Int J Earth Sci (Geol Rundsch)

DOI 10.1007/s00531-014-1006-9

ORIGINAL PAPER

Volatile (sulphur and chlorine), major, and trace element

geochemistry of mafic to intermediate tephras from the
Chilean Southern Volcanic Zone (33–43°S)
Heidi Wehrmann • Kaj Hoernle • Guillaume Jacques •

Dieter Garbe-Schönberg • Kai Schumann •

Julia Mahlke • Luis E. Lara

Received: 31 May 2013 / Accepted: 12 February 2014

Ó Springer-Verlag Berlin Heidelberg 2014

Abstract Here we present the first systematic investiga- depleted mantle sources. The size of the volcanic edifices
tion of volatile geochemistry along the Southern Volcanic correlates inversely with Cl abundances in the melt. This
Zone (SVZ) of Chile. Holocene olivine-hosted melt could reflect more extensive degassing during ascent
inclusions in the most mafic tephras sampled from 16 through the complex magma plumbing systems beneath the
volcanoes along the volcanic front of the SVZ between stratovolcanoes or greater dilution during larger degrees of
33°S and 43°S were analysed for pre-eruptive sulphur, melting of more depleted sources, or a combination of
chlorine, and major element contents. These results are these factors. Compared to other subduction zones, the
combined with trace element compositions of the host SVZ melt inclusions exhibit Cl and S abundances in the
whole rocks. The highest fractionation-corrected gas con- same range as most of those from the Central American
tents occur in the least-degassed melt inclusions from small and those from the Marianas arcs.
monogenetic cones of Los Hornitos, Cabeza de Vaca, and
Apagado from both the transitional and the southern-cen- Keywords Chilean Southern Volcanic Zone  Melt
tral SVZ, reaching *3,000 lg/g S and 1,400 lg/g Cl, inclusions  Sulphur  Chlorine  Subduction zone
while the lowest abundances of *1,100 lg/g S and
*600 lg/g Cl were found in the central SVZ at Volcán
Lonquimay, Volcán Llaima, and Volcán Villarrica. Chlo- Introduction
rine co-varies with trace element indicators for the degree
of melting and/or source enrichment, such that the lowest Subduction zones are sites of strong material turnover,
Cl contents are found in high-degree melts from the most where the incoming oceanic lithosphere (consisting of
sediments, seawater-altered and unaltered oceanic crust,
and variably serpentinised mantle) is subducted into the
Electronic supplementary material The online version of this
article (doi:10.1007/s00531-014-1006-9) contains supplementary Earth’s mantle and material is selectively recycled back to
material, which is available to authorized users. the Earth’s surface by arc magmatism. Volatile phases
such as water, carbon dioxide, sulphur, and chlorine are
H. Wehrmann (&)  K. Hoernle  G. Jacques 
released during explosive arc eruptions. Input of gases
K. Schumann  J. Mahlke
SFB 574, GEOMAR Helmholtz Centre for Ocean Research Kiel, such as S, halogens, and CO2 into the stratosphere can
Wischhofstr. 1-3, 24148 Kiel, Germany have a significant impact on the climate on both regional
e-mail: [email protected] to global scales, if the eruptions have high eruption col-
umns and if they erupt large volumes of magma (e.g.
D. Garbe-Schönberg
SFB 574, Institute of Geosciences, Christian-Albrechts- Tabazadeh and Turco 1993; Robock 2000). Pre-eruptive
University of Kiel, Ludewig-Meyn-Str. 10, 24118 Kiel, volatile contents can be preserved in melt inclusions in
Germany rapidly quenched tephras (e.g. Kent 2008). When com-
bined with trace element data, factors controlling melt
L. E. Lara
Sernageomı́n (Servicio Nacional de Geologı́a y Minerı́a), volatile contents, such as degrees of melting and enrich-
Av. Santa Marı́a 0104, Santiago, Chile ment processes can be identified, which is essential to

123
 Int J Earth Sci (Geol Rundsch)

improve our understanding of volatile budgets and the different arc segments, which is then related to dif-
degassing histories. ferences in large-scale tectonic parameters. Another goal
This study is focused on the S, Cl, major, and trace of this paper is to compare these data with a transitional
element geochemistry of an active continental margin (Central America) and an oceanic (Marianas) arc to see
subduction system, the Chilean Southern Volcanic Zone whether nature and thickness of the overlying crust have a
(SVZ). The SVZ is characterised by a high eruption fre- significant control on volatile contents of arc melts.
quency despite magma extrusion rates that are similar to
other arcs (Völker et al. 2011). Large-scale studies of the
oceanic Izu–Bonin–Marianas subduction system (e.g. Geologic setting
Straub and Layne 2003; Shaw et al. 2008) and of the
transitional to oceanic Kamchatka (Portnyagin et al. 2007) The SVZ, which extends from latitude 33°S to 46°S, results
and Central American (CAVA) (Sadofsky et al. 2008; and from subduction of the Nazca Plate beneath the South
references therein) subduction systems have been pub- American Plate at convergence rates of 70–90 mm/year
lished. The Chilean SVZ allows us to explore the volatile (Stern 2004). The SVZ contains more than 60 large vol-
inventory of an active margin that accommodates a conti- canic centres, most of which have been active in the
nental end-member with respect to the thickness of the Holocene, in addition to numerous small cones and satellite
overriding plate’s continental crust. vents. Volcanism ceases north of 33°S, where the sub-
To date, only few melt inclusion investigations have duction angle decreases to \24° forming a region of flat
been carried out at the SVZ, most of which were focused subduction (Fig. 1). The SVZ includes a continental end-
on individual volcanoes. Witter et al. (2004) worked on a member in arc systems with respect to the crustal thickness
young deposit of Volcán Villarrica, which was hampered of the overriding plate. Beneath the active volcanic arc, the
by the fact that all melt inclusions except one suffered crust is 25–30 km thick in the south between latitudes
severe degassing. The authors attribute the degassing to 42.5–45°S, 35–40 km between latitudes 37–42.5°S and it
Villarrica’s present-day open vent situation with a lava lake increases systematically to a thickness of up to 60 km
in its crater, allowing the lava to constantly degas. Zajacz northwards to latitude 33°S. The incoming plate shows an
and Halter (2009) analysed melt inclusions from Villarrica age gradient from *15 Ma in the south at latitude
with a focus on the transport of copper and silver and the *42.5°S to *37 Ma in the north at latitude 33°S. The
formation of ore deposits. Bouvet de Maisonneuve et al. southern part of the incoming plate hosts a number of
(2012) interpreted compositions of a melt inclusion suite fracture zones (from north to south between latitudes *38
from Volcán Llaima with regard to magma chamber pro- and 44°S, they are the Mocha, Valdivia, Chiloé, and Guafo
cesses at shallow levels, i.e. crystallisation, reservoir Fracture Zones) (Fig. 1). With the exception of the Mocha
replenishment, and mixing. Rea (2009) and Rea et al. Fracture Zone, these are extensive fracture zone systems
(2009) published volatile concentrations in several andes- with multiple splays. The fracture zones are likely to pro-
itic inclusions in olivines, clinopyroxenes, and orthopy- mote an enhanced transport of water via altered oceanic
roxenes from Volcán Callaqui. Linked thermal and crust and possibly serpentinised mantle into the subduction
chemical gradients across the southern SVZ were con- system (Sellés et al. 2004; Rodrı́guez et al. 2007; Dzierma
strained by Watt et al. (2013) based on melt inclusion et al. 2012).
major, trace, and volatile element compositions. The SVZ has been divided into arc segments based on
The igneous rocks of the SVZ have been fairly well incoming fracture zones, spatial volcano distribution, and
researched in terms of whole-rock major and trace ele- differences in geochemistry in the erupted rocks [recently
ment and isotope geochemistry of arc lavas (e.g. Hildreth revised by Völker et al. (2011) with respect to earlier
and Moorbath 1988; Hickey et al. 1984; López-Escobar definitions of Hildreth and Moorbath (1988) and Dungan
et al. 1992; Jacques et al. 2013; Wehrmann et al. 2014, et al. (2001)]. In this paper, we use a simplified subdivi-
this volume). Here we provide an overview of the SVZ by sion, in which we slightly shifted two boundaries based on
characterising sixteen volcanic centres for their pre- geochemical observations: a northern SVZ (NSVZ;
eruptive melt sulphur and chlorine concentrations. Our 33–34.5°S), a transitional SVZ (TSVZ; 34.5–38°S; of
investigation is based on olivine-hosted melt inclusions which the southern boundary was initially defined at 37°S),
from Holocene to historical mafic to intermediate tephras. a central SVZ (CSVZ; 38–42°S; here extended from the
We provide new analyses of *300 melt inclusions and former boundary of 41.5°S to include Volcán Apagado at
combine them with trace element proxies for hydrous flux 41.9°S, since it geochemically belongs to the CSVZ), and a
melting. By filtering the results for degassing processes southern SVZ (SSVZ), here represented by a volcano in the
and magma differentiation at the individual volcanoes, area of Volcán Minchinmávida at 43°S. The boundaries are
our main focus is the variation in volatile geochemistry in illustrated in Fig. 1.

123
Int J Earth Sci (Geol Rundsch)

Although several smaller cones were shown to be geo-

chemically distinct from the main vents and, therefore,
33°S their representativeness for the respective arc system may
Tupungatito be discussed (e.g. Hickey-Vargas et al. 1989, 2002, 2003;
andez Ridge
an Fern Santiago

N SVZ
Ju
San Jose Mahlke 2009; Schumann 2010), in most respects these
Maipo
34°S differences are minor with regard to the large-scale com-
nch
parison we aim at in this paper. Where the differences are
Tinguiririca
Tre

significant, this will be indicated in the respective section.

