations in mineralogy and chemistry of sediments (especially their C load) subducted at the corresponding trenches [6,7]. A better understanding of these global correlations between arc inputs (sediments) and  [33,34] and Yasur [31] are also plotted for comparison, as well as Manam (Papua New Guinea) [37]. For within-plot field delimitations (e.g., hydrothermal gases and arc volcanism) see [28].

Along-Arc and Inter-Arc Volcanic Gas CO 2 /S T Variations
Recent global volcanic gas compilations have revealed systematic inter-arc and along-arc variations in gas CO 2 /S T ratios [15,21] which are thought to be mainly governed by variations in mineralogy and chemistry of sediments (especially their C load) subducted at the corresponding trenches [6,7]. A better understanding of these global correlations between arc inputs (sediments) and outputs (volcanic gases) requires extending volcanic gas measurements to remote, poorly studied arc regions, such as Melanesia. Indeed, as noted by Aiuppa et al. [15,21] several of the top volcanic arc emitters have yet to be measured for their CO 2 /S T ratios, due to their remote location and/or high level of volcanic activity.
The results we report here add new data to this global dataset to allow a more complete characterization of the source, fate and cycling of volatiles, especially C, in subduction zones. In this section, we focus the discussion of our results within the context of recently collected gas data for other Melanesian volcanoes.
Aiuppa et al. [15,21] noted that this CO 2 -poor volcanic gas composition for the Vanuatu arc, as inferred from observations at Ambrym and Yasur, is somewhat at odds with the relatively high C content of sediment in the subducting slab (bulk sediments in the Vanuatu trench contain up to 4 wt.% CO 2 ) [6]. In other arc sectors where C is subducted via sediments (e.g., Central and northern Southern America) [7], volcanic gases are typically more C-rich, and display characteristics of those volcanoes in the Group 2 category (2 < CO 2 /S T < 4) [15,21,50,58]. In contrast, the Vanuatu gas compositions measured thus far resemble more closely those of subduction zones of the NW Pacific (the Kuril-Kamchatka and Marianas-Japan arc segments) [15], all of which exhibit low C content in subducting sediments [6].
Our results here provide a first insight into the CO 2 /S T signature of volcanic gases in the northern part of the Vanuatu arc. Our average volcanic gas CO 2 /S T estimate at Gaua is~0.9, for measurements performed in very dense plume conditions (up to >50 ppmv SO 2 ). The 2018 Gaua gas composition also plots at the low-CO 2 end of the population of volcanic arc gases worldwide ( Figure 5). Notably, a similarly C-poor gas measurement was recently obtained further to the north at Manam volcano, in Papua New Guinea, for which Liu et al. [37] quoted a CO 2 /S T ratio of~1.
The relatively high C content of sediments at the Vanuatu trench led Aiuppa et al., [21] to predict a CO 2 /S T ratio of~2.5 for Gaua, well in excess of our 2018 measured ratio of~0.9. Thus, the C-rich nature of the subducted sedimentary input in the region does not appear to translate into a C-rich volcanic gas output, as seen elsewhere (e.g., central and northern Southern America). The possibility that the sedimentary C input into the subduction zone is accreted to the accretionary wedge (rather than being subducted) is inconsistent with the results of Von Huene and Scholl [59], who classified the Vanuatu arc as a non-accreting margin, with effectively no accretionary prism present. Further, seismic reflection data indicate that most of the sediments are indeed subducted [60]. In fact, the same authors argue that, in the region, the addition of a significant subducted sediment component has variably modified the mantle wedge over time. Within the less-surveyed northern Vanuatu region, distinct geochemical patterns can be observed. For instance, lavas from Gaua have lower 143 Nd/ 144 Nd than lavas from Merelava, despite similar Sr-and Pb-isotope compositions. This can be explained in terms of a common Indian-MORB-like mantle source but would indeed require a higher contribution from subducted sediments to Gaua lavas [60]. Ultimately, the mismatch between the high C input (sediments) and the low C output (volcanic gases) may simply reflect a limited C recycling by fluids delivered by the slab, leading to sedimentary C transport deep in the mantle [7].
