Pacific offshore record of plinian arc volcanism in Central America: 2. Tephra volumes and erupted masses



[1] Sediment gravity cores collected from the Pacific seafloor offshore Central America contain numerous distal ash layers from plinian-type eruptions at the Central American Volcanic Arc dating back to more than 200 ka. In part 1 of this contribution we have correlated many of those ash layers between cores and with 26 tephras on land. The marine ash layers cover areas of up to 106 km2 in the Pacific Ocean and represent a major fraction (60–90%) of the erupted tephra volumes because the Pacific coast lies within a few tens of kilometers downwind from the volcanic arc. Combining our own mapping efforts on land and published mapping results with our marine data yields erupted volumes of all major tephras along the arc that range from ∼1 to 420 km3. Recalculated to erupted magma mass, the widespread tephras account for 65% of the total magma output at the arc. Complementing our tephra data with published volumes of the arc volcanic edifices and volcano ages, we calculate the long-term average magma eruption rates for each volcano. Moreover, we use incompatible element variations to calculate the cumulate masses that were fractionated during variable degrees of differentiation. This yields a minimum estimate of long-term average magma production rate at each volcano, because intrusives without surface expression and losses by erosion are not accounted for. Peak magma production rates increase from Costa Rica to Guatemala, but there is considerable scatter within each region and large differences even between neighboring volcanoes.

1. Introduction

[2] Ash plumes of numerous plinian, phreatoplinian and ignimbrite-forming eruptions from calderas and stratocones along the Central American Volcanic Arc (CAVA) were dispersed westward across the Pacific at stratospheric heights [Kutterolf et al., 2007a, 2008]. The resulting marine ash layers cover areas of up to 106 km2 in the Pacific Ocean (Figure 1) and represent a major fraction of the erupted tephra volumes because the CAVA volcanoes lie within a few tens of kilometers east of the Pacific coast (Figure 1).

Figure 1.

Shaded and colored SRTM elevation model of Central America (NASA/JPL/NGA, 2000) and high-resolution bathymetry along the Middle America Trench (MAT) from Ranero et al. [2005]. The line of Central American arc volcanoes runs through the two large lakes and parallel to the trench at about 200 km distance. Names of numbered volcanoes are listed at the bottom left, also giving major tephras: WFT, W-fall Tephra; LCY, Los Chocoyos Tephra; LFT, L-fall Tephra; EFT, E-fall Tephra; MFT, Mixta Tephra; PAT, Pinos Altos Tephra; ACT, Arce Tephra; CGT, Congo Tephra; CCT, Conacaste Tephra; OPI, Older pumice Ilopango; TB4, Terra Blanca 4 Tephra; TBJ, Terra Blanca Joven Tephra; BRT, Blanca Rosa Tephra; TT/AT, Twins/A-fall Tephra; MCO1-3, Mafic Cosigüina tephras; Laq, Lower Apoyeque Tephra; Uaq, Upper Apoyeque Tephra; CT, Chiltepe Tephra; FT, Fontana Tephra; SAT, San Antonio Tephra; MTL/LCT, Masaya Triple Layer/La Concepción Tephra; MT/TIL, Masaya Tuff/Ticuantepe Lapilli; LAT, Lower Apoyo Tephra; UAT, Upper Apoyo Tephra; UOT, Upper Ometepe Tephra. Dots show core positions of R/V METEOR cruises M66/3a + b and M54/2 and R/V SONNE cruise SO173/3 along and across the trench.

[3] In part 1 we used our database of bulk-rock, glass and mineral major and trace element compositions, petrographic and lithologic characteristics, and eruption ages of all plinian and comparatively large, Pleistocene to recent tephras of the CAVA from Costa Rica to Guatemala to correlate ash layers in the Pacific seafloor with 26 tephras on land and thereby obtained a tephrostratigraphic framework for 1100 km length of the CAVA [Kutterolf et al., 2008]. In this second part of our contribution, we use these correlations, our mapping results on land, and published isopach maps to better constrain erupted volumes of the widespread tephras and thus the magnitudes of the large eruptions that occurred at the CAVA during the past 200 ka.

[4] A fundamental problem with subduction zones is the budgeting of material input versus material output. The most significant output is the flux of magma through the volcanic arc, because this also largely determines the associated output fluxes of water and other volatiles. Previous attempts to determine magma fluxes at the CAVA considered magma masses stored in the volcanic edifices. Here we improve such estimates by including the magma masses represented by the widespread tephras which, as will be shown below, constitute a large part of the total magma output.

2. Geological Setting

[5] The Central American Volcanic Arc (CAVA) extends from Panama to Guatemala and runs roughly parallel to, and 150–200 km away from, the deep-sea trench. This middle America trench results from the subduction of the Cocos plate beneath the Caribbean plate at a convergence rate of 70–90 mm/a [Barckhausen et al., 2001; DeMets, 2001]. The volcanic arc resulting from this subduction is one of the most active arcs on Earth and produced numerous plinian eruptions in the last several hundred thousand years. Easterly winds prevailing in the lower stratosphere distributed the ash across the Pacific Ocean where resulting ash layers provide marker beds in the mostly non-erosive submarine environment.

[6] The observations of numerous active bend faults across the outer rise of the Cocos plate penetrating the crust and uppermost mantle [Ranero et al., 2003], and the anomalous heat flow and seismic velocities indicative of substantial hydration by seawater invading the faults [Grevemeyer et al., 2005], suggest hydrated crust and serpentinized mantle as major carriers of water that ultimately drives melting in the mantle wedge and arc volcanism [Rüpke et al., 2002]. The volcanic front in Nicaragua shifted to its present position about 8 Ma ago probably in response to re-arrangement of the subduction angles [Barckhausen et al., 2001; DeMets, 2001; Ehrenborg, 1996] whereas it has had a more or less stable position in Costa Rica and Guatemala. The arc is tectonically segmented by Caribbean tectonic structures as well as by strike-slip tectonics caused by slightly oblique subduction [DeMets, 2001; Ranero et al., 2005]. Slab dip varies between 40° and 75° along the subduction zone [Cruciani et al., 2005; Protti et al., 1995; Syracuse and Abers, 2006].

