SEARCH

SEARCH BY CITATION

Keywords:

  • Izu-Bonin-Mariana;
  • nitrogen isotopes;
  • nitrogen fluxes;
  • subduction zones;
  • volatiles

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[1] We report new chemical and nitrogen isotopic data from 29 volcanic and hydrothermal gas samples covering eight centers in the Izu-Bonin-Mariana (IBM) arc to investigate the sources, flux, and mass balance of nitrogen at a “cool” convergent margin. The majority of samples have high N2/He (1217–17,300) and low CO2/N2,exc. (78–937), implying addition of nitrogen from the subducting slab. This inference is supported by positive (i.e., sediment-like) δ15N values (up to 5.5‰) in most samples. The exception to these trends is Agrigan in the Mariana arc, with low N2/He (∼200), high CO2/N2,exc. (∼1500), and negative δ15N. Mixing calculations suggest an average of 34% of the nitrogen in our samples is derived from subducted sediment, or 75% after correction for atmospheric contamination. Sediment-derived N2 fluxes estimated by three different methods range from 0.25 × 108 to 1.11 × 108 mol yr−1 N2, representing 4%–17% of the total nitrogen input flux or 11%–51% of the sedimentary nitrogen input flux. The altered oceanic crust is identified as an important contributor to the arc nitrogen budget, and the δ15N of the residual nitrogen subducted into the mantle is estimated at approximately −1.9‰. Despite similarities in gas chemistry and δ15N values, our conclusions regarding nitrogen recycling for IBM are markedly different than those for the Nicaraguan segment of the Central American arc, and we suggest that thermal regime is the major control on nitrogen recycling within subduction zones. The global nitrogen cycle is estimated to be in steady state, suggesting either that subducted sediments are an unlikely source for heavy nitrogen in plume-related rocks or secular variation in the isotopic composition of subducted sediments. Better constraints on nitrogen recycling at other arcs are required to test these conclusions.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[2] Subduction zones are sites of material transfer between the Earth's internal (i.e., mantle) and external (i.e., crust, oceans and atmosphere) reservoirs and as such exert a fundamental control on the distribution of elements within the Earth. Due to contrasting isotopic compositions nitrogen has been shown to be a useful tracer of material exchange between these reservoirs [e.g., Sano et al., 1998; Fischer et al., 2002]. Constraints on the amount of nitrogen degassing from the mantle and recycled to the atmosphere through subduction zones are crucial components of models [e.g., Javoy, 1997, 1998; Tolstikhin and Marty, 1998; Marty and Dauphas, 2003] that seek to explain the isotopic imbalance between the Earth's internal (δ15N < 0) and external (δ15N ≥ 0) reservoirs [Busigny et al., 2003]. Efficient recycling between external and internal reservoirs will act to homogenize their isotopic composition, as is the case for carbon and sulfur [Hilton et al., 2002]. In contrast, the large discrepancy between atmospheric and mantle 40Ar/36Ar and 3He/4He is consistent with a subduction barrier for the noble gases [Staudacher and Allègre, 1988]. Fischer et al. [2002] suggested that this barrier may exist for nitrogen as well.

[3] The nitrogen isotope composition of the various reservoirs is harder to interpret, with apparent evidence for recycling to both the atmosphere and mantle. The source of MORB and diamonds seem to have the same isotopic composition, approximately −5‰ [Marty and Dauphas, 2003], implying little recycling of nitrogen over the Ga time scale of diamond formation and storage [Hilton et al., 2002]. Recent studies of the Central American arc [Fischer et al., 2002; Zimmer et al., 2004; Elkins et al., 2006] support this idea, with the majority of subducted nitrogen being returned to the atmosphere via arc volcanism, bypassing recycling to the deeper mantle. The hypothesis of efficient recycling is at odds with studies of high-pressure (HP) and ultrahigh-pressure (UHP) oceanic metasediments from the Western Alps, however, which conclude that in a “cool” subduction environment (as currently exists in most western Pacific subduction zones such as the Kamchatka-Kurile-Honshu and Izu-Bonin-Mariana systems [Peacock, 2003]) nitrogen is retained in the subducted sediments, at least to the depths of arc magma genesis [Busigny et al., 2003]. High nitrogen concentrations in UHP metamorphic diamonds [Busigny et al., 2003], and the discovery of positive δ15N values in plume-related volcanic rocks (e.g., the Kola magmatic province, Loihi Seamount, Hawaii), support the contention that sedimentary nitrogen can be recycled into the deep mantle [Dauphas and Marty, 1999; Marty and Dauphas, 2003; Fischer et al., 2005].

[4] In this study, we aim to complement the existing work on the Central American Volcanic Arc and characterize the mass balance of N2 across the Izu-Bonin-Mariana (IBM) subduction system. Volatile chemistry and nitrogen stable isotope data are used to constrain the sources of nitrogen within, and the flux of nitrogen from, the arc system, and comparisons are then developed with the Central American (CA) arc. Finite element thermal models [e.g., Peacock, 2003; Peacock et al., 2005; Wada and Wang, 2009] indicate that slab-mantle interface temperatures are predicted to be greater at a given depth within the “warm” Central American arc system (∼620–800°C) compared to the relatively “cool” IBM system (∼540°C), due primarily to the age and velocity of the subducting slab (∼25 Ma and 80 mm yr−1 versus ∼140 Ma and ∼40 mm yr−1, respectively). Our comparison between IBM and CA suggests that subduction zone thermal structure may be a major control on the efficiency of volatile recycling at arcs.

2. Background

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[5] The IBM arc is the type example of an intraoceanic convergent margin. It is located in the western Pacific, where the Pacific Plate subducts westward beneath the Philippine Sea Plate, and stretches ∼2800 km from near Tokyo, Japan (∼35°N), to beyond Guam, USA (∼11°N) (Figure 1). The thin arc crust (∼20 km), lack of an accretionary prism, and simple variation in subducting sediment lithology, slab dip and slab age [Stern et al., 2003], make IBM an ideal locality for studying volatile recycling. Furthermore, the subducted oceanic crust is the oldest of any subduction zone, providing an end-member scenario for investigating a “cool” convergent margin.

image

Figure 1. Sketch map of the Izu-Bonin-Mariana arc system. Black triangles are sampled arc volcanoes. Heavy black lines are trenches, with teeth on the overriding plate. Single line denotes the volcanic front, double line is the active back-arc spreading ridge, and dashed lines represent older volcanic arcs.

Download figure to PowerPoint

[6] The nitrogen concentration and isotopic composition of the ∼500 m thick sedimentary sequence being subducted at the IBM margin has been well characterized at ODP Site 1149 (outboard of the Izu-Bonin arc) [Sadofsky and Bebout, 2004], and the uppermost ∼500 m of altered oceanic crust (AOC) was drilled at Site 801 (outboard of the Mariana arc) [Li et al., 2007]. The sedimentary sequence at both sites is dominated by chert and pelagic clay, with little carbonates except near guyots [Stern et al., 2003]. Sediment composition differs markedly between the southern and northern arc due to the presence in the south of a ∼200 m thick volcaniclastic sequence, which is absent in the north [Stern et al., 2003].

[7] Fluids and magmas are discharged from the arc in four different settings; the fore arc, the active volcanic front, rear-arc cross chains and back-arc basins. Degassing subaerial volcanoes, the focus of this study, are restricted to the volcanic front and rear-arc cross chains.

3. Samples and Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[8] A total of 29 volcanic and hydrothermal gas samples were collected from 6 centers in the Izu-Bonin arc (Hakone, Niijima, Shikinejima, Oshima, Hachijojima and Aogashima) and 2 in the Mariana arc (Pagan and Agrigan) (Figure 1). All gas samples were collected in evacuated ∼200 mL pyrex flasks (“Giggenbach bottles”) containing ∼60 mL of 5N NaOH solution using standard techniques as described by Giggenbach and Goguel [1988] (Appendix A).

[9] Chemical analyses of all gas samples were made in the Volcanic and Geothermal Fluid Analysis Laboratory at the University of New Mexico (UNM) using standard gas chromatography (GC) and wet chemical techniques described by Giggenbach and Goguel [1988] and Zimmer et al. [2004] (Appendix A). Uncertainties on GC measurements are estimated at ±5% [Fischer et al., 2002; T. P. Fischer and M. Simoes, unpublished report, 2002]. Chloride, fluoride and sulfate concentrations were also analyzed at UNM in the Analytical Chemistry Laboratory by ion chromatography (IC) on a Dionex DX500 (Appendix A). Based on prerun and postrun analyses of standards, uncertainty on IC measurements is less than ±5%.

[10] Nitrogen isotope analyses were conducted in the Stable Isotope Laboratory at UNM on a Finnigan Mat Delta Plus continuous flow mass spectrometer with a Finnigan Mat Gas Bench interface [Fischer et al., 2002; Zimmer et al., 2004] (Appendix A).

4. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

4.1. Reactive Gases

[11] As is typical of subduction zone volcanoes, gases are dominated by water, with most samples containing >99 wt % H2O. Only four samples, Uto-1/2 and Jin-1/2, contain <95% H2O (Table 1). CO2 is typically the second most abundant gas species, with concentrations ranging from 344 (Uto-2) to 973 (Mar-9) mmol/mol dry gas. Total sulfur concentrations vary over nearly three orders of magnitude, and also vary as a function of sample type, with the fumarole samples spanning the largest range, from 0.55 (Aog-3) to 436 (Sou-2) mmol/mol. HCl, HF and NH3 concentrations are generally low (<10 mmol/mol in most cases). H2O and HCl concentrations are not reported for low gas flow rate geothermal well samples where water was deliberately allowed into the flask during sampling to enhance exsolution of gas from the water into the evacuated headspace.