35°S In most cases, the small centres provide the opportunity to
Nazca Planchon
Peteroa obtain the most primitive samples and therefore the best
Plate Talca C. Azul
San Pedro insight into mantle processes and magma source compo-
TS V Z 36°S
L.d.Maule nents. Although the stratovolcanoes should be understood
Longavi to better represent the magmatic system at the respective
Nevados de
Concepcion Chillan 37°S location, their melts often have undergone a certain degree
Antuco of magma differentiation, accumulation, and/or crustal
Copahue contamination. These processes make it more difficult to
38°S capture a primary geochemical composition, and therefore,
Lonquimay
we often must rely on small volcanic centres for such
FZ Llaima
Mo
cha primitive compositions.
Temuco
Villarrica 39°S
CSVZ

Valdivia Lanin
Mocho− Sample preparation and analytical methods
Choshuenco 40°S
Rinninahue
Valdivia FZ Puyehue Major volcanic centres and their satellite cones along the
Casablanca
Osorno Osorno 41°S
SVZ were sampled for the most mafic, olivine-bearing,
Cabeza de Vaca young (Holocene and historical if possible) tephras. Oliv-
Puerto Montt Calbuco South
Yate American ine crystals were handpicked from small scoria clasts
Hornopiren Plate
Chiloe F
Z Apagado 42°S (maximum 20–30 mm in length, usually around 10 mm),
SSVZ

Huequi mounted in epoxy, and polished to expose their melt

Minchinmavida/ inclusions for analysis. In addition, thin sections were
Chaiten 43°S prepared for some of the samples, and exposed melt
76°W 75°W 74°W 73°W 72°W 71°W 70°W inclusions therein were also included in the analytical runs.
Fig. 1 Location map of the Chilean Southern Volcanic Zone. Red All analysed melt inclusions were glassy, and the presence
triangles depict those volcanoes for which volatile data were of shrinkage bubbles was variable. Major element and S
generated in this study. Yellow triangles indicate other young and Cl concentrations in the inclusion glasses, the matrix
volcanic centres. Image courtesy SRTM team NASA/JPL/NIMA, glasses, as well as the major element compositions of the
modified
host olivines, were determined by electron microprobe
using a Jeol JXA 8200 at GEOMAR. The inclusions were
analysed with an accelerating voltage of 15 kV and 10 nA,
Samples and sample classification and the beam was set to 10 lm in diameter with a count
time of 20 s per major element and 40 s per volatile ele-
This study is based on young, mafic to intermediate tephra ment. Na was analysed first to minimise element loss due
samples obtained from 16 volcanic centres of the volcanic to migration. For olivine analyses, the beam was focused to
front 33–43°S (from north to south): Tupungatito, Maipo, \1 lm and measurements were performed at 15 kV and
Tinguiririca, Planchón–Peteroa, Los Hornitos near Cerro 100 nA, counting Mg, Fe, Si, Al for 20 s, Mn for 30 s, and
Azul, San Pedro, Antuco, Lonquimay, Llaima, Villarrica, Ca for 60 s. The instrument was calibrated and monitored
Mocho-Choshuenco, Casablanca, Osorno, Cabeza de Vaca, using natural reference materials ALV-981 (R23-5), VGA-
Apagado, and a vent in the Minchinmávida area. For each 99 (USNM113498), VG-2 (USNM 111240/52), and CFA-
volcano, a series of melt inclusions (typically *10–30 47 for the glass analyses and a San Carlos olivine (USNM
inclusions per volcano/tephra layer) has been analysed 111312/444) for the olivines (Online Resource 1). A subset
along with matrix glasses, host olivines, and bulk rock. of the larger melt inclusions were measured two or three
Both major volcanic centres and minor peripheral or times at different spots to check for potential heterogeneity.
satellite monogenetic cones are included in this study. Previous studies showed that compositional gradients or

123
 Int J Earth Sci (Geol Rundsch)

element diffusion may occur, possibly leading to changing contents at low silica (B53 wt%). In the K2O versus silica
redox conditions (Gaetani et al. 2012). Our repeat mea- diagram, most of the CSVZ inclusions show medium-K2O,
surements confirmed the determined compositions. This is calc-alkaline compositions, whereas some inclusions from
in line with the findings of, e.g. Baker et al. (2005), Carroll Volcán Lonquimay, Volcán Llaima, Volcán Villarrica, and
(2005), Freda et al. (2005), Behrens and Stelling (2011), Volcán Apagado fall in the low-K2O, tholeiitic field.
who state that processes of diffusive re-equilibration in Mafic TSVZ and NSVZ inclusions generally fall in the
melt inclusions are of minor importance regarding S and medium-K calc-alkaline field, while trending over the
Cl. medium- and high-K calc-alkaline fields at higher SiO2.
For the whole-rock analyses, the tephra samples were Samples from Volcán Tinguiririca and Volcán Planchón–
handpicked and powdered. Major element analyses were Peteroa in the TSVZ have the highest K2O within the
carried out on fused homogeneous glass discs made from NSVZ–TSVZ trend. Overall, the CSVZ has the lowest
the rock powders, using a Philips X’Unique PW 1480 XRF K2O, the TSVZ, and NSVZ have intermediate K2O, while
instrument at GEOMAR. Trace element compositions were the volcano at Minchinmávida has higher K2O values than
obtained with an Agilent 7500cs ICP-MS instrument at the other SVZ arc rocks at a given SiO2 content.
Institute of Geosciences at the University of Kiel. For the Trace element compositions exhibit typical subduction
ICP-MS analyses, conventional table-top digests were zone patterns, with elevated concentrations of fluid-mobile
made in PFA vials at 160 °C with HF-aqua regia-per- elements such as Ba, U, Pb, and Sr, and depletion in high
chloric acid following the multi-step procedure established fields strength elements (HFSE), such as Nb and Ta (see
by Garbe-Schönberg (1993). Analytical trueness was Wehrmann et al. 2014, this volume for figures and further
monitored by including a set of reference materials in the details). Volatile elements vary considerably. S is most
sample batches: BHVO-2, AGV-1, JA-2, BIR-1, and JG- enriched in melt inclusions in the most forsteritic (Fo)
1a, and results are provided in Online Resource 2. olivines from the small monogenetic centres Volcán Los
Repeatability was verified through threefold to fourfold Hornitos, Volcán Cabeza de Vaca, and Volcán Apagado,
measurements of the same digestion of randomly chosen while the large stratovolcanoes have lower S concentra-
samples (Online Resource 2). Blanks for background tions, but no systematic variation with respect to the major
monitoring were included in each analytical run (Online element geochemistry (Fig. 3a, b). Cl is highest in the
Resource 2). volcanoes in the northern part of the arc (Volcán Tupun-
gatito, Volcán Maipo, and Volcán Tinguiririca) in melt
inclusions of basaltic to andesitic composition that are
Results contained in relatively evolved olivines (Fo74–80) and at the
small centres in low-SiO2 inclusions in primitive, high-Fo
Major and volatile element compositions of the melt olivines (Fig. 3c, d). The large stratovolcanoes of the
inclusions, their host olivine crystals, and matrix glasses CSVZ and the TSVZ have the lowest Cl in olivines of
are provided in Online Resources 3 and 4, including Fo70–85 (Fig. 3c, d).
recalculation of inclusions to equilibrium with their host
crystal and estimation of parental melt compositions at
Fo91, using Petrolog 2.1 software (Danyushevsky et al. Discussion
2002) with the assumption of oxygen fugacity at Ni–NiO
buffer. Whole-rock major and trace element concentrations Melt inclusion major element geochemistry
are presented in Online Resource 5. For a general classi-
fication, the melt inclusions are presented here in total As shown in Figs. 2 and 3, the melt inclusions are variably
alkali (TAS; Fig. 2a, b) and K2O (Fig. 2c, d) versus silica evolved. While basaltic inclusions are abundant in samples
diagrams. In the TAS diagram, most of the TSVZ and from the small, monogenetic cones and in the CSVZ
CSVZ samples occupy the basaltic and basaltic–andesite stratovolcanoes, crystal fractionation particularly affected
field, while the NSVZ trends into the field of trachy- inclusions of the NSVZ. Two (trachy-)basaltic–andesitic
andesites. Melt inclusions from the vent at Minchinmávida inclusions have, however, been found in the tephras from
have the highest Na2O ? K2O contents in general for Volcán Tupungatito and Volcán Maipo, which provide
samples with SiO2 B 53 wt% with the exception of two insight into reasonably mafic compositions in the NSVZ.
melt inclusions from Planchón–Peteroa in the TSVZ. On Based on a northwards increase in K2O in volcanic front
the other hand, melt inclusions from Volcán Llaima, Vol- lavas (whole-rock data normalised to SiO2 = 57.5 wt%)
cán Villarrica, Volcán Osorno, and in part of Volcán Ap- correlating with increasing crustal thickness along the arc,
agado in the CSVZ and most of the melt inclusions from Hildreth and Moorbath (1988) developed a model of
Volcán Los Hornitos in the TSVZ display the lowest alkali melting, assimilation, storage, and homogenisation

123
Int J Earth Sci (Geol Rundsch)

10 5.5
NSVZ
TSVZ
Trachy− a b
andesite Trachybasalt
CSVZ
at Minch. 5.0
lic Basalt
Trachy− Alka
lic/
8 basaltic alka
andesite Sub oleiitic 4.5
Th
Na 2O + K2 O

Na 2O + K2 O
VZ 4.0
TS VZ
6 NS
Trachy− 3.5
n.
basalt hi Z
i nc V
Basalt tM CS
a
3.0
4

2.5 Tupungatito
Maipo Basaltic
Basaltic Planchon andesite
andesite Andesite Los Hornitos
San Pedro
Shoshonitic c 1.4 Antuco d
Lonquimay
High−K Medium−K
High−K Llaima
calc−alkaline Villarrica
1.2 Mocho−Ch.
3 Casablanca
VZ

Osorno
NS

1.0 Cabeza d. V.
Apagado
at Minchin.
0.8
K2 O

2
Z
K2 O

TSV
. 0.6
in
nch Medium−K
Mi calc−alkaline
at
1 0.4
Low−K
0.2
Low−K tholeiitic
Z
CSV
0 0.0
45 50 55 60 65 46 48 50 52 54 56 58
SiO2 SiO 2 (wt. %)