We caution, however, that our discussion is based on one single, brief campaign at Gaua, while more continuous and prolonged gas records would be required to fully capture the volcano's time-averaged degassing behavior and gas composition. Indeed, gas compositions exhibit both long-term changes in response to changes in volcano behavior (e.g., as persistent degassing activity transitions into unrest and finally eruption) [19], and also more subtle short-term changes related to rapid events of gas segregation and explosive burst. At Yasur volcano, for example, the bulk composition appears to have remained relatively stable since the 2000s, yet with CO 2 /S T ratios spanning between 1.6 and 2.2 [30][31][32]. In addition, short-term (timescales of seconds to minutes) fluctuations in gas compositions between 'passive' (CO 2 /S T~1 .9) and 'active' (CO 2 /S T~2 .9; during strombolian explosions) degassing regimes have also been observed [30,32]. Ultimately, additional gas observations at Gaua and other Vanuatu volcanoes is required for a more robust assessment of C recycling efficiency along the arc.

Implications for the Vanuatu Arc Volatile Budget
Vanuatu hosts two of the largest volcanic gas emission sources globally. Ambrym, a volcanic island known for frequently hosting the lava lakes of Benbow and Marum until its 2018 eruption [61] appears first on the list of subaerial SO 2 emissions measured from space between 2005-2015 [23]. The 10-year averaged SO 2 emission inventory ranks Yasur volcano (located in the island of Tanna; see Figure 1) at number 11 in the list. Notably, Bagana, Tavurvur and Manam, from the nearby archipelago of Papua New Guinea, are all in the top 10. Clearly, the southwest pacific is a major contributor to global volcanic gas release.
Vigorous volcanic SO 2 degassing in Vanuatu had been reported previously by Bani et al. [36]. Using repeated DOAS (differential optical absorption spectroscopy) ground measurements (2004)(2005)(2006)(2007)(2008)(2009)) and satellite OMI data (2004-2011), (Bani et al., 2012) estimated that on average 3 Tg y −1 of volcanic SO 2 are emitted into the atmosphere along the Vanuatu arc, to which Ambrym reportedly contributed to about a third. Such estimates did not include sporadic eruptions and/or extreme passive degassing events, but nonetheless represented about one-fifth of currently estimated global volcanic SO 2 emissions. The study also marked the first ground-based SO 2 flux estimates for Mount Garet volcano. During October and December 2009, seven DOAS traverses yielded a mean SO 2 emission rate of 2959 t d −1 [36]. These measurements were taken during the 2009-2010 eruption that started in September (see cfr. 2 for details) and most probably reflected the high degassing rates registered in the first few months of the eruption [36,44].
In fact, Carn et al. [23] found much lower SO 2 emission rates for the period of 2005-2015 at Gaua. They reported a 10-year maximum of~1418 t d −1 in 2012, and a 10-year average of~434 t d −1 of SO 2 . Mismatches between ground and satellite-based emission inventories may arise from several factors [62]. Especially during the course of the 2009-2010 eruption (OMI satellite data yielded yearly averaged emissions of~490 and 605 t d −1 for 2009 and 2010, respectively) [23], discrepancies may originate from difficulties in accurately measuring SO 2 in proximal ash-laden plumes, like the ones originated, for instance, in the explosive events of November 2009 [44]. Due to the low altitude of Mount Garet (~797 m a.s.l.), it is also possible for some passive degassing emission rates measured via satellite to be underestimated. That is because the detection limit will be lowest for low-latitude volcanoes, which benefit from more satellite observations under optimal conditions (e.g., low solar zenith angles) [23,62].