[7] Variations in the nature of the incoming plate [Hoernle et al., 2002], in crustal thickness and composition [Carr, 1984] and the tectonic setting, are paralleled by along-arc variations in the composition of the volcanic rocks [Carr et al., 2003, 2007a; Carr, 1984; Feigenson and Carr, 1986; Feigenson et al., 2004; Hoernle et al., 2002; Patino et al., 1997, 2000] and the magnitudes of eruptions [Rose et al., 1999]. In addition, variable degrees of magmatic differentiation led to compositions ranging from basalt through rhyolite. Such compositional variations greatly assisted the geochemical correlations with marine ash beds presented in part 1 [Kutterolf et al., 2008].

3. CAVA Tephrostratigraphy

[8] A number of publications have investigated tephrostratigraphic successions in middle to northern Central America [e.g., Commision Ejecutiva Hidroelectrica del Rio Lempa [CEL] 1992; CEL, 1995; Drexler et al., 1980; Freundt et al., 2006; Hart, 1983; Koch and McLean, 1975; Kutterolf et al., 2007b; Newhall, 1987; Pérez and Freundt, 2006; Peterson and Rose, 1985; Pullinger, 1998; Rose, 1987; Rose et al., 1999; Wehrmann et al., 2006; Wundermann and Rose, 1984, Scott et al., 2006; Self et al., 1989].

[9] As described in part 1, we have used these studies to collect samples for our compositional data-base in Central America, and extended earlier studies in El Salvador and Nicaragua by own mapping and stratigraphic work in collaboration with the local geological services (SNET, San Salvador; INETER, Managua; INSIVUMEH, Guatemala City).

[10] Upper Pleistocene (since ∼500 ka) to Holocene arc volcanism of Central America formed a number of large caldera volcanoes which produced large-magnitude eruptions of highly evolved, silicic magmas [Rose et al., 1999]. Although large calderas are less common in Nicaragua and Costa Rica, volcanoes that generated major plinian eruptions were frequently active in Nicaragua since the Upper Pleistocene. Therefore the Pacific submarine sediment successions sampled offshore Central America contain ash layers from particularly large eruptions at the Ayarza, Amatitlán and Atitlán calderas in Guatemala, the Berlin-Chinameca complex, Ilopango Caldera, and Coatepeque Caldera in El Salvador and Apoyo Caldera, Masaya Caldera, Chiltepe volcanic complex, and Cosigüina volcano in Nicaragua (Figure 1). The cores offshore southern Nicaragua and Costa Rica show mostly ash layers from particularly large eruptions at Concepción volcano, as well as eruptions of Barva volcano in Costa Rica.

4. Methods

4.1. Marine Ash Layer Correlations

[11] During R/V METEOR cruises M54/2 and M66/3 as well as R/V SONNE cruise SO173/3, we collected 56 sediment gravity cores offshore Central America. These cores were located between 9°12′N/84°39′E and 12°15′N/91°30′E on the oceanic plate and the continental slope, at distances of 150–530 km from the CAVA (Figure 1). The cores contain 213 ash horizons including primary ash layers and slightly reworked ash that retained its compositional integrity and stratigraphic position. Criteria identifying primary and variably reworked ash horizons, and the methods employed to correlate them with deposits on land are described and discussed in part 1 [Kutterolf et al., 2008].

4.2. Tephra Distribution, Volumes, and Masses

[12] To construct the isopach maps of the tephra layers onshore, we logged ∼100 outcrops in El Salvador and Guatemala to complement thickness data from earlier studies [CEL, 1992, 1995; Rose et al., 1987; Wundermann, 1982; Wundermann and Rose, 1984] (Figure 2). We also include unpublished thickness data collected and kindly provided by Carlos Pullinger, Dolors Ferres and Walter Hernandez of the Servicio Nacional de Estudios Territoriales (SNET) in El Salvador. In west-central Nicaragua, we have revised and extended earlier work by Bice [1985], identified and dated tephras not previously recognized, and constructed isopach and isopleth maps of all these deposits [Freundt et al., 2006; Kutterolf et al., 2007b; Pérez and Freundt, 2006; Wehrmann et al., 2006].

Figure 2.

Selected isopach maps for the Tiribi Tuff, Upper Apoyo Tephra (UAT), and Congo Tephra (CGT), with isopachs on land taken from the references mentioned in the text. For clarity, we do not distinguish between well-constrained and poorly constrained isopach sections but note that the pattern of offshore isopachs is constrained by a few available data only. Isopach maps of every correlated tephra can be found in auxiliary material Figure S1.

[13] Here we further extend land-based results by including the offshore data. Since the offshore thickness data is sparse, the shape of the distal isopachs can only be estimated and introduces some error into the volume calculations. Total tephra volumes are obtained by fitting straight lines to data on plots of ln [isopach thickness] versus square root [isopach area] following Pyle [1989] and Fierstein and Nathenson [1992] and integrating to infinity. In most cases, the data required two straight-line segments to fit proximal to medial and distal data separately. We use the point of intersection of these line segments to distinguish between the volume contributions from proximal to medial and distal facies.

[14] We also use different approaches to convert tephra volume to magma mass over proximal to medial and distal regions. The proximal to medial volume of each tephra was reduced by 50% to account for interparticle pore space (space between pumice clasts) and lithic contents. Measured bulk densities of pumice and scoria lapilli range from 400 to 810 kg/m3 [e.g., Kutterolf et al., 2007b], depending on composition. We used average values of 600 kg/m3 for felsic and 800 kg/m3 for mafic tephras to convert the remaining volume to magma mass.

[15] In the distal, marine region primary ash layers have a sharp base but gradually change into mixtures of ash and pelagic sediment at the top. From analyses of such mixed sediments we conclude that, on average, 20% should be added to the primary ash layer thickness. Balancing this with the interparticle space (space between glass shards filled with water; measured average of ∼40%) to be subtracted, we have reduced distal tephra volumes by an average of ∼30% (25–35% per tephra). Measured average ash-particle densities of 2100 kg/m3 for felsic and 2400 kg/m3 for mafic marine tephras are used to convert distal volumes to magma masses.

5. Tephra Volumes and Magma Masses

[16] Selected isopach maps and ln (thickness) versus square root (isopach area) variations are shown in Figures 2 and 3. The respective figures, Figures S1 and S2, for the remaining tephras are available as auxiliary material. Single tephra volumes are summarized in Table 1 and are shown schematically along the CAVA in Figure 4, according age and regional distribution. Very fine ash from large eruptions is transported to much larger distances than investigated here and may even circle the globe. Such very distal deposits may produce a still shallower slope in ln (thickness) versus square root (isopach area) diagrams such that the volumes and masses we have determined are still minimum estimates. We now discuss every tephra according to the source volcanic centers in geographic order from north to south, using the correlations to marine ash layers established in part 1 [Kutterolf et al., 2008].