Table 1. Chemistry of Izu-Bonin-Mariana Fluids
IslandSample LocationSample IDLatitude (°N)Longitude (°E)Sampling DateSample TypeaTemperature (°C)pHH2ObCO2StSO2H2SHClcHFdNH3ArN2HeH2O2CH4COδ15N (‰)
  • a

    Sample types: F, fumarole; G+W, geothermal well plus excess water; G, geothermal well; BS, bubbling spring.

  • b

    H2O reported as mmol/mol total gas; all other species as mmol/mol dry gas.

  • c

    Values not reported for samples containing excess water from the sampling well.

  • d

    n.m., not measured.

  • e

    Indicates a corrected value on the basis of [N2].

HakoneSounjigoku Fumarole FieldSou-135 14.492′139 01.925′8/5/2005F96.15.5990.61516.46432.2964.36366.360.9890.1880.0570.06914.085<0.0004834.915<0.0040.949<0.000125.5 ± 0.5
  Sou-235 14.492′139 01.925′8/5/2005F96.15.5992.82522.56436.4163.28371.543.0210.3480.2020.05311.102<0.0006425.713<0.0050.585<0.000163.5 ± 0.6
 Owakdani Fumarole FieldOwa-135 14.488′139 01.230′8/5/2005F96.55.5992.95613.05341.4249.52290.6510.4710.5310.2440.07314.593<0.0004619.509<0.0040.106<0.000123.0 ± 0.5
 Yunohanazawa Fumarole FieldNoh-135 13.442′139 01.948′8/5/2005F95.85.0996.59698.32219.7191.87127.236.2790.6870.6790.60466.998<0.002295.754<0.0190.943<0.00057−1.4 ± 0.02
OshimaOhtsu WellOht-134 45.581′139 21.358′7/21/2005G+W36.46.7 902.2655.3535.4119.83 9.6080.7250.490e20.993<0.002580.01110.558<0.001<0.000680.0 ± 0.1
 Koshimizu WellKos-134 45.521′139 21.704′7/21/2005G+W48.56.7 881.8271.6962.149.45 10.3000.8090.588e25.399<0.002980.0109.380<0.001<0.00077−0.6 ± 0.1
 Univ. of Tokyo Obs. WellUto-134 45.243′139 24.125′7/22/2005G81.04.0744.81521.881.13 1.1310.3201.4671.1054.661e384.388<0.005910.02675.020<0.001<0.001430.9 ± 0.05
  Uto-234 45.243′139 24.125′7/22/2005G81.04.0416.51344.060.48 0.484.2020.3460.2016.529e540.343<0.003590.019103.814<0.001<0.000860.9 ± 0.1
 Hotel Onsen WellOns-134 45.213′139 24.231′7/22/2005G+W84.96.7 965.7512.76 12.76 3.3161.3640.289e12.368<0.002270.1263.9180.108<0.00059−2.5 ± 0.1
  Ons-234 45.213′139 24.231′7/22/2005G+W84.96.7 944.5217.43 17.43 2.5822.2560.472e20.788<0.007390.27711.4250.237<0.00184−2.2 ± 0.2
NiijimaMamashita Hot Spring WellMam-134 21.670′139 14.699′7/20/2005G+W62.07.0 737.05218.56 218.56 8.6760.6430.711e30.804<0.002840.0093.542<0.001<0.00073−0.3 ± 0.2
  Mam-234 21.670′139 14.699′7/20/2005G+W62.07.0 804.36185.05167.3417.47 n.m.0.0990.250e9.421<0.003100.0040.8090.001<0.000140.6 ± 0.2
ShikinejimaJinata Hot SpringsJin-134 19.128′139 12.882′7/19/2005BS66.27.0771.05908.891.26 1.2629.6410.1560.5830.707e58.1900.024950.0040.4010.140<0.000281.6 ± 0.2
  Jin-234 19.128′139 12.882′7/19/2005BS66.27.0705.65827.100.78 0.7819.0050.1270.0681.765150.0580.064080.0060.6790.348<0.000252.0 ± 0.2
HachijojimaSueyoshi Hot Spring WellSue-133 04.889′139 50.985′8/2/2005G+W43.98.0 787.87144.55 144.55 n.m.0.8300.71365.985<0.002930.014<0.0250.012<0.00078−0.1 ± 0.1
 Kashidate Hot Spring WellKas-133 04.562′139 47.423′8/2/2005G+W44.67.0 926.1561.06 61.06 n.m.0.0810.290e12.243<0.000950.0040.1680.007<0.000260.1 ± 0.1
 Geothermal Power PlantGeo-133 04.494′139 48.742′8/3/2005G170.85.5990.76864.94121.5642.6978.511.3630.3560.0510.0051.3110.0010810.389<0.0030.022<0.000084.8 ± 0.1
  Geo-233 04.494′139 48.742′8/3/2005G170.85.5990.68820.18107.5663.4643.862.2920.2541.4380.26346.598<0.0010021.3180.0480.0430.003543.0 ± 0.1
 Ogoshi Hot Spring WellOgo-133 03.830′139 49.002′8/2/2005G+W53.37.5 918.3047.50 47.50 1.3650.5760.692e29.041<0.001760.0442.4070.069<0.000481.9 ± 0.1
 Yasuragi Hot Spring WellYas-133 03.623′139 48.894′8/2/2005G+W54.46.5 854.43118.30113.234.94 4.4180.7830.510e20.7900.007290.0080.7090.040<0.000330.3 ± 0.1
AogashimaFunakoyama Fumarole FieldAog-332 27.435′139 45.717′7/29/2005F99.45.0996.20861.860.55 0.553.2190.7680.9451.240e101.3330.014191.17927.8631.028<0.00054−2.5 ± 0.1
  Aog-3a32 27.435′139 45.717′7/29/2005F99.45.0996.78835.460.68 0.6812.1161.7640.7111.379e112.4250.017441.30433.0681.074<0.00049−2.1 ± 0.1
 Maruyama Volcano Central ConeAog-132 27.182′139 45.915′7/29/2005F85.85.3898.22504.561.27 1.2710.6331.2563.4574.679e386.547<0.008810.02387.562<0.002<0.002131.3 ± 0.1
 Misone Fumarole FieldAog-232 27.136′139 45.735′7/29/2005F99.25.0986.79965.526.31 6.310.8620.1930.0900.13822.7450.010523.3880.0110.733<0.000081.6 ± 0.2
  Aog-2a32 27.136′139 45.735′7/29/2005F99.25.0988.68961.546.902.644.233.5851.1410.1990.13121.6110.012553.963<0.0040.916<0.000131.3 ± 0.1
PaganOld CraterMar-318 04.265′145 43.487′4/18/2004F96.15.0964.18942.848.11 8.112.2180.0050.0310.39835.3620.004664.1096.8730.049<0.000100.7 ± 0.2
  Mar-918 04.265′145 43.487′4/18/2004F96.15.0949.37972.569.07 9.072.5270.0050.0160.0139.7300.003884.9181.1050.056<0.000231.6 ± 0.4
AgriganSummit CraterMar-718 46.694′145 40.2824/21/2004F98.03.0957.91937.0556.0211.0744.762.5010.0050.0180.0382.8120.013321.5160.0150.0110.00323−2.7 ± 0.3
  Mar-818 46.694′145 40.2824/21/2004F98.03.0890.19883.17109.20 109.204.7960.0030.0060.0291.8740.007630.9050.0030.0070.00141−2.5 ± 0.5

4.2. Nonreactive Gases

[12] Nitrogen is typically the most abundant nonreactive gas; however, in some samples (e.g., Sou-1/2 and Geo-1/2), hydrogen concentrations are comparable or higher. Oxygen concentrations can reach high values in severely air contaminated samples (e.g., Uto-1/2, Aog-1 and Aog-3/3a).

[13] Argon in volcanic gases is assumed to be entirely of atmospheric origin [Giggenbach, 1996], thus N2/Ar ratios should not be lower than the air or air-saturated groundwater (asw) values of 83 and 45, respectively [Fischer et al., 1998]. Lower values may indicate a severely air-contaminated sample with high oxygen concentrations. Argon and oxygen are not separated by our GC column, so an O2 trap is used in series with the first GC column to remove O2 in order to measure Ar. However, where O2 concentrations are high (greater than ∼15 mmol/mol; 5 of 29 samples), the O2 trap can become saturated, leading to some O2 being measured as Ar [Elkins et al., 2006]. In these cases the reported Ar value in Table 1 (denoted by footnote d) is calculated from the N2 concentration together with the air or asw N2/Ar ratio [Zimmer et al., 2004; Clor et al., 2005].

[14] The inert gases N2, He and Ar can be used to qualitatively estimate the relative contribution of different source components, as illustrated in Figure 2 (following Giggenbach [1996]). Mantle-derived gases have relatively low N2 concentrations and N2/He ratios <200 [Giggenbach, 1996; Fischer et al., 1998]. In contrast, arc-related volcanic gases have higher N2 concentrations and N2/He > 1000, due to the addition of sedimentary nitrogen from the subducting slab [Matsuo et al., 1978; Kita et al., 1993; Giggenbach, 1996; Fischer et al., 1998; Sano et al., 2001; Fischer et al., 2002]. As can be seen in Figure 2, the majority of samples lie in the arc-related field, with varying degrees of air contamination drawing samples toward the air or air-saturated groundwater points on the N2-Ar axis. The only samples that unambiguously plot in the field of mantle-derived gases are the two samples from the island of Agrigan in the Mariana arc, with N2/He of 211 and 246. The N2-He-Ar results are consistent with addition of sedimentary nitrogen from the subducting slab to all centers sampled in the IBM arc with the exception of Agrigan. This is surprising given that prior geochemical studies [e.g., Elliott et al., 1997; Kent and Elliott, 2002] have argued that melts from Agrigan are thought to be dominated by a sedimentary component.

image

Figure 2. N2-He-Ar ternary plot showing fields for mantle-derived and “arc-type” gases, as well as N2-Ar ratio of air (83) and air-saturated groundwater (ASW) (45). Colors of symbols are as follows: black, fumaroles; green, bubbling springs; red, geothermal wells. Symbols outlined in red are geothermal well samples where water was deliberately allowed into the flask during sampling. Grey circles are data for the Nicaraguan segment of the Central American arc from Elkins et al. [2006]. After Giggenbach [1996].

