Replenishment of volatile-rich mafic magma into a ...

Bull Volcanol (2014) 76:872 DOI 10.1007/s00445-014-0872-0

RESEARCH ARTICLE

Replenishment of volatile-rich mafic magma into a degassed chamber drives mixing and eruption of Tungurahua volcano Madison L. Myers & Dennis J. Geist & Michael C. Rowe & Karen S. Harpp & Paul J. Wallace & Josef Dufek

Received: 19 March 2014 / Accepted: 10 September 2014 # Springer-Verlag Berlin Heidelberg 2014

Abstract In July and August of 2006 and May of 2010, Tungurahua volcano, Ecuador, produced pyroclastic flowforming eruptions, representing increased explosivity compared to the Strombolian events that characterized its behavior since its renewal in 1999. Volatiles (H2O, CO2, S, Cl) and major elements were analyzed in 35 melt inclusions hosted in olivine and pyroxene phenocrysts in tephra from both events to reconstruct the pre-eruptive magmatic conditions and mechanisms that led to these more explosive episodes. Melt inclusion composition paired with host phenocryst zonation indicate mixing of two distinct magmas: a volatile-rich (∼4.0 wt% H2O and ∼1,800 ppm S) basaltic andesite containing olivine phenocrysts and a degassed (∼1.0 wt% H2O and 100–500 ppm S) andesite with plagioclase and pyroxene phenocrysts that contain andesitic to dacitic melt inclusions. We attribute the lower volatile concentrations in the evolved melt inclusions to degassing that occurred during residence in Editorial responsibility: J.E. Gardner M. L. Myers (*) : P. J. Wallace Department of Geological Sciences, University of Oregon, Eugene, OR 97403, USA e-mail: [email protected] D. J. Geist : K. S. Harpp Department of Geological Sciences, University of Idaho, 3022, Moscow, ID 83843, USA M. C. Rowe School of Environment, University of Auckland, Auckland 1142, New Zealand K. S. Harpp Department of Geology, Colgate University, Hamilton, NY 13346, USA J. Dufek School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA 30332, USA

a shallow reservoir, where fractional crystallization led to the production of dacitic melt. Our melt inclusion data confirm the hypothesis made on the basis of phenocryst zoning profiles (J Volcanol Geotherm Res 199:69–84, 2011) that the intrusion of a volatile-rich basaltic andesite into a more evolved chamber and subsequent mixing led to explosive eruption in 2006. Melt inclusions from the 2006 and 2010 eruptive products have comparable volatile and major element compositions. High H2O concentrations in melt inclusions from 2010 olivine indicate little diffusive loss from the melt inclusions following mixing with the degassed andesitic reservoir, which requires that the 2010 eruption be the result of a new recharge event and not remobilization of the 2006 hybrid. Keywords Tungurahua . Melt inclusions . Recharge . Magma mixing

Introduction Understanding the processes that control the explosivity of a volcanic system, especially those that control the change from a more effusive to explosive eruption, is crucial for hazards monitoring. It is well known that the dynamics and eruptive strength of all volcanic eruptions are strongly controlled by the concentrations of water (H2O), and to a lesser extent carbon dioxide (CO2), within the magma (Wilson et al. 1980; Sparks et al. 1997; Davidson and Kamenetsky 2007). Thus, volatile concentrations are important for understanding the increased activity associated with more explosive episodes. Melt inclusions provide a way of measuring the preeruptive volatile concentrations within deeply stored magma, as after entrapment, melt inclusions are largely isolated from the host melt (e.g., Anderson 1975; Roedder 1984; Wallace 2005; Métrich and Wallace 2008; Blundy et al. 2010, and references therein). Typically, melt inclusions preserve

872, Page 2 of 17

volatile concentrations that are higher than those of the eventually erupted magma (Wallace 2005; Métrich and Wallace 2008). Melt inclusions have also been found to record the progressive degassing of magma over the course of an eruption, which in turn can govern the explosivity (Anderson et al. 1989; Roggensack et al. 1997; Oppenheimer et al. 2003; Sparks 2003; Blundy and Cashman 2005). Furthermore, comparisons of crystal chemistry and the compositions of glassy melt inclusions can provide information on the presence of different magma batches, the timing of degassing and mixing, and other magmatic evolutionary processes that occur before eruption (Danyushevsky et al. 2000; Blundy and Cashman 2005; Nichols and Wysoczanski 2007; Johnson et al. 2011). This paper expands on the petrologic data presented by Samaniego et al. (2011) for rocks recently erupted from Tungurahua volcano by interpreting compositional data from melt inclusions in juvenile scoria bombs from the 2006 and 2010 pyroclastic density current deposits (Hall et al. 2013). Samaniego et al. (2011) used petrological evidence to show that the 2006 eruptions were the result of mafic recharge into a shallow, more evolved chamber, which drove higher energy eruptions. In the present study, melt inclusions hosted in olivine, orthopyroxene, and clinopyroxene were analyzed for H2O, CO2, sulfur (S), and chlorine (Cl), along with major elements, with the purpose of reconstructing the processes that occurred in Tungurahua’s magma chamber prior to these more explosive eruptions and thereby testing the recharge hypothesis. Our data support and expand upon the Samaniego et al. (2011) interpretation by quantifying the volatile concentrations associated with the recharged mafic magma and shallowly stored andesitic magma erupted in 2006. We suggest that this model can also be applied to the 2010 event and adds to growing evidence from melt inclusion studies for the importance of recharge and mixing of magmas in the evolution of andesitic volcanoes and the role of recharge as an eruption trigger (e.g., Carrasco-Núñez and Rose 1995; Murphy et al. 2000; Edmonds et al. 2001; Roberge et al. 2009).

