SEARCH

SEARCH BY CITATION

Keywords:

  • Central America;
  • stable isotopes;
  • subduction zone;
  • chlorine

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

Volcanic outputs (ash, tephra, and lava samples) of 23 volcanic centers from the volcanic front, secondary front, and back arc in Central America were analyzed for their δ37Cl values with the goal of using chlorine isotopes as a tracer of fluids sourced from the subducting Cocos slab. δ37Cl values range from −2.6 to +3.0‰ with systematic variations along the length of the front. Values from the northernmost (Guatemala and El Salvador) and southernmost (Costa Rica) ends of the front have mantle-like signatures. In contrast, δ37Cl values are both positive and negative in the center of the front (Nicaragua), implying a sediment and/or serpentinite-derived component. Geophysical observations are consistent with extensive hydration of the Cocos plate offshore of Nicaragua, in support of a serpentinite-derived fluid source. Fluids from dehydrating serpentinites may also incorporate Cl from overlying sediments, resulting in a multiple-source chlorine signature.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

The geochemical characteristics of volcanic material can be used to identify the source(s) of fluids and volatiles derived from a subducting slab. Chlorine is a major component (at the weight percent level) in fluids derived from the subducting slab [Kent et al., 2002]. Although Cl concentrations may change dramatically due to dehydration of the subducting slab and/or hydration in the overlying mantle wedge, Cl isotope ratios should retain the signature of their source, as Cl is strongly partitioned into an aqueous fluid or melt, and fractionation is thought to be small at high temperatures [Schauble et al., 2003; Barnes et al., 2006; Bonifacie et al., 2008a]. Despite the need for additional work (experimental and empirical) to fully understand potential effects of chlorine fractionation, chlorine isotope geochemistry remains a tantalizing, and potentially powerful, tool for identifying fluid sources in subduction zones.

Numerous isotopic and geochemical tracers (e.g., Ba/La, U/Th, 10Be/9Be, δ15N, δ18O, U series nuclides, Sr-Nd-Pb isotopes) have been used to identify fluid sources in arc systems [e.g., Carr et al., 1990; Morris et al., 1990; Giggenbach, 1992; Reagan et al., 1994; Herrstrom et al., 1995; Thomas et al., 2002; Shaw et al., 2008]. However, no single system can give a unique answer regarding fluid sources in magmatic systems, because ambiguities and potential mixing of multiple sources, dramatic variation in the concentration of tracers in different potential fluids, and crustal contamination and modification (e.g., isotopic fractionation) can occur. Adding another complementary geochemical tracer, in this case chlorine isotopes, further constrains fluid sources and pathways in subduction zones and offers clues to how fluids are returned to the surface in arcs.

Chlorine is hosted in a number of different subducting materials: sediments, pore fluids, altered oceanic crust, and serpentinites. Most of these materials have distinct Cl stable isotope signatures, with some overlap (Figure 1). Seafloor sediments from ODP cores in the western Pacific have δ37Cl values between −2.5 and +0.3‰ (n = 11) [Barnes et al., 2008]; nearly identical to marine and nonmarine Jurassic sediments in Chile (−2.6 to +0.5‰, n = 24) [Arcuri and Brimhall, 2003]. Sedimentary pore fluids range from −7.8 to +0.3‰ [Ransom et al., 1995; Hesse et al., 2000; Spivack et al., 2002; Godon et al., 2004a; Bonifacie et al., 2007b; Wei et al., 2008]. The δ37Cl values of three altered oceanic crust samples range from −1.6 to −0.9‰ (n = 3) [Bonifacie et al., 2007a]. The δ37Cl values of seafloor serpentinites are bimodal: samples ranging from −1.3‰ to −0.5‰ are related to serpentinization by overlying sedimentary pore fluids, whereas those ranging from +0.1‰ to +0.4‰ are serpentinized directly by seawater [Barnes and Sharp, 2006]. Other potential chlorine sources are seawater, which is homogeneous and defined to be 0‰ [Kaufmann et al., 1984; Godon et al., 2004b], and the mantle. The mantle is reported as having values ranging from near 0‰ [Sharp et al., 2007] to ≤−1.6‰ [Bonifacie et al., 2008b] to ≤−3.0‰ [Layne et al., 2009]. The δ37Cl values of the deep mantle, based on mantle plume and diamond samples average −0.15 ± 0.22‰ (Z. D. Sharp, unpublished data, 2009). Given that chondrites also have near-zero δ37Cl values [Sharp et al., 2007], no secular variations are seen in crustal materials [Sharp et al., 2007], and deep mantle samples have near-zero values (Z. D. Sharp, unpublished data, 2009), we prefer the near-zero value as representative of the mantle. Additional Cl sources to the volcanic arc lavas may come from incorporation of Cl in to the magma source through crustal assimilation and subduction erosion.

image

Figure 1. Range of δ37Cl values of potential fluid reservoirs and volcanic ash and gas outputs in a subduction zone. Individual analyses are shown for mantle, altered oceanic crust, sediment, ash, and gas samples. Sediment data from ODP cores offshore of Central America (this study) are shown as diamonds, and sediment data from ODP cores offshore of the Izu-Bonin-Mariana chain [Barnes et al., 2008] are shown as circles. IBM volcanic ash/lava and gas (collected in Giggenbach bottles and gas condensates)/geothermal well data are from Barnes et al. [2008]. CA gas and hydrothermal well data are from Sharp et al. [2009]. CA ash/lava data are from this study. Other data are as follows: pore fluids [Ransom et al., 1995; Hesse et al., 2000; Spivack et al., 2002; Godon et al., 2004a; Bonifacie et al., 2007b], altered oceanic crust [Bonifacie et al., 2007a], serpentinites [Barnes and Sharp, 2006; Bonifacie et al., 2008a], and sediment [Arcuri and Brimhall, 2003; Barnes et al., 2008]. The δ37Cl values of the mantle range from near 0‰ [Sharp et al., 2007] to ≤−1.6‰ [Bonifacie et al., 2008b] to ≤−3.0‰ [Layne et al., 2009]. The latter two mantle values are shown by red bars and an arrow.

