SEARCH

SEARCH BY CITATION

Keywords:

  • oceanic crust;
  • composition;
  • alteration;
  • seismic structure;
  • gabbro;
  • seismic velocity

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[1] The objective of this study was to determine whether the effects of low Mg content and/or cracks that typically populate olivine grains can explain why P and S wave velocities in samples of oceanic gabbro do not increase with increasing olivine content. The Mg content of olivine in these rocks is known to be appreciably less than Fo90, and optical, SEM, and electron microprobe analyses indicate that the “olivine” is actually an aggregate of olivine and 5–15% crack fill consisting of talc and/or serpentine plus magnetite. Voigt average models show that there is a strong, positive correlation between P wave velocities and Fo90 content but no correlation with Fo73. Because of its Mg content, olivine that is typical of oceanic gabbros does not influence seismic velocities. Adding 11% alteration of olivine to serpentine plus magnetite in the form of crack fill to the Fo73 model yields a nearly exact match between the model and measured P and S wave velocities. The fact that the crack fill consists largely of talc and/or serpentine indicates that these minerals formed by hydrothermal alteration in young, hot (>150°C) lower crust. Hence our results apply to gabbros in situ, as well as to the laboratory samples; seismic velocities in lower oceanic crust produced at high and intermediate spreading rates are controlled principally by the degree of hydrothermal alteration of the gabbros that compose the lower crust.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[2] By the mid 1970s, the ophiolite or “Penrose” model, consisting of extrusive basalt overlying dikes overlying gabbro, had become the accepted model for the lithologic structure of the oceanic crust [e.g., Coleman, 1977], and laboratory studies demonstrated that the seismic properties of samples from ophiolites are consistent with the seismic velocity structure of the crust [Christensen and Salisbury, 1975; Spudich and Orcutt, 1980]. Subsequently, submersible surveys (at Hess Deep and the Blanco Fracture Zone, for example) and scientific ocean drilling at Sites 504, 894, and 1256 [Becker et al., 1992; Gillis et al., 1993; Karson et al., 2002; Wilson et al., 2006] have confirmed that the architecture of crust produced at intermediate and fast rates of seafloor spreading in the Pacific is consistent with the ophiolite model. The lower oceanic crust, which is 4 to 5 km thick, and has seismic P wave velocities in the range 6.7 to about 7 km/s, is almost certainly composed of gabbro.

[3] Two compositional factors are known or thought to affect the seismic properties of oceanic gabbros. One is the alteration of primary phases (plagioclase and pyroxene ± olivine) to phyllosilicates and amphibole by hydrothermal fluids [Salisbury and Christensen, 1978; Christensen and Smewing, 1981; Detrick et al., 1994; Salisbury et al., 1996], and the second is olivine content, which has been cited to explain the commonly observed gradual increase of velocity with depth in the lower crust [Salisbury and Christensen, 1978; Christensen and Smewing, 1981; Collins et al., 1989].

[4] From their study of a suite of gabbros recovered from ODP Sites 735, 894 and 923 [Robinson et al., 1989; Gillis et al., 1993; Iturrino et al., 1991, 1996; Miller and Christensen, 1997], Carlson and Miller [2004] demonstrated a systematic decrease of seismic velocities with increasing degree of alteration, and found that the modal mineralogy of gabbro samples with P wave velocities that are typical of the lower crust (6.7 to 7 km/s) includes 5–15% hydrous mineral phases: phyllosilicates and amphiboles. They did not, however, observe the expected increase of seismic velocity with increasing olivine content (Figure 1). They found instead that the effective elastic moduli of olivine in these samples proved to be anomalously low (see Figure 2).

image

Figure 1. P wave velocities measured at 200 MPa in gabbro samples from ODP Holes 735B, 894G, and 923A versus reported modal olivine contents.