Bull Volcanol (2014) 76:872

to be a concern because of its steep relief (∼3,000 m), posing a threat to the population of Baños, a town of ∼18,000 inhabitants located on the lower northern flanks of the volcano (Arellano et al. 2008). This threat is magnified by the structure of Tungurahua’s crater, which causes pyroclastic flows to descend the north and west flanks of the edifice (Kelfoun et al. 2009). Tungurahua volcano has been monitored since 1988 by the Instituto Geofisico de la Escuela Politécnica Nacional using seismological, geochemical, thermal, geodetic, acoustic, and other observational techniques (Fee et al. 2010), and a comprehensive hazard map was published by Hall et al. (2002). Historic eruptions and 1999 renewal Over historic times, Tungurahua has experienced an eruptive cycle almost every century, each lasting about a decade. The current cycle has persisted beyond this predicted interval, making understanding the causes of its more explosive phases and continued activity crucial for ongoing hazards monitoring. Historically, activity tends to be dominated by Strombolian and Vulcanian eruptions, although Plinian eruptions are also documented (Hall et al. 1999). The Strombolian and Vulcanian events are characterized by regional tephra fallout, pyroclastic and debris flows, and blocky lava flows (Samaniego et al. 2011), but pyroclastic products dominate each cycle (Hall et al. 1999; Arellano et al. 2008). Seismic and infrasonic stations were installed on the flanks of the volcano after activity recommenced in 1999 (Arellano et al. 2008). Arellano et al. (2008) found that passive degassing prior to 2006 accounted for ∼90 % of the observed SO2 emissions at Tungurahua, most of which occurs during the frequent effusive eruptions. During 1999 and 2005, low eruption rates suggest a moderate magma ascent rate from the reservoir and through the conduit system (Samaniego et al. 2011). Pairing gas compositions and seismicity indicate that volcanic activity during these opening eruptive phases originated from a conduit system with a depth range between 1 and 5 km below the summit (Molina et al. 2005; Arellano et al. 2008).

Regional setting and eruptive history

The 2006 and 2010 eruptions

Tungurahua (elevation 5,023 m) is located within the eastern Cordillera of the Ecuadorian Andes (1.467° S, 78.442° W), 120 km SE of Quito (Fig. 1). In 1999, Tungurahua renewed volcanic activity after 80 years of quiescence. Eruptions during the first 7 years were dominated by Strombolian activity, with frequent ash and tephra emissions (Arellano et al. 2008). In 2006, Tungurahua experienced an increase in eruptive strength, resulting in several pyroclastic-flow-forming eruptions (Samaniego et al. 2011). These more energetic eruptions also occurred in 2008, 2010, 2011, 2012, and 2013. Tungurahua remains active as of August 2014 and continues

In early April 2006, long-period seismic activity was detected 5 to 15 km below the summit, followed by an increase in SO2 discharge (Arellano et al. 2008; Carn et al. 2008; Fee et al. 2010; Samaniego et al. 2011). From May 10–16, 2006, several explosive eruptions were recorded, with ash columns reaching 19 km above the volcano. Cataclysmic pyroclastic-flow-forming eruptions occurred on July 14 and August 16–17 of 2006, with flows reaching the base of the volcano (Samaniego et al. 2011; Hall et al. 2013). Following the emplacement of pyroclastic flows, lavas erupted (Hall et al. 2013). The cumulative eruptive volume for the July 14 event totals ∼6×106 m3 of pyroclastic and fallout


Page 3 of 17, 872

Fig. 1 Aerial image of Tungurahua volcano (Google Earth); stars indicate sample locations. The town of Baños, Ecuador, is located along the northern flanks of Tungurahua, about 8 km from the main vent. The dark areas radiating from the summit are pyroclastic density deposits, mostly

from the 2006 eruption. Location of Tungurahua volcano (bottom right inset), where triangles depict some of Ecuador’s most well-known volcanoes. Dashed lines represent the Western (WC) and Eastern Cordilleras (EC), with Tungurahua falling along the EC

material, about 12 % the volume of the August 16–17 eruption (∼50–55×106 m3, Troncoso et al. 2006; Kelfoun et al. 2009; Eychenne et al. 2012; Hall et al. 2013). Between 2006 and 2010, activity returned to the lowerenergy Strombolian eruptions, similar to those observed from 1999 to 2006, although one major eruption occurred in 2008, producing 1.5×106 m3 of material and 17.5 cm of uplift on the upper western flank (Biggs et al. 2010). On May 26 and 28, 2010, another strong eruption sequence occurred, producing 1–3 km long pyroclastic flows on the N, NW, W, and SW flanks of Tungurahua. Although the 2010 eruption produced pyroclastic flows, it was not nearly as voluminous as the 2006 climactic events.