Download figure to PowerPoint

[15] Another tracer of gas sources is the CO2/N2 or CO2/N2,exc. ratio. CO2 is the most conservative major species in volcanic gases [Giggenbach, 1996] and variations in CO2/N2 ratios therefore generally relate to variable N2, not CO2, concentrations, although calcite precipitation at depth can be a factor in some geothermal systems [Snyder et al., 2001]. To circumvent the problem of air contamination, a more meaningful ratio is CO2/N2,exc., where N2,exc. is the amount of nitrogen in excess of that contributed from air, calculated according to the following relationship [Fischer et al., 1998]:

  • equation image

where (N2/Ar)air is 83 and (N2/Ar)asw is 45. The (N2/Ar)asw value is used to correct samples where water was allowed into the sampling flask or where the measured N2/Ar is less than 83, while the (N2/Ar)air value was used to correct all other samples. Typical CO2/N2,exc. values of mantle-derived gases are greater than ∼700 and “arc-type” gases are generally 100 ± 60 [Fischer et al., 1998]. Table 2 shows that CO2/N2,exc. values generally support the N2/He-based interpretations, with low values in the typical “arc-type” gas range for most samples and higher values (>1000) for the Agrigan samples, consistent with a dominantly mantle wedge source for these gases.

Table 2. Ratios and Calculations Used to Evaluate Data and in the Discussiona
IslandSample IDN2/HeN2/ArN2,exc.CO2/N2CO2/N2,exc.δ15NMSAMcSc%Sδ15Nc%AOCM*S*AOC*
  • a

    Mc and Sc, fractions in a mantle-sediment binary mixture; δ15Nc, corrected value following the method of Fischer et al. [2002]; %AOC, AOC contribution in a sediment-AOC binary mixture required to explain δ15Nc values (see section 6.2.5 for details); M*, S*, and AOC*, results of mixing calculations using mantle, sediment, and AOC end-members and δ15Nc values (see section 6.2.5 for details) (na indicates sample fell out with mixing lines).

HakoneSou-217,3002096.747.178.03.5 ± 1.10.000.500.500.001.001007.00nanana
NiijimaMam-23,039  85.4 0.6 ± 0.40.050.120.830.290.71713.53536532
ShikinejimaJin-12,332  15.6 1.6 ± 0.40.060.270.670.180.82824.82257916
 Jin-22,342853.25.5231.72.0 ± 0.40.060.330.610.150.85855.21958312
HachijojimaGeo-11,2172810.9659.6936.84.8 ± 0.30.110.770.120.120.88885.61511881
 Ogo-116,544  31.6 1.9 ± 0.30.000.280.720.001.001007.00nanana
 Yas-12,852  41.1 0.3 ± 0.10.050.080.870.380.62622.44635443
AogashimaAog-22,16116511.342.585.41.6 ± 0.50.060.280.660.180.82824.92158015
 Aog-2a1,72216510.744.589.81.3 ± 0.20.080.250.670.240.76764.12977221
PaganMar-92,5087498.7100.0112.41.6 ± 0.80.050.270.680.160.84845.11948214
AgriganMar-7211740.9333.21016.3−2.7 ± 0.70.710.120.170.860.1414−3.3 701119
 Mar-8246650.4471.32175.3−2.5 ± 1.00.610.080.310.880.1212−3.6 60634
Mantle (M) 150    −5.0           
Sediment (S) 10,500    7.0           
Air (A) 148,900    0.0           
AOC 4,500    −2.9           
Average (IB)    168333  40   85  67420
Average (M)    217564  20   49  453322
Average (IBM)    182410  34   75  176221

4.3. Nitrogen Isotopic Composition

[16] Nitrogen isotope measurements are reported in the delta notation, where δ15N is the per mil (‰) deviation of the sample 15N/14N ratio from that of air, the standard for N isotope measurements. Thus, by definition, air has δ15N = 0. Nitrogen isotope values of the IBM samples range from −2.7 to +5.5‰, with 1σ errors ranging from ±0.02 to 0.6‰ (Table 1). The most positive values are those from Hakone, with sample Sou-1 having a δ15N value of +5.5‰. The other Hakone samples are also positive, with the exception of sample Noh-1, with δ15N = −1.4‰. This particular sample also has unusually high δ13C values [Hilton et al., 2006] suggesting that the original magmatic carbon and nitrogen isotope values were overprinted by shallower level sources. The island of Agrigan has the most negative values of −2.7‰ and −2.5‰, and is the only volcano of the present study to have exclusively negative values. Duplicate samples are generally similar (i.e., Mar-7, Mar-8), although if one sample is more air contaminated than the duplicate, its value is closer to zero, consistent with the occurrence of air contamination during sampling (e.g., Geo-1, Geo-2).

[17] Two volcanoes, Oshima and Aogashima, have samples with both positive and negative values. In both cases, the negative values (Ons-1/Ons-2 from Oshima and Aog-3/Aog-3a from Aogashima) are from samples that are severely air contaminated (see discussion below). This is unexpected as air is, by definition, 0‰. The fact that both duplicate samples from each site have similar values suggests that these results are not due to fractionation during the collection or analysis procedure. Both samples seem to be sampling a severely air-contaminated component with δ15N < 0. This observation could indicate a small fractionation between air and water. In this respect, we note that δ15N values varying by ±2‰ from 0‰ have been observed previously for severely air contaminated samples [Clor et al., 2005].

[18] The δ15N values for all samples fall within the range suggested for possible sources contributing nitrogen to these gases; upper mantle with δ15N of −5 ± 2‰ [e.g., Marty, 1995; Marty and Humbert, 1997; Cartigny et al., 1998] and hemipelagic sediments with δ15N of +7 ± 4‰ [e.g., Peters et al., 1978; Sadofsky and Bebout, 2004; Li and Bebout, 2005]. The generally positive δ15N values of the present sample suite, and the negative values for the Agrigan samples, are consistent with the gas chemistry results and support a large sedimentary contribution to all IBM arc centers except Agrigan.

5. Data Integrity

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[19] Before addressing issues related to nitrogen sources, fluxes and mass balance, it is essential to evaluate the database to ensure that samples are representative of subarc sources and processes, and have not been compromised by air or crustal contamination. The criteria used here for including/excluding samples are based on those used by previous workers as reliable indicators of air or crustal contamination [e.g., Zimmer et al., 2004; Clor et al., 2005; Elkins et al., 2006].

5.1. Air Contamination

[20] A number of indicators can be used to identify severely air-contaminated samples, including high N2/He (e.g., Uto-1 = 65,040, Uto-2 = 150,513), low He/Ar (e.g., Kas-1 = 0.0033, Aog-1 = 0.0019) and high O2 concentrations (e.g., Aog-3 = 27.9, Aog-3a = 33.1). On the basis of these indicators the following samples are excluded from further consideration because they possess one or more of the characteristics described above: Sou-1, Owa-1, Noh-1, Oht-1, Kos-1, Uto-1, Uto-2, Ons-1, Ons-2, Mam-1, Sue-1, Kas-1, Geo-2, Aog-3, Aog-3a, Aog-1, Mar-3.

[21] In addition, samples from localities Oht, Kos, Uto, Ons, Kas, Sue and Aog-1 have 3He/4He ratios less than 5RA (where RA is the atmospheric ratio of 1.4 × 10−6 [Poreda and Craig, 1989; Hilton et al., 2002]), and in most cases less than 3RA [Hilton et al., 2006], indicating severe air contamination. The He isotope measurements were made on samples collected at the same time and place but in different sampling flasks. These samples are therefore interpreted to have been air contaminated before sampling, and not during sample collection. In contrast, the remaining air-contaminated samples (Sou-1, Owa-1, Noh-1, Mam-1, Geo-2, Aog-3, Aog-3a and Mar-3) are interpreted to have suffered air contamination during sample collection.

5.2. Crustal Contamination

[22] A reliable indicator of crustal contamination is the 3He/4He ratio [Hilton et al., 2002], which is 8 ± 1 RA for the upper mantle and ∼0.02 RA for crustal material [Ballentine et al., 2002]. All samples that are not considered air contaminated on the grounds discussed above have 3He/4He between 6 and 8 RA [Hilton et al., 2006], indicating they have experienced minimal crustal contamination, as would be expected from an arc built on thin, oceanic arc crust.

[23] In summary, the following samples are omitted from further discussion because they are considered to be too severely contaminated by air or crustal volatile additions to yield meaningful information on subarc sources and processes: Sou-1, Owa-1, Noh-1, Oht-1, Kos-1, Uto-1, Uto-2, Ons-1, Ons-2, Mam-1, Sue-1, Kas-1, Geo-2, Aog-3, Aog-3a, Aog-1, Mar-3. Rejecting these 17 samples from further consideration leaves the following 12 only minimally contaminated samples, covering 7 volcanic centers (Table 2): Sou-2 (Hakone), Mam-2 (Niijima), Jin-1 and Jin-2 (Shikinejima), Geo-1, Ogo-1 and Yas-1 (Hachijojima), Aog-2 and Aog-2a (Aogashima), Mar-9 (Pagan), and Mar-7 and Mar-8 (Agrigan).

6. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

6.1. Nitrogen Sources

[24] Before any attempt is made to constrain the output flux of nitrogen from the IBM arc, and to evaluate the mass balance of the system, it is critical to quantitatively resolve the various contributions to the nitrogen output measured at each volcano. Mixing equations developed by Sano et al. [1998, 2001] and applied by Fischer et al. [2002] and Zimmer et al. [2004] to N2/He ratios can be used to calculate the relative contributions from air (A), upper mantle (M) and sediments (S) using the measured δ15N and N2/He values. Table 2 gives the results of these mixing calculations according to:

  • equation image
  • equation image
  • equation image

where fA, fM and fS are the fractions of the measured N2 derived from air, mantle and sediment, respectively, and δ15NA/M/S and (N2/He)A/M/S are the respective values for air, mantle and sediment.

[25] End-member δ15N values are 0‰ [Ozima and Podosek, 2002], −5 ± 2‰ [e.g., Marty, 1995; Marty and Humbert, 1997; Cartigny et al., 1998] and +7 ± 4‰ [e.g., Peters et al., 1978; Sadofsky and Bebout, 2004; Li and Bebout, 2005] for air, upper mantle and sediment, respectively. The end-member N2/He values are 1.49 × 105 [Ozima and Podosek, 2002], 150 [Marty and Zimmerman, 1999] and 1.05 × 104 [Matsuo et al., 1978] for air, upper mantle and sediment, respectively. See Appendix B for a discussion of the end-member values adopted here. These end-member values, as well as other ratios and data used in the subsequent flux calculations, are presented in Table 2.

[26] Isotopic measurements and the results of mixing calculations are also reported in Table 2 and the samples are plotted in Figure 3, along with the end-members and binary mixing trajectories. Sediment-derived nitrogen fractions (S) range from 0.08 to 0.77 of the total nitrogen. These fractions can be corrected for air-derived nitrogen to yield Sc and Mc, the fractions of S and M in a binary mixture, where Sc = S/(S+M) and Mc = 1 − Sc. The corrected N isotope composition is calculated according to:

  • equation image

where δ15NM = −5‰, δ15NS = +7‰ and f is the fraction of mantle-derived nitrogen (Mc) [Fischer et al., 2002]. This corrected value (Table 2) is an attempt to constrain the isotopic composition of each sample prior to any atmospheric contamination. The validity of this correction is dependent upon the inherent assumption that, apart from air, only two components (upper mantle and sediment) are contributing nitrogen to the sample. This assumption will be explored further in sections 6.2.3 and 6.2.5.

image

Figure 3. Plot of δ15N versus N2/He showing the three end-members of Sano et al. [2001]: air (A), upper mantle (M), and subducted sediments (S). Symbols are as for Figure 2. Solid lines are mixing curves between the end-members, and dotted lines are mixing curves between air and mantle-sediment mixtures containing 25%, 50%, 75%, and 95% sediment. Error bars are 1σ errors reported in Table 1.

Download figure to PowerPoint

[27] Sediment-derived nitrogen fractions in a binary mantle-sediment mixture range from 0.12 to 1.0. All volcanoes show a large proportion of sedimentary nitrogen (62%–100%), with the exception of the two samples from Agrigan, which show that only 12% and 14% of the nitrogen in these gases is sedimentary in origin. These isotopic measurements and calculations quantitatively support the conclusions presented above, on the basis of gas chemistry, that there is a strong slab signature in gases from all islands except Agrigan. On average, 34% of the total nitrogen in IBM gases is sediment-derived (n = 12), while considering only the Izu-Bonin segment of the arc increases this fraction to 40% (n = 10). This corresponds to 75% and 85% sediment in an air-free mantle-sediment binary mixture.

[28] In order to investigate the controls on isotopic variability within the arc system, Figure 4 shows N2/He, CO2/N2,exc., δ15N and %S as a function of latitude, or distance along strike of the arc. If external forcing functions such as sediment composition (see section 2), slab dip (which varies from ∼45° beneath Izu-Bonin to almost 90° beneath the Marianas) or slab age (∼130 Ma beneath Izu-Bonin, ∼160 Ma beneath the Marianas) were the primary control on isotopic variation, one would expect to observe clear and systematic differences between the Mariana and Izu-Bonin segments of the arc. As can be seen in Figure 4, this is not the case.

image

Figure 4. Along-arc variation in N2/He, CO2/N2,exc., δ15N, and %S. Each circle is a single gas sample. Black dashed line in %S plot is average IBM value (75%), and gray dashed line is average Nicaragua value (71% [Elkins et al., 2006]). Open circles in δ15N plot are corrected values (δ15Nc); see section 6.1 for explanation. Dashed line in δ15N plot is the average value (+5‰) of the subducting sediments [Sadofsky and Bebout, 2004].

Download figure to PowerPoint

6.2. Nitrogen Flux and Mass Balance

[29] One of the ultimate aims of this work is to calculate a nitrogen flux for the IBM arc system and compare this to estimates of the subduction input flux. Previous work on N2 mass balance at subduction zones (e.g., the Central American arc [Zimmer et al., 2004; Elkins et al., 2006]) have utilized COSPEC SO2 flux measurements combined with regional gas chemistry to estimate a N2 output flux. However, the paucity of SO2 flux data from degassing subaerial volcanoes along the IBM volcanic front necessitates the use of a different approach. Furthermore, the study of Hilton et al. [2007] cautions against the use of data based on short-term observations of actively degassing volcanoes to extrapolate to arc-wide volatile fluxes. In light of this caveat, volatile fluxes were calculated using three different approaches, each of which is described below, with results summarized in Table 3.

Table 3. Summary of Flux Calculationsa
 Izu-BoninMarianaIBM
Outputs
Method 1   
   CO20.740.87 
   N20.440.39 
   N2,sed.0.180.080.25
Method 2   
   3He2.242.98 
   CO23.433.13 
   N22.041.44 
   N2,sed.0.820.291.11
Method 3   
   3He1.341.57 
   CO22.051.65 
   N21.220.76 
   N2,sed.0.490.150.64
Inputs
Sedimentb0.941.252.19
AOCc1.912.554.46
Total2.853.806.65
6.2.1. Nitrogen Output Flux

[30] The first method utilizes estimates of the magma production rate (or crustal growth rate) and primary magmatic volatile contents inferred from melt inclusion data. The maximum magma production rate estimates of Taira et al. [1998] of 80 km3 Myr−1 km−1 and 70 km3 Myr−1 km−1 for the Izu-Bonin and Mariana segments, respectively, have been used in the following calculations to provide an upper limit on the calculated output flux. However, we concede that better constraints on the magma production rate of the IBM arc are crucial for accurate estimates of output fluxes and subsequent mass balance calculations.

[31] The second crucial constraint needed to estimate the N2 flux by the first method is the volatile content of the primary undegassed magmas generated in the mantle wedge. One estimate of primary volatile content comes from melt inclusion data. Shaw et al. [2008] report H2O, CO2, F, S and Cl concentrations from olivine-hosted melt inclusions in recent tephras from the Mariana arc. The highest-concentration sample is used on the assumption that it is the least degassed, and therefore most representative of the primary volatile content of the original melt. However, we cannot rule out the possibility that all melt inclusions have suffered some level of degassing, which would result in a minimum estimate of source CO2 content. Using a CO2 concentration of 1384 ppm [Shaw et al., 2008], the production rates of Taira et al. [1998], a melt density of 2.8 g cm−3, trench lengths of 1050 and 1400 km for the Izu-Bonin (IB) and Mariana (M) segments, respectively [von Huene and Scholl, 1991], and average CO2/N2 ratios of 168 and 217 (Table 2) yields an N2 flux of 0.44 × 108 mol yr−1 for IB and 0.39 × 108 mol yr−1 for M. Using average sediment fraction (S) values from the mixing calculations allows us to calculate sediment-derived N2 fluxes of 0.18 × 108 mol yr−1 for IB and 0.08 × 108 mol yr−1 for M, which give a combined flux for the entire arc of 0.25 × 108 mol yr−1. Alternatively, the CO2/N2 and S values can be replaced with CO2/N2,exc. and %S values, respectively, to arrive at the same output flux.

[32] The second method for estimating N2 fluxes uses 3He fluxes, based on the assumption that all 3He being discharged from the arc originates in the mantle wedge [Hilton et al., 2002; Fischer and Marty, 2005]. The global arc 3He flux of 92.4 mol yr−1 [Hilton et al., 2002] is scaled to IB and M using a global trench length of 43,400 km and the IBM trench lengths used earlier. This 3He flux is converted to CO2 and N2 fluxes using the average CO2/3He ratios of 15.3 × 109 for IB and 10.5 × 109 for M [Hilton et al., 2006] and the CO2/N2 ratios used earlier. These calculations give N2 fluxes of 2.04 × 108 mol yr−1 (IB) and 1.44 × 108 mol yr−1 (M) and subsequently sediment-derived N2 fluxes of 0.82 × 108 mol yr−1 (IB) and 0.29 × 108 mol yr−1 (M). The sum for the entire arc is 1.11 × 108 mol yr−1, a factor of ∼4 greater than the flux calculated from magma production rates. For comparison, using the global arc average CO2/3He of 12 × 109 [Sano and Marty, 1995] rather than the IBM average also yields a total sediment-derived N2 flux for IBM of 1.11 × 108 mol yr−1.