Fig. 2 a Total-alkali versus silica (TAS) diagram b K2O versus silica each arc segment. c, d Blow-ups of the green boxes in a and
(wt%) as a general major element classification of the SVZ melt b showing melt inclusion suites of individual volcanoes
inclusions, corrected for post-entrapment crystallisation, grouped for

(MASH) to explain the lava compositions. When compar- remainder of the SVZ arc rocks. However, the least
ing melt inclusions with SiO2 \ 53 wt%, this northwards evolved NSVZ samples overlap with the more evolved
increase in K2O, however, largely disappears (Fig. 4). This rocks from the TSVZ and CSVZ in their Mg-number and
implies that even if the MASH model is valid for rocks of SiO2 concentration, allowing us to make comparisons at
elevated SiO2, mafic melts may ascend without extensive similar Mg-numbers and/or SiO2 contents. In most cases,
interaction with the crust of the overriding plate in the the evolved NSVZ rocks do not extend the differentiation
NSVZ. trends formed by the TSVZ or CSVZ (Online Resource 6),
pointing to magma genesis to explain the origin of these
Whole-rock trace element geochemistry differences. Crystal fractionation processes will not affect
ratios of trace elements of similar incompatibility.
The samples in this study span a range in degree of magma In past studies, the variations in minor and trace element
differentiation. Since more incompatible elements undergo geochemistry have been related to differences in a number
a stronger enrichment than the less incompatible ones of subduction parameters along the arc (Stern 2004;
during magma differentiation, some of the trace element Wehrmann et al. 2014, this volume; and references
ratios discussed in the following sections are likely to have therein). They were interpreted to reflect (1) variable
been affected by differentiation processes. The lack of degrees of interaction with the overlying crust during dif-
basaltic whole-rock samples from Volcán Tupungatito and fering time periods of magma storage (e.g. Hildreth and
Volcán Maipo places some limitations on evaluating their Moorbath 1988), (2) compositional heterogeneity of the
petrogenesis and on comparisons of their origin with the mantle wedge (e.g. López-Escobar et al. 1977), (3)

123
 Int J Earth Sci (Geol Rundsch)

a b Tupungatito
3000 Maipo
Tinguiririca
Planchon
Los Hornitos
San Pedro
Antuco
Lonquimay
2000
Llaima
S (µ g/g)

Villarrica
Mocho−Ch.
Casablanca
Osorno
Cab. de Vaca
1000 Apagado
at Minchin.

0
c d
1500

1000
Cl (µg/g)

500

0
70 75 80 85 90 48 50 52 54 56 58 60 62
Forsterite content of host olivine SiO2

Fig. 3 Melt inclusion volatile abundances at the different volcanic versus Fo, and d Cl versus SiO2. The arrows at the lowest-SiO2
centres in relation to the Fo content of their host olivines and the melt Maipo inclusion indicate exceptionally high volatile contents
inclusion SiO2 concentration. a S versus Fo, b S versus SiO2, c Cl

2.5 variations in the material input into the mantle wedge from
N NSVZ TSVZ CSVZ SSVZ S
Tupungatito
Maipo
the subducting plate and through forearc subduction ero-
Planchon

2.0
Los Hornitos
Antuco
sion (e.g. Hickey et al. 1986; Tormey et al. 1991; Kay et al.
Lonquimay
Llaima 2005; Jacques et al. 2013), and (4) differences in the degree
K2

Villarrica
of mantle melting (e.g. Hickey et al. 1986; Hickey-Vargas
O5

Mocho−Ch.
Cab. de Vaca
7.
K 2O (wt.%)

Apagado
1.5 at Minchin. et al. 1989; Muñoz and Stern 1989; López-Escobar et al.
1995a, b; Tormey et al. 1991, 1995). We will discuss these
processes in more detail below.
1.0

Crustal contamination
0.5
Many authors favour a control on the changing geochem-
Distance along the arc [km]
0 200 400 600 800 1000 istry northwards through increasing interaction with the
0.0
32 33 34 35 36 37 38 39 40 41 42 43 44 progressively thickening crust (e.g. Hildreth and Moorbath
Degree latitude [°S] 1988; Davidson et al. 1987; Ferguson et al. 1992) or dif-
ferent crust to mantle wedge thickness ratios (Guthrie and
Fig. 4 Along-arc variation in K2O in melt inclusions with Wulff 2002). Also in the CSVZ, where crustal thickness is
SiO2 \ 53 wt%. Green field shows K2O in whole rocks normalised
to SiO2 = 57.5 wt%, taken from Hildreth and Moorbath (1988). In
uniformly thinner (*40 km; Tassara et al. 2006; Völker
contrast to the whole rocks at 57.5 wt% SiO2, the K2O of the et al. 2011), processes in the crust such as assimilation of
primitive melt inclusions does not increase northwards crustal material with isotopic signatures similar to the

123
Int J Earth Sci (Geol Rundsch)

magmas were proposed to explain the restricted isotopic 0.6

Tupungatito t n g
variations over a wide range of bulk magma compositions Maipo en io tin
Tinguiririca h m tiat el
r ic ren f m
(Hickey et al. 1984; Hickey-Vargas et al. 1995; McMillan 0.5
Planchon
en fe o
Los Hornitos
e if ee
et al. 1989; Reubi et al. 2011). In contrast, Saal et al. (1993) San Pedro
u rc a d egr
Antuco so agm g d
and Tormey et al. (1991) attributed the northwards increase Lonquimay m asin
0.4 Llaima e
cr

Z
in incompatible element contents and La/Yb and Rb/Cs to a Villarrica
de

SV
Mocho−Ch.

N
northwards decreasing contribution of slab-derived fluids Casablanca
tin
g

Nb/Y
Osorno
io
n el
and thus lower degrees of melting rather than increased 0.3 Cabeza d. V. et m
pl of
de
Apagado e
crustal assimilation. Other authors argue that increased at Minchin.
ce gr
e
ur de

VZ
subduction erosion of crustal materials in the north can best so g

TS
0.2 n
si
explain the enriched incompatible element and isotope ea
cr
in
ratios (Stern 2004; Kay et al. 2005).
0.1 AOC
As exemplified at Volcán Mocho-Choshuenco by
VZ
McMillan et al. (1989), in some/many cases, trace elements DMM CS

are a more powerful tool for identifying crustal contami- 0.0

1 2 3 4 5 6
nation than isotopes. This is particularly the case when La/Sm
subducted materials, the crust of the overriding plate, and
erupted products have identical isotopic signatures through Fig. 5 Positively correlating whole-rock Nb/Y versus La/Sm of the
SVZ tephras, complemented by additional tephra and lava compo-
long-term subduction recycling. Such recycling is consid- sitions of the respective volcanoes from Jacques et al. (2013) and
ered to have been intense at the SVZ, since the marine Wehrmann et al. (2014, this volume). CSVZ samples display the
sediments are largely of volcanic origin from the active arc, lowest Nb/Y and La/Sm, the TSVZ intermediate values, and the
and the continental crust consists, to a large extent, of NSVZ the highest. The CSVZ magmas therefore have formed by the
highest degrees of melting of the most depleted source [DMM
plutonic rocks derived from the same sources as the vol-
depleted MORB mantle (Salters and Stracke 2004), AOC altered
canic rocks. Assimilation of plutonic rocks was inferred ocean crust (Sadofsky et al. 2009)]. The higher La/Sm and Nb/Y of
from U-series disequilibria in the young magmatic system the TSVZ and NSVZ can reflect a combination of greater degrees of
of Llaima by Reubi et al. (2011). They proposed that it was differentiation, smaller degrees of melting, and/or derivation from
more enriched source material
responsible for elevating trace element ratios such as Ba/
Th, which is often interpreted to reflect an enhanced slab
fluid signal. Therefore, in older rocks where U-series iso- proceeds. Both Nb/Y and La/Sm of the TSVZ samples
tope ratios are in secular equilibrium, identification of increase with differentiation, but form arrays elevated from
crustal assimilation remains complicated at the SVZ, the CSVZ, supporting decreasing degrees of melting and
because of the compositional similarity between the sub- derivation from more enriched sources towards the north.
ducted marine sediments and the overlying crust. In the NSVZ, these ratios show a diffuse relationship to the
Mg-number and an increase at increasing SiO2 (Online
Mantle wedge heterogeneity Resource 6). These latter trends, however, do not extend
the trends formed by the TSVZ, but show higher Nb/Y and
Elevated Nb/Y and La/Sm can be used to trace low degrees La/Sm for a given SiO2 concentration. Therefore, despite
of melting and/or the presence of enriched components in the obvious effect of differentiation within the northern and
the mantle wedge. The SVZ rocks form a clear positive co- the transitional arc segment sample suites, the differences
variation, in which the CSVZ rocks have the lowest Nb/Y between the arc segments appear to be primary character-
and La/Sm, the TSVZ rocks have intermediate ratios, and istics of the parental melts, reflecting lower degrees of
the NSVZ rocks have the highest Nb/Y and La/Sm ratios. melting and/or derivation from a more enriched source
These variations are consistent with the CSVZ melts being towards the north besides the potential interaction with the
derived from a depleted mantle source through high crust. Isotopic signatures presented in previous studies also
degrees of partial melting. The NSVZ can be explained by pointed to a distinct, isotopically more enriched source, or
the lowest degrees of melting from an enriched source contribution of continental crust via assimilation (Hildreth
component in the mantle wedge, with or without assimi- and Moorbath 1988) or subduction erosion into the source
lation of crustal material (Fig. 5). However, increasing beneath the NSVZ (Stern et al. 1984; Kay et al. 2005).
Nb/Y and La/Sm could also simply result from higher
degrees of magma differentiation. When comparing the Fluid signals and hydrous flux melting
Nb/Y and La/Sm to Mg-number and SiO2 to assess the
effect of magma evolution (Online Resource 6), the data High ratios of highly fluid-mobile over less fluid-mobile
show a negative or no trend for the CSVZ as differentiation or fluid-immobile elements, e.g. Pb/Ce and Ba/Nb, are