change from more explosive to effusive eruptive behavior (Hall et al. 1999). Samaniego et al. (2011) published a comprehensive petrologic study of the eruptive material produced between 1999 and 2006, focusing on the July and August 2006 eruptions. All of the 1999–2005 rocks analyzed, including the lava, pyroclastic flow deposits, and tephra, have andesitic compositions (58–59 wt% SiO2) and contain plagioclase (5–15 vol.%), clinopyroxene (2–4 vol.%), orthopyroxene (2–4 vol.%), and magnetite ± trace olivine as phenocrysts. This is interpreted as indicating the existence of a homogeneous andesitic reservoir over the entire eruptive cycle (Samaniego et al. 2011). Samaniego et al. (2011) noted the presence of resorption textures in plagioclase and pyroxene phenocrysts from the climactic 2006 eruptions, along with 20–100 μm thick anorthite (An) and Mg-rich rims. Pairing this information with deep (8–10 km) seismicity and increased gas emissions, Samaniego et al. (2011) conclude that the explosive eruptions in 2006 were caused by the intrusion of a mafic magma into an andesitic reservoir, with subsequent mixing and pressurization

Petrology Most of Tungurahua’s historic eruptive cycles have begun with dacitic compositions (64–66 wt% SiO2) and ended with the eruption of andesite (56–59 wt% SiO2), corresponding to a

872, Page 4 of 17

of the chamber. Inputs of mafic magma are commonly observed as eruptive triggers in intermediate systems (CarrascoNúñez and Rose 1995; Murphy et al. 2000; Edmonds et al. 2001; Roberge et al. 2009 and references therein), making recharge a plausible hypothesis for the cause of these suddenly more explosive eruptions in 2006. No chemical or petrographic differences were observed in juvenile samples from the July 14 and August 16–17, 2006, eruptive products (Samaniego et al. 2011); however, the August event produced a tephra layer that contained a trace amount of dacitic pumice (50 μm), making them preferred for this study due to analytical constraints. Plagioclase-hosted inclusions from these eruptions tend to contain daughter crystals and are typically smaller (∼20 μm) compared to the olivine- and pyroxene-hosted MIs. Nearly all of the analyzed MIs are vapor and shrinkage bubble free, suggesting that volatile loss from entrapment concentrations is minimal (Severs et al. 2007; Gaetani et al. 2012). Individual pyroxene and olivine grains were doubly polished to a 1 μm finish for analysis by Fourier transform infrared spectroscopy (FTIR) at the University of Oregon. Infrared spectra were collected between 6,000 and


1,000 cm−1, where the main peaks of interest for this study were the total OH peak at 3,550 cm−1 and the carbonate doublet at 1,435 and 1,515 cm−1. The concentration of CO2 was calculated using an average height of the two peaks in the carbonate doublet (Fine and Stolper 1985). At more evolved compositions, CO2 is stored as molecular CO2 (2,350 cm−1), however this peak was absent from all spectra. Replicate spectra were acquired for each MI as well as a spectrum for the host mineral (see below). Thickness was measured using both a digital micrometer (±2 μm) and the reflectance interference fringe method described by Wysoczanski and Tani (2006). One difficulty we encountered was that some MIs could not be doubly exposed, either because of fracturing of the mineral host or to preserve two MIs within the same crystal. To correct for interference by the host mineral, a correction ratio was applied by subtracting the host mineral spectrum from the MI spectrum, using the method described by Nichols and Wysoczanski (2007). The reproducibility associated with this correction was found to be excellent, with the standard deviation before and after double exposure of 0.08 wt% (Nichols and Wysoczanski 2007). Although this method was created to deal with olivine interferences, the same principles were applied to correct for pyroxene interferences, using multiple wavelength overlaps to constrain variability. Absorbances were converted to H2O concentrations using the Beer-Lambert law, where ci =Mi ×A/ρ×d×ε. In this, ci is the concentration of the absorbing species (in wt%), Mi is the molecular weight of the species (g mol−1), A is the absorbance height of the relevant vibration band, ρ is the sample density (g L−1), d is the thickness of the wafer analyzed (cm), and ε is the molar absorption coefficient (L mol−1 cm−1). Total H2O concentration was calculated using the total OH− peak at 3,550 cm−1. We used an ε value of 70±6.9 L mol−1 cm−1 for the pyroxene-hosted andesitic to dacitic MIs, taken from King et al. (2002), whereas the more mafic andesitic MIs in olivine require an ε value of 63±3 L mol−1 cm−1 (Dixon et al. 1988). An ε value of ∼350 L mol−1 cm−1 was used for the CO2 doublet, calculated using the MI composition (Dixon and Pan 1995). Peak heights were calculated using a straight-line background correction (Dixon et al. 1995). Glass density was calculated from the major element compositions as described by Luhr (2001). Combined uncertainties in density, absorption coefficient, and thickness cause uncertainty to be approximately 10 % (Dixon et al. 1988), where most of this error is due to the thickness measurement. Based on these uncertainties, average 1 standard deviation uncertainties for H2O are ±0.25 wt% and for CO2 ±20 ppm. After FTIR analysis, the crystal wafers were mounted and analyzed for major elements (Mg, Fe, Si, Mn, Al, Ca, Na, K, P, Ti, Ni, Cr) using the JEOL JXA-8500 F electron microprobe at Washington State University. Analysis was conducted on MIs, major mineral phases, including core to rim transects,