[33] The final method also utilizes magma production rates, but in combination with an estimate of the subarc mantle 3He concentration. Assuming 30% partial melting of the IBM volcanic front source [Peate and Pearce, 1998], and using the subarc mantle 3He concentration estimate of Fischer and Marty [2005] (1.79 × 10−15 mol g−1) gives a 3He concentration in the undegassed mantle melt of 5.7 × 10−15 mol g−1. Combining this concentration with the magma production rates, melt densities and trench lengths used above gives 3He fluxes of 1.34 mol yr−1 and 1.57 mol yr−1 for IB and M. These 3He fluxes were then converted to N2 fluxes as described above. This final method yields N2 fluxes of 1.22 × 108 mol yr−1 (IB) and 0.76 × 108 mol yr−1 (M) and sediment-derived N2 fluxes of 0.49 × 108 mol yr−1 (IB) and 0.15 × 108 mol yr−1 (M). The flux for the entire arc is 0.64 × 108 mol yr−1, approximately half that of the second method and ∼2.5x that of the first method.

[34] The fluxes calculated by the three methods are all significantly different, with the first method yielding particularly low fluxes. There are two reasons why the first method may be underestimating the output flux. First, magma production rates may have been underestimated. Although we use the maximum estimate of Taira et al. [1998], their crustal growth rate is based on a single seismic line through the northern Izu arc and a number of assumptions, lending significant uncertainty to their estimate. However, the fourfold increase in magma production rate required to match the flux of method 2 seems unlikely given the other estimates of magma production rate. A greater magma production rate would also increase the flux calculated by method 3. Second, melt inclusion data may not be representative of initial melt volatile concentrations. Fischer and Marty [2005] note that measured melt inclusion volatile concentrations are lower than concentrations calculated using their estimate of the subarc mantle 3He concentration. Furthermore, “excess volatile” fluxes, where the amount of SO2 measured in eruption columns is approximately an order of magnitude greater than could have been degassed from the erupted magma, have been observed at many arc volcanoes, including Anatahan in the Marianas [de Moor et al., 2005; Pallister et al., 2005]. However, the volatile “excesses” observed at arc volcanoes could be accounted for if a C-O-H-S vapor phase existed in the magma before melt inclusion entrapment. This conclusion is consistent with the results calculated here, and suggests that melts exsolve a fluid phase before crystallization of mineral phases that trap melt inclusions, e.g., olivine [Fischer and Marty, 2005]. CO2 has low solubility in melts relative to other major volatile components, and is therefore one of the first species to exsolve as pressure drops. Consideration of both these points leads us to the conclusion that the output flux estimated by method 1 is probably not representative of the IBM arc system. If the magma production rate used in method 1 is an underestimate then this may partly explain the discrepancy between methods 2 and 3. Alternatively, the difference may be due to other factors such as overestimating the global 3He flux or 3He subarc mantle concentration, or because simply scaling the global 3He flux by trench lengths is inappropriate.

6.2.2. Sedimentary Input and Mass Balance

[35] The flux of nitrogen into the arc system is needed in order to evaluate the N2 mass balance of the IBM arc. A detailed investigation of the sedimentary section recovered outboard of the Izu-Bonin arc at ODP Site 1149 Section A (Figure 1) concluded that subduction of the section there would deliver 2.5 × 106 g yr−1 km−1 N, with average δ15N of +5.0‰ [Sadofsky and Bebout, 2004]. Scaling this value to the whole arc yields an input flux of 2.19 × 108 mol yr−1 N2 (0.94 × 108 mol yr−1 for Izu-Bonin alone).

[36] Comparing input/output fluxes for the entire Izu-Bonin-Mariana margin shows that the input flux is significantly greater than any of the calculated output fluxes (Figure 5a). Estimates of the amount of sedimentary nitrogen recycled to the atmosphere range from 11% (method 1) to 51% (method 2). As there is no evidence of sediment off-scraping at the trench [Bellaiche, 1980; von Huene and Scholl, 1991] the entire sedimentary package is assumed to be subducted to subarc depths [Stern et al., 2003]. Thus, it appears as though a large fraction of the sedimentary nitrogen entering the trench is being delivered past the depths of arc magma genesis into the upper mantle.

image

Figure 5. Schematic illustration showing N2 input and output fluxes for (a) the entire IBM arc and (b) the Izu-Bonin segment alone. Red, blue, and green arrows correspond to fluxes calculated by methods 1, 2, and 3, respectively. Arrow width is proportional to magnitude of flux. See section 6.2.1 for details of different output flux calculations. Values in parentheses are the calculated sediment-derived N2 outputs as a percentage of the total input flux. Sediment input flux is from Sadofsky and Bebout [2004]. AOC input flux is from Li et al. [2007]. Modified from Eiler [2003].

Download figure to PowerPoint

6.2.3. AOC Input and Mass Balance

[37] Previous studies of arc systems (e.g., the Central American arc) have focused solely on the sedimentary input [e.g., Li and Bebout, 2005] (see section 6.3). However, it has been recently recognized that the altered oceanic crust (AOC) is a potentially significant source of nitrogen in subduction zones, and needs to be included in estimates of volatile recycling [e.g., Fischer et al., 2002; Li et al., 2005]. Recent work on the nitrogen concentration and isotopic composition of AOC recovered from ODP Site 801 near the Mariana trench suggests that, despite the far lower concentration (max. 18 ppm compared to >200 ppm in some sediment samples), the large volume and higher density of the AOC, relative to the overlying sediments, results in an estimated input flux from AOC of 5.1 × 106 g yr−1 km−1 N, significantly greater than that from the overlying sediments, and with average δ15N of −2.9‰ for the uppermost 470 m and −5.2‰ for the entire section [Li et al., 2007]. This corresponds to a flux of 4.46 × 108 mol yr−1 N2 for the entire arc (1.91 × 108 mol yr−1 for Izu-Bonin), more than double the sedimentary flux.

[38] Since the AOC has been identified as a major carrier of nitrogen, we can now reevaluate mass balance of the IBM system. Consideration of the new sediment- and AOC-hosted input flux of 6.65 × 108 mol yr−1 shows that the amount of nitrogen recycled to the atmosphere is now estimated to be as little as 4% (method 1) and no more than 17% (method 2) (Figure 5a). If the recycling values estimated above are taken at face value, then it can be concluded that, even if all of the output flux is derived from sedimentary nitrogen, between 49% and 89% of the 15N-enriched sedimentary nitrogen entering the trench is delivered to the deep mantle (where “deep mantle” refers to the mantle beyond the influence of the subduction system, i.e., depths greater than ∼150–200 km, not just lower mantle below 670 km). If, as seems likely, some of the output flux is AOC derived the amount of 15N-enriched sedimentary nitrogen delivered past the subduction zone into the deep mantle will be even greater.

[39] Considering mass balance for the Izu-Bonin arc alone we see trends similar to the whole margin, but importantly, the greater proportion of sediment-derived nitrogen in the output (40% compared to 20%; Table 2) means that the overall flux of recycled nitrogen to the atmosphere is greater. Considering the sedimentary input alone, the uncertainty in the different estimates of output flux becomes a major factor, with different methods suggesting that anywhere between 19% and 87% of the sedimentary nitrogen might be recycled to the atmosphere. When considering the AOC as well, between 6% and 29% of the total input flux may be recycled to the atmosphere (Figure 5b). This greater sediment-derived flux implies that for the Izu-Bonin segment, even more than for the arc as a whole, the nitrogen carried into the mantle is likely to be dominated by 15N-depleted nitrogen from the AOC.

6.2.4. Other Arc Outputs

[40] This simplified scenario excludes consideration of the output at the fore arc, rear-arc cross chains, or back-arc basins, all of which may sample slab-derived material. On the basis of chlorine isotopes, Barnes et al. [2008] suggest that fluid sources vary as a function of depth within the subduction system, and they identify slab signatures in fore-arc seamounts, the volcanic front and the deeply sourced Guguan cross chain. Furthermore, Hochstaedter et al. [2001] identify an AOC-derived fluid component in Western Seamount (Izu-Bonin cross chain) lavas. Studies of glasses from the back-arc rifts in the Izu-Bonin segment of the arc also show evidence of a weak arc component [Fryer et al., 1990; Hochstaedter et al., 1990a, 1990b]. Finally, a number of studies of submarine glasses from the Mariana Trough back-arc basin show the presence of arc signatures [Hawkins et al., 1990; Stern et al., 1990; Gribble et al., 1996, 1998; Kelley et al., 2006]. The nitrogen concentration and isotopic composition of these fluids and magmas is unknown, yet they may have important consequences for the amount and isotopic composition of nitrogen being delivered to the mantle, as discussed below.

6.2.5. Input Isotopic Composition and Residual Slab Nitrogen

[41] The isotopic composition of the subducting sediments is not uniform but varies with depth at ODP Site 1149 from δ15N of +8.2‰ at the top of the core to values of +4.7‰ at 120 m [Sadofsky and Bebout, 2004]. The remainder of the core has values ranging from +2.5‰ to +4.9‰ [Sadofsky and Bebout, 2004]. The total sedimentary section has mean δ15N of +5.0‰ [Sadofsky and Bebout, 2004]. There is also considerable variability in isotopic composition as a function of depth within the AOC, with values ranging from +1.2‰ to −11.6‰ [Li et al., 2007]. The uppermost 470 m of basement sampled at Site 801 has mean δ15N of −2.9‰ [Li et al., 2007]. Assuming the remaining 6.5 km of crustal section has MORB-like δ15N of −5.6‰ gives an average δ15N for the entire AOC of −5.2‰ [Li et al., 2007]. Combining the contribution from sediments [Sadofsky and Bebout, 2004] gives the entire sediment and crustal section average δ15N of −1.8‰ [Li et al., 2007].