Download figure to PowerPoint

Cl isotope compositions should not change during release and ascent from a subducting slab. Theoretical calculations [Schauble et al., 2003] and experiments [Eggenkamp et al., 1995; Liebscher et al., 2006] suggest very minor fractionation (<0.8‰ between silicates and NaCl at 300°C [Schauble et al., 2003]) between Cl-bearing phases, a conclusion supported by recent work on obducted serpentinites and metaperidotites, in which the Cl isotope composition remained unchanged during prograde subduction metamorphism [Barnes et al., 2006; Bonifacie et al., 2008a]. Additional Cl isotope fractionation work is clearly needed, but based on current data, the chlorine isotope ratio of ash, tephra, and lava samples should faithfully record the isotopic composition of the chloride source in the subducting slab. In this paper, we present chlorine stable isotope compositional data for volcanic ashes from along the strike of the Central American volcanic front and within the back arc. We link changes in chlorine stable isotope composition to variations in forcing functions and other geochemical tracers in order to identify fluid sources to the arc magmas.

2. Geologic Setting

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

The Central American volcanic front, extending from Guatemala to Costa Rica, is the result of the Cocos Plate actively subducting beneath the Caribbean Plate (Figure 2). Variations in forcing functions (e.g., convergence rates, angle of subduction, crustal thickness, sediment input) occur along the strike of the subduction zone. For example, the dip of the subducting Cocos Plate beneath the volcanic front is steepest under Nicaragua (75°) and shallows to ∼35–40° under Guatemala and Costa Rica [Carr and Stoiber, 1990; Protti et al., 1995]. In addition, the thickness of the overlying crust is at a minimum beneath Nicaragua (∼32 km) and reaches a maximum toward either end of the front (thickness beneath Guatemala is ∼48 km) [Carr et al., 1990].

image

Figure 2. Map of the Central American arc system. Black circles are DSDP/ODP Sites of sediment and basalt samples. Triangles are volcanoes analyzed in this study.

Download figure to PowerPoint

Such variations in forcing functions result in distinct geochemical variations along the volcanic front [e.g., Carr, 1984]. For example, 10Be is a short-lived cosmogenic radionuclide that is concentrated in oceanic sediments, therefore enrichments of 10Be in arc lavas are an indication of subducted sediment contribution [Morris et al., 1990]. The elemental ratios Ba/La, B/La, and U/Th ratios are also excellent proxies for sediment contributions to arc magmas [Leeman et al., 1994; Patino et al., 2000]. Ba/La, U/Th, and 10Be/9Be ratios are highest in Nicaraguan lavas where the slab dip is steepest and the crustal thickness is at a minimum, indicating that recycling of subducted sediments is maximized beneath Nicaragua [Carr et al., 1990; Morris et al., 1990; Herrstrom et al., 1995; Patino et al., 2000]. In contrast, the ends of the arc have low sediment contributions, but with subtle geochemical differences. In Costa Rican lavas Ba/La and U/Th ratios are both low (<50 and <0.5, respectively), similar to those of mid-ocean ridge and ocean-island basalts, whereas Guatemalan lavas have low U/Th ratios (<0.5), as in Costa Rica, but slightly higher Ba/La ratios (∼50–70) [Carr et al., 1990; Patino et al., 2000].

Other isotopic systems (e.g., 3He/4He, δ13C, δ15N, δ18O) have been used to investigate volatile cycling through the Central American subduction zone and to correlate along-arc geochemical variations with forcing functions. For example, Shaw et al. [2003] and de Leeuw et al. [2007] show that CO2 is sourced predominantly from subducted oceanic sediments and that only a small fraction of the subducted CO2 (<20%) is returned via arc volcanism. Large isotopic differences among nitrogen reservoirs allow N isotopes to be a powerful volatile tracer in subduction zones. MORB is depleted in 15N (∼−5‰) relative to air [Marty, 1995; Marty and Zimmerman, 1999], whereas sedimentary material is enriched in 15N (∼+3 to +7‰) [Peters et al., 1978; Bebout, 1995; Kienast, 2000; Sadofsky and Bebout, 2004; Li and Bebout, 2005]. Volcanic gases from Guatemala and Nicaragua have positive δ15N values and high N2/He ratios indicating a hemipelagic sediment contribution [Fischer et al., 2002; Elkins et al., 2006]. In contrast, volcanic gases from Costa Rica have negative δ15N values and low N2/He ratios, indicating a mantle N source [Fischer et al., 2002; Zimmer et al., 2004]. Snyder et al. [2003] reported a similar arc-parallel trend for nitrogen isotope ratios. δ18O values of olivine phenocrysts from basalts and basaltic andesites also vary along the volcanic front, from MORB δ18O values (∼5.0 to 5.2‰) in Costa Rica, 4.6 to 4.9‰ in Nicaragua, and 5.1 to 5.7‰ in Guatemala and El Salvador [Eiler et al., 2005]. Based on these trends, Eiler et al. [2005] identified two principal slab-derived components contributing to the overlying mantle wedge: 1) a solute-rich aqueous fluid that dominates in the center of the arc (Nicaragua) and originates from the dehydration of the altered oceanic basement and 2) a partial melt of subducted sediment that dominates in the northern segment of the arc (Guatemala, El Salvador).

3. Samples and Analytical Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

3.1. Sample Descriptions

3.1.1. Inputs: Sediments and Altered Oceanic Crust

In order to better define the characteristic δ37Cl values of subducting marine sediments and altered oceanic crust (important potential fluid sources in subduction zones), we analyzed 11 sediment and 2 basalt samples (Table 1) collected offshore from the Central American volcanic front. Samples are from Deep Sea Drilling Program (DSDP) Sites 495 and 499 off the coast of Guatemala and from the Ocean Drilling Program (ODP) Site 1039 off the coast of Costa Rica (Figure 2). All sites have similar sedimentary sequences (∼175 m of hemipelagic mud overlying ∼250 m of pelagic carbonates) on the subducting Cocos Plate [Aubouin et al., 1982; Plank and Langmuir, 1993; Kimura et al., 1997; Valentine et al., 1997], implying little along arc variation in sediment input. There is no geophysical evidence for sediment accretion in the fore arc, thereby making all the sediments on the subducting Cocos Plate potential contributors to the slab volatile flux [Aubouin et al., 1982; Valentine et al., 1997; von Huene et al., 2000].

Table 1. Chloride Components in Sediments and Altered Oceanic Crusta
Sample (Leg-Hole-Core (Type)-Section, Interval (cm))Lithologyb,cDepth (mbsf)ClWSC (wt %)ClSBC (wt %)δ37ClWSC (‰)δ37ClSBC (‰)Bulk δ37Cld (‰)δ15NAirc (‰)
  • a

    WSC, water-soluble chloride components; SBC, structurally bound chloride components; NA, not analyzed; BDL, below detection limit. Chlorine isotope data reported in italics were analyzed in continuous flow; rest of analyses were in dual inlet.