123
 Int J Earth Sci (Geol Rundsch)

0.5
Tupungatito
amounts of seawater to hydrate the lower oceanic crust and
Maipo
Tinguiririca
uppermost mantle, converting the peridotite in the upper-
Planchon most mantle of the incoming plate to serpentinite (Con-
0.4 Los Hornitos
San Pedro treras-Reyes et al. 2008). During subduction, dehydration
Antuco
Lonquimay allowed this excess water to be released beneath Volcán
Llaima
0.3 Villarrica and Volcán Llaima (Dzierma et al. 2012).
Pb/Ce

l Villarrica
na Mocho−Ch.
sig The sample from the volcano in the Minchinmávida area
ui
d Casablanca
fl
Osorno
g
sla
b
Cab. de Vaca
in the South shows intermediate values with respect to
0.2 sin
cr
ea
Apagado
at Minchin.
Nb/Y and La/Sm (Fig. 5), associated with a relatively weak
In
fluid signal from the trace element fluid proxies (Fig. 6).
0.1 Since the latter is consistent with the melt having the
highest concentrations of the incompatible major elements
20 40 60 80 100 120 140 160 K2O, Na2O (Figs. 2a–d, 4), TiO2, and P2O5, this suggests
Ba/Nb small degrees of flux melting. The elevated Nb/Y and
Fig. 6 Whole-rock Pb/Ce versus Ba/Nb for the SVZ arc rocks (data
La/Sm point to derivation from a mantle source that is less
from this study, Jacques et al. 2013; Wehrmann et al. 2014, this depleted than that of the adjacent CSVZ, whereby the more
volume). The positive co-variation points to low fluid signals in the limited depletion is likely related to a lower melt formation
NSVZ, at Minchinmávida, and at the small cones of Apagado and beneath the vent at Minchinmávida. The low subduction
Cabeza de Vaca, while reaching peak values at Llaima and Villarrica;
exceeded only by the Los Hornitos cinder cones
signal in this melt is consistent with the young age and high
temperature of the subducting plate, which might yet have
undergone only initial serpentinisation (Völker et al. 2011),
and hence less water was entrained into the mantle beneath
observed at Volcán Llaima and Volcán Villarrica (both in the Minchinmávida area.
the CSVZ), which are amongst the currently most active
volcanoes of South America (Fig. 6; see also Dzierma et al. Melt inclusion volatile geochemistry
2012). Samples from Volcán Osorno, Volcán Mocho-
Choshuenco, Volcán Lonquimay, and Volcán Los Hornitos Variations in the volatile abundances are related to variable
also have high fluid-mobile to fluid-immobile element source contents, subsequent enrichment by magma differ-
ratios. The peak values of the CSVZ decrease towards the entiation, and differences in degassing intensities during
TSVZ (with the exception of the Los Hornitos cones), storage and ascent at medium to shallow depths in the
where it appears that a lower hydrous flux is related to crust.
lower degrees of melting. At Volcán Tupungatito and
Volcán Maipo in the NSVZ and in the small cones of Magma evolution, volatile solubility/saturation,
Cabeza de Vaca and Apagado in the South, low degrees of and volatile degassing
melting (Fig. 5) are associated with low fluid signatures, as
expressed by low Pb/Ce and low Ba/Nb (Fig. 6). At most volcanic systems, volatile degassing occurs in
Unexpectedly, Ba/Th does not correlate well with other conjunction with crystallisation. Volatiles exsolve to form
highly fluid-mobile to less fluid-mobile trace element a fluid or gas phase when their concentration exceeds the
ratios. Reubi et al. (2011) proposed assimilation of plutonic saturation limits, which is a function of pressure, temper-
rocks in the magmatic system of Volcán Llaima to have ature, and compositional properties of the melts, or through
caused a reduction in Ba/Th, thus providing a possible partitioning into a coexisting fluid phase when it is present.
explanation as to why Ba/Th behaves differently from the Furthermore, volatile elements may be removed from the
other fluid proxies. They propose that assimilation of plu- melt through precipitation of volatile-bearing mineral
tonic basement may also be important in other arc volca- phases (e.g. Wallace and Edmonds 2011; Métrich and
noes. However, the presence of an intense slab signal at Wallace 2008; Kent 2008; Wallace 2005; and references
Volcán Llaima, as we show here to be reflected by the fluid therein).
tracers mentioned above, is strongly supported by geo-
physical constraints of Dzierma et al. (2012), which Chlorine For chlorine, comparison of the melt inclusion
revealed peak fluid flux at Volcán Llaima and Volcán and matrix glass compositions with the Cl solubility cal-
Villarrica in an arc regional context. This spike at Volcán culations of Webster et al. (1999) reveals that all tephras
Llaima and Volcán Villarrica has been associated with the investigated here are undersaturated with regard to Cl
Valdivia Fracture Zone, a morphologically large feature on (Fig. 7). This is a common feature in arc basalts (e.g.
the incoming Nazca plate, which likely allowed enhanced Wallace 2005). Even at atmospheric pressures, the Cl-

123
Int J Earth Sci (Geol Rundsch)

2500
San Pedro least deg. incs
a Maipo least degassed incs b

Pa

Pa
Pa

Pa
Maipo degassed incs San Pedro degassed incs

y at 200 M

y at 200 M
y at 0.1 M

y at 0.1 M
Maipo matrix glass San Pedro matrix glass
Los Hornitos least deg. incs Llaima (CL 035) least deg. incs
2000 Los Hornitos degassed incs Llaima (CL 035) degassed incs
Los Hornitos matrix glass Llaima (CL 035) matrix glass

Cl solubilit

Cl solubilit
Cl solubilit

Cl solubilit
1500

Cl fere
Cl ( µg/g)

Di

en n t i
f

ric ati
assing
act.

hm o n
en
apa. fr

t
Cl deg
n
1000 tio
ia
nt
er
e i ng
iff a ss
D
eg
ld
C
500
Co−crystallisation
Cl degassing

0.0
0 0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0 1.2
Al+Na+Ca+Mg Al+Na+Ca+Mg
molar molar
Si Si

Fig. 7 Melt inclusion and matrix glass Cl concentration as a function a Maipo and Los Hornitos, and b San Pedro and Llaima (sample CL
of composition in comparison with Cl solubility limits after Webster 035). All SVZ tephras investigated here are undersaturated with
et al. (1999) at 200 and 0.1 MPa, illustrating Cl under saturation of respect to Cl

saturation limits are not exceeded, implying that in those mixing, reservoir replenishment and eruption described by
volcanic systems in the SVZ where Cl degassed, this took Mahlke (2009) and Bouvet de Maisonneuve et al. (2012).
place through partitioning of Cl into a hydrous fluid. At
Volcán Tupungatito, Volcán San Pedro, Volcán Lonqui- Sulphur Tracing the behaviour of S in a melt is a long-
may, the vent at Volcán Minchinmávida, and the small standing challenge in magmatic petrology, since S can
cones of Los Hornitos and Cabeza de Vaca, there is a trend exist in multiple valence states and species, and reacts very
towards the saturation limit as the melt evolves from the sensitively to changes of oxygen fugacity (fO2), tempera-
melt inclusions to the matrix glasses (exemplarily shown ture, and melt composition (e.g. Wallace 2005; Jugo 2009;
for San Pedro and Los Hornitos in Fig. 7a, b). This is Baker and Moretti 2011; Beermann et al. 2011; Binder and
clearly associated with Cl degassing at Los Hornitos, Keppler 2011; Wallace and Edmonds 2011; Klimm et al.
whereas the matrix glasses of the other volcanoes show a 2012). Sulphur contents generally decrease dramatically
passive Cl enrichment (possibly degassing-accompanied) during magma differentiation (e.g. Wallace and Edmonds
as the residual melt evolves. The Cl contents of the dry 2011). Besides exsolution of S-bearing fluids or vapour
magma at Tupungatito most closely approach the satura- phases, S may be lost from a melt by fractionation of
tion limit, possibly due to a lack of fluid available for Cl to S-bearing minerals. Under reducing conditions, S may
partition into, and/or due to the decreasing Cl solubility as precipitate into e.g. pyrrhotite [stable at fO2 \ NNO and
the melt evolves. No apatite, which could also have 600–900 °C (Parat et al. 2011)] when the S content at
accounted for Cl removal, was found in the thin section of sulphide saturation (SCSS; Baker and Moretti 2011) is
the Tupungatito sample, and negative P2O5 trends are only reached. Under oxidising conditions upon crossing the S
observed in matrix glasses with SiO2 [64 wt%, not in the content at anhydrite saturation (SCAS, Baker and Moretti
Tupungatito melt inclusion suite. Volcán Maipo, Volcán 2011), S may be lost from the melt through crystallisation
Mocho-Choshuenco, and Volcán Casablanca show melt of anhydrite or incorporation in S-bearing apatite, which
inclusion–matrix glass trends paralleling the saturation are stable at fO2 [ NNO ? 1 and 700–1,000 °C (Parat
limit (exemplarily shown for Maipo in Fig. 7a), implying et al. 2011). In turn, degassing of S and of water affects the
that the decreasing solubility of Cl as the magma differ- oxygen fugacity of a melt. Details of this process and its
entiates is anticipated by Cl removal from the melt despite consequences were subject of debate in recent literature
the undersaturation. The different tephra layers from Lla- (Klimm et al. 2012; Kelley and Cottrell 2012; and refer-
ima and Villarrica show both increasing and decreasing ences therein).
trends (exemplified for one of the Llaima tephras in At the SVZ, fO2 values were determined at several
Fig. 7b), pointing to alternating phases of Cl enrichment TSVZ and CSVZ volcanoes. Ruprecht et al. (2012) esti-
and degassing during the cycles of magma evolution, mate NNO ? 0.24 to NNO ? 0.53 for mafic melts at