[42] The nitrogen isotope composition of the residual slab material is an important component of a number of models that attempt to explain the isotopic imbalance between the Earth's external and internal reservoirs [e.g., Javoy, 1997; Cartigny et al., 1998; Javoy, 1998; Tolstikhin and Marty, 1998; Marty and Dauphas, 2003]. The residual nitrogen isotope composition is also crucial to the interpretation that positive δ15N values observed in plume-related rocks, believed to have been derived from the deep mantle [e.g., Tolstikhin and Marty, 1998; Marty and Dauphas, 2003], reflect sampling of recycled sedimentary material. Significantly, in both these cases, the nitrogen delivered to the mantle in subduction zones is assumed to be 15N enriched. We have already demonstrated that a significant proportion of the sedimentary nitrogen subducted at the IBM trench is recycled beyond the subduction zone into the upper mantle (section 6.2.2). Further sedimentary nitrogen may be lost to the fore arc, rear-arc cross chains or back-arc basins (section 6.2.4), although at present we lack the output flux data necessary to quantify this loss.

[43] To further explore the isotopic composition of the residual slab nitrogen, the δ15N values (Table 2) can also be modeled in terms of just two slab-derived components. Given that nitrogen in most Izu-Bonin samples is <20% mantle-derived according to the mixing equations above, we use δ15Nc values listed in Table 2 as an estimate of the magmatic value (i.e., before air contamination), and model this in terms of just two slab components: AOC-derived nitrogen with average δ15N of −2.9‰ (for the uppermost ∼500 m of crust) and sedimentary nitrogen with δ15N of +7‰. The results of this calculation are shown in Table 2 as %AOC. The calculations show that most samples could be explained by a contribution of ∼20%–40% AOC-derived nitrogen.

[44] These calculations are extended by modeling the δ15Nc values as a three component mixture of mantle, sediments and altered oceanic crust, using an approach analogous to that outlined in section 6.1 (Figure 6). The AOC is modeled with δ15N of −2.9‰ and N2/He of 900 and 4500. The N2/He of the altered oceanic crust is largely unconstrained in the literature so these values are necessarily approximate. Marty and Zimmerman [1999] measured an average N2/He in fresh MORB of 45. Nitrogen is unequivocally enriched in AOC during alteration, with nitrogen concentrations in AOC at Site 1149 of ∼2–5 times that of fresh MORB [Li et al., 2007]. Helium concentrations in MORB and AOC are variable but AOC typically appears to be depleted in 4He by a factor of 10–100 relative to MORB [e.g., Staudacher and Allègre, 1988; Moreira and Sarda, 2000]. The values of 900 and 4500 cover a reasonable range of likely values and, with the exception of Agrigan samples, return very similar results. The results discussed here were obtained using a value of 4500 to maximize the number of samples that fall within the mixing lines. The results of these calculations imply contributions of ∼75%–80% sediment, 15–20% AOC and 5% mantle for the majority of samples (Figure 6 and Table 2). The samples from Agrigan appear to result from mixing of ∼60%–70% mantle with ∼5%–10% sediment and ∼20%–30% AOC. Whether the Agrigan samples really do contain more of an AOC component relative to the rest of the arc, or whether this is just a consequence of the chosen N2/He end-member value, is uncertain. Using an N2/He value of 4500 gives AOC fractions for these samples similar to those for the rest of the arc, while using the lower value of 900 results in an apparently larger AOC contribution to Agrigan. It is worth noting, however, that Agrigan melt inclusions have the highest water contents of any island in the entire Mariana section of the arc [Shaw et al., 2008], consistent with a greater AOC contribution.

image

Figure 6. Mixing plot of δ15Nc versus N2/He showing the upper mantle (M), subducted sediment (S), and altered oceanic crust (AOC) end-members. See section 6.2.5 for discussion of mixing calculations. Solid lines are mixing curves for an AOC end-member N2/He value of 4500, and dashed lines are mixing curves for an AOC end-member N2/He value of 900. Symbols are as for Figure 2. Error bars are 1σ errors reported in Table 1.

Download figure to PowerPoint

[45] The constraints on nitrogen sources and mass balance presented above can be combined to estimate the isotopic composition of the residual nitrogen that is subducted into the mantle. The total N2 flux calculated by each of the three methods described earlier (Table 3) is averaged and converted to a non-air-derived flux using the average M, S and A fractions for the Izu-Bonin and Mariana segments listed in Table 2. The average non-air-derived flux for the two segments is then converted to sediment- and AOC-derived fluxes using the average S* and AOC* values of 74% and 20%, respectively, for IB and 33% and 22%, respectively, for M (Table 2). The Izu-Bonin and Mariana components are then combined to obtain a total sediment-derived flux for the entire arc of 0.51 × 108 mol yr−1 N2 and a total AOC-derived flux for the entire arc of 0.21 × 108 mol yr−1 N2. The total sediment-derived output is then subtracted from the total sediment-hosted input flux of 2.19 × 108 mol yr−1 N2 [Sadofsky and Bebout, 2004] to obtain a residual sediment-hosted flux into the mantle (with δ15N of +7‰) of 1.68 × 108 mol yr−1 N2. The total AOC-derived output is subtracted from the input flux in the upper 470 m of altered crust (with δ15N of −2.9‰) of 0.7 × 108 mol yr−1 N2 [Li et al., 2007] to obtain an AOC-derived flux of 0.49 × 108 mol yr−1 N2. Combining these residual fluxes with the input flux in the remaining 6.5 km of crust (with δ15N of −5.6‰) of 3.76 × 108 mol yr−1 N2 [Li et al., 2007] yields a total flux into the deep mantle of 5.93 × 108 mol yr−1 N2 with average δ15N of −1.9‰, virtually indistinguishable from the bulk subducting slab input value of −1.8‰ [Li et al., 2007]. Further loss of nitrogen through the other arc outputs (fore arc, rear-arc cross chains and back-arc basin), from either the sediments or AOC, would decrease the average δ15N of the residual material subducted into the mantle. However, with increasing metamorphic grade the average δ15N of the residual material is likely to increase [Bebout and Fogel, 1992; Halama et al., 2010].

6.3. Comparison With the Central American Arc

[46] The Central American Volcanic Arc is the counterpart MARGINS Focus Site to IBM and has been the subject of several recent volatile (i.e., N2, He-C) studies [Snyder et al., 2001; Fischer et al., 2002; Shaw et al., 2003; Snyder et al., 2003; Zimmer et al., 2004; Li and Bebout, 2005; Elkins et al., 2006; de Leeuw et al., 2007]. The Central American (CA) margin shows distinct along-arc segmentation, and a characteristic “chevron pattern” in crustal thickness, slab dip and many geochemical indices [Carr, 1984; Carr et al., 1990; Morris et al., 1990; Eiler et al., 2005; Barnes et al., 2007]. These geochemical variations have been interpreted as reflecting a stronger slab signature in the center of the arc under Nicaragua, where slab dip is steepest [Carr et al., 1990]. The weak slab signal in Costa Rican volcanics has been interpreted to be the result of erosion and accretion of the uppermost sediment column at the trench [Leeman et al., 1994].

6.3.1. Central America Volatile Systematics

[47] This tectonic and geochemical segmentation is reflected in volatile studies, which reveal significant differences between the northern/central arc in Guatemala and Nicaragua, and the southern arc in Costa Rica. Volcanic and hydrothermal gases in Nicaragua and Guatemala display a similar chemistry to most IBM gases, having high N2/He, low CO2/N2,exc. and positive δ15N [Fischer et al., 2002; Elkins et al., 2006]. These characteristics indicate a large contribution of sediment-derived nitrogen in Nicaraguan gases, averaging 71% [Elkins et al., 2006]. Calculated outputs range from similar to, to significantly higher than, estimated sedimentary inputs, suggesting that some AOC-derived nitrogen may be required to balance output fluxes, and that nitrogen is efficiently recycled to the atmosphere, with little or no sedimentary nitrogen being carried into the mantle [Elkins et al., 2006] (Figure 7). Despite the similarity in volatile chemistry, this conclusion regarding nitrogen recycling is entirely opposite to that proposed here for the IBM margin, where mass balance considerations suggest large-scale transport of nitrogen into the deep mantle. This observation suggests that major differences exist between the two arcs, either in the relative amount of the sedimentary inputs or in the way in which nitrogen is processed within the subduction zone (see section 6.3.2).

image

Figure 7. Schematic illustration showing N2 input and output fluxes for the Nicaraguan segment of the Central American arc. Arrow width is proportional to magnitude of flux. Output fluxes are from Elkins et al. [2006]. Input fluxes are from Li and Bebout [2005]. Inputs in red are IBM AOC inputs scaled to the Nicaraguan margin (see section 6.3.1 for details). Modified from Eiler [2003].

Download figure to PowerPoint

[48] Importantly, the AOC has not been considered as a significant carrier of nitrogen in the CA margin. Li and Bebout's [2005] study of the sedimentary input to the margin acknowledged that the AOC could contain some nitrogen, but without AOC samples from the margin they estimated only 5 ppm N in the uppermost 2 km of crust and 0.5 ppm N in the remaining 3.5 km of crust. In comparison, N concentrations in the upper 470 m of AOC drilled at Site 801 near the Mariana arc range from 1.3 to 18.1 ppm [Li et al., 2007]. Thus, CA nitrogen inputs are dominated by the sedimentary units (∼80% of subducted nitrogen), while in IBM the sediments are estimated to carry only ∼33% of the total subducted nitrogen [Li et al., 2007].