No SO 2 flux record was obtained during our 2018 survey. We therefore combine the 10 Our inferred CO 2 fluxes are especially relevant because they contribute new data to recent cataloguing efforts [19][20][21]29] aimed at a more robust quantification of the global volcanic CO 2 output. Such global estimates have remained poorly constrained until recently, because technical challenges in volcanic CO 2 measurements have resulted in sparse and incomplete global CO 2 flux catalogues. With a CO 2 flux of~259 t d −1 , Mount Garet ranks as intermediate-size volcanic CO 2 source (the median volcanic CO 2 flux in catalogues [20,29] is~320 t d −1 ). Notably, our measured Gaua CO 2 flux is a factor~3 lower than the CO 2 flux (~745 t d −1 ) predicted by Aiuppa et al. [21]. The latter authors circumvented the lack of direct gas measurements at Gaua (unavailable at the time of their study) by predicting the time-averaged volcanic gas CO 2 /S T ratio from trace element compositions (Ba/La and Sr/Nd ratio) of Gaua lavas and global gas vs. trace element relationships. The mismatch between the direct (this study) and indirect [21] estimates reflect the difference between measured and predicted CO 2 /S T ratio (0.9 vs. 2.5), as discussed above (cfr. 5.1).

Conclusions
Global inventories of volcanic gas data remain sparse and incomplete, especially for remotely located volcanic centers such as Mount Garet, in Gaua. Less than 60 of the several hundred currently degassing Holocene volcanoes have been characterized geochemically for their gas emissions and less than a few (<10) volcanoes have permanent in-situ instrumentation operating in real-time.
In this study we presented volcanic gas composition data from Mount Garet volcano. Our survey provides the first characterization of plume chemistry for this volcano. Although results reported here are punctual, they set unprecedented information on the composition and flux of volcanic gases at Mount Garet during a quiescent activity period (December 2018). By combining our results with satellite-based SO 2 flux measurements (average of 434 t d −1 over the 2005-2015 period) we estimated H 2 O, CO 2 , H 2 S, and H 2 outputs of~5758 t d −1 ,~259 t d −1 ,~30 t d −1 , and~0.5 t d −1 . These results rank Mount Garet as an intermediate-size volcanic CO 2 point source, well below the intense CO 2 emissions derived for nearby Ambrym (until its 2018 eruption) and Yasur volcanoes further south in the same Vanuatu arc. We caution that our observations are only likely to be reprehensive of Mt. Garet's quiescent emissions, while much higher CO 2 and total volatile fluxes are likely to be associated with eruptive phases, as for instance supported by the high SO 2 flux emissions measured during the 2009-2010 period of unrest (for which no gas composition is unfortunately available). Our results thus confirm the need to better constrain the temporal variability of gas compositions through periodic and/or continuous surveys. Resolving the volatile output for both quiescent and unrest phases, along with newly reported volcanic gas plume compositions, would foster the ability to interpret volcanic gas monitoring signals and perform hazard assessments in the future.
Our volcanic gas survey has also revealed an especially C-poor composition (CO 2 /ST~0.9) for Mount Garet volcano, well below the compositions derived for nearby Ambrym and Yasur (CO 2 /ST~1.5 to 1.6), and un-matching with the previously predicted volcanic gas compositions for the same volcano (CO 2 /ST~2.5). The contrast between the relatively high C content of sediments at the Vanuatu trench and the CO 2 -poor nature of volcanic gases measured at Mount Garet during our 2018 field campaign is intriguing. In other volcanic arcs where C is subducted through sediments (northern segments of the Andean Volcanic Belt, for example), volcanic gases typically display higher CO 2 /ST ratios (~2-4). Furthermore, despite the fact that recent gas studies at other persistently degassing Melanesian volcanoes have reported similar in-plume volatile signatures, more volcanic gas surveys are needed to accurately assess the apparent mismatch between the high C input (sediments) and the low C output (volcanic gases). That will ultimately allow for more robust arc-scale interpretations of the nature of gas emissions from Southwest Pacific volcanism, a region of much relevance to arc CO 2 budgets globally, and their implications for our overall understanding of global geochemical cycles and atmospheric evolution over time.