Figure 3.

Natural logarithm of isopach thickness versus square root of isopach area for three selected tephras: Fontana tephra (FT), Congo Tephra (CGT), and L-fall Tephra (LFT). For all other mapped tephras, see auxiliary material Figure S2. As in the examples, most tephra data can be fitted by two line segments.

Figure 4.

Composite tephrostratigraphy of Central America showing the position of source volcanoes along the arc versus age of tephras as in Kutterolf et al. [2008] but with violet symbol size scaled to erupted volume (diameter of volume-scaled sphere). Numbers next to each circle give the age of the tephra. See Figure 1 for acronyms. Green circles above each volcanic center represent the cumulative tephra volumes, the arrow indicating an increase northward.

Table 1. Summary of Correlated Tephras With Core Positions, Ages, Maximum Distance to Source, and Volume Estimations of Investigated Fallout Tephras
TephraAgeaCorrelation NumberDistance to Source, kmProximal Tephra Fallout Volume, km3Distal Tephra Volume, km3Total Fallout Tephra Volume, km3Estimated Flow Volumes After LiteratureApprox. Total Magma Mass, 1013 kg
  • a

    D, dating field tephra; S, dating from sedimentation rates.

TBJ1.6 ka; DC13903238,670,6(na)6.6
MT/TIL1.8 ka; SC22004.
CT1.9 ka; SC35703.91417.92.2
MTL/LCT2.1 ka; DC41700.
SAT6 ka; SC53300.51313.52.2
Uaq12.4 ka; DC63002.
LAq17 ka; SC72100.8+3,13.90.5
UOT19 ka; SC82802.9+
MCO21–23 ka; SC92201.5+4.560.4
UAT24.5 ka; DC105307.235.742.987.2(y)
LAT25 ka; DC1127033.56.50.6
PAT23 ka; D 460—–(+)(+)2(#)0.3
TB4∼36 ka; S(228) CGTC1238025,910,436,32.3
MFT39 ka after S(228) CGTC139402,96,191
CCT∼51 ka; S(228)C14320(+)(+)11,210.3
EFT51 ka; S(D-3; Bowles) 860540456
CGT∼53 ka; DC153205.5(+)12.6(+)18.15∼2
FT60 ka; SC163301.
TT/AT60 ka; S(222)C1727019.410.41.4
ACT75 ka; DC183209.66.616.2101.3
BRT75 ka; DC192001.92.74.620.7
OPI75–84 ka; SC20470(na)(na)(na)(na)(na)
LCY84 ka; DC211900(na)(na)420(*)20059(*)
TFT119 ka; D 940∼8(X)9(na)173.7
WFT158 ka; DC2256013.69.723.331.8
LFT191 ka; DC2381018.544.763.2127.5
Tiribi322 ka; D 23035(2)4278 2.2

5.1. Atitlán Caldera

[17] Atitlán Caldera is the source of an at least 200 ka old tephra succession comprising from old to young W-fall and flow tephra, Los Chocoyos H-fall and flow tephra, D-fall deposits, F-fall deposits, and I-fall deposits [Rose et al., 1987]. We recognized two of these tephras in the Pacific record. The W-tephra on land comprises fall and pyroclastic flow deposits extending across ∼10,000 km2 with an estimated volume of 12 km3 tephra (9 km3 fall, 3 km3 flow) [Rose et al., 1999]. Correlation of 4 cm thick marine tephra C22 and an ash layer in the near-by core RC-12-32 of Bowles et al. [1973] to the 158 ± 3 ka old W-tephra enlarges the minimum distribution area (up to 5 cm isopach) to 6x104 km2 and the erupted tephra volume to ∼23.3 km3 or ∼1.8x1013 kg magma mass, respectively (Figures 4, S1, and S2 and Table 1).

[18] The 84 ± 0.5 ka old Los Chocoyos eruption produced the largest known Quaternary tephra in Central America with a tephra volume of 420 km3 (200 km3 flow and 220 km3 fall), i.e., 280 km3 DRE, that covers an minimum area of ∼6 × 106 km2 extending from the Pacific to the Gulf of Mexico [e.g., Drexler et al., 1980; Rose et al., 1987, 1999]. Our new data of correlated ash layer C21 to Los Chocoyos tephra confirms, but does not extend, these estimates and emphasizes the usefulness of this layer as a marker bed across the whole region (Figure 4 and Table 1).

5.2. Amatitlán Caldera

[19] Amatitlán Caldera is the source of at least six tephras which are from bottom to top the L-flow and fall, Z-falls, T-flow and fall, C-fall, E-fall and J-falls [Koch and McLean, 1975; Wundermann, 1982; Wundermann and Rose, 1984]. L-tephra is found in our marine cores as correlated ash layer C23 and, using documented compositional data, also in DSDP Leg 67 as well as in cores of Bowles et al. [1973]. The 191 ± 11 ka old L-tephra on land [Rose et al., 1999] has a total tephra volume of 30–40 km3 estimated by Wundermann and Rose [1984] and includes 12 km3 DRE pyroclastic flow and 6 km3 DRE fall deposits. It is spread out over an area of 1600 km2. With the marine addition, we now estimate a minimum distribution area 7 × 105 km2 (up to 2 cm isopach), corresponding to a tephra volume of 63.2 km3 and a magma mass of 7.5 × 1013 kg including the flow volume (Figures 4, S1, and S2 and Table 1). Correlation of 119 ka old T-Fall Tephra to ash layers in DSDP core (Leg 66/493B-1-3) lead also to very rough estimates of ∼17 km3 and corresponding magma mass of ∼3.7 × 1013 kg.

[20] The E-tephra is a coarse grained reversely graded fall of white pumice clasts distributed across an area of ∼1300km2 on land, corresponding to a tephra volume of >5 km3 (2.5 km3 DRE) [Wundermann, 1982; Wundermann and Rose, 1984]. Correlations based on published data of an ash layer at DSDP Leg 84 (570-2-3), the D3 ash layer of Bowles et al. [1973] as well as the Y5 ash layer in the Gulf of Mexico documented by Rabek et al. [1985], which all have the E-tephra glass composition [Kutterolf et al., 2008], yield a minimum distribution area (up to 1 cm isopach) of 6.3 × 105 km2 (Figure S1), corresponding to a tephra volume of 45 km3 (Figures 4 and S2 and Table 1; 6 × 1013 kg magma mass).