Download figure to PowerPoint

image

Figure 2. Shear versus bulk modulus for mineral phases that are common in oceanic gabbros. Solid triangles with error bars are estimated moduli of minerals in the gabbro samples [Carlson and Miller, 2004]. Open triangle indicates the values of olivine moduli estimated by linear regression using the Voigt model. Also shown are the model values for serpentine plus 5% magnetite, Fo73, and Fo73 with 11% alteration to serpentine and magnetite.

Download figure to PowerPoint

[5] These observations are significant for the following reason. If olivine concentration does not affect the seismic properties of the lower oceanic crust, it follows that in situ P wave velocities in the lower crust are a measure of the degree of hydrothermal alteration and hence of bound water content. However, before this interpretation can be applied with confidence, it is essential to establish how olivine affects the seismic properties of oceanic gabbros, and why. The purpose of this study is determine why olivine does not affect the seismic velocities in samples of oceanic gabbro measured in the laboratory, and whether this condition applies in situ. We find that the reported abundance of olivine in the samples includes 5–15% alteration of olivine to serpentine and/or talc plus a small amount of magnetite. The combined effects of olivine composition (Fo73, previously noted by Carlson and Miller [2004]) and 11% alteration of olivine to serpentine plus magnetite are sufficient to explain the measured P and S wave velocities.

2. Hypotheses

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[6] Carlson and Miller [2004] suggested the elastic moduli of the olivine in the samples might be lower than expected because of their Mg content and/or owing to the effects of microcracks. The density and elastic moduli of olivine vary significantly with Mg content (Figure 2). Consequently, seismic P and S wave velocities decrease with decreasing Mg [Carlson and Miller, 2004, Figure 8]. Olivine grains in the samples from Site 923, for example, have an average composition of Fo73 ± 5, while the olivines from Site 894 (Hess Deep) have Mg contents ranging from Fo63 to Fo71 with a mean of 65.5 [Pederson et al., 1996], and whole-rock values from Site 1256 have a mean Mg content near Fo60 [Wilson et al., 2006]. Cracks are also known to have a profound effect on the elastic properties of rocks [e.g., Kuster and Toksöz, 1974; Carlson and Gangi, 1985; Wilkens et al., 1991]. The network of microcracks that is characteristic of olivine but not other minerals in oceanic gabbros (see Figure 3) might therefore contribute to the low effective elastic moduli of olivine in the laboratory samples.

image

Figure 3. Photomicrograph and backscattered electron (BSE) image chosen to illustrate the important characteristics of olivine in gabbro samples. (top) Photomicrograph showing a portion of thin section 923A-R5-1-107-6. The field of view includes two highly fractured olivine grains surrounded by uncracked plagioclase grains. Alteration products are amphibole and mixtures of talc and/or serpentine. Cracks filled with talc/serpentine occupy 5–15% of the olivine grains. (bottom) Backscattered electron (BSE) image of sample 923A-4R-1, showing a fractured olivine grain (light gray on left) and unfractured plagioclase grain (dark gray, right). The field of view is 600 μm. The highly reflective crack fill (white) is magnetite; crack fill (gray) consists of serpentine > talc > tremolite. Talc and tremolite often occur in the centers of cracks, with serpentine along the margins. Black areas represent voids, possibly caused by plucking during sample preparation.

Download figure to PowerPoint

[7] A critical question is how the combined effects of Mg content and cracks affect the properties of oceanic gabbros in situ. Cracks would affect the seismic properties of the lower oceanic crust only if they form and remain open, at least to some degree, in situ. We reason that if the cracks that populate the samples formed in the lower crust, and not during recovery of the core or preparation of the samples or thin sections, we expect them to be lined or filled with hydrous alteration products that form at temperatures appropriate to the lower oceanic crust, specifically talc and/or serpentine. To test this hypothesis, we used optical microscopy, SEM and electron microprobe analyses of several gabbro samples from the suite of samples included in the Carlson and Miller [2004] study.

3. Alteration Products

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[8] Carlson and Miller [2004] determined modal mineralogy in these gabbros through optical microscopy with point counting, averaging 1500 points per thin section. However, we found it challenging to resolve the composition of crack fillings or even to determine if many cracks are filled or empty through optical microscopic methods alone. In addition, underestimation of low-concentration minerals is common to many methods of determining modal mineralogy [Harvey et al., 1998; Maloy and Treiman, 2007].