  • b
  • c
  • d

    Bulk δ37Cl = [(XWSC)(δ37ClWSC)+ (XSBC)(δ37ClSBC)].

  • e

    Rinsed 15 times in deionized water to ensure removal of water-soluble chloride.

Altered Oceanic Crust
67-495-48R-1, 83-86olivine-plag phyric basalt with smectite veins437.330.02BDLNANA  
67-499D-1W-1, 30-34weathered basalt1.300.00BDLNANA  
Sediments
67-499-7R-3, 73-75turbidite52.231.580.05−0.23, 0.16, −0.21−0.29  
67-499-19R-2, 99-101biogenic siliceous mud164.991.920.04−0.35−0.80−0.36 
67-495-14R-3, 69-71diatomaceous mud127.193.010.14−0.46−0.60−0.47 
67-495-32R-3, 62-64nannofossil ooze298.021.350.03−0.140.68, 0.31−0.13 
170-1039B-10H-4, 138-140diatomaceous ooze with ash83.911.580.030.31−2.05 6.4
170-1039B-11H-5, 138-140silty clay with ash94.881.710.010.16−1.030.155.3
170-1039B-17X-4, 134-136calcareous clay147.642.020.020.49−1.570.478.2
170-1039B-19X-5, 134-136siliceous nannofossil ooze and clay168.442.030.050.41−0.18, −0.16e0.406.1
170-1039B-25X-5, 131-133siliceous nannofossil ooze226.011.310.040.110.67 3.4
170-1039B-32X-2W, 127-129calcareous diatomaceous ooze and breccia289.171.220.060.48−0.490.433.8
170-1039C-2R-1W, 63-65calcareous diatomaceous ooze and breccia373.331.350.110.420.030.39NA
3.1.2. Outputs: Volcanic Gases and Ashes

Ash, tephra, and lava samples were analyzed from 23 volcanic centers (Table 2). Ash samples are from T. Fischer and M. Carr's collections. Gas samples were collected during 2001 to 2008 field seasons and are from T. Fischer's collection. Volcanic gas samples were collected in evacuated glass flasks containing ∼60 mL of a 4N NaOH solution (Giggenbach bottles). Gases are draw into the flask through titanium and Teflon tubing. H2O, CO2, SO2, and HCl dissolve in the NaOH solution; whereas, N2, Ar, and He fill the headspace. A few fumaroles gases were collected via condensation flasks cooled with ice water. Once condensed, the liquid is transferred to a clean Nalgene bottle (see Sharp et al. [2009] for more details).

Table 2. Chlorine Stable Isotope Values of Ash and Lava Samples From Central America
Volcanic SitesSample NameDistancea (km)Latitudea (°N)Longitudea (°W)δ37Clb (‰)
Volcanic front     
   FuegoFuego170.114.4890.88−0.47
   IzalcoIzalco-1 (18-06-02)321.213.8289.630.57, 0.11
   BoqueronSAL-B-18c358.013.7389.28−0.07, −1.19, 0.30
   San MiguelSan Miguel467.813.4388.270.94
   CosiguinaNIC-COS5c556.912.9887.57−2.34, −2.26
   San CristobalNIC-32-10137Ac624.712.7087.00−0.62
   TelicaNIC-TE-115c644.012.6086.85−1.66
   Las PilasLas Pilas-1 (1-18-02)664.912.5086.68−1.73
   Cerro NegroCN-92-1663.312.5086.70−0.03
   Cerro NegroCN-95-1663.312.5086.700.77, 0.36
   MomotomboMomo 1905683.312.4286.530.83, 0.40
   NejapaNejapa-pit (1-9-02)720.012.1186.32−0.86
   MasayaMasaya (Nindiri sect. 1-7-02)742.711.9886.15−1.53
   GranadaNIC-GR-6c762.211.8886.000.05
   ConcepciónConcepcion-1 (1-11-02)816.911.5385.62−2.57
   MiravallesMiravalles 11.1.01-1903.910.7585.151.85, 2.96
   ArenalArenal Nov06958.310.4784.73−1.08, −0.51
   ArenalArenal 10.1.01-3958.310.4784.730.52, −0.19
   LagunaLaguna 8.1.01-3 10.2984.21−0.40
   PoásPoas 7.01.01-1102110.2084.220.74
   IrazúIrazu 5.1.01-31067.39.9883.850.37
   TurrialbaTurrialba 6-20-071072.110.0383.771.00
Secondary front     
   Cerro los MartinezCerro de Martinez 1-9-02 lower layer 12.1686.320.14
   El TigreEl Tigre HO3-RO8 13.2787.63−0.47
   El TigreHON-AM-1c 13.2787.63−1.77
Back arc     
   AgromosaAgromosa HO-RO3 14.9688.01−0.87
   YojoaYojoa HO3-RO1 14.9887.98−0.59
   La BarcaLa Barca HO3-R06 15.1187.93−0.77

3.2. Analytical Methods

3.2.1. Sample Preparation

Ash, tephra, and basalt samples were washed 5 times for 15 min each in 18MΩ deionized water in an ultrasonic bath and then dried at low heat on a hot plate to remove any surface chloride. Well-lithified sediments were ultrasonicated to remove surficial chloride and poorly lithified sediments were rinsed with 18MΩ deionized water. Samples from Site 1039 (provided courtesy of G. Bebout) had been previously dried and crushed before storing. These samples remained untreated until subsequent crushing at the University of New Mexico. Note that one powdered sediment sample (170-1039B-19X-5, Table 1) was ultrasonicated 15 times, at 15 min for each treatment in order to evaluate the efficacy of our standard method of five washes. No difference in the δ37Cl value of the structurally bound Cl was seen between the two washing methods, indicating that our standard 5-times washing procedure effectively removes all water-soluble chloride.

Once cleaned and dried, basalt and sediment samples were crushed to pass through a 100 mesh sieve (<0.149 mm) and leached in 18MΩ water for 24 h at room temperature to extract water-soluble salts. The leached sample and solution were then filtered during which the sample was rinsed several times and all the leachate was collected. The leachate was analyzed for Cl concentration on the Dionex DX-100 Ion Chromatograph, which has a minimum detection limit of 0.22 ppm for Cl. The leached sample was then dried and pressed into a pellet for chlorine concentration analysis by X-ray fluorescence (XRF) using the Rigaku RIX 2100 wavelength dispersive X-ray spectrometer at the University of New Mexico. Chloride standards NIM-L2, MAG-1, RGM-1, AC-E, NIM-D, and FK-N were used for calibration [Govindaraju, 1994]. The minimum detection limit for Cl is 10 ppm. Ash and tephra samples were crushed after cleaning and drying in preparation for pyrohydrolysis. Cl concentrations were not determined. Approximately 2 to 8 g of material was processed for each sample.