[49] Given uncertainties in AOC N contents, we apply the IBM AOC nitrogen input flux to Nicaragua and add it to the sedimentary input to investigate whether assuming generic AOC nitrogen concentrations has any effect on the recycling calculations. These revised fluxes are shown in Figure 7 (in red). Interestingly, the effect on the AOC flux is relatively small, and the sedimentary flux still dominates in Nicaragua (∼67% of subducted nitrogen). This is due to the high nitrogen concentrations (up to 2382 ppm N) [Li and Bebout, 2005] in the sediments subducting at the CA margin.

[50] Volatile chemistry of Costa Rica is strikingly different from that of Nicaragua and Guatemala. Costa Rican gases have low N2/He, high CO2/N2,exc. and negative δ15N, reflecting a low sediment-derived nitrogen contribution, averaging only 37% [Zimmer et al., 2004]. The output/input ratio of nitrogen is less than unity, suggesting that more nitrogen is subducted than is released through the arc [Zimmer et al., 2004]. This could indicate that a large amount of nitrogen is transported into the deep mantle, possibly due to limited fluid availability under the arc. However, the lack of a strong slab signature is normally interpreted as reflecting a lack of sedimentary material reaching the subarc due to sediment offscraping or fore-arc devolatilization [Zimmer et al., 2004].

[51] Thus, in contrast to IBM, nitrogen systematics in CA appear to be dominantly controlled by along-strike variation in external forcing functions. However, the major difference in nitrogen systematics between the two margins is in the recycling efficiency of nitrogen. In CA, little or no sedimentary nitrogen is carried into the deep mantle, despite the high nitrogen concentration in the sediments, and the margin acts as a “subduction barrier” for nitrogen [Fischer et al., 2002]. In contrast, the most extreme output/input ratios in IBM suggest that as much as 89% of sedimentary nitrogen, and by inference the majority of AOC nitrogen, may be delivered past the zone of arc magma generation and into the mantle.

6.3.2. Nitrogen Release Beneath the Arc

[52] Despite the similarity in gas chemistry between IBM and Nicaragua samples, the nitrogen output fluxes are strikingly different. The annual sediment-derived flux to the atmosphere for Nicaragua is 14.8 × 105 mol yr−1 km−1 N2, while for IBM it is 0.26 × 105 mol yr−1 km−1 N2. Thus, despite the output flux of nitrogen being sediment dominated in both margins (71% and 75% in CA and IBM, respectively) the trench length normalized output flux from Nicaragua is almost 60 times greater than that of IBM. Comparing Nicaragua to the flux solely from the Izu-Bonin segment of the arc, which has a stronger sediment signature (0.47 × 105 mol yr−1 km−1 N2), results in a factor of ∼30 difference between Nicaragua and Izu-Bonin. These comparisons suggest a significant difference in the efficiency of nitrogen release and/or transport from sediments beneath these two arc segments.

[53] The observations above could be explained if the CA margin simply subducts 30–60 times more nitrogen than the IBM margin. However, published sediment subduction fluxes [Sadofsky and Bebout, 2004; Li and Bebout, 2005] show that the CA margin only subducts ∼5x more nitrogen (11.8 × 106 versus 2.5 × 106 g yr−1 km−1), or an order of magnitude less than the ∼60x discrepancy in output flux. Two further, potentially linked explanations exist for the discrepancy. First, nitrogen may be carried in different mineral phases, which breakdown differently, in the sediments at each margin. Second, the different thermal regimes of the two margins may facilitate the breakdown of mineral phases (whether they are the same or not) and release of nitrogen more readily at the CA margin than IBM.

[54] Nitrogen in subducted sediments occurs mainly as ammonium cations (NH4+), produced during diagenesis of organic matter in oceanic sediments [Busigny et al., 2003]. During subduction metamorphism, NH4+ substitutes for K+ and is incorporated into micas and feldspars [Bebout and Fogel, 1992; Bebout et al., 2007]. There is no evidence to suggest that nitrogen in sediments at either margin is not carried as ammonium cations [Sadofsky and Bebout, 2004; Li and Bebout, 2005]. Therefore, it appears as though an explanation involving different mineral phase hosts for nitrogen is an unlikely possibility for the different fluxes. This leaves the differing thermal regimes, as illustrated by two-dimensional, finite element thermal models of the two margins, as the most plausible explanation for the different behavior of nitrogen. The rapid subduction of old, cool oceanic lithosphere at the IBM margin results in a relatively “cool” subduction environment and lower slab-mantle interface model temperatures, ∼540°C for Izu-Bonin at depths of ∼100 km [Peacock, 2003]. In contrast, the CA margin subducts young, warm lithosphere, which results in a warmer subduction environment and higher slab-mantle interface model temperatures, ∼620–800°C for Central America at equivalent depths [Peacock et al., 2005].

[55] Studies of metasedimentary palaeoaccretionary suites, such as the Catalina Schist, suggest that in “cool” subduction zones, fluid mobile elements may be far more efficiently retained in sediments, at least to subarc depths if not beyond, while in warmer subduction zones, these elements are more likely to be lost beneath the arc or fore arc [Bebout and Fogel, 1992; Bebout, 1996; Bebout et al., 1999, 2007] (Appendix B). Therefore, we suggest that the greater sediment-derived nitrogen flux at the CA margin relative to IBM is a result of the more efficient release of nitrogen from sediments, as a consequence of the warmer thermal regime in the Central America subduction zone. In light of recent experimental studies [Spandler et al., 2007], we acknowledge, however, that the exact mechanism(s) by which such fluid mobile elements are stripped from the slab is still uncertain.

[56] A recent N and He-C study of the Sangihe arc, Indonesia [Jaffe et al., 2004; Clor et al., 2005], the only other such study of nitrogen systematics, demonstrates clear along-arc variations in slab volatile contributions to arc volcanoes. Although tectonic complexity prevents the identification of specific causes for these differences, along-arc variation in thermal regime, specifically slab temperature, is suggested as one possible explanation [Jaffe et al., 2004; Clor et al., 2005].

[57] Thus, the most significant finding of our study is that the thermal regime is likely the dominant factor controlling the recycling efficiency of nitrogen (and volatiles in general) at a given arc; this has important implications for the global nitrogen budget. The thermal structure of IBM is more typical of the majority of western Pacific subduction zones than CA [Peacock, 2003], suggesting that these subduction zones could act as conduits for large-scale recycling of nitrogen into the mantle, rather than back to the atmosphere, with all of the associated implications for terrestrial nitrogen cycling and ocean island basalt sources discussed above. The data required to test and/or reinforce this conclusion, such as the flux and nitrogen concentration and isotopic composition of fluids discharged in the fore arc and back arc of each margin, are sorely lacking and should be considered a priority for future studies.

6.4. Long-Term Recycling Implications

[58] The recycling estimates and fluxes calculated here can be extrapolated globally to address issues of long-term volatile exchange between the atmosphere and mantle. Marty and Dauphas [2002] used the correlation between mean degassing duration (MDD) and the degree of recycling at arcs (output:input ratio) to calculate mantle-atmosphere exchange efficiency for a number of volatile species. Nitrogen has a MDD of 2.74 × 1011 years, significantly longer than the age of the Earth, implying either a decreasing MORB degassing rate with time, or a contribution to the surface inventory that is not mantle derived, or both [Marty and Dauphas, 2002].

[59] Given the total annual subduction input flux at the IBM margin (6.65 × 108 mol yr−1 N2) and a median recycling efficiency of 10% (Figure 5 and section 6.2.5), results in the IBM subduction system returning on the order of 5.9 × 108 mol yr−1 N2 to the deep mantle. At this rate, it would require 4.57 × 1011 years to subduct the total surface N2 inventory. Extrapolating this flux of N2 into the deep mantle at IBM to all arcs globally results in 1.06 × 1010 mol N2 being returned to the deep mantle annually. At this rate the total surface inventory would be subducted in 2.58 × 1010 years. However, at arcs such as Central America, little or no nitrogen is being recycled to the deep mantle [e.g., Fischer et al., 2002; Zimmer et al., 2004; Elkins et al., 2006] and thus the time for the total surface inventory to be subducted into the deep mantle could be significantly longer. Further estimates of nitrogen recycling at other arcs are required to better constrain long-term recycling budgets.

[60] Thus, the MDD for nitrogen is of the same order of magnitude as the time required to subduct all surface nitrogen, or, expressed differently, the annual MORB degassing rate and the annual subduction flux back into the mantle are approximately equal. Therefore, if the majority of nitrogen being returned to the deep mantle in subduction zones is carried in the AOC, and has δ15N ∼ −5‰, then the global nitrogen cycle is in steady state, and there is no net change in the size or isotopic signature of the internal or external reservoirs. This is consistent with the observation that MORB and diamonds both sample a common reservoir that currently has, and has had for 2–3 Ga, δ15N ∼ −5‰. This observation is also consistent with the model of Marty and Dauphas [2002], suggesting that the Earth had an initially chondritic nitrogen abundance and isotopic signature (∼103 ppm and δ15N ∼ 0 ± 50‰), and that nitrogen was efficiently exchanged between atmosphere and mantle prior to ca. 3 Ga and that the MORB flux has decreased since. However, if all the nitrogen currently being subducted into the deep mantle has negative δ15N, as for IBM, then the source of heavy nitrogen in plume-related rocks must lie elsewhere. Alternatively, the δ15N of subducted nitrogen may be variable at the present time, or may have varied in the past. Once again, constraints on nitrogen recycling at other arcs are required to test this.

7. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[61] Volcanic and hydrothermal gases from eight centers along the Izu-Bonin-Mariana arc are typical of subduction zone volcanoes, and are dominated by H2O, CO2 and S species. The majority of samples have high N2/He and low CO2/N2,exc., reflecting addition of sedimentary nitrogen from the subducting slab. The only exception is Agrigan in the Mariana arc, with mantle-like characteristics of low N2/He and high CO2/N2,exc..