5.3. Ayarza Caldera

[21] Two nested calderas characterize the Ayarza Caldera in the south of Guatemala and produced the 27 ± 1.6 ka old Mixta Tephra, the Pinos Altos Tephra, and the younger (23 ±0.5 ka) Tapalapa Tephra [Peterson and Rose, 1985] (Table 1). The Mixta Tephra is a compositionally zoned tephra with pale brown to black and banded pumice clasts. Outcrops limited to near the source poorly constrain a volume of 0.1 km3 DRE [Peterson and Rose, 1985]. Correlation to ash layer C12 in our core M66/3-228, to glass composition data of the C-layer of Bowles et al. [1973] reported by Drexler et al. [1980] as well as to glass composition data from DSDP Leg 84 (570-2-1/141) of an ash layer at 621 cm bsf, yield a new minimum distribution (up to 1 cm isopach) of 1.3 × 105 km2 (Figure S1) and a tephra volume of ∼9 km3 (Figures 2, 4, and S2 and Table 1), which corresponds to 9.9 × 1012 kg erupted magma mass for the Mixta Tephra.

[22] The Pinos Altos Tephra is a thick pumice fall deposit, and Peterson and Rose [1985] estimated the erupted volume as at least 2 km3 DRE on the basis of its identification at two distal sites on land and in marine core RC12-32 from the Pacific Ocean. Our trace element data confirm the correlation with an ash layer 50 cm bsf in core RC12-32 of Bowles et al. [1973] but we did not find Pinos Altos Tephra in our cores.

5.4. Coatepeque Caldera

[23] A tephra succession of four widespread tephras can be found at Coatepeque Caldera in northern El Salvador starting with the Bellavista eruption at 77 ± 2 ka [Rose et al., 1999] and followed by the 72 ± 3 ka old Arce Tephra, the Congo Tephra (53 ± 3 ka; own radiocarbon dating) and the ∼51 ka old Conacaste Tephra. The Arce tephra is the largest of the four Coatepeque tephras and includes plinian fall beds and ignimbrite. Mapping for a geothermal reconnaissance project [CEL, 1992] yielded a distribution area of 2000 km2 and a total tephra volume of ∼40 km3 (17 km3 DRE). Correlation to ash layer C18 in our cores gives the new minimum distribution area (up to 2 cm isopach) of 1 × 105 km2, and the minimum tephra fall volume is 16.2 km3 (∼1.3 × 1013 kg magma mass) (Figures 4, S1, and S2 and Table 1) to which the volume of the ignimbrite (>10 km3), which we continue to map, must be added. Compositional data of ash beds sampled in the Caribbean by Rabek et al. [1985] (K131-446 cm bsf; TR126-22 321 cm bsf) suggests that these are also distal Arce Tephra but we presently cannot validate this correlation with certainty. If these correlations were true, the Arce tephra volume would be ∼70 km3.

[24] The Congo Tephra is a complex succession of plinian fall, ignimbrite and surge deposits and reconnaissance studies of this tephra suggested a distribution area of ∼900 km2 and a tephra volume of ∼15 km3 (6 km3 DRE [CEL, 1992]). We correlate ash layer C15 of our cores and an ash layer at DSDP Leg 84 [Pouclet et al., 1985] with the Congo Tephra, which leads to 2.8 × 105 km2 area of minimum distribution (up to 1 cm isopach; Figure 2) and a total tephra volume of 18.1 km3, which corresponds to 2 × 1013 kg magma mass (Figures 3 and 4 and Table 1). Again, the volume of the ignimbrite and surge deposits, which we continue to map, has to be added to this fall volume.

[25] The newly described Conacaste Tephra comprises a lower fall section of two pumice lapilli beds bracketing a central fine ash fall extremely rich in accretionary lapilli, and an upper surge package, which were produced by a phreatoplinian eruption. Ash layer C14 in our cores can be correlated to Conacaste Tephra. We still need to map the Conacaste Tephra in more detail; preliminary estimates including the offshore data suggest a minimum distribution (up to 1 cm isopach) of 6 × 104 km2 and a tephra volume of 11.2 km3 corresponding to 3.4 × 1012 kg magma mass (Figures 4, S1, and S2 and Table 1).

5.5. Ilopango Caldera

[26] At least five tephra deposits were produced by the central El Salvadorian Ilopango caldera since the Upper Pleistocene. From youngest to oldest, these are the Tierra Blanca Joven (TBJ) and the TB2, TB3 and TB4 Tephras [Rose et al., 1999]; in addition, there are remains of a deposit from an older eruption at the shore of the caldera lake [Mann et al., 2004]. The A.D. 429 ± 107 old Terra Blanca Joven eruption (TBJ) [Dull et al., 2001] comprises a succession of fall, ignimbrite and surge deposits [Hart and Steen-McIntyre, 1983]. These authors estimated a distribution area of 10,000 km2 and a volume of 18 km3 DRE. The TBJ Tephra can be correlated to ash layer C1 in the upper few decimeters of our Pacific sediment cores and we also recognized it in the core data of Bowles et al. [1973]. Combining the onshore and offshore thickness data we obtain a tephra volume of 70.6 km3, which corresponds to a magma mass of 6.6 × 1013 kg that is distributed across an minimum area minimum (up to 3 cm isopach) of 3 × 105 km2 (Figures 4, S1, and S2 and Table 1).

[27] The ∼36 ka old TB4 Tephra is a prominent white massive pumice lapilli fall deposit widely distributed over El Salvador. We identified ash layer C12 in our cores and ash layers at DSDP Leg 67 as the distal equivalent of TB4. Using reconnaissance mapping on land by CEL [1992], unpublished data by SNET, and the marine data, the TB4 tephra volume becomes 36.3 km3 (2.3 × 1013 kg magma mass; Table1 and Figures 4 and S2) distributed across an minimum area (up to 5 cm isopach) of 7 × 104 km2 (Figure S1).

[28] Additionally we found a thin ash layer in core 226 which has the composition of the older pumice deposit inside Ilopango caldera described by Mann et al. [2004]. These two far-apart data points do not allow us to estimate the tephra volume but do demonstrate that this was another large eruption that occurred at Ilopango between 73–84 ka ago [Kutterolf et al., 2008].