3.2.2. Stable Chlorine Isotope Analysis

Solid samples (ashes, lavas, basalts, and sediments) were first pyrohydrolysized (melted in a water vapor stream) to extract the Cl from the sample into an aqueous solution [Magenheim et al., 1994; Barnes and Sharp, 2006]. The solutions were reacted with AgNO3 to produce AgCl. The AgCl is then reacted with CH3I to produce CH3Cl, the gas that is ultimately analyzed in the mass spectrometer [Eggenkamp, 1994; Eggenkamp et al., 1995]. Volcanic gas samples were prepared following the method of Eggenkamp [1994], except for one minor modification. Due to the high level of sulfur in the volcanic gas and ash/lava samples, the methods of Eggenkamp [1994] and Eggenkamp et al. [1995] failed to completely remove all the sulfur in the sample. Trace levels of contamination by sulfur in the sample interferes with Cl extraction and causes irreproducible Cl isotope determinations [Sharp et al., 2009]. Therefore, prior to reaction with AgNO3, sulfur is removed in the form of SO2 by the addition of 10 mL 50% nitric acid and gentle heating for several hours on a hot plate. The beaker containing the sample is covered with a watch glass filled with deionized water to prevent chloride loss. After complete removal of S and extraction of Cl, samples are converted to CH3Cl as described above (see Sharp et al. [2009] for more details). Prior to introduction into the mass spectrometer, samples are purified of excess CH3I on a gas chromatographic column in a continuous He flow using the mass spectrometer as a detector. After the CH3Cl passes through the GC column, the system is back-flushed to remove CH3I to a waste trap.

Isotope ratios are determined using a Delta XL Plus mass spectrometer configured for collecting masses 50 and 52 simultaneously. For dual inlet analysis, samples are collected in a liquid nitrogen cooled trap and then expanded into the Finnigan Delta XL Plus mass spectrometer. For smaller samples (∼20 μg Cl), the gas entering the mass spectrometer from the GC column is analyzed directly in continuous flow mode [Barnes and Sharp, 2006; Sharp et al., 2007]. δ37Cl values are reported relative to standard mean ocean seawater (SMOC). Samples analyzed using the dual inlet and continuous flow have errors of ±0.12‰ and ±0.33‰ (1σ), respectively, based on the long-term average of 3 internal laboratory standards.

4. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

4.1. Inputs: Sediments and Altered Oceanic Crust

Chloride concentrations and δ37Cl values of sediments and altered oceanic crust are reported in Table 1. For the purposes of identifying potential Cl sources in subduction zones, we will only consider structurally bound chloride as possible contributors to arc magmatism. We assume that water-soluble chloride (adsorbed material on mineral surfaces leachable in water) is lost at shallow levels in a subduction zone, during the expulsion of pore fluids. Structurally bound chloride ranged from 0.01 to 0.14 wt% in sediment samples. Neither basalt sample contained structurally bound chloride above the minimum detection limit (10 ppm) for the bulk rock sample. Previously reported δ37Cl values of altered oceanic crust are limited to three analyses (−0.9‰, −1.3‰, and −1.6‰) from ODP Hole 504B in the East Pacific Rise [Bonifacie et al., 2007a]. The δ37Cl values of the sediments range from −2.1 to +0.7‰ (average = −0.5 ± 0.8, n = 11), similar to the δ37Cl values of other seafloor sediments from the W. Pacific (−2.5 to +0.3‰, n = 11, average = −1.1 ± 0.7‰) [Barnes et al., 2008] and marine and nonmarine sediments from Chile (−2.6 to +0.5‰, n = 24, average = −0.6 ± 1.0‰) [Arcuri and Brimhall, 2003].

4.2. Outputs: Ashes, Tephras, and Lavas

δ37Cl values of ash and tephra samples range from −2.6 to +3.0‰ (Table 2 and Figures 3 and 4) and show systematic and sometimes abrupt changes along the strike of the arc. There is some scatter in the data from Costa Rica, but there is a general decrease from close to +1‰ in the southeast to −1‰ toward the northwest in Costa Rica, with the exception of Miravalles. Miravalles represents a distinct break from the Costa Rican volcanoes with values of +1.9‰ and +3.0‰. An even more dramatic break occurs across the Costa Rica–Nicaragua border with the transition from Miravalles volcano with the highest δ37Cl values in the arc to Concepción with the lowest δ37Cl values at −2.6‰. δ37Cl values are slightly positive at the southern end of the northern Nicaragua section and decrease to −2.3‰ at Cosigüina near the border with El Salvador. Data are limited toward the northern end of the front, but scatter between +1 and −1‰, similar to the range observed in southern Costa Rica. Samples from the back arc in Honduras are consistently negative, averaging −0.7‰. Samples from the secondary front in Honduras are also negative, averaging −1.1‰.

image

Figure 3. Plot showing the variation in δ37Cl values of ash/tephra, volcanic gas, and crater lake data along the length of the arc system. Volcanic gas and lake data are from Sharp et al. [2009]. High temperature fumaroles (>100°C), Santa Ana, Momotombo, and Poás, with anomalously high δ37Cl values due to isotopic fractionation are circled in gray (see Sharp et al. [2009] for more details).

Download figure to PowerPoint

image

Figure 4. Geochemical trends along the arc. Nitrogen isotope data are corrected for air contamination [Fischer et al., 2002; Zimmer et al., 2004; Elkins et al., 2006]. Nitrogen isotope data from Synder et al. [2003] are not included because not enough data are presented to correct for air contamination. Ba/La data are from M. Carr's online database (http://www-rci.rutgers.edu/∼carr/index.html). Oxygen isotope data are from Eiler et al. [2005]. Open circles are back-arc samples; gray circles are secondary front samples.