[62] Nitrogen isotope values are generally positive, up to +5.5‰, reflecting a large contribution of sedimentary nitrogen. Agrigan, with negative values more indicative of an upper mantle source for nitrogen, is again the exception. Mixing calculations indicate that an average of 34% of nitrogen in the samples is sediment derived, corresponding to 75% in a sediment-mantle binary mixture after correction for atmospheric contamination. Further mixing calculations incorporate the altered oceanic crust (AOC) as an end-member for the first time, permitting the contribution from the slab to be separated into sedimentary and AOC contributions. These calculations suggest ∼75%–80% of the nitrogen is supplied by the subducting sediments, ∼15%–20% is supplied from the altered oceanic crust, and the mantle contributes only ∼5%.

[63] Nitrogen fluxes from the arc calculated by three different methods are 0.25 × 108 mol yr−1 N2, 0.64 × 108 mol yr−1 N2 and 1.11 × 108 mol yr−1 N2, representing 4%, 10% and 17% of the total input flux or 11%, 29% and 51% of the sedimentary input flux. These calculations suggest that a large amount of nitrogen is transported into the deep mantle (where “deep mantle” refers to the mantle beyond the influence of the subduction system, i.e., depths greater than ∼150–200 km, not just lower mantle below 670 km). The majority of this nitrogen is likely 15N-depleted material broadly similar in isotopic composition to the present-day upper mantle. The estimated isotopic composition of the residual slab nitrogen subducted into the mantle is approximately −1.9‰. Predicted nitrogen loss from other arc outputs (i.e., the fore arc) would likely further decrease this value.

[64] The conclusions regarding nitrogen recycling from our IBM study are very different from those concerning the Central American arc. Nitrogen fluxes in CA are far higher, with AOC-derived nitrogen being required to balance the output flux, and it is suggested here that the thermal regime of the subduction zone is the primary control on the recycling efficiency of nitrogen, and likely other volatiles, at convergent margins.

[65] Global extrapolation of our deep mantle subduction flux suggests that the annual MORB degassing rate and the annual subduction flux to the deep mantle are approximately equal, and thus the global nitrogen cycle is in steady state. If true, and the nitrogen returned to the mantle has negative δ15N, then the source of heavy nitrogen in plume-related rocks is not from recycled sediments. Alternatively the δ15N of subducted sediments may have varied through time. Constraints on nitrogen recycling at other arcs are required to test these conclusions.

Appendix A:: Sampling and Analytical Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[66] Fumarolic gases were collected by inserting a titanium tube into the fumarole and connecting the titanium tube to the Giggenbach flask containing 5N NaOH with silicon tubing. Bubbling spring samples were collected by placing an inverted plastic funnel over the bubbling area, and connecting the funnel to the flask with silicon tubing. Geothermal wells were sampled by either inserting a titanium tube into the well pipe and connecting the tube to the flask with silicon tubing, or by using a steam separator. At all sites the sampling equipment was flushed with sample gas for at least 5–10 min before collecting the sample.

[67] Headspace gases were analyzed by gas chromatography using a Gow-Mac Series 600 gas chromatograph (GC). The GC is equipped with a Hayes Sep precolumn, two 5Å molecular sieve columns and two detectors arranged in series. Samples are initially measured by a nondestructive thermal conductivity detector (TCD) for analysis of Ar, N2, He, H2 and O2 and then subsequently on a flame ionization detector (FID) for analysis of CO and CH4. Carbon dioxide, chloride and sulfur oxidation state were determined in the caustic solutions by titration, total sulphur gravimetrically and ammonia and fluoride with anion specific electrodes. Water contents were calculated as the difference between sample bottle presampling and postsampling weights, minus the weights of all other species.

[68] Chloride, fluoride and sulfate concentrations were also determined by ion chromatography following sample pretreatment with Dionex OnGuard® II H cartridges for removal of hydroxyl groups. The caustic solution was first diluted to a concentration of 0.38N NaOH or less to avoid oversaturating the cartridges. Cartridges were flushed with 10 mL of DI water followed by 5 mL of sample at a flow rate of ∼4 mL/min. The first 2 mL was discarded to avoid accidental dilution by the DI water and the remaining 3 mL of sample was loaded onto the IC for analysis.

[69] Nitrogen isotope analyses were conducted on aliquots of the headspace gas expanded into glass breakseals. Helium carrier gas was used at a flow rate of 0.9 mL/min. The breakseal headspace was evacuated and filled with helium before the sample was introduced into the mass spectrometer. Sample gas was passed through a NAFION™ water trap, into a Hayes Sep D 5Å molecular sieve column via a 50 μL sample loop on a six-port valve, and through a second NAFION™ water trap before being introduced into the mass spectrometer. Reported values are the mean of between 3 and 7 measurements and errors are the 1σ standard deviation of those 3 to 7 measurements. Air was analyzed as a standard at the beginning and end of, and at intervals during, each analytical run.

Appendix B:: Modeling End-Members

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[70] Sadofsky and Bebout [2004] note that the δ15N values for sediments subducting at the IBM margin are highly variable, and dependent on lithology. Measured values range from −0.2 to +8.2‰, although only two samples have values less than +2.5‰ [Sadofsky and Bebout, 2004]. The average value for the whole section at Site 1149 is +5.0‰ [Sadofsky and Bebout, 2004]. Although this value is lower than the end-member value used here, Sadofsky and Bebout [2004] concede that use of a value of +7‰ is not unreasonable due to the uncertain effects of metamorphism during subduction.

[71] Studies of low grade metasedimentary palaeoaccretionary suites, including the Catalina Schist and the Franciscan Complex, California and the Western Baja Terrane, Mexico, offer an insight into the effects of metamorphism and devolatilization beneath the fore arc on N2 concentrations and isotopic compositions in a “cool” subduction zone [e.g., Bebout and Fogel, 1992; Bebout et al., 1999; Busigny et al., 2003; Sadofsky and Bebout, 2003; Bebout et al., 2007]. Calculations by Bebout and Fogel [1992] and Bebout et al. [1999] based on studies of the Catalina Schist (metamorphosed at 350–750°C and pressures corresponding to 15–45 km depths) suggest a decrease in N2 concentration and an increase in δ15N of at least 3%–4‰ with increasing metamorphic grade (up to epidote-amphibolite and amphibolite facies). In contrast, studies of the Franciscan Complex and Western Baja Terrane (peak temperatures of 250–300°C and ∼40 km depth) suggest little or no N2 loss or isotopic modification [Sadofsky and Bebout, 2003]. Studies of the Schistes Lustrés/Cignana nappe in the western Alps, subducted along a “cold” geothermal gradient (∼8°C/km) to ∼90 km, also suggests no loss of N2 (or other fluid mobile elements) or isotopic modification to depths approaching those of subarc magma genesis [Busigny et al., 2003; Bebout et al., 2008]. Until data are available on the N2 concentrations and isotopic compositions of fore-arc fluids, these studies and theoretical calculations represent the state of current knowledge of the fate of subducted sediments during subduction. Due to lack of evidence to the contrary, and taking into account that δ15N values higher than +5‰ were measured during the course of this work (Table 1), the use of +7‰ as an end-member is adopted here.

[72] In addition to studies cited above that demonstrate that the effects of prograde metamorphism on the nitrogen isotopic composition of metasedimentary rocks are uncertain, recent work shows that other considerations, involving metabasaltic rocks, must be included [Halama et al., 2010]. HP/UHP mafic metamorphic rocks studies by Halama et al. [2010] demonstrate gradually decreasing N concentrations, and gradually increasing N isotopic compositions, with increasing fluid-induced eclogitization, suggesting that residual AOC transported past the depths of arc magma genesis into the deep mantle could evolve to heavier, more sediment-like, δ15N values.

[73] In addition to uncertainty over the sediment end-member value, the nitrogen isotope composition of the upper mantle has also recently been questioned, with Mohapatra and Murty [2004] suggesting that the true value may be closer to −15‰, rather than the commonly used value of −5‰. However, the validity of this argument has been challenged on the basis that the technique of Mohapatra and Murty [2004] used a molybdenum furnace during nitrogen extraction experiments, which may cause a large isotopic fractionation (on the order of 10‰ at typical experimental conditions) [Yokochi and Marty, 2006].

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

[74] Field work in the Izu islands would not have been possible without the logistical assistance of many Japanese colleagues, including N. Morikawa and M. Takahashi from the Geological Survey of Japan, A. Shimizu and Y. Sano from the University of Tokyo, and T. Ohba from the Tokyo Institute of Technology. The guidance of Mehdi Ali and Viorel Atudorei with the IC and stable isotope analyses, respectively, is gratefully acknowledged. Comments from G. Bebout, two anonymous reviewers, and Associate Editor J. Ryan on an earlier draft helped improve the manuscript. This research was supported by NSF grants OCE 0305218 and 0642832 (MARGINS, T.P.F.) and OCE 0305248 (MARGINS, D.R.H.).

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Background
  5. 3. Samples and Methods
  6. 4. Results
  7. 5. Data Integrity
  8. 6. Discussion
  9. 7. Conclusions
  10. Appendix A:: Sampling and Analytical Methods
  11. Appendix B:: Modeling End-Members
  12. Acknowledgments
  13. References
  14. Supporting Information
FilenameFormatSizeDescription
ggge1644-sup-0001tab01.txtplain text document6KTab-delimited Table 1.
ggge1644-sup-0002tab02.txtplain text document2KTab-delimited Table 2.
ggge1644-sup-0003tab03.txtplain text document1KTab-delimited Table 3.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.