123
 Int J Earth Sci (Geol Rundsch)

Quizapu. The adakitic volcano Nevado de Longavı́ is percentages of the host olivines. This indicates a rough
described with NNO ? 1 for a mafic enclave (Sellés et al. balance between Cl degassing and Cl enrichment through
2004), while Rodrı́guez et al. (2007) show highly oxidising differentiation for this tephra. S shows strongly decreasing
conditions of NNO ? 1.5 to NNO ? 2 for Longavı́. Costa concentration values with increasing K2O and decreasing
and Singer (2002) measured NNO to NNO ? 1 for Fo-% of the olivine, pointing to degassing and/or precipi-
andesitic and dacitic compositions at San Pedro, while tation of S-bearing minerals, the latter likely happening as
Witter et al. (2004) deal with a range from QFM to the melt oxidation state moves across the sulphide/sulphate
NNO ? 1 at Villarrica. Bouvet de Maisonneuve et al. redox boundary.
(2012, 2013) assume QFM to NNO for Llaima, and Watt From these plots, we infer that in most of the SVZ
et al. (2013) assume QFM ? 1 for Apagado and volcanoes, Cl degassing is limited, while S is removed
Minchinmávida. from the melt during magma differentiation. The only
We have not determined fO2 or S speciation in this exception detected at the SVZ is a melt inclusion suite
study, precluding detailed reconstructions of the relation- from Volcán Planchón–Peteroa, which forms unclear
ship between degassing paths, S-bearing mineral fraction- trends and contains high-S inclusions with high FeO and
ation along with changes in oxygen fugacities in the low SiO2 in evolved olivines (Online Resource 7).
individual systems. In the absence of constraints on sul- We have assessed degassing for each suite of melt
phide or sulphate saturation in the SVZ magmas, we refrain inclusions from every tephra deposit investigated in this
from assessing if or to what extent S loss from the melt study. Although it remains a challenge to firmly assess
may have occurred prior to melt inclusion entrapment. whether or how much S loss might have potentially
However, with regard to the exsolution of a S-bearing fluid occurred prior to melt inclusion entrapment, we employ
phase, previous studies at other arcs revealed that mafic these plots in order to identify the inclusions least affected
magmas in general are often not saturated with S, and the by degassing. The following considerations are based on
‘‘excess sulphur problem’’ was found to be minor for constraints from the subset of melt inclusions that are least
explosive basaltic volcanism (Wallace and Edmonds volatile saturated and least degassed, representing the
2011). Since this is the predominant type of eruption we minimum initial volatile concentration for each tephra.
look at in this study, it is likely that such processes have not
dramatically overprinted the S concentrations in the studied Parental melt volatile concentrations
samples. ‘‘Excess sulphur’’, furthermore, is considered to
require oxidising conditions, because under reducing con- Based on the evaluation outlined above, the least-degassed
ditions, S would be consumed by precipitation of phases melt inclusions of each volcanic system were selected here
such as pyrrhotite (Scaillet et al. 1998; Wallace and to approximate the volatile concentrations of the parental
Edmonds 2011). Therefore, also for the stratovolcanoes melt. Passive enrichment of S and Cl during fractional
Tupungatito and Maipo, a potential ‘‘excess sulphur prob- crystallisation was accounted for by iteratively adding
lem’’ will be of minor importance, since the enriched olivine to the melt composition until equilibrium with
mantle source beneath the NSVZ (Hickey et al. 1986; mantle olivine of Fo91 was reached, using Petrolog 2.1
Wehrmann et al. 2014, this volume) would be associated software of Danyushevsky et al. (2002), as described by
with reducing conditions (Cottrell and Kelley 2013). Portnyagin et al. (2007); Sadofsky et al. (2008); and Watt
To identify the most primitive melt inclusions that et al. (2013). For this, we assumed an oxygen fugacity of
approach initial pre-eruptive volatile contents, the crystal- NNO. To evaluate the effect a different fO2 assumption
lisation-related gas release is assessed here by comparing would have, the calculation was repeated with an oxygen
the volatile concentrations of individual melt inclusion fugacity of NNO ? 1, which is within the range provided
suites with indicators of magma differentiation, such as the in the literature for SVZ volcanoes. NNO ? 1 would cause
K2O-concentration of the melt or the Fo content of the only a minor shift in the amount of olivine added into the
inclusion-hosting olivine crystal. If no degassing or volatile inclusions and thus of the estimated parental melt volatile
fractionation into minerals took place in a theoretical content, amounting to on average *4, but no more than 9
closed system scenario, both S and Cl would be expected to relative-% of the fractionation-correction. The along-arc
increase as the magma evolves because of their incom- pattern remains whether NNO or NNO ? 1 is assumed. All
patible behaviour. Decreasing volatile concentrations, in corrected values we report here are based on NNO.
contrast, point to crystallisation-related gas loss. A set of Intrinsic uncertainties of this method are small for melt
such plots is provided here for a melt inclusion suite from a inclusions hosted in very primitive olivines such as at
tephra deposit from Volcán Llaima as an example (Fig. 8a– Apagado or the Los Hornitos peripheral cinder cones of
f). Cl concentrations remain fairly constant over a range of Cerro Azul. Uncertainties increase for the more evolved
melt K2O-contents, as well as over decreasing Fo- olivines, since larger amounts of the phase need to be

123
Int J Earth Sci (Geol Rundsch)

Fig. 8 Assessing the degree of degassing for one melt inclusion suite equal rate as Cl enrichment by differentiation. The co-variation in
of a tephra layer from Volcán Llaima (sample CL 035): a S e and f underscores concurrent degassing of S and Cl, with greater S
concentration in melt inclusions versus Fo content of the host olivine than Cl degassing. This is revealed by steep trends in the S versus Cl
crystal, b S versus K2O, c Cl versus Fo of the host olivine crystal, d Cl plot, a negative intercept of the K2O-normalised trend with the S/K2O
versus K2O, e S versus Cl, f S/K2O versus Cl/K2O. Closed symbols axis, and by relatively high concentrations of Cl in the residual matrix
melt inclusions, open symbols matrix glasses. Plots a and b reveal glasses
strong S degassing, while c and d show Cl loss at an approximately

numerically added, and the crystallisation of pyroxene and captured melt with SiO2 = 52–53 wt%, representing rela-
plagioclase is not taken into account. This applies in par- tively primitive compositions, closer to a parental melt.
ticular to most of the inclusions in the more differentiated Additional uncertainty originates from volatile degassing
tephras of Volcán Tupungatito and Volcán Maipo in the during differentiation, in particular of S, which cannot be
NSVZ. In those tephras, however, two olivines of Fo75–80 accounted for by this method.

123
 Int J Earth Sci (Geol Rundsch)

Fig. 9 Fractionation-corrected N NSVZ TSVZ CSVZ SSVZ S Tupungatito

concentrations of a S and b Cl 3000
Maipo
in melt inclusions along the arc Tinguiririca
Planchon
2500 Los Hornitos
San Pedro
Antuco

S at Fo 91 (µ g/g)
Lonquimay
2000 Llaima
Villarrica
Mocho−Ch.
1500
Casablanca
Osorno
Cab. de Vaca
Apagado
1000 at Minchin.

500

0
a
1400

1200

1000
Cl at Fo91 (µg/g)

800

600

400

200
Distance along the arc [km]
b 0 200 400 600 800 1000
0
32 33 34 35 36 37 38 39 40 41 42 43 44
Degree latitude [°S]

In a comparison of S and Cl along the SVZ (Fig. 9a, b), inclusions are twice as high (H. Wehrmann, unpublished
the lowest parental Cl concentrations, *300–600 lg/g, data). The one basaltic–andesitic inclusion of Maipo con-
occur in the centre of the arc at Volcán Villarrica, Volcán tains exceptionally high volatile contents [up to 4,700 lg/g
Llaima, Volcán Lonquimay, and Volcán Antuco. The Cl S and 7,900 lg/g Cl in a Fo79 olivine, corresponding to
concentrations increase both towards the North and 3,900 lg/g S and 6,600 lg/g Cl at Fo91 (indicated by the
towards the South of the CSVZ, while the volcano at arrows in Figs. 3, 9)], possibly related to redox conditions
Minchinmávida in the South is similarly low in Cl as the different from that of Volcán Tupungatito. The dimension
central arc. Peak S contents were found at the small scoria of the differences between the individual volcanoes, and
cones of Apagado, Cabeza de Vaca, and Los Hornitos (up the occurrence of the highest volatile contents in the most
to *3,000 lg/g), in agreement with them having the most primitive of the NSVZ melt inclusions precludes magma
primitive compositions, least fractionation of S-bearing differentiation from being the only process responsible for
minerals at probably the highest SCSS, as well as reflecting the observed differences at least at Maipo, and suggests
minimum degassing. In the NSVZ, despite larger uncer- influence of the changing subduction processes along the
tainties, Tupungatito shows relatively low S at high Cl, arc. The along-arc trends remain when data filtering to
which could reflect degassing of S and/or dropping of inclusions with SiO2 \ 53 wt% is applied to exclude
SCSS and enrichment of Cl during magma differentiation. evolved inclusions (Online Resource 8). When excluding
Volcán Maipo, in contrast, typically displays both high S the small cones and considering only inclusions with
and Cl (*1,000 lg/g in olivines); pyroxene-hosted SiO2 \ 53 wt%, there is a marked difference in Cl

123
Int J Earth Sci (Geol Rundsch)

concentration between high values at the northernmost Volatile concentrations in relation to subduction
volcanoes and generally lower values through the southern parameters
TSVZ and CSVZ.
To infer potential controls on the systematics of the volatile
1600 abundances in the SVZ, the melt inclusion data are inte-
Tupungatito Lonquimay
Maipo
Tinguiririca
Llaima
Villarrica
grated here with volcano size and selected whole-rock trace
Planchon Mocho−Ch.
1400 Los Hornitos
San Pedro
Casablanca
Osorno
element ratios (Figs. 10, 11a, b).
Antuco Apagado
As indicated above, several of the small eruptive centres
1200 have geochemical characteristics distinct from their
neighbouring large volcanoes. For example, Hickey-Var-
Cl at Fo 91 (µ g/g)