5.6. Berlin-Pacayal-Volcan Group

[29] Ian Nairn and coworkers of DSIR, New Zealand, performed a stratigraphic study of the volcanic deposits from this group of volcanoes in southern El Salvador [CEL, 1995] and identified six major tephras in the Upper Pleistocene succession. These are from old to young: the Blanca Rosa Tephra (75 ± 10 ka), Twins/A-Tephra (∼61 ka), Pacayal 1 Tephra, Volcan Tephra and Pacayal 3 Tephra, but we only could correlate the Blanca Rosa and Twins/A-tephra to Pacific sediment cores. The Twins and A tephras have been previously described as three separate units but Kutterolf et al. [2008] interpret the three layers as the deposits of one eruption producing three thick fall lapilli beds as well as a pyroclastic flow deposit. Nairn and coworkers have estimated the areal distribution as 900 km2 with a total tephra volume of 12 km3 (∼6 km3 DRE). Since Layer C17 in our cores correlates to Twins/A-Tephra the minimum distribution area (up to 3 cm isopach) becomes 1 × 105 km2 (Figure S1) and the tephra volume accounts for 10.4 km3 (1.4 × 1013 kg magma mass; Table 1 and Figures 4 and S2).

[30] Additionally major element glass compositions suggest that an ash layer in core SO173/3-18 may correlate with the Blanca Rosa Tephra, which would lead to an minimum aerial distribution (up to 5 cm isopach) of 2.5 × 104 km2 and a total fall volume of 4.6 km3, yielding 6.5 × 1012 kg magma mass, when proximal pyroclastic flow deposits are included (Figures 4, S1, and S2 and Table 1).

5.7. Cosigüina Caldera

[31] Cosigüina volcano at the northern end of the Nicaraguan volcanic arc produced several widespread tephras prior to the last plinian eruption in A.D. 1835 which is the only one studied [Williams, 1952; Self et al., 1989; Scott et al., 2006]. We sampled three mafic fall tephras (MCO1 to MCO3) and two overlying dacitic falls, the Lower and Upper Cosigüina tephras (LCO and UCO), but could correlate marine ashes only to the mafic falls. Ash layer C9 in our cores offshore Nicaragua and Southern Salvador correlate compositionally with the ∼21 to 23 ka old MCO tephras. Using also compositional data of cores V-15-26 (510 and 539 cm bsf) and V-15-22 (112 cm bsf) documented by Bowles et al. [1973] extends the minimum distribution (up to 1 cm isopach) of those tephras 200 km to the west and 180 km to the south of Cosigüina volcano (6 × 104 km2) but, due to the few available thickness data, only allows a very rough estimate of the tephra volume as around 6 km3 (4 × 1012 kg magma mass) at least for the uppermost tephra MCO1 (Figure 4 and Table 1).

5.8. Chiltepe Volcanic Complex

[32] The Chiltepe volcanic complex includes Apoyeque stratocone, the Xiloa maar, at least two more, now hidden, vents of plinian eruptions and several basaltic cinder cones and maars [Kutterolf et al., 2007b]. During the past 15 ka, six (phreato-)plinian dacitic tephras erupted from this area: the Lower (∼17 ka) and Upper Apoyeque (12.4 ka) tephras, the 6.1 ka Xiloa Tephra, the Mateare and Los Cedros tephras, and finally the 1.9 ka old Chiltepe Tephra. In part 1 we have been able to correlate ash layer C3 in the pacific sediment cores and an ash layer 75 cm bsf in core V-15-26 of Bowles et al. [1973] to the Chiltepe Tephra. Considering these distal thickness data, the tephra volume of 4 km3 estimated from on-land data by Kutterolf et al. [2007b] must now be increased to 17.9 km3 corresponding to 2.2 × 1013 kg magma mass, distributed across an minimum area (up to 1 cm isopach) of 1.7 × 105 km2 (Figures 4, S1, and S2 and Table 1).

[33] The 12.4 ka old, coarse grained, reversely graded Upper Apoyeque Tephra pumice fall is correlated to reworked ash pods (ash layer C6) and it probably correlates also with a >1-cm-thick ash layer 580 cm bsf in core V-15-26 of Bowles et al. [1973]. The minimum distribution area (up to 1 cm isopach) thus is 5 × 104 km2 and the tephra volume of 4.3 km3 corresponds to 3.7 × 1012 kg magma mass (Figures 4, S1, and S2 and Table 1).

[34] The Lower Apoyeque Tephra compositionally corresponds to ash layer C7, which allows a minimum estimate of 3.9 km3 (Table 1 and Figures 4 and S2) of erupted tephra volume corresponding to 5.4 × 1012 kg magma mass distributed across an minimum area (up to 1 cm isopach) of 5 × 104 km2 (Figure S1).

5.9. Masaya Caldera

[35] The Masaya Caldera is a volcano that has produced large-magnitude plinian and phreatomagmatic eruptions of mafic composition [Bice, 1980, 1985; Williams, 1983]. The Fontana Tephra (FT) is a layered sequence of scoria lapilli fall beds that have a wide, plinian dispersal and erupted from a vent northwest of the Masaya Caldera [Wehrmann et al., 2006] about 60 ka ago [Kutterolf et al., 2008]. Vents within Masaya Caldera erupted the plinian to phreatomagmatic San Antonio Tephra (SAT, ∼6 ka), the Masaya Triple Layer/La Concepción Tephra (MTL/LCT; 2.1 ka), and finally the Masaya Tuff/Ticuantepe Lapilli (MT/TIL), the product of a huge surtseyan eruption 1.8 ka ago that ended in a plinian phase [Kutterolf et al., 2007b; Pérez and Freundt, 2006]. Correlations of marine ash layer C2 to MT/TIL facilitate the extension from 2 × 103 km2 to 4.3 × 104 km2 minimum distribution area (up to 1 cm isopach) by including the distal data. This yield a new tephra volume of ∼6.6 km3 (∼5 × 1012 kg magma mass; Table 1 and Figures 4, S1, and S2).

[36] The MTL/LCT plinian deposit onshore can be correlated to marine mafic ash layer C4 which leads to a minimum distribution area (up to 3 cm isopach) of 2.2 × 104 km2 and new total tephra volume of 3.4 km3 (∼4.8 × 1012 kg magma mass; Table 1 and Figures 4, S1, and S2).