and the matrix glass from several locations within each sample. For host crystal analysis, beam current varied between 30 and 50 nA, using a beam diameter of ≤7 μm. Beam conditions were set with an accelerating voltage of 15 kV, a beam current of 30 nA, and a beam diameter of 5–7 μm for all glass analyses. To minimize the effects of Na migration in the glasses, Na was among one of the first elements analyzed and a zero-time intercept correction was applied. Matrix glass and MIs were also analyzed for S and Cl. All concentrations were above detection limits, although the matrix glass S concentrations approach these values. Reproducibility and accuracy of melt inclusion major element, S, and Cl abundances is based on repeated analysis of VG-2 standard glass (Table 2). Error analysis (reported as % relative standard deviation) indicates that melt inclusion compositions are generally reproducible to better than 2 %, with higher RSD for S (3.4 %), Cl (7.8 %), Na2O (6 %), P2O5 (12 %), and MnO (21 %). Although none of the plagioclase-hosted MIs were prepared for analysis by FTIR because of their small size, glassy MIs in plagioclase in thin section were also analyzed by EMPA. It should be noted however that plagioclase-hosted MI were mostly contained within larger sieve textured phenocrysts and thus represent only one phase of plagioclase crystallization.

Results Whole-rock and phenocryst compositions Whole-rock major element compositions of all three scoria samples are within analytical uncertainty, with SiO2 =58.40± 0.05 wt% (Table 1). These compositions fall within the range of the andesites erupted during Strombolian events from Tungurahua between 1999 and 2005 (58–59 wt% SiO2) and are close to the range of andesites from previous historic eruptions (56.3–58.2 wt% SiO2, Samaniego et al. 2011). Plagioclase occurs in three textural populations in samples from 2006 to 2010: euhedral phenocrysts, strongly sieved phenocrysts, and microlites. The phenocryst textures described by Samaniego et al. (2011) are also present in our 2006 and 2010 samples, including resorption embayments to differing degrees in all phenocrysts, and strong, stepwise reverse zonation within 20–100 μm of the rim in plagioclase (average of An64±3 core and An72±4 rim) and clinopyroxene phenocrysts (average of Mg#=71±3 core and Mg#=77±4 rim, where Mg#=100×MgO/(MgO+FeOT). These disequilibrium textures are independent of phenocryst size. One noteworthy difference in the 2010 samples is their finer grain size, due to the smaller population of the 0.6–2.0-mm size fraction. Clinopyroxene phenocrysts are mostly euhedral, whereas orthopyroxene tends to be embayed. A few 2010 orthopyroxene crystals have slight normal zoning (∼1 Mg#

Page 5 of 17, 872 Table 1 Whole-rock major elemental (wt%) analyses from 2006 to 2010 eruptive products measured in this study Sample name Sample type Eruption year

TU10-02 PJ Scoriae/bomb 2006

TU10-06 PJ Scoriae/bomb 2006

SiO2 TiO2 Al2O3 FeO* MnO MgO CaO Na2O K2O P2O5 Total Ni Cr Sc

58.37 0.90 16.74 6.80 0.11 4.37 6.84 3.88 1.73 0.24 99.56 45.7 100 18.2

58.45 0.91 16.70 6.76 0.11 4.31 6.85 3.91 1.76 0.25 98.64 44.1 101.2 16.3

V Ba Rb Sr Zr Y Nb Ga Cu Zn Pb La Ce Th Nd U

173.3 755.8 47.1 596.1 144.1 17.3 7.2 19.4 55.8 87.9 10.5 21.4 37.6 4.4 22.1 2.5

175.4 753.4 48 581 144.7 17.1 6.2 18.3 57.6 85.9 10.8 19.2 40.8 5.3 20.6 2.2

TU10-2010 PJ Scoriae/bomb 2010 58.45 0.89 16.92 6.61 0.11 4.23 6.86 3.95 1.74 0.24 99.05 39.9 90.4 15.8 168.4 754 47.5 606.2 138.9 16.2 6.9 19.4 55.1 85 10.6 16.8 39 5 21 3

unit), but otherwise orthopyroxene is unzoned (Fig. 2). Olivine is normally zoned in both the 2006 and 2010 samples, but to different extents. 2006 olivine phenocrysts have cores of Fo80±2 that sharply transition within the outer 20–60 μm to Fo75±2. The 2010 olivine cores have core compositions of Fo79±0.5 dropping to Fo77±2 within 20 μm of the rim. Melt inclusion compositions The major element compositions of the 52 analyzed MIs (26 each from 2006 to 2010) range from basaltic andesite to dacite (Table 2). Melt inclusion compositions are normalized on an anhydrous basis to allow for direct comparison between inclusions. Measured analytical totals (major element oxides + S, Cl, H2O, and CO2), range from 96 to 100 wt% (98.1±