1000
gas et al. (1989, 2002, 2003) described a reduced subduc-
800
tion signal at the small cones surrounding Villarrica. The
small centres of Cabeza de Vaca near Osorno also have a
600
1500
reduced slab signature (Schumann 2010). In contrast, the
1000
small Los Hornitos cones at Cerro Azul show very strong
Cl ( µg/g)

400 subduction signatures (Fig. 6). Despite this fundamental

500 difference, these small cones, including the cinder cone of
200
Apagado in the south, share a common feature of very
0
0.01 0.1 1 10
Volcano edifice size [km3 ]
100 1000
primitive compositions and stand out because of their high
0.0
0.01 0.1 1 10 100 1000 melt S and Cl concentrations. A sample from a small cone
Volcano edifice size (km3 ) located at mid-altitude on the flank of Osorno and a sample
from a satellite cone of Mocho-Choshuenco, in contrast,
Fig. 10 Melt inclusion Cl at Fo91 concentrations (average of the are geochemically indistinguishable from Osorno’s and
least-degassed inclusions for each tephra deposit) versus volcanic
edifice size. Small volcanic cones are associated with the highest Cl Mocho-Choshuenco’s main vent deposits, confirming a
contents in the melts, while the Cl abundances appear to be strongly greater slab input into the stratovolcano melts compared to
diluted at the large volcanoes. The inset (Cl not fractionation- the small volcanic vents between the stratovolcanoes.
corrected to Fo91, of individual inclusions) illustrates that this In addition to the primitive compositions and related
relationship also holds when considering only mafic melt inclusions
with SiO2 \ 53 wt%. Volcano edifice volumes from Völker et al. high SCSS, the higher S abundances in the small cones
(2011) compared to those in the large centres likely reflect

2000
CAVA Lonquimay
Tupungatito Llaima
Maipo Villarrica
Tinguiririca Mocho−Ch.
Planchon Casablanca
San Pedro Osorno
Antuco at Minchin.
1500

VA
Cl at Fo 91 (µg/g)

CA CAVA
1000 VZ VZ
NS NS

VZ VZ
TS TS
VZ
500 V Z CS
CS

a b
0.0
0 1 2 3 4 5 6 0 0.1 0.2 0.3 0.4 0.5
La/Sm Nb/Y

Fig. 11 Melt inclusion Cl at Fo91 versus a La/Sm and b Nb/Y form melt inclusions, in comparison, also form a positive trend between Cl
positive correlations for the tephras from SVZ stratovolcanoes, at Fo91 and La/Sm, but at higher Cl concentrations, possibly
suggesting that the degree of melting is the primary control on Cl in suggesting that the source beneath the CAVA contains more Cl than
primitive melts and on the whole-rock La/Sm and Nb/Y ratios, rather the source beneath the SVZ. Cl at Fo91 does not show a correlation
than differentiation or source composition, especially when taken with Nb/Y at the CAVA. CAVA data from Sadofsky et al. (2008) and
together with the correlation with edifice size. CAVA (green circles) Wehrmann et al. (2011)

123
 Int J Earth Sci (Geol Rundsch)

stronger degassing of the stratovolcano melts during ascent subduction zone (i.e. the nature and thickness of the crust
through complex pathways in the stratovolcano plumbing of the overriding plate that the melts must traverse) is not
system. Considering the undersaturation of Cl, Cl should, the factor exerting the major control on Cl inventories in
in contrast, be enriched in the more differentiated magmas primitive melts in equilibrium with Fo91, but rather degree
of the stratovolcanoes. Contrary to what is expected, the of melting and source composition. S concentrations are
melt Cl concentrations (fractionation-corrected to Fo91) lower in the SVZ, with the exception of small peripheral
correlate inversely with the size of the volcanic edifices cones of Los Hornitos next to Cerro Azul, Cabeza de Vaca
from Völker et al. (2011) (Fig. 10). near Osorno, and Apagado, indicating that degassing and/
When comparing only the major volcanic centres or removal through fractionation of an S-bearing phase is
(excluding the small intra-arc centres Los Hornitos, Cabeza the major control on S contents in these stratovolcanoes.
de Vaca, and Apagado), Cl at Fo91 correlates positively Based on the correlation between fractionation-cor-
with more to less incompatible element ratios of fluid- rected melt volatile concentrations and melt inclusion trace
immobile elements, such as La/Sm and Nb/Y (Fig. 11a, b). element chemistry, Sadofsky et al. (2008) defined Ba/La as
As discussed above and by Wehrmann et al. (2014, this a proxy for water in the parental melts, and La/Nb for S and
volume) at a given SiO2, both La/Sm and Nb/Y vary with Cl. These proxies were shown to be valid along the CAVA.
degree of melting and source enrichment/depletion. Inte- For the Marianas, however, the correlation between H2O
grating the negative correlation of Cl at Fo91 with volcanic and Ba/La was not confirmed (Shaw et al. 2007). A com-
edifice size and the positive correlation of Cl at Fo91 with parison to the SVZ has its limitations at this stage, because
more to less incompatible element ratios suggests that the trace element compositions are only available for bulk
degree of melting and source enrichment also controls the rock. Nevertheless, those should be largely mimicked by
Cl content of the melts. melt inclusion data (as shown, e.g. by Sadofsky et al.
For S, the concentration values do not show any marked 2008). As at the Marianas, the SVZ contrasts with the
relationship to potentially controlling subduction parame- CAVA such that fractionation-corrected melt inclusion S
ters. High melt inclusion S concentrations are confined to and Cl concentrations do not correlate with the La/Nb of
the most primitive tephras from small cinder cones, the host rock (Fig. 13), emphasising the need to test the
whereas the lowest values are found in high-degree melts applicability of volatile proxies for each arc system
of Volcán Osorno and in the more evolved stratovolcanoes. individually.
Two potential explanations for this correlation are (1) S
loss through degassing during differentiation and/or (2) S
removal in a fractionating S-rich phase. The only S-rich Conclusion
phase observed in the melts, however, was apatite in the
sample from Volcán Maipo. Therefore, it is likely that Mafic to intermediate olivine-hosted melt inclusions con-
degassing played the dominate role in depleting the more tained in small tephra particles erupted from young vol-
evolved magmas in S. canoes along the Chilean Southern Volcanic Zone were
investigated for their major element, S and Cl geochem-
Cl and S inventories of the SVZ compared to the Central istry. The pre-eruptive melt volatile contents, combined
American Volcanic Arc and to the Marianas Arc, with whole-rock trace element geochemistry, allow us to
and the validity of volatile proxies examine the potential relationship between differences in
the subduction system to the magmatic volatile inventory.
As at the SVZ, Cl correlates positively with La/Sm in the In the NSVZ, where magmas ascend through thick crust,
melt inclusions of the Central American Volcanic Arc melt inclusions captured the most primitive compositions
(CAVA), whereby the Cl abundances exceed those of the found to date. The lack of correlation between crustal
SVZ at any given La/Sm (Fig. 11a). In contrast, no co- thickness and K2O in the SiO2 B 53 wt% melt inclusions
variation is seen at the CAVA between Cl and Nb/Y analysed in this study contrasts with the whole-rock com-
(Fig. 11b). Therefore, La/Sm seems to be a more universal positions normalised to SiO2 = 57.5 wt% of Hildreth and
proxy of Cl abundance in arc melts than Nb/Y. Figure 12 Moorbath (1988), suggesting that crustal assimilation may
illustrates the variation in S and Cl concentrations along- not be the process controlling the composition of more
strike of the SVZ, the CAVA, and the Marianas subduction mafic melts along the SVZ. Melt inclusion S contents
system. Melt inclusions from the SVZ have Cl abundances correlate inversely with indicators for magma evolution
in the same range as those from the Central American arc such as K2O, suggesting removal of S during differentia-
[except for few high-Cl inclusions from Irazú (Sadofsky tion by degassing/partitioning of S into a fluid/vapour
et al. 2008; Wehrmann et al. 2011)] and to the Marianas phase and/or fractionation of S-bearing minerals such as
arc, implying that the continental versus oceanic nature of a apatite. The SVZ is characterised by hydrous flux melting,

123
Int J Earth Sci (Geol Rundsch)

Fig. 12 Along-arc variations in 4000

Cl and S at Fo91 contents of the CONTINENTAL TRANSITIONAL OCEANIC
SVZ CAVA Marianas
SVZ (left, red diamonds)
N S NW SE N S
compared to the CAVA (middle,
green circles) and to the 3000
Marianas (right, blue squares,
as an example for an oceanic

S at Fo (µg/g)
arc). CAVA data from Sadofsky
2000

91
et al. (2008) and Wehrmann
et al. (2011), Marianas data
from Shaw et al. (2008)
1000

0
3000

2000
Cl at Fo (µg/g)
91

1000

0
0 200 400 600 800 1000 0 200 400 600 800 1000 0 200 400
Distance along the arcs (km)

2000 located between the stratovolcanoes, whereby Cl at Fo91

was found to inversely correlate with volcanic edifice size.
With the exception of the Los Hornitos cones, the inte-
grated volumetric and geochemical systematics along the
1500
SVZ show that the small inter-stratovolcano vents have a
weaker slab signature than the stratovolcanoes, substanti-
Cl at Fo 91(µg/g)

ating the edifice-size-to-slab-signal relationship previously

1000
demonstrated for Villarrica (Hickey-Vargas et al. 2002,
2003). This implies that stratovolcanoes form where there
SVZ is an enhanced subduction input into the mantle wedge,
A
AV resulting in larger degrees of melting and dilution of
C
500 incompatible elements including Cl. Beneath the small
intra-arc centres, in contrast, fluid activity is reduced,
SVZ
leading to an increase in incompatible elements including
CAVA Cl, in response to low degrees of melting.
0.0 In comparison to the CAVA and to the Marianas, the
0 1 2 3 4 5 6
SVZ shows similar magmatic Cl and somewhat reduced S
La/Nb
contents. This indicates a predominant lower plate control
Fig. 13 Melt inclusion Cl at Fo91 versus La/Nb of the SVZ (red on Cl geochemistry as recorded by melt inclusions, with no
diamonds) compared to the CAVA (green circles). While La/Nb was fundamental difference between continental, transitional,
validated as a trace element proxy for Cl at the CAVA (Sadofsky and oceanic arc systems.
et al. 2008), no such correlation can be stated for the SVZ
Acknowledgments We thank Paul van den Bogaard, Jorge Clavero,
showing peak slab fluid proxies such as Pb/Ce and Ba/Nb Katja Hockun, and Rayén Rivera-Vidal for their assistance in the
field. Sampling campaigns were also supported by helicopter pilot
associated with the highest degrees of melting in the central
Eduardo Boisset and by the Chilean arrieros Titin, Gerardo, Valde-
SVZ. The highest volatile concentrations occur in low- mar, Roberto, Nano, and their colleagues with horses and guidance.
degree melts erupted at small monogenetic cinder cones Credit is due to Philipp Rohde and Silvia Gütschow for their support