Download figure to PowerPoint

The chlorine isotope composition of volcanic gas, vapor condensate, bubbling spring, geothermal well, and crater lake samples from nine volcanic centers along the Central American arc have previously been reported [Sharp et al., 2009]. The data range from −4.0 to +18.8‰ (Figure 3). All δ37Cl values greater than ∼+1.5‰ are restricted to fumarole samples from three volcanic centers, Poás, Momotombo, and Santa Ana; each of which has fumarole temperatures >100°C and a crater lake and/or large hydrothermal system. These anomalously high δ37Cl values are thought to be the results of isotopic fractionation in the HCl liquid-vapor system [Sharp et al., 2009]. While fractionation in volcanic gases has important implications for processes occurring in hydrothermal systems, the near-surface fractionation observed in boiling fumaroles complicates the use of chlorine stable isotopes in some volcanic systems as a direct tracer for the subducted component in volcanic gases. In order to avoid these near-surface fractionation effects, only ashes, tephras, and lavas were used to trace sources of the subducted chlorine in this study.

5. Interpretations and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

5.1. Comparison of Chlorine Isotope Composition With Other Geochemical Tracers

The chlorine isotope data are compared to eight other geochemical variables for the same volcanoes in Figure 5. Elevated Ba/La, U/Th, U/La, 10Be/9Be, and 87Sr/86Sr ratios, are all indicative of slab input, in most cases sediment [e.g., Carr et al., 1990; Morris et al., 1990; Feigenson et al., 2004]. High U/La ratios are thought to trace the hemipelagic component of sediment [Patino et al., 2000]. High 87Sr/86Sr values likely reflect hydrothermal alteration of the subducting Cocos plate [Carr et al., 2003]. La/Yb and Nd isotopes are used to show mantle composition and degree of melting [e.g., Herrstrom et al., 1995]. High La/Yb ratios indicate a lower degree of partial melting or derivation from a more enriched source. δ18O values between 5.0 and 5.2‰ reflect MORB values, values >5.2‰ are thought to reflect a subducted sediment melt component, and values <5.0‰ are thought to reflect a hydrothermally altered rock-derived fluid (possibility serpentinite-derived) from the subducting slab [Eiler et al., 2005].

image

Figure 5. Correlations of δ37Cl values of average ash samples per volcano from this study compared to other geochemical parameters: (a) Ba/La, (b) U/Th, (c) U/La, (d) La/Yb, (e) 10Be/Be, (f) 87Sr/86Sr, (g) 143Nd/144Nd, and (h) δ18O. Oxygen isotope data are from Eiler et al. [2005], 10Be/Be data are from Morris et al. [1990], and all other data are from M. Carr's online database (http://www-rci.rutgers.edu/∼carr/index.html). Data are averaged for each volcano.

Download figure to PowerPoint

In general, there are no striking correlations between δ37Cl values and the eight other geochemical variables depicted in Figure 5. The most distinctive trend is high 10Be/9Be ratios in samples from Nicaragua are correlated with δ37Cl values <0‰. Both high 10Be/9Be ratios and negative δ37Cl values are indicative of a sediment component to the melt. Similarly, the lowest δ37Cl values also have the highest Ba/La, U/Th, and U/La ratios. Thus far, it appears that the negative δ37Cl values reflect a fluid mobile sediment derived slab component. However, negative δ37Cl values are also correlated with low δ18O values (<5.0‰) (Figure 5h). This correlation indicates that chlorine stable isotopes also trace hydrothermally altered rock, including serpentinites. The plots are presented as a means to compare various data sets; however, great care must be taken when making overall correlations to avoid comparing tracers that are controlled differently. The subduction system is influenced by the chemistry of multiple inputs therefore correlations are not necessarily linear.

5.2. Along-Arc Variations in Ash, Tephra, and Lava Samples

5.2.1. Southern End of the Volcanic Front: Costa Rica

δ37Cl values of volcanic ashes from Costa Rica range from −1‰ to +1‰, averaging +0.1‰, excluding samples from Miravalles. Although there is some scatter in the data, the near zero δ37Cl values (when interpreted in conjunction with other geochemical tracers) is thought to represent a mantle-like signature (Figure 4). Almost all oxygen isotope analyses on olivine phenocrysts from basalts and basaltic andesites in Costa Rica lie within the range of NMORB values [Eiler et al., 2005]. Negative nitrogen isotope data also support a strong mantle component beneath Costa Rica [Fischer et al., 2002; Zimmer et al., 2004]. Low Ba/La and 10Be/9Be ratios also imply that a sediment component is limited or lacking in Costa Rican magmas [Carr et al., 1990; Morris et al., 1990; Patino et al., 2000]. Based on Sr and Nd isotopes and low Ba/La ratios and high La/Yb ratios, enriched mantle minimally modified by the subducting slab exists below Costa Rica [Carr et al., 1990; Herrstrom et al., 1995].

Subduction erosion is thought to be a major process occurring along the Central American convergent margin [Ranero and von Huene, 2000]. Vannucchi et al. [2001] estimate that ∼560–600 km3 of material per kilometer along the trench has been lost along the Costa Rica margin in the past 17 million years (equivalent to ∼34–36 km3 km−1 Myr−1); therefore, fore-arc debris may be a significant component of the subducting material. Based on Pb isotopes, Goss and Kay [2006] suggest that subduction erosion and incorporation of ophiolitic fore-arc material into the mantle wedge may occur in Costa Rica. However, subsequent work has shown that central Costa Rican lavas are more enriched in 208Pb/204Pb and lower in 143Nd/144Nd at a given 206Pb/204Pb than eroded Costa Rican fore-arc terrains; whereas, the isotope chemistry of the currently subducting seamount province and Cocos and Coiba ridges can explain the geochemistry of Costa Rican arc lavas [Gazel et al., 2009].

Obducted serpentinites from the Santa Elena Peninsula, Costa Rica, have δ37Cl values ranging from +0.71 to +1.26‰ (n = 6), some of the highest values for either seafloor or obducted serpentinites [Barnes, 2006]. The explanation for the high δ37Cl values of the Santa Elena serpentinites is unclear. However, based on Sr, Nd, and Pb isotope ratios, the ultramafic and mafic rocks of the Santa Elena Nappe are substantially different from other mafic terrains along the Pacific marginal zone of Costa Rica. It is hypothesized that the Santa Elena Nappe represents a primitive arc basement [Gazel et al., 2006]. Therefore, the unique geochemistry of the Santa Elena serpentinites indicates that their subducted equivalents may be responsible for the high δ37Cl values in some of Costa Rican volcanoes. These high values overlap with the slightly positive δ37Cl values of ashes from Poás and Turrialba volcanoes, but of particular interest is Miravalles volcano in northern Costa Rica.