872, Page 6 of 17


85 83 81

Mg# Core

79

Olivine OPX CPX

77 75

Normal Zoning

Reverse Zoning

73 71 69 67 65

Mg# Rim

Fig. 2 The Mg# (where Mg#=100×MgO/(MgO+FeOT) in mol.%) of rim compositions plotted against the Mg# of the core for individual crystals illustrates zonal patterns in phenocrysts, where a 1:1 ratio represents unzoned crystals. Symbol shapes represent different minerals, whereas shading represents either the 2006 or 2010 eruption. Most clinopyroxenes are reversely zoned, whereas olivine is either unzoned or weakly normally zoned. Orthopyroxene falls mostly along the 1:1 line. No significant differences exist between 2006 and 2010 samples

1.1 wt% on average). Calibration of major element oxide analysis by EMPA on anhydrous standards likely contributes significantly to low analyzed totals. Major element compositions are found to relate strongly to the host phase, where clinopyroxene contains the most evolved MIs, olivine contains the most mafic MIs, and inclusions hosted in plagioclase and orthopyroxene fall in-between (Fig. 3). For example, MIs in pyroxene from the 2006 eruption range to higher silica concentrations (60–67 wt%) than those hosted within plagioclase (61–63 wt%) or olivine (54–61 wt%) and are also more evolved than the bulk rock (58 wt%). The 2010 eruptive products produce similar ranges, but extend to slightly more evolved compositions (Fig. 3). The matrix glass composition in all samples is homogeneous relative to the variation in the MIs, with SiO2 concentrations ranging from 60 to 62 wt%. Most of the MIs have compositions within the range of wholerock data from historical eruptions (Hall et al. 1999; Samaniego et al. 2011), with the exception of TiO2 and FeO (Fig. 3): both concentrations become increasingly offset (TiO2 ∼1.0 wt% and FeO ∼3.0 wt%) from the historical WR trend with increasing silica concentration. H2O, CO2, S, Cl Of the 35 MI measured for H2O and CO2 concentrations, 16 are from 2006 and 19 from 2010 samples (Table 2). Water concentrations of 2006 pyroxene-hosted MIs range from 0.4 to 2.4 wt%, a wider range than the tight clustering of 0.8– 1.6 wt% for MIs in the 2010 pyroxenes, with an average of ∼1.0 wt% for both years. The larger range in 2006 is due to orthopyroxene-hosted MIs containing greater H2O concentrations than those hosted in clinopyroxene, whereas in 2010

these two H2O ranges are indistinguishable. Olivine-hosted MIs have H2O concentrations that extend to 3.4 wt% for 2006 samples and 4.1 wt% for 2010, with both years containing a significant spread that extends down to ∼1 wt% H2O (Fig. 4). Only two of the MIs have CO2 concentrations above detection limits (∼50 ppm), both from the same 2006 sample. One is an olivine-hosted MI with 200 ppm CO2 and the other is an orthopyroxene-hosted MI with 60 ppm CO2 that also contains the highest H 2 O concentration found in any pyroxene-hosted MI. The absence of CO2 in all but two MIs coupled with the fact that the melt inclusions do not contain vapor bubbles suggests that degassing of CO2 occurred from ascending magmas at greater depths, before the melt inclusions were trapped (Luhr 2001; Métrich and Wallace 2008). Inclusions from both the 2006 and 2010 samples fall within two distinct populations on the basis of S concentrations (Fig. 4). One population contains 1,000–1,800 ppm S and is solely found in the more mafic olivine-hosted MIs (olivine host Mg# 78–82). The second group contains 50–500 ppm S and is hosted within more evolved olivine- (Mg# 73–77), pyroxene-, and plagioclase-hosted inclusions. No sulfur contents fall between 500 and 1,000 ppm (Fig. 4). The concentration of Cl increases with differentiation in all MIs. In both years, the matrix glasses contain the lowest concentrations of Cl, as well as S, falling slightly below the overall trend of the MIs (Figs. 4 and 5). This suggests late-stage degassing of Cl and S from the system, most likely during eruption (Métrich and Wallace 2008).

Discussion The majority of olivine- and pyroxene-hosted MIs are in MgO-FeO equilibrium with their host (Kd =(MgO/FeO)liq / (MgO/FeO) xl, where Kd = 0.26 ± 0.03 for clinopyroxene (Putirka 2008) and 0.26 ± 0.06 for orthopyroxene (von Seckendorff and O’Neill 1993). For olivine-hosted MIs, Kd depends on melt composition (Table 2), where Kd =0.36±.02 for more evolved olivine-hosted inclusions (Fo≤77) and 0.30 ±.02 for the mafic population (Fo>78, Ford et al. 1983). Olivine-hosted MIs that are out of equilibrium with their host were corrected by either adding or subtracting host compositions to the original melt composition (Danyushevsky et al. 2000; Table 2); no inclusion required more than ∼5 % modification. Most orthopyroxene-hosted MIs are also within the range of published values for Kd. Kd corrections were not applied to clinopyroxene- and plagioclase-hosted inclusions, as these calculations are complicated by the complexity of the host composition, meaning that some of the scatter observed within these MIs could be attributed to uncorrected postentrapment modification.

Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol Ol OPX OPX OPX OPX OPX OPX CPX CPX CPX CPX CPX CPX CPX CPX Plag Plag Plag Plag Plag

2006 sample names Tu10-02 #2 Tu10-02 #5 Olivine06 #4 Olivine06 #9 Olivine06 #10, MI1 Olivine06 #10, MI2 Olivine06 #3 Tu10-06 #1 Tu10-02 #4 Olivine06 #1 Olivine06 #2, MI1 Olivine06 #2, MI2 Tu10-02 #102 Tu10-02 #105a Tu10-02 #106 Tu10-02 #106

OPX Box 28 #1 OPX Box 1 #1 Tu10-02 #107 Tu10-06 #102 MI 1 Tu10-06 #102 MI 2 Tu10-02 #1

Tu10-06 #1, MI1 Tu10-06 #4 Tu10-02 #2 CPX Box 28 #2 CPX Box 87 #1 CPX Box 40 #1 Tu10-06 #1, MI2 Plag Box 68 #1 Plag Box 90 #1 Plag Box 106 #1 Plag Box 5 #1 Plag Box 1 #2

Host

GM GM GM TS TS TS GM TS TS TS TS TS

TS TS GM GM GM GM

GM GM GM GM GM GM GM GM GM GM GM GM GM GM GM GM

65.72 66.68 65.00 66.43 61.83 62.87 – 61.94 62.03 62.08 62.94 61.25

64.36 61.56 – – – 64.74

55.20 54.91 54.08 56.61 55.97 56.84 54.62 – 61.18 60.19 61.18 59.86 – – – 59.85

Mount SiO2

1.30 1.27 1.32 0.96 1.31 1.23 – 1.24 1.65 1.16 1.04 1.29

1.07 1.32 – – – 1.30

1.01 1.02 1.00 1.00 1.04 1.06 1.05 – 1.18 1.40 1.28 1.27 – – – 1.28

TiO2

15.17 15.34 15.48 17.07 17.00 16.43 – 16.51 13.71 15.98 15.92 15.53

16.48 15.79 – – – 15.83

19.36 19.24 20.23 19.03 19.05 19.73 19.67 – 16.64 16.49 16.95 16.89 – – – 16.50

5.66 5.01 6.15 3.46 6.15 6.58 – 5.97 8.07 6.34 5.97 6.77

4.21 7.12 – – – 5.48

6.50 6.37 7.48 6.19 7.09 6.60 6.47 – 6.70 6.93 6.83 6.92 – – – 7.39

Al2O3 FeOT

0.08 0.09 0.10 0.03 0.07 0.09 – 0.11 0.10 0.08 0.06 0.12

0.07 0.10 – – – 0.09

0.10 0.09 0.13 0.07 0.06 0.08 0.05 – 0.11 0.13 0.15 0.07 – – – 0.09

1.40 1.42 1.59 0.90 1.57 1.31 – 2.06 3.06 2.13 2.10 2.34

1.28 1.79 – – – 1.73

4.25 4.53 4.29 4.47 4.15 3.26 4.53 – 2.58 2.18 2.36 2.44 – – – 2.01

3.41 3.12 3.62 2.33 4.36 4.15 – 4.39 4.26 4.55 3.99 4.32

4.29 4.28 – – – 3.50

8.26 8.72 7.30 7.43 6.95 6.81 8.52 – 5.61 5.50 5.49 6.06 – – – 5.75

3.39 3.49 3.45 4.11 4.71 3.76 – 4.22 4.13 4.34 4.28 4.64

4.84 4.82 – – – 3.53

3.68 3.61 4.38 3.95 4.16 4.34 4.02 – 3.35 4.32 4.28 4.55 – – – 4.30

3.35 3.21 2.84 4.41 2.67 3.14 – 3.20 2.57 3.00 3.38 3.41

3.06 2.84 – – – 3.17

1.38 1.27 0.83 1.05 1.23 1.02 0.86 – 2.31 2.48 1.13 1.63 – – – 2.48

0.51 0.37 0.45 0.30 0.33 0.44 – 0.36 0.41 0.33 0.33 0.33

0.34 0.39 – – – 0.61

0.27 0.23 0.30 0.21 0.29 0.26 0.22 – 0.34 0.38 0.35 0.32 – – – 0.35

MnO MgO CaO Na2O K2O P2O5