123
 Int J Earth Sci (Geol Rundsch)

with sample preparation. Mario Thöner and Ulrike Westernströer slab and continental plate in the region of the 1960 Valdivia
contributed valuable help with performing the electron microprobe (Chile) slip maximum—insights into fluid release and plate
and ICP-MS analyses. We wish to thank Patricia Sruoga and an coupling. Earth Planet Sci Lett 331(332C):164–176. doi:10.
anonymous reviewer for their helpful suggestions, and we are also 1016/j.epsl.2012.02.006
grateful to an anonymous reviewer for comments on an earlier version Ferguson KM, Dungan MA, Davidson JP, Colucci MT (1992) The
of this manuscript. This paper is contribution No. 240 of Sonderf- Tatara-San Pedro Volcano, 36°S, Chile: a chemically variable,
orschungsbereich 574 ‘‘Volatiles and Fluids in Subduction Zones’’ dominnatly mafic magmatic system. J Petrol 33(1):1–43
funded by the German Research Foundation. Freda C, Baker DR, Scarlato P (2005) Sulfur diffusion in basaltic
melts. Geochim Cosmochim Acta 69(21):5061–5069
Gaetani GA, O’Leary JA, Shimizu N, Bucholz CE, Newville M
(2012) Rapid reequilibration of H2O and oxygen fugacity in
olivie-hosted melt inclusions. Geology 40:915–918. doi:10.1130/
References G32992.1
Garbe-Schönberg CD (1993) Simulataneous determination of thirty-
Baker DR, Moretti R (2011) Modeling the solubility of sulfur in seven trace elements in twenty-eight international rock standards
magmas: a 50-year old geochemical challenge. Rev Mineral by ICP-MS. Geostand Newslett 17:81–97. doi:10.1111/j.1751-
Geochem 73:167–213. doi:10.2138/rmg.2011.73.7 908X.1993.tb00122.x
Baker DR, Freda C, Brooker RA, Scarlato P (2005) Volatile diffusion Guthrie KM, Wulff AH (2002) Magma petrogenesis of stratigraph-
in silicate melts and its effets on melt inclusions. Ann Geophys ically collected flows from Volcán Cerro Azul, Southern
48:699–717 Volcanic Zone, Chile. Abstract, GSA joint annual meeting,
Beermann O, Botcharnikov RE, Holtz F, Diedrich O, Nowak M Lexington, Kentucky, USA
(2011) Temperature dependence of sulfide and sulfate solubility Hickey RL, Gerlach DC, Frey FA (1984) Geochemical variations in
in olivine-saturated basaltic magmas. Geochim Cosmochim Acta volcanic rocks from central-south Chile (33°–42°S). In: Harmon
75:7612–7631. doi:10.1016/j.gca.2011.09.024 RS, Barreiro BA (eds) Andean magmatism: chemical and
Behrens H, Stelling J (2011) Diffusion and redox reactions of sulfur isotopic constraints. Shiva Publishing Ltd, Nantwich, UK,
in silicate melts. Rev Mineral Geochem 73:79–111. doi:10.2138/ pp 72–95
rmg.2011.73.4 Hickey RL, Frey FA, Gerlach DC, López-Escobar L (1986) Multiple
Binder B, Keppler H (2011) The oxidation state of sulfur in magmatic sources for basaltic arc rocks from the Southern Volcanic Zone
fluids. Earth Planet Sci Lett 301:190–198. doi:10.1016/j.epsl. of the Andes (34–41 S): trace element and isotopic evidence for
2010.10.042 contributions from subducted oceanic crust, mantle and conti-
Bouvet de Maisonneuve C, Dungan MA, Bachmann O, Burgisser A nental crust. J Geophys Res 91:5963–5983
(2012) Insights into shallow magma storage and crystallization Hickey-Vargas R, Sun M (2003) Decompression and flux melting in
at Volcán Llaima (Andean Southern Volcanic Zone, Chile). the Chilean Southern Volcanic Zone: implications for slab-
J Volcanol Geotherm Res 211–212:76–91. doi:10.1016/j.jvol wedge and wedge-crust mass transfer. State of the Arc confer-
geores.2011.09.010 ence, Timberline, Oregon, USA
Bouvet de Maisonneuve C, Dungan MA, Bachmann O, Burgisser A Hickey-Vargas R, Moreno-Roa H, López-Escobar L, Frey FA (1989)
(2013) Petrological insights into shifts in eruptive styles at Geochemical variations in Andean basaltic and silicic lavas from
Volcán Llaima (Chile). J Petrol 54(2):393–420. doi:10.1093/ the Villarrica-Lanı́n volcanic chain (39.5°S): an evaluation of
petrology/egs073 source heterogeneity, fractional crystallization, and crustal
Carroll MR (2005) Chlorine solubility in evolved alkaline magmas. assimilation. Contrib Mineral Petrol 103:361–386
Ann Geophys 48(4–5):619–631 Hickey-Vargas R, Abdollahi MJ, Parada MA, López-Escobar L, Frey
Contreras-Reyes E, Grevemeyer I, Flueh ER, Reichert C (2008) FA (1995) Crustal xenoliths from Calbuco Volcano, Andean
Upper lithospheric structure of the subduction zone offshore of Southern Volcanic Zone: implications for crustal composition
southern Arauco peninsula, Chile, at 38°S. J Geophys Res and magma-crust interaction. Contrib Mineral Petrol
113:B07303. doi:10.1029/2007JB005569 119:331–344
Costa F, Singer B (2002) Evolution of Holocene dacite and Hickey-Vargas R, Sun M, López-Escobar L, Moreno-Roa H, Reagan
compositionally zoned magma, Volcán San Pedro, Southern MK, Morris JD, Ryan JG (2002) Multiple subduction compo-
Volcanic Zone, Chile. J Petrol 43(8):1571–1593 nents in the mantle wedge: evidence from eruptive centers in the
Cottrell E, Kelley KA (2013) Redox heterogeneity in mid-ocean ridge Central Southern volcanic zone, Chile. Geology 30(3):199–202.
basalts as a function of mantle source. Science 340:1314–1317. doi:10.1130/0091-7613(2002)030
doi:10.1126/science.1233299 Hildreth W, Moorbath S (1988) Crustal contributions to arc magma-
Danyushevsky L, McNeill AW, Sobolev AV (2002) Experimental tism in the Andes. Contrib Mineral Petrol 98:455–489
and petrological studies of melt inclusions in phenocrysts from Jacques G, Hoernle K, Gill J, Hauff F, Wehrmann H, Garbe-
mantle-derived magmas: an overview of techniques, advantages Schönberg D, van den Bogaard P, Bindeman I, Lara LE (2013)
and complications. Chem Geol 183:5–24 Across-arc geochemical variations in the Southern Volcanic
Davidson JP, Dungan MA, Ferguson KM, Colucci MT (1987) Crust- Zone, Chile (34.5- 38.0°S): constraints on mantle wedge and
magma interactions and the evolution of arc magmas: the San input compositions. Geochim Cosmochim Acta 123:218–243.
Pedro-Pellado volcanic complex, southern Chilean Andes. doi:10.1016/j.gca.2013.05.016
Geology 15:443–446 Jugo PJ (2009) Sulfur content at sulfide saturation in oxidized
Dungan MA, Wulff A, Thomson R (2001) Eruptive stratigraphy of magmas. Geology 37:415–418. doi:10.1130/G25527A.1
the Tatara-San Pedro Complex, 36°S, Southern Volcanic Zone, Kay SM, Godoy E, Kurtz A (2005) Episodic arc migration, crustal
Chilean Andes: reconstruction method and implications for thickening, subduction erosion, and magmatism in the south-
magma evolution at long-lived arc volcanic centers. J Petrol central Andes. Geol Soc Am Bull 117(1-2):67–88. doi:10.1130/
42(3):555–626 B25431.1
Dzierma Y, Rabbel W, Thorwart M, Koulakov I, Wehrmann H, Kelley KA, Cottrell E (2012) The influence of magmatic differen-
Hoernle K, Comte D (2012) Seismic velocity structure of the tiation on the oxidation state of Fe in a basaltic arc magma. Earth