[37] The widespread San Antonio Tephra onshore (∼8500 km2) is a sequence of black scoria falls overlain by surge deposits. A tephra volume of ∼0.5 km3 has been estimated by Pérez and Freundt [2006]. Geochemical fingerprinting shows that marine ash layer C5 and additionally a >1-cm-thick ash layer 270 cm bsf in core V-15-26 of Bowles et al. [1973] is equivalent to the SAT. The corresponding new minimum distribution (up to 3 cm isopach) is then 1.2 × 105 km2 (Figure S1). The erupted tephra volume of 13.5 km3 corresponds to 2.2 × 1013 kg magma mass for the San Antonio Tephra, which thus represents one of the biggest mafic eruptions in Central America since the Upper Pleistocene (Table 1 and Figures 4 and S2).

[38] The Fontana Tephra compositionally corresponds to marine ash layer C16 in the Pacific sediment cores. This allows a minimum estimate of 2.7 km3 (Figures 4 and S1) of erupted tephra volume, which doubles the previous estimate by Wehrmann et al. [2006] and corresponds to 2.9 × 1012 kg magma mass distributed across an minimum area (up to 1 cm isopach) of 3.4 × 104 km2 (Figure 3).

5.10. Apoyo Caldera

[39] Apoyo Caldera, in Central Western Nicaragua, generated two large plinian eruptions in rapid succession 24.5 ka ago, the Lower (LAT) and Upper (UAT) Apoyo tephras which are separated by an incipient paleosol [Kutterolf et al., 2007b]. The LAT is a pumice fall deposit, the UAT also comprises coarse-grained pyroclastic surge deposits and about 8 km3 of ignimbrite distributed mainly to the east of the caldera and into Lake Nicaragua [Sussman, 1985]. The UAT correlates with marine ash layer C10 as well as with ash beds in cores V-15-27, 18, 22 and 26 of Bowles et al. [1973] offshore Central America. The resulting UAT fall tephra volume is 42.9 km3 (Table 1 and Figures 4 and S2) distributed across an minimum area (up to 3 cm isopach) of at least 3.7 × 105 km2 (Figure 2). Including also the 8 km3 proximal ignimbrite and its density of 2200 kg/m3 estimated by Sussman [1985] gives an erupted magma mass of 7.2 × 1013 kg.

[40] The Lower Apoyo Tephra correlates with marine ash layer C11 which extends its minimum distribution (up to 1 cm isopach) across 5 × 104 km2 (Figure S1) resulting in a total tephra volume of 6.5 km3 (∼6 × 1012 kg magma mass; Table 1 and Figures 4 and S2).

5.11. Concepción Volcano

[41] Concepción on Ometepe island produced several basaltic to dacitic pyroclastic tephras [Borgia and van Wyk de Vries, 2003]. One of those, the ∼19 ka old Upper Ometepe Tephra, is correlated to marine ash layer C8 offshore Nicaragua and can also be found in cores V-15-18, 27 and 19 of Bowles et al. [1973]. The resulting UOT minimum aerial distribution (up to 1 cm isopach) of 7 × 104 km2 (Figure S1) and the tephra volume of 5.2 km3 (∼4.2 × 1012 kg magma mass; Table 1 and Figures 4 and S2) are mainly based on offshore data since exposure on land is poor.

5.12. Tiribi Tuff From Costa Rica

[42] The Tiribi Tuff, the largest Costa Rican eruption in the last 350 ka, has erupted from Barva volcano. Compared to other CAVA rocks, it has a unique chemical composition [Pérez et al., 2006] which makes it easy to correlate with marine ash beds. We find the distal ash of the Tiribi Tuff in core M66/3a-147 and also as the I6-Layer of Ledbetter [1985]. Peréz et al. [2006] estimated a tephra volume of 35 km3 DRE on the basis of land data, which is now increased to ∼80 km3 tephra volume (2,2 × 1013 kg magma mass; Figures 4 and S2 and Table 1) considering the offshore minimum distribution (up to 1 cm isopach) of 3.6 × 105 km2 (Figure 2).

6. Erupted Masses and Mass Fluxes Along the CAVA

[43] Previously, erupted masses along the Central American volcanic arc have been calculated from the volumes of volcano edifices [Carr, 1984]. This approach has been revised by Carr et al. [1990, 2007b], who already calculated magma mass fluxes but without including the volumes of widespread tephras. We extend this approach by adding erupted masses of the widespread tephras to those volcanoes that produced them. Moreover, while average edifice compositions are basaltic to andesitic, tephras mostly have dacitic to rhyolitic compositions and we calculate the masses of fractionated cumulates trapped in the crust to estimate the total magma production for each volcano. Using the edifice volumes determined by Carr et al. [1990, 2007b], we calculated the corresponding magma masses using a density of 2800 kg/m3 assuming that the edifice material is well compacted; this actually yields maximum estimates of the edifice magma masses while the tephra masses added tend to be minimum estimates.

[44] The wide compositional range of the CAVA volcanic rocks from basalt through rhyolite is largely due to different extents of fractional crystallization. Incompatible elements thus typically increase linearly to higher silica contents although the gradients vary between magmatic series. We use the variations of K2O with silica (Figure 5) because this has been analyzed in all samples but other incompatible elements give similar results. Employing our own compositional data (Table S1) and that of Carr and coworkers [Carr et al., 2003; Patino et al., 2000] compiled at∼carr/, we determine an average ratio of observed K2O content to the K2O content at 50 wt% SiO2 for each tephra as well as an average value for all samples available from each volcanic edifice. These average fractionation factors of 1.2–4.8 for the tephras and 0.7–3.8 for the edifices allow us to calculate fractionated cumulate masses from erupted magma masses for differentiation to >50 wt% SiO2, which is a minimum estimate because we ignore the significant cumulate mass produced during fractionation from primitive compositions which, however, are not easily determined.

Figure 5.

K2O versus SiO2 diagram showing the compositional range of the CAVA volcanic rocks from basalt through rhyolite due to different extents of fractional crystallization (XRF data from Carr et al. database; Patino et al. [2000] and Carr et al. [2003] (gray dots); and own data (colored dots)). Incompatible K2O increases linearly to higher silica contents within magmatic series as indicated by the lines. Vertical blue line at SiO2 = 50 wt% intersects K2O contents used to determine fractionation factors.