Miravalles volcano has anomalously high δ37Cl values compared to other Costa Rican volcanoes. These values could potentially be explained by incorporation of isotopically heavy serpentinites from the Santa Elena Peninsula into the subduction channel and ultimately the mantle wedge. However, the poor reproducibility of δ37Cl values (+1.9‰ and +3.0‰) is outside analytical error and may imply either chlorine isotope heterogeneity in the ashes or hydrothermal alteration overprinting the original Cl isotope signature. Miravalles has a large, active hydrothermal system and is the site of a geothermal energy plant. Vapor condensates from hydrothermal gases may have altered ash samples from this volcano. The Miravalles sample was washed to remove any surficial contamination/deposition, but alteration due to vapor condensates is in agreement anomalously high δ37Cl values (up to +18.8‰) of volcanic gases from Central American volcanoes with crater lakes and large hydrothermal systems [Sharp et al., 2009].

5.2.2. Northern End of the Volcanic Front: El Salvador and Guatemala

Isotopic and elemental ratio data give conflicting results at the northern end of the front. High δ18O and δ15N values from Guatemala and El Salvador point to a sedimentary source [Fischer et al., 2002; Eiler et al., 2005]. In this region, Ba/La ratios are more variable, with some samples having high values suggestive of a sedimentary component, but others have values as low as any measured in the arc consistent with a dominant mantle input (Figure 4) [Patino et al., 2000]. U/Th and 10Be/9Be ratios are low (∼<0.6 and ∼<23), also implying a low sedimentary input [Morris et al., 1990; Patino et al., 2000].

Eiler et al. [2005] proposed that the apparent inconsistency regarding the degree of sediment contribution can be explained by a slab-derived melt, rather than aqueous fluid. A slab-derived melt component, sourcing the Guatemalan and El Salvadoran magmas, would retain the high-δ18O value from sediment melts, but would fail to transport water-soluble elements such as U and Ba. A strong sedimentary component should impart a negative δ37Cl value to the lavas. The δ37Cl values from this region are near zero (average = +0.02‰). Only two analyses of samples from Guatemala and El Salvador are distinctly negative (outside analytical error of ±0.3‰). One of those negative values is from Boqueron and is not reproducible based on duplicates of the same sample. The near-zero data argue against the involvement of an aqueous fluid derived from sediments. While Cl solubility is a function of P, T, and melt composition, in general Cl is preferentially concentrated in an aqueous fluid compared to a melt [Carroll and Webster, 1994]. Therefore, the mantle signature recorded by the chlorine isotope data at the ends of the arcs may be the “mantle background” value due to the virtual absence of a water-rich slab-derived transport phase.

Based on volatile and trace element concentrations in melt inclusion from the Central American volcanic arc, Sadofsky et al. [2008] also propose that melts may control the geochemistry of Guatemalan and Costa Rican magmas. These melts have high LREE concentrations, high La/Nb ratios, low Ba/La ratios, but may also be enriched in Cl, S, and F. Cl is shown to correlate with La/Yb and not Ba/La or H2O content [Sadofsky et al., 2008]. If Cl was sourced from a sediment-derived melt, the δ37Cl value would be negative and not the near zero values observed.

Assimilation of crustal material from the overlying plate must also be considered as a component of the source magma. Sr and Nd isotopic data have been used to argue for assimilation of basement rock in lavas from behind in the volcanic front in Guatemalan lavas; however, high-Mg# lavas throughout the region, including the volcanic front, show no evidence of crustal assimilation [Walker et al., 1995]. Contamination of crustal material was considered, and rejected, as a means to explain high δ18O values in Guatemalan lavas [Eiler et al., 2005]. Eiler et al. [2005] argue that potential assimilants (likely quartzo-feldspathic gneisses [Walker et al., 1995]) in the Guatemalan lavas have high 87Sr/86Sr ratios, therefore a positive correlation between 87Sr/86Sr and δ18O values would be seen if there was crustal contamination. There is also an expected trend of low Mg# lavas correlating with high δ18O values if the Guatemalan lavas were contaminated. Neither of these trends is observed. A plot of Mg# versus δ37Cl values has almost no correlation (R2 for all data = 0.12) (Figure 6). If crustal contamination occurred, and the δ37Cl value of the crustal contaminant was different from the mantle value, then the two should correlate. Therefore, either there is no crustal contamination or the two reservoirs have the same δ37Cl value.

image

Figure 6. Comparison of δ37Cl values and Mg# (molar Mg/(Mg + Fe)) for samples from the same volcano. Chlorine isotope data are from this study. Mg and Fe data used to calculate Mg# are from M. Carr's online database (http://www-rci.rutgers.edu/∼carr/index.html). Data are averaged for each volcano.

Download figure to PowerPoint

5.2.3. Center of the Front: Nicaragua

δ37Cl values of ashes and lavas from the center of the volcanic front in Nicaragua span almost the entire range seen in the arc, from −2.6 to +0.8‰. This range is nearly identical to that of subducting sediments on the Cocos plate (−2.1 to +0.7‰, Table 1) and to the global range of sediment data (−2.6 to +0.7‰) [Arcuri and Brimhall, 2003; Barnes et al., 2008; this study]. Note, however, that only 4 of 24 ODP seafloor sediment samples analyzed to date have positive δ37Cl values [Barnes et al., 2008; this study] and only 1 of 17 marine and nonmarine siltstones, mudstones, and limestones from Chile has a positive δ37Cl value [Arcuri and Brimhall, 2003]. The majority of sediments are isotopically negative. Therefore, the negative δ37Cl values of ash and lava imply the contribution of a sedimentary-derived component. However, the positive values are better explained by a serpentinite source [e.g., Barnes et al., 2008], a conclusion supported by the low δ18O values for olivine separates [Eiler et al., 2005]. Serpentinites formed by seawater hydration of peridotite have positive δ37Cl values [Barnes and Sharp, 2006], a signature that is visible in the Izu-Bonin-Mariana arc lavas [Barnes et al., 2008]. A serpentinite source is in apparent contradiction with nitrogen isotope data from Nicaraguan volcanic gases, in which high δ15N values correspond to a dominant (>70%) sediment contribution [Elkins et al., 2006]. In addition, Ba/Th, U/Th, and 10Be/9Be ratios all peak in Nicaraguan lavas, implying that the center of the front is characterized by the greatest sediment signature [Carr et al., 1990; Morris et al., 1990; Patino et al., 2000].