123
Int J Earth Sci (Geol Rundsch)

Planet Sci Lett 329–330:109–121. doi:10.1016/j.epsl.2012.02. Saal AE, Frey FA, Delpino D, Bermudez A (1993) Geochemical
010 characteristics of alkalic basalts erupted behind the Andean
Kent AJR (2008) Melt inclusions in basaltic and related volcanic rocks. volcanic front (35–37°S): constraints on sources and processes
Rev Mineral Geochem 73:273–331. doi:10.2138/rmg.2011.69.8 involved in continental arc magmatism. EOS Trans. AGU 74,
Klimm K, Kohn SC, Botcharnikov RE (2012) The dissolution mechanism (43), Fall Meet Suppl, Abstract F652
of sulphur in hydrous silicate melts. II: solubility and speciation of Sadofsky SJ, Portnyagin MV, Hoernle K, van den Bogaard P (2008)
sulphur in hydrous silicate melts as a function of fO2. Chem Geol Subduction cycling of volatile and trace elements through the
322–323:250–267. doi:10.1016/j.chemgeo.2012.04.028 Central American volcanic arc: evidence from melt inclusions.
López-Escobar L, Frey FA, Vergara M (1977) Andesites and high- Contrib Mineral Petrol 155(4):433–456. doi:10.1007/s00410-
alumina basalts from the Central-South Chile high andes: 007-0251-3
geochemical evidence bearing on their petrogenesis. Contrib Sadofsky SJ, Hoernle K, Duggen S, Hauff F, Werner R, Garbe-
Mineral Petrol 63:199–228 Schönberg D (2009) Geochemical variations in the Cocos Plate
López-Escobar L, Parada MA, Moreno H, Frey FA, Hickey-Vargas R subducting beneath Central America: implications for the
(1992) A contribution to the petrogenesis of Osorno and Calbuco composition of arc volcanism and the extent of the Galápagos
volcanoes, Southern Andes (41°–41°300 S): comparative study. Hotspot influence on the Cocos oceanic crust. Int J Earth Sci
Rev Geol Chile 19:211–226 98:901–913. doi:10.1007/s00531-007-0289-5
López-Escobar L, Cembrano J, Moreno H (1995a) Geochemistry and Salters VJM, Stracke A (2004) Composition of the depleted mantle.
tectonics of the Chilean Southern Andes basaltic Quaternary Geochem Geophys Geosyst 5(5):Q05004. doi:10.1029/
volcanism (37°–46°S). Rev Geol Chile 22(2):219–234 2003GC000597
López-Escobar L, Kempton PD, Moreno H, Parada MA, Hickey- Scaillet B, Clemente B, Evans BW, Pichavant M (1998) Redox
Vargas R, Frey FA (1995b) Calbuco volcano and minor eruptive control of sulfur degassing in silicic magmas. J Geophys Res
centers distributed along the Liquine-Ofqui fault zone, Chile 103:23937–23949
(41°–42°S): contrasting origin of andesitic and basaltic magma Schumann K (2010) Geochemical investigation of the three volcanoes
in the Southern Volcanic Zone of the Andes. Contrib Mineral Casablanca, Osorno and Cabeza de Vaca, Southern Volcanic
Petrol 119:345–361 Zone, Chile. Unpublished Diploma thesis, University of Halle-
Mahlke, J (2009) Geochemical evolution of Llaima and Villarrica Wittenberg, Germany
Volcanoes, Central Chile. Unpublished diploma thesis, Univer- Sellés D, Rodrı́guez C, Dungan MA, Naranjo JA, Gardeweg M (2004)
sity of Kiel, Germany Geochemistry of Nevado de Longavı́ volcano (36.2°S): a
McMillan NJ, Harmon RS, Moorbath S, López-Escobar L, Strong DF compositionally atypical arc volcano in the Southern Volcanic
(1989) Crustal sources involved in continental arc magmatism: a Zone of the Andes. Rev Geol Chile 31:293–315
case study of Volcán Mocho-Choshuenco, southern Chile. Shaw AM, Hauri EH, Fischer TP, Hilton DR, Stern RJ (2007)
Geology 17:1152–1156 Geochemical studies of the Izu-Bonin-Mariana subduction
Métrich N, Wallace P (2008) Volatile abundances in basaltic magmas system: highlights, progress and future directions. Eos Trans.
and their degassing paths tracked by melt inclusions. Rev AGU, 85(50), Fall Meet Suppl, Abstract T44C-01
Mineral Geochem 69:363–402. doi:10.2138/rmg.2008.69.10 Shaw AM, Hauri EH, Fischer TP, Hilton DR, Kelley KA (2008)
Muñoz J, Stern CR (1989) Alkaline magmatism within the segment Hydrogen isotopes in Mariana arc melt inclusions: implications
38°–39° S of the Plio-Quaternary volcanic belt of the southern for subduction dehydration and the deep-Earth water cycle. Earth
South American margin. J Geophys Res 794(B4):4545–4560 Planet Sci Lett 275:138–145. doi:10.1016/j.epsl.2008.08.015
Parat F, Holtz F, Streck MJ (2011) Sulfur-bearing accessory minerals. Stern CR (2004) Active Andean volcanism: its geologic and tectonic
Rev Mineral Geochem 73:285–314. doi:10.2138/rmg.2011.73.10 setting. Rev Geol Chile 31(2):161–206
Portnyagin M, Hoernle K, Plechov P, Mironov N, Khubunaya S Stern CR, Futa K, Muehlenbachs K, Dobbs FM, Muñoz J, Godoy E,
(2007) Constraints on mantle melting and composition and Charrier R (1984) Sr, Nd, Pb, and O isotope composition of late
nature of slab components in volcanic arcs from volatiles (H2O, Cenozoic volcanics, northernmost SVZ (33°–34°S). In: Harmon
S, Cl, F) and trace elements in melt inclusions from the RS, Barreiro BA (eds) Andean magmatism, chemical and
Kamchatka Arc. Earth Planet Sci Lett 255:53–69. doi:10.1016/j. isotopic constraint. Shiva Publ, Cheshire, pp 96–105
epsl.2006.12.005 Straub SM, Layne GD (2003) The systematics of chlorine, fluorine,
Rea JC (2009) The petrology and geochemistry of Volcán Callaqui, and water in Izu arc front volcanic rocks: implications for
Chile. Dissertation, Weselyan University, 184 p volatile recycling in subduction zones. Geochim Cosmochim
Rea JC, Varekamp JC, Mandeville C, Goss AR, Colvin A (2009) Acta 67(21):4179–4203. doi:10.1016/S0016-7037(00)00307-7
Regional magmatic setting of Callaqui Volcano, (S-Andes, Tabazadeh A, Turco RP (1993) Stratospheric chlorine injection by
Chile). Abstract, Portland GSA annual meeting, 33-14 volcanic eruptions: HCl scavenging and implication for ozone.
Reubi O, Bourdon B, Dungan MA, Koornneef JM, Sellés D, Science 260:1082–1086
Langmuir CH, Aciego S (2011) Assimilation of the plutonic Tassara A, Götze HJ, Schmidt S, Hackney R (2006) Three-dimen-
roots of the Andean arc controls variations in U-series disequi- sional density model of the Nazca plate and the Andean
libria at Volcán Llaima, Chile. Earth Planet Sci Lett 303:37–47. continental margin. J Geophys Res 111(B09404). doi:10.1029/
doi:10.1016/j.epsl.2010.12.018 2005JB003976
Robock A (2000) Volcanic eruptions and climate. Rev Geophys Tormey DR, Hickey-Vargas R, Frey FA, López-Escobar L (1991)
38:191–219 Recent lavas form the Andean volcanic front (33 to 42°S);
Rodrı́guez C, Sellés D, Dungan M, Langmuir C, Leeman W (2007) interpretations of along-arc compositional variations. Geol Soc
Adakitic dacites formed by intrcrustal crystal fractionation of Am Spec Pap 265:57–77
water-rich parent magmas at Nevado de Longavı́ volcano Tormey DR, Frey FA, López-Escobar L (1995) Geochemistry of the
(36.2°S; Andean Southern Volcanic Zone, Central Chile). active Azufre–Planchon–Peteroa volcanic complex, Chile
J Petrol 48/11:2033–2061. doi:10.1093/petrology/emg049 (35°150 S): evidence for multiple sources and processes in a
Ruprecht P, Bergantz GW, Cooper KM, Hildreth W (2012) Quizapu, cordilleran arc magmatic system. J Petrol 36(2):265–298
Chile, and the effects of magma mixing on magma diversity. Völker D, Kutterolf S, Wehrmann H (2011) Comparative mass
J Petrol 53(4):801–840. doi:10.1093/petrology/egs002 balance of volcanic edifices at the Southern Volcanic Zone of the

123
 Int J Earth Sci (Geol Rundsch)

Andes between 33°S and 46°S. J Volcanol Geotherm Res zone inferred from melt inclusions in olivine crystals from mafic
205:114–129. doi:10.1016/j.volgeores.2011.03.011 tephras. Geochem Geophys Geosys 12(6):Q06003. doi:10.1029/
Wallace PJ (2005) Volatiles in subduction zone magmas: concentra- 2010GC003412
tions and fluxes based on melt inclusions and volcanic gas data. Wehrmann H, Hoernle K, Garbe-Schönberg D, Jacques G, Mahlke J,
J Volcanol Geotherm Res 140:217–240. doi:10.1016/j.volgeores. Schumann K (2014) Insights from trace element geochemistry as
2004.07.23 to the roles of subduction zone geometry and subduction input
Wallace PJ, Edmonds M (2011) The sulfur budget in magmas: on the chemistry of arc magmas. Int J Earth Sci. doi:10.1007/
evidence from melt inclusions, submarine glasses, and volcanic s00531-013-0917-1
gas emissions. Rev Mineral Geochem 73:215–246. doi:10.2138/ Witter JB, Kress VC, Delmelle P, Stix J (2004) Volatile degassing,
rmg.2011.73.8 petrology, and magma dynamics of the Villarrica Lava Lake,
Watt SFL, Pyle DM, Mather TA, Naranjo JA (2013) Arc magma Southern Chile. J Volcanol Geotherm Res 134:303–337. doi:10.
compositions controlled by linked thermal and chemical gradi- 1016/j.volgeores.2004.03.002
ents above the subducting slab. Geophys Res Lett Zajacz Z, Halter W (2009) Copper transport by high temperature,
40(11):2550–2556. doi:10.1002/grl.50513 sulfur-rich magmatic vapor: evidence from silicate melt and
Webster JD, Kinzler RJ, Mathez EA (1999) Chloride and water vapor inclusions in a basaltic andesite from the Villarrica
solubility in basalt and andesite melts and implications for volcano (Chile). Earth Planet Sci Lett 282:115–121. doi:10.
magmatic degassing. Geochim Cosmochim Acta 63:729–738 1016/j.epsl.2009.03.006
Wehrmann H, Hoernle K, Portnyagin M, Wiedenbeck M, Heydolph K
(2011) Volcanic CO2 output at the Central American subduction

123