[45] The resulting total magma masses produced vary greatly between volcanoes, which reflects different modes of eruption and volcano ages. Edifice ages range up to 600 ka (Table 2) while the tephras we have studied cover the last 200 ka (except Tiribi Tuff at 320 ka). The tephra ages are well known but many of the edifice ages are not well constrained. We follow Carr et al. [2007b] in using 600 ka and 350 ka ages of undated volcanoes in Costa Rica and Nicaragua, respectively. Unpublished age data available to the Servicio Nacional de Estudios Territoriales (SNET) suggest 230 ka as a reasonable maximum age of undated volcanoes in El Salvador (C. Pullinger, SNET, personal communication, 2007). For Guatemala, we make the conservative assumption that undated volcano ages are equal to the oldest age yet determined (240 ka at Agua volcano [Wundermann, 1982]).

Table 2. Summary of Ages, Edifice Volumes, and Tephra Volumes as Well as Respective Extrusive Masses and Calculated Fluxes Along the Central American Volcanic Arca
NumberVolcanoesOldest Age at Complex, yearsAge Ref.Oldest Tephra at Complex, yearsAge Ref.Distance Along Arc, kmEdifice Volume, km3Edifice Magma Mass, kgMagma flux 1, g/sK2O/K2O50Magma Flux 2, g/sTephra Magma Mass, kgMagma Flux 3, g/sK2O/K2O50Magma Flux 4, g/sMagma Flux 1+3, g/sMagma Flux 2+4, g/s
1Tacana40,000GP  22.3205.60E+1344,3633.00133,088    44,363133,088
2Tajumulco240,000X  46.7451.26E+1416,6363.1051,572    16,63651,572
3Cicabál240,000X  82.7123.36E+134,4362.3010,203    4,43610,203
4Siete Orejas240,000X  84.7401.12E+1414,7882.9042,884    14,78842,884
5Cerro Que240,000X  94.851.40E+131,8483.205,915    1,8485,915
6Santa Maria240,000X140,000R193.9205.60E+137,3942.4017,7451.96E+134,4253.8617,08011,81934,825
7San Pedro84,000N  125.5277.56E+1328,5192.3065,594    28,51965,594
8Tolimán84,000N  135.3185.04E+1319,0132.4045,630    19,01345,630
10Acatenango230,000V  169.2621.74E+1423,9172.4057,402    23,91757,402
11Fuego100,000MR  170.1732.04E+1464,7701.4090,677    64,77090,677
12Agua240,000W  183.0681.90E+1425,1392.3057,819    25,13957,819
13Pacaya191,000W  201.6174.76E+137,8971.5011,846    7,89711,846
14Amatitlán  191,000R1205.0     1.71E+1428,4504.02114,36828,450114,368
15Tecuamburo100,000C2  231.0391.09E+1434,6031.8062,285    34,60362,285
16Moyuta100,000C2  266.5154.20E+1313,3092.3030,610    13,30930,610
17Ayarza  30,000C2, PR260.0     1.73E+1318,3054.5082,37418,30582,374
El Salvador
18Apaneca230,000P  304.51253.50E+1448,2201.6077,153    48,22077,153
19San Mareclino230,000P  318.012.80E+123861.90733    386733
20Santa Ana230,000P80,000CEL319.42206.16E+1484,8682.00169,7368.00E+113174.501,42685,185171,162
21Conejo230,000P  320.012.80E+123861.80694    386694
22Cerro Verde230,000P  321.125.60E+127721.901,466    7721,466
23Izalco300R2  321.212.80E+12295,7521.65487,990    295,752487,990
24Coatepeque 80,000CEL330.0     3.63E+1314,3764.0257,79314,37657,793 
25San Salvador100,000CEL30,000P358.0631.76E+1455,8972.00111,7943.24E+123,4224.3014,71659,319126,510
27San Vincente230,000P  404.1601.68E+1423,1462.3053,235    23,14653,235
29Tigre150,000CEL  441.0205.60E+1311,8301.6018,928    11,83018,928
30Taburete230,000P  442.782.24E+133,0861.604,938    3,0864,938
31Tecapa75,000CEL  443.0651.82E+1476,8951.60123,033    76,895123,033
32Usulatán220,000CEL  449.8154.20E+136,0491.207,259    6,0497,259
33Chinameca230,000P  460.0102.80E+133,8581.505,786    3,8585,786
35San Miguel230,000P80,000CEL467.0581.62E+1422,3741.3029,0871.20E+134,7532.009,50627,12738,593
36Conchagua230,000P  514.0277.56E+1310,4161.5015,623    10,41615,623
37Conchaguita230,000P  524.012.80E+123861.30501    386501
38Meanguera230,000P  531.038.40E+121,1571.301,504    1,1571,504
40San Cristobal350,000C1  624.01103.08E+1427,8851.5041,828    27,88541,828
41Casita350,000C1  627.0451.26E+1411,4081.5017,111    11,40817,111
42Telica350,000C1  644.0308.40E+137,6051.309,887    7,6059,887
43St. Clara350,000C1  648.025.60E+125071.30659    507659
44Rota350,000C1  655.0123.36E+133,0421.404,259    3,0424,259
45Malpaisillo 300,000vWdV660.0     3.00E+133,1693.4210,8373,16910,837 
46Cerro Negro150MW  663.00.25.60E+11118,3011.00118,301    118,301118,301
47Las Pilas350,000C1  665.0287.84E+137,0982.0014,196    7,09814,196
48El Hoyo350,000C1  667.0143.92E+133,5491.906,743    3,5496,743
49Momotombo350,000C1  683.3185.04E+134,5631.607,301    4,5637,301
51Nejapa30,000F  720.038.40E+128,8731.4012,422    8,87312,422
54Mombacho350,000C1  762.2205.60E+135,0700.703,549    5,0703,549
55Granada350,000C1  766.438.40E+127612.601,977    7611,977
56Zapatera350,000C1  784.851.40E+131,2681.802,282    1,2682,282
58Maderas350,000C1  834.0226.16E+135,5771.8010,039    5,57710,039
Costa Rica
59Orosi600,000C1  861.5752.10E+1411,0911.0011,091    11,09111,091
60Cacao    863.0           
61Rincón de la Vieja600,000C130,000Ch882.31022.86E+1415,0832.1031,6755.00E+115282.501,32015,61132,995
62Miravaelles600,000C1  903.9601.68E+148,8732.1018,632    8,87318,632
63Tenorio600,000C1  920.4491.37E+147,2462.1015,216    7,24615,216
65Alto Palomo 550,000Vl980.0     1.81E+1410,4173.0031,25210,41731,252 
66Platanar600,000C1  1001.7842.35E+1412,4222.4029,812    12,42229,812
68Tiribi  320,000Pz1030.0     4.97E+134,9173.3016,2254,91716,225
71Turialba600,000C1  1072.0872.44E+1412,8652.5032,163    12,86532,163

[46] Dividing the magma mass produced by each volcano by its lifetime yields long-term average magma fluxes. Figure 6 compares flux values obtained from edifice volumes alone, from combining edifice and tephra volumes, and from including masses of fractionated cumulates. The along-arc variation is highly irregular but peak magma fluxes from volcanoes that stand out by their high productivity increase northward along the CAVA. Overall, magma fluxes are higher in El Salvador than in Nicaragua and Costa Rica.