The data can be reconciled if we consider that elemental concentrations can vary significantly between different reservoirs. When Elkins et al. [2006] estimated a 71% sediment component to explain the positive δ15N values, they only considered the percentage of the total nitrogen from sediment. This calculation does not give relative total masses of different materials. N concentrations in sediments from the Ocean Drilling Program sites offshore of Costa Rica and the Izu-Bonin-Mariana arc are 62–2382 ppm and 5–661 ppm, respectively [Sadofsky and Bebout, 2004; Li and Bebout, 2005]. Cl concentrations (structurally bound Cl) of sediments from these same localities are 144–1391 ppm and 36–1240 ppm, respectively [Barnes et al., 2008; this study]. In contrast, N concentrations of seafloor serpentinites from the South West Indian Ridge (SWIR) and metaperidotites from the Erro Tobbio Massif range from 2.0 to 2.9 ppm and 3.9–5.9 ppm, respectively [Philippot et al., 2007]. Published Cl concentrations of serpentinites range from <0.1 wt% to over 1 wt%, with an estimated average of about 0.2–0.25 wt% [e.g., Anselmi et al., 2000; Barnes and Sharp, 2006]. Therefore, it is possible to have a large contribution of N from sediments and Cl from serpentinites in the same sample. Even a large addition of serpentinite-derived fluid would have little effect on the N isotope composition of the volcanic gas. The δ15N value of serpentinites from the SWIR and Erro Tobio range from +6.3 to +7.5 and +8.9 to +15‰, respectively [Philippot et al., 2007]. These values overlap those of sediments [e.g., Sadofsky and Bebout, 2004; Li and Bebout, 2005]. Therefore, the relative contributions of serpentinite and sediments to the δ15N of a volcanic gas are indistinguishable. Evidence for the serpentinite contribution is provided by both the high δ37Cl and low δ18O values samples from Nicaragua.

A second type of serpentinite with negative δ37Cl values is related to hydration of peridotite by pore water [Barnes and Sharp, 2006]. One of the two DSDP/ODP sites where negative δ37Cl values have been found is offshore Guatemala (DSDP Leg 84 Hole 570 [Barnes and Sharp, 2006]). Although the low δ37Cl values seen in Nicaragua can be explained by dehydration of sediments, dehydration of this second type of serpentinite could also explain the low δ37Cl values. Recent geophysical evidence also supports a serpentinite-derived fluid phase contributing to Nicaraguan magmatism. Multichannel seismic reflection data have documented 18–20 km deep faults offshore Nicaragua due to plate bending at the outer trench [Ranero et al., 2003]. These faults provide conduits for fluid infiltration allowing for significant amounts of serpentinization (up to ∼30%) in the subducting lithosphere [Rüpke et al., 2002; Ranero et al., 2003; Grevemeyer et al., 2007]. Moreover, low seismic velocities in the subducting Cocos slab beneath Nicaragua suggest significant hydration of the slab (≥5 wt% water, 2–3 times the hydration inferred for other slabs) [Abers et al., 2003]. Low heat flow values also provide evidence for large amounts of hydrothermal circulation in the crust [Fisher et al., 2003; Grevemeyer et al., 2005].

There is some speculation as to the presence of evaporites in Nicaragua based on brief mention of locally dispersed evaporites [Weyl, 1980] and analogy of the Nicaragua Depression to an inland sea that partially dried up due to evaporation; however, regionally occurring evaporite deposits have not been identified. Evaporites have well-known δ37Cl values of 0 ± 0.5‰ [Eastoe et al., 2007]; therefore, assimilation of crustal evaporite material would drive the δ37Cl values toward 0‰. The distinctly negative δ37Cl values observed in the Nicaraguan ashes argues against assimilation of significant crustal evaporites.

5.2.4. Back Arc and Secondary Front: Honduras

The three samples from the back arc in Honduras (Lake Yojoa region, 170 km behind the volcanic front [Patino et al., 1997]) and two samples from the secondary front, also in Honduras, have δ37Cl values ranging from −1.8 to −0.5‰. These data lie within the range of sediment, altered oceanic crust, and/or serpentinite chlorine isotope compositions.

Previous work on alkali basalts from Lake Yojoa interprets low 87Sr/86Sr and high La/Yb ratios in the basalts as evidence of melt derivation from a normal mid-ocean ridge basalt source with low degrees of melting with little input from slab-derived components [Patino et al., 1997]. However, Eiler et al. [2005] observe anomalously high δ18O values in Yojoa olivines, as in olivines from Guatemala and El Salvador. Although data are limited, they could be explained by deeply subducted sediment melts that have lost their water-soluble elements [Eiler et al., 2005]. However, these high δ18O values may also reflect assimilation of crustal material. Additionally, de Leeuw et al. [2007] report δ13C values of −3.7 to −0.3‰ for geothermal samples of bubbling hot springs from Sula Graben, 200 km behind the volcanic front in Honduras, implying a sedimentary slab source for carbon. It is, therefore, plausible that the negative δ37Cl values from samples behind the volcanic front in Honduras reflect a slab-derived sedimentary component.

Alternatively, isotopically negative, dehydrating serpentinites may be responsible for the chlorine chemistry of the Honduran back arc. Thermal modeling across several transects of the Costa Rica–Nicaragua subduction zone suggests hydrated oceanic mantle beneath Nicaragua and Costa Rica may still contain significant amounts of water at depths >240 km [Peacock et al., 2005]. The modeling also predicts only minor along-strike variations in the thermal structure of the subduction zone [Peacock et al., 2005], implying that the thermal structure is not significantly different at the northern end of the arc and serpentinites may indeed carry water to the back-arc melting region. A serpentinite source for back-arc material is suggested also in the Izu-Bonin-Mariana arc [Stern et al., 2006; Barnes et al., 2008], where basalts from the Gugan cross chains have δ37Cl values consistent with breakdown of antigorite at depth [Barnes et al., 2008]. A similar scenario could be envisioned here, but the negative δ37Cl values in the Honduran back arc require that the serpentinites were hydrated by pore waters due to flexure-related faulting of a subducting slab. Variability in boron isotope data from El Salvadoran lavas has also been explained by a 10–50% serpentinite-derived fluid component. The serpentinite source is believed to be the mantle wedge dragged down, rather than subducted oceanic lithosphere [Tonarini et al., 2007]. However, the serpentinite contribution is thought to be reduced toward the back arc [Tonarini et al., 2007].