97.25 97.26 97.60 96.62 100.29 99.53 – 96.48 98.19 97.02 98.39 99.09

97.75 99.43 – – – 96.37

96.25 96.39 97.58 97.09 95.92 96.56 96.23 – 96.14 98.11 96.29 97.15 – – – 96.74

0.55 0.73 0.45 – – – 0.53 – – – – –

– – 2.43 1.72 1.60 0.41

2.40 1.91 – – – – – 3.42 2.30 – – – 0.62 1.73 1.03 0.87

Analytical total H2O

BDL BDL BDL – – – BDL – – – – –

– – 62 BDL BDL BDL

BDL 203 0 – – – – BDL BDL – – – BDL BDL BDL BDL

CO2

191 164 282 200 119 284 – 163 336 173 153 237

385 285 – – – 193

1,688 1,458 1,774 1,200 1,041 1,136 1,525 1,090 374 122 457 359 – – – 410

S

1,304 1,290 1,244 1,367 857 1,102 – 835 1,175 1,011 956 1,107

1,320 1,111 – – – 1,418

783 755 914 847 858 810 915 618 889 806 942 864 – – – 941

Cl

68.2 69.7 70.9 76.0 76.0 72.9 – – – – – –

73.7 68.0 – – – 70.9

80.2 81.6 78.1 81.8 78.4 78.4 81.2 79.7 76.7 73.3 75.1 75.8 – – – 72.8

NA NA NA NA NA NA NA NA NA NA NA NA

NA NA NA NA NA NA

0.40 −0.20 3.00 1.60 2.00 2.00 4.40 NA −0.20 −1.40 −1.80 −0.40 NA NA NA NA

Mg# Host OCa

Table 2 EMPA anhydrous normalized major element concentrations (wt%) and measured S and Cl (ppm), with H2O (wt%) and CO2 (ppm) by FTIR, separated by year, host, and mounting format

Bull Volcanol (2014) 76:872 Page 7 of 17, 872

GM GM GM GM GM GM GM GM GM GM GM GM GM GM

Ol Ol Ol Ol Ol Ol Ol Ol Ol OPX OPX OPX OPX OPX GM OPX OPX CPX CPX CPX CPX CPX CPX CPX CPX CPX CPX CPX

Tu10-10L #1 OPX Tu10-10L #110, MI2 Tu10-10L #111 Pyroxene B 18

Pyroxene B 18#2 Pyroxene B 18 #3 Pyroxene B 89 #1 Pyroxene B 89 #2 Tu10-10L #103, MI1 Tu10-10L #103, MI2 Tu10-10L #101, MI1 Tu10-10L #102, MI1 Tu10-10L #102, MI2 Tu10-10L #104, MI1

TS TS TS TS GM GM GM GM GM GM

66.10 GM GM TS

TS TS TS

TS TS TS

TiO2

66.59 68.11 67.12 66.92 64.44 64.54 – – – –

1.02 – – 61.67

58.59 56.22 57.31 62.93 60.46 59.83 60.45 60.85 – 65.49 64.25 60.64 64.97 65.28

1.15 0.92 1.10 0.90 1.24 1.11 – – – –

15.29 – – 1.03

1.00 0.90 0.95 1.18 1.13 1.17 1.23 1.21 – 1.07 1.04 1.82 1.20 1.45

60.37 1.28 61.45 1.22 60.89 1.27

61.93 1.28 62.63 1.25 61.63 1.32

Mount SiO2

MG MG MG

MG MG MG

Host

Tu10-02 MG1 Tu10-02 MG2 Tu10-02 MG3 2010 sample names Tu10-10L #108 Tu10-10L #105, MI1 Tu10-10L #105, MI2 Tu10-10 #2 Olivine06 #5 Olivine06 #6, 2 Olivine06 #7 Olivine06 #8 Tu10-10L #107 Tu10-10L #110, MI1 Tu10-10L #113, MI1 Tu10-10L #113, MI2 Tu10-10L #114, MI1 Tu10-10L #114, MI2

2006 MG #2 2006 MG #3 2006 MG #1

Table 2 (continued)