Figure 6.

Magma mass fluxes from CAVA volcanoes averaged over volcano age and based on edifice volumes (green) and on edifice plus tephra volumes (red). Extrusive mass fluxes (derived from edifice plus tephra volumes, in red) are compared with total magma fluxes including fractionated cumulate masses trapped in the crust (for differentiation above 50 wt% SiO2, in blue). The red arrow shows the increase of peak total magma fluxes from Costa Rica to Guatemala.

[47] The sum of total magma fluxes (including cumulates) for the entire 1100 km length of the CAVA studied is 4.5*106 g/s, which divided by the length gives an average flux per unit arc length of 4.2 g/s/m. The long-term average eruptive magma fluxes of CAVA volcanoes range across 400–296,000 g/s which agrees with global estimates for oceanic arcs (8,200–667,000 g/s; avg. 177,000 g/s) as determined by White et al. [2006] but is lower than respective values for continental arcs (940–6,344,000 g/s; avg. 481,000 g/s). Wadge et al. [2006] determined a magma flux of about 1,400,000 g/s for Arenal volcano, Costa Rica, from the volumes of lava extruded from 1980 until 2004. Our value for the long-term (600 ka) erupted magma flux of Arenal is 3,739 g/s. This example emphasizes the temporal variability of magma fluxes at the volcanoes. While short-term measurements are important for topics such as hazard assessment, the long-term behavior is the more important in studying how arc volcanism relates to subduction processes. The long-term magma fluxes determined here are a prerequisite to determine elemental fluxes through the volcanic arc such as those of H2O and other volatiles which, in turn, can be compared with subduction input fluxes to better understand the transfer processes operating in the subduction zone.

[48] However, the magma mass fluxes we have determined suffer from a number of uncertainties. The ages of some of the CAVA volcanoes are only poorly constrained. The known tephra record does not cover the entire lifetime of some volcanoes because older tephras or poorly exposed or preserved. More important, however, are two other factors. First, the total magma flux should also include intrusive magmas that never reached the surface. Intrusive bodies may eventually be visualized by geophysical methods but still their ages and compositions would remain unknown. Surface deformation measured by satellite radar interferometry can provide information on crustal inflation by intruding magmas and allows to determine magma flow rates [Pritchard and Simons, 2004] but such momentary data may not be representative of the long-term arc evolution as indicated by the Arenal example mentioned above. If there is some constant ratio between extrusive and intrusive magma flux, the absolute values but not the pattern of along-arc variation shown in Figure 6 would change. The second factor not accounted for is erosion which can be substantial during volcano lifetimes of order 105 years. Volcaniclastic detritus eroded from the CAVA is ultimately delivered into the Pacific Ocean. From a reconnaissance study of forearc sediments sampled in our gravity cores offshore Nicaragua we estimate that a minimum of 30% should be added to the long-term magma fluxes to account for the erosive losses. However, more detailed work on the marine and terrestrial volcaniclastic sediments is needed to really quantify erosion rates and their variation along the CAVA which rises to greater elevations above sea level northward.

7. Summary

[49] Using the correlations of marine tephras to onshore eruptions from part1, we have constructed isopach maps of the widespread tephras produced at the CAVA during the last 200 ka. The offshore thickness data allowed us to determine more realistic volumes of these correlated tephras than was possible from on-land exposures. In Central America, where the volcanic arc lies only a few tens of kilometers upwind from the Pacific coast, the tephra volume emplaced in the ocean, on average, makes up 60% of the total erupted volume which is still a minimum estimate since we only considered isopachs of ≥1 cm thickness. The largest single eruptions, such as those of the Los Chocoyos ash and L-Tephra in Guatemala, occurred at the northern part of the CAVA but are not as frequent as the eruptions from Nicaragua and Southern Salvador with an overall smaller magnitude (Figure 4). This is also reflected in the greater thickness and coarser grain size of the Pacific ash layers offshore Guatemala and Northern El Salvador compared to those near Nicaragua and Costa Rica.

[50] We have integrated tephra volumes and published volumes of the volcanic edifices, and we have calculated the associated masses of fractionated cumulates from chemical compositions to determine the total magma mass produced by each CAVA volcano during its lifetime. Division by volcano ages yielded the long-term average magma flux at each volcano. The resulting data show that, averaged over the CAVA, highly explosive eruptions generating widespread tephras account for at least 65% of the total magma output. Magma fluxes of neighboring volcanoes are often vastly different and peak fluxes increase northward along the CAVA. Compared to estimates of global average volcanic magma fluxes, the long-term fluxes of CAVA volcanoes reach comparable magma fluxes regarding oceanic island arcs and slightly lower values for continental arcs. The magma flux values we have compiled here form a useful basis to calculate elemental fluxes of volcanic output as an important aspect in the overall flux budget of the Central American subduction zone.


[51] We gratefully acknowledge the support during field work by the local geological surveys: SNET in El Salvador, INETER in Nicaragua, and INSIVUMEH in Guatemala. We particularly appreciate the permission to use unpublished thickness data by Carlos Pullinger, Dolores Ferres, Walther Hernadez, and Heidi Wehrmann. Wendy Peréz acknowledges a Ph.D. stipend by the Deutscher Akademischer Austauschdienst (DAAD). Many thanks also to all members of the scientific parties of the R/V METEOR and R/V SONNE cruises for the assistance in drilling, logging, and sampling the gravity cores. R/V METEOR and R/V SONNE cruises were funded by the Deutsche Forschungsgemeinschaft (DFG). We also appreciate the helpful comments and suggestions of Vincent Salters, Stephen Blake, Mike Carr, Phil Shane, and Tom Vogel, who reviewed this paper. This publication is contribution 98 of the Sonderforschungsbereich 574 “Volatiles and Fluids in Subduction Zones” at Kiel University.