5.3. Physical Volcanic Segmentation Linked to Chemistry

The Central American volcanic front can be divided into eight different volcanic segments based on dextral offsets by as much as 40 km or changes in strike along the front [e.g., Carr et al., 2003]. Segments are shown in Figure 7a. Some of these physical breaks in the volcanic front are linked to variations in geochemical trends (gradient changes in trends and offsets in Ba/La ratios between the segments) (Figure 7c) [Patino et al., 2000].

image

Figure 7. (a) Segmentation map modified from Carr et al. [2003]. Gray bars denote segments defined by Stoiber and Carr [1973]. QSC, Quesada Sharp Contortion [Protti et al., 1995]. Subducting ridges (Fisher and Quepos) are from von Huene et al. [2000]. Dotted line delineates the boundary of crust formed from the East Pacific Rise (EPR) from crust formed by the Cocos-Nazca spreading center (CNS) [von Huene et al., 2000]. LP, Las Pilas; CN, Cerro Negro; Iz, Izalco. (b) The δ37Cl values of ash samples versus distance along the arc. For the purposes of this study, the division between northern and central Costa Rica is made south of Arenal. (c) Trends in Ba/La ratios versus distance along the arc. Ba/La data are from Patino et al. [2000].

Download figure to PowerPoint

Offsets in δ37Cl values are also correlated with some of the physical segmentation breaks (Figure 7b), however some of these correlations may be an artifact given the limited amount of Cl isotope data. For example, the δ37Cl values increase at the boundary of Nicaragua and El Salvador, where the volcanic chain is offset by ∼10 km [Carr et al., 2003]. This trend is mirrored by a drop in the Ba/La ratios [Carr et al., 2007]. At the boundary from southern to northern Nicaragua, the chain is again offset by at least 10 km [Carr et al., 2003]; however, both δ37Cl values and Ba/La ratios continue to increase at the border. The largest physical offset along the front is between Costa Rica and southern Nicaragua [e.g., Carr et al., 2003]. Here the Wadati-Benioff zone contorts from a steeper dip under Nicaragua to a shallower dip under Costa Rica [Protti et al., 1995]. This offset is also marked by a drastic decrease in δ37Cl values and a decrease in Ba/La ratios from Costa Rica to Nicaragua. The segmentation break between central and northern Costa Rica is defined by a tear in the subducting plate, known as the Quesada Sharp Contortion (QSC) [Protti et al., 1995], which correlates with subduction of the Fisher Seamount and an offset and gap between Platanar and Arenal volcanoes [von Huene et al., 2000]. Here there is a slight increase in Ba/La ratios; however, δ37Cl values remain constant across the boundary.

The reason for the correlation between Ba/La gradient changes at some of the physical offsets in the volcanic chain, but not at others, is not well understood [Carr et al., 2003]. Patino et al. [1997] observe a decrease in Ba/La across the arc in Honduras implying that changes in the depth to the Wadati-Benioff zone (WBZ) may control Ba/La ratios. Therefore, Ba/La ratios should consistently decrease from SE to NW across segment breaks due to an increased distance from the trench; however, observations fail to support this hypothesis [Carr et al., 2003]. Similarly, there is an inconsistency in the chlorine isotope trends across segment breaks. Of particular interest are the high δ37Cl values recorded in the Miravalles ashes. These high values may reflect alteration by a vapor condensates (see above), but it is also worth noting that Miravalles volcano closely aligns with the boundary of a fossil triple junction in the subducting oceanic crust, which separates crust formed by the Cocos-Nazca spreading center (CNS) and the East Pacific Rise (EPR) [von Huene et al., 2000]. However, Ba/La ratios and δ37Cl values need not correlate (as seen in Figure 5a). Elevated Ba/La reflect a sediment input into the source magma. Negative δ37Cl values may reflect sediment input, but can also reflect hydrothermally altered rock. It is possible to change the δ37Cl value by changing the source material and not alter the Ba/La ratio.

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

δ37Cl values of ash, tephra, and lava samples from along the Central American volcanic front vary systematically due to changes in forcing functions.

1) Costa Rican samples, with the exception of unusually heavy samples from Miravalles, have mantle-like δ37Cl values, consistent with results from δ18O, δ15N, Ba/La, U/Th, La/Yb, 10Be/9Be data and Nd isotopes. Incorporation of eroded serpentinite fore-arc material into the mantle wedge may be reflected in some positive δ37Cl values from Costa Rica.

2) Chlorine isotope data from El Salvadoran and Guatemalan samples also suggest a mantle source, a conclusion supported by low U/Th and 10Be/9Be ratios. However, δ18O and δ15N data imply a large sediment component to Guatemalan and El Salvadoran lavas. These conflicting results could reflect an influence of shallow level processes on the O and N isotope systems (e.g., crustal assimilation and/or alteration) or may reflect the transport mechanism from the subducting slab to the region of magma genesis and not the original signature itself. Sediment melts may transport the O and N signature, but fail to transport fluid mobile elements, such as Ba, U, and Cl, resulting in a mantle background signature for these tracers.

3) Nicaraguan ashes and lavas have both isotopically positive and negative δ37Cl values that can be due to either sediment and/or hydrothermally altered rock (altered oceanic crust and/or possibly serpentinites) contribution. Ba/La, U/Th, and 10Be/9Be indicate a sediment component; whereas, high 87Sr/86Sr ratios imply altered oceanic crust-derived fluids and oxygen isotope data distinctly fingerprints a hydrothermal alteration signature. Recent work has highlighted the likely presence of serpentinites, and hence a significant fluid source, in the Cocos slab subducting beneath Nicaragua. Fluid from the slab would mobilize water-soluble elements from subducting sediments. The positive δ37Cl values in Nicaragua, and likely the negative ones as well, are due to serpentinite-derived fluid, possibly with some sediment-scavenged chlorine.

4) Negative δ37Cl values in the Honduran back arc may also be explained by serpentinite-derived fluids.

5) Recent work has highlighted geochemical segmentation along the volcanic front correlating with physical segmentation. Chlorine isotope geochemistry shows similar segmentation in portions of the arc; however, the reasons for this correlation are not known and represent an area of required future work.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References

The authors thank G. Bebout for sediment samples and the Ocean Drilling Program (ODP) for providing samples and data. ODP is sponsored by NSF and participating countries under the management of the Joint Oceanographic Institutions (JOI), Inc. This manuscript was greatly improved by thoughtful reviews from T. Hansteen, B. Leeman, M. Reagan, and K. Hoernle (Associate Editor). V. Salters is thanked for editorial handling. This work was supported by NSF grants (EAR-SGER-0620160 and EAR-0711533) and by a L'Oréal USA Fellowship for Women in Science to J.B.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Geologic Setting
  5. 3. Samples and Analytical Methods
  6. 4. Results
  7. 5. Interpretations and Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References