15.23 14.07 14.47 15.13 15.56 15.54 – – – –

5.84 – – 16.71

19.85 18.07 18.13 16.38 16.70 17.96 17.77 16.34 – 15.61 15.26 15.97 15.60 15.43

16.48 16.52 16.50

15.79 15.72 15.94

5.16 5.77 5.54 4.67 6.55 6.60 – – – –

0.08 – – 6.08

4.88 7.65 7.37 5.81 6.07 5.83 6.77 6.61 – 5.97 6.79 7.22 6.19 6.32

6.64 6.42 6.75

6.37 6.01 6.52

Al2O3 FeOT

0.11 0.15 0.11 0.08 0.10 0.08 – – – –

1.43 – – 0.11

0.10 0.12 0.08 0.09 0.11 0.07 0.08 0.11 – 0.09 0.12 0.16 0.11 0.11

0.06 0.04 0.09

0.08 0.08 0.09

1.09 1.23 1.22 1.13 1.82 1.74 – – – –

3.27 – – 1.82

3.31 4.84 4.70 2.30 2.49 2.45 2.64 2.50 – 1.60 1.71 2.47 1.73 1.77

2.63 2.31 2.51

2.19 2.13 2.23

2.58 2.94 2.89 3.06 3.86 3.73 – – – –

3.59 – – 4.05

7.39 8.03 6.87 4.99 5.72 5.57 3.76 5.32 – 3.36 3.63 5.17 3.47 3.22

5.22 4.67 5.02

4.73 4.42 4.78

4.24 4.23 3.94 4.11 3.84 4.05 – – – –

3.04 – – 3.99

3.77 3.22 3.49 3.69 5.13 4.42 5.45 4.67 – 4.45 3.94 4.06 4.04 3.84

4.31 4.49 4.25

4.25 4.33 4.26

3.53 2.42 3.45 3.84 2.07 2.15 – – – –

0.35 – – 3.90

0.82 0.73 0.88 2.37 1.84 2.41 1.54 2.00 – 1.90 2.77 2.17 2.08 2.04

2.66 2.49 2.39

3.05 3.08 2.87

0.32 0.15 0.15 0.16 0.52 0.47 – – – –

98.27 – – 0.64

0.29 0.22 0.21 0.27 0.33 0.28 0.31 0.40 – 0.45 0.49 0.32 0.61 0.54

0.35 0.38 0.34

0.33 0.35 0.38


97.46 98.05 97.72 98.44 95.33 95.82 – – – –

1.57 – – 97.97

96.60 96.27 92.31 97.80 100.29 99.84 97.61 97.63 – 97.10 98.16 96.48 97.18 97.09

99.33 99.50 99.23

98.88 99.46 99.39

BDL BDL BDL BDL – – – – BDL BDL – BDL BDL BDL

– – –

– – –

CO2

1,363 1,337 1,364 488 262 – 427 202 – 186 305 425 265 326

189 179 141

103 71 109

S

949 923 944 1,051 803 834 1,081 1,038 – 1,122 1,357 1,013 1,331 1,365

742 681 698

749 650 735

Cl

– – – – – 1.02 0.55 1.19 0.84 1.03 1.21

– – – – – BDL BDL BDL BDL BDL BDL

413 176 118 171 103 298 220 – – – –

2,237 1,457 989 1,237 1,502 1,366 1,353 – – – –

BDL 217 1,385 68.6 1.21 BDL – – 0.86 BDL – –

1.28 3.87 4.05 2.01 – – – – 1.18 0.96 – 1.26 1.12 1.09

– – –

– – –


74.3 72.3 69.0 72.0 74.1 71.5 70.8 – – – –

NA – –

80.7 79.6 79.8 77.2 77.8 78.2 76.7 76.2 – 68.9 69.8 68.5 71.2 70.9

– – –

– – –

NA NA NA NA NA NA NA NA NA NA NA

NA NA

2.40 5.20 2.80 −0.20 1.00 −0.40 1.60 0.80 NA NA NA NA NA NA

NA NA NA

NA NA NA

Mg# Host OCa

872, Page 8 of 17 Bull Volcanol (2014) 76:872

96 161 109 24 48 96 1,424 3.4 1,416

827 1,028 869 688 750 856 304 7.8 303

– – – – – – – – –

NA NA NA NA NA NA NA NA NA

Melt inclusion populations

Olivine corrections (OC) show how much olivine was added or subtracted from MIs to achieve compositional equilibrium a

99.45

“–” not measured, Ol olivine, OPX orthopyroxene, CPX clinopyroxene, Plag plagioclase, MG matrix glass, BDL below detection limits, NA not applicable

– – – – – – – – – – – – – – – – – – 98.78 97.49 97.61 99.72 99.37 97.50 98.81

3.37 0.33 3.04 0.31 3.07 0.27 3.26 0.30 3.13 0.39 2.97 0.40 0.20 0.20 4.6 12.0 0.19 0.19 4.07 4.51 4.35 4.70 4.89 5.14 11.0 1.9 11.1 1.82 2.07 1.92 2.17 2.25 2.48 6.98 0.7 6.72 Plag Plag Plag MG MG MG Plagioclase B23 #1 Plagioclase B44 #1 Plagioclase B138 #1 2010 matrix glass #2 2010 matrix glass 2010 matrix glass #3 Avg. VG-2 (N=5) VG-2 %RSD Accepted (De Hoog et al. 2001)

TS TS TS TS TS TS

64.15 62.72 63.19 62.01 61.71 60.58 50.10 0.5 50.63

1.07 1.16 1.01 1.28 1.26 1.28 1.87 1.1 1.92

15.83 16.00 16.04 15.77 15.70 15.81 13.92 0.6 13.97

4.70 6.01 5.53 6.40 6.47 6.78 11.72 0.9 11.88

0.09 0.11 0.11 0.10 0.07 0.09 0.19 21.4 0.22

4.57 4.07 4.50 4.01 4.12 4.48 2.63 6.6 2.61

NA NA NA – – – – 883 984 BDL – – 156 – 215 1.14 – – – 97.89 98.08 – 0.22 0.35 – 3.03 2.95 – 4.10 4.71 – 1.95 2.23 CPX Plag Plag

GM TS TS

– – 63.25 0.97 61.88 1.19

– 15.99 16.03

– 5.70 6.22

– 0.05 0.09

– 4.75 4.35

Page 9 of 17, 872

Tu10-10L #109, MI1 Plagioclase B138 #2 Plagioclase B140 #1

Table 2 (continued)

Host

Mount SiO2

TiO2

Al2O3 FeOT



CO2

S

Cl

Mg# Host OCa


Major element compositions of all MIs can be separated into two distinct groups on the basis of zoning characteristics of the host phenocryst and S concentrations. The more mafic inclusions have silica concentrations ranging between 54 and 59 wt% and contain the highest H2O (∼1–4 wt%) and S concentrations (1,000–1,800 ppm). These MIs are preserved solely in olivine phenocrysts, which display normal zoning (Figs. 2 and 5). More evolved MIs have silica concentrations ranging between 60 and 67 wt% contain lower H2O (average ∼1 wt%) and S concentrations (