[9] Therefore, we used BSE imaging with energy dispersive analysis of X rays (EDS) (Zeiss FE-SEM with EDAX at Texas A&M University) to determine the nature of the crack fill. Imaging revealed that alteration products, including serpentine and/or talc, magnetite and trace amounts of tremolite, line the cracks (Figure 3). Typically, talc and tremolite occur in the centers of cracks, with serpentine along the margins. Alteration products are also found along olivine grain boundaries. Amphibole is also observed locally along some olivine grain boundaries. Estimation of percent alteration, based on visual examination of olivine grains during BSE imaging, suggests that alteration of olivine grains within and between samples ranges from 5 to 15%. The observation that the cracks are filled with serpentine and/or talc plus tremolite and magnetite indicates that the cracks formed in situ at comparatively high temperatures, i.e., in newly formed lower crust.

[10] In an effort to refine our estimate of the degree of alteration of the olivine grains, we used a combination of techniques, including BSE imaging and high-resolution X-ray mapping (1024 × 1024 over a 2 mm2 area; one sample location every 0.2 μm), and spot EDS (Cameca electron microprobe at Texas A&M University). X-ray maps (Mg, Fe, Ca, Si) and BSE images were prepared for 3 olivine grains each, in 3 samples recovered by drilling from the MARK area of the Mid-Atlantic Ridge [Miller and Christensen, 1997]: 923A-7R-1, 81 cm, 9R-1, 130 cm and 12R-1, 120 cm. Following identification of phases from maps, and confirmed using spot EDS, image analyses on BSE images were used to estimate the relative proportions of olivine and alteration products.

[11] This method for determining the proportion of alteration minerals is subject to some uncertainties:

[12] 1. While most of the alteration is to talc and serpentine, we cannot determine the relative proportions of these two phases from the available data.

[13] 2. Because of alteration, grain boundaries can be difficult to define. This ambiguity leads to some uncertainty in the relative proportions of olivine to alteration products.

[14] 3. There are voids in some images evidenced by black areas that amount to 0–1% of the image area. While we suspect that these voids were caused by plucking during sample preparation, we cannot say with certainty that they were not present in situ.

[15] The abundances of the alteration products are summarized in Table 1. On average, the alteration assemblage within olivine grains consists of 13% talc/serpentine, 1% tremolite and 1% magnetite. Significantly, these results suggest that the abundance of talc/serpentine in the samples was underestimated in the reported modal analyses used by Carlson and Miller [2004]. In fact, the reported modal analyses include only small amounts of talc and essentially no serpentine. For example, Miller and Christensen [1997] report the composition of sample 12R-1, 120 cm as 70% plagioclase, 2% pyroxene, 26% olivine and 2% alteration products, magnetite, chlorite and talc. Our data (Table 1) suggest that 13% of the original olivine in these rocks is typically altered to talc and/or serpentine that fills the cracks. The true olivine content would then be 23%, and crack fill would thus contribute about 4% talc/serpentine to the modal analysis of the sample from core 12R-1. We infer that the material that fills the cracks in the olivine grains was not counted in the modal analyses; what was reported as olivine is actually an aggregate of olivine plus talc/serpentine crack fill.

Table 1. Modal Mineralogy of Olivine Crack Fillings and Grain Boundaries
SampleTotal Alteration (%)Magnetite (%)Tremolite (%)Talc/Serpentine (%)
923A-7R-1, 81 cm152113
121111
131112
923A-9R-1, 130 cm101110
150013
9118
923A-12R-1, 120 cm202218
221120
10009
Mean141113
SD5  4

[16] The observation that the cracks in the olivine grains are largely filled by talc and serpentine, with minor tremolite and magnetite suggests that the cracks formed in situ at temperatures within the tremolite and serpentine stability field, probably not less than about 400°C. Some of the cracks in the BSE image are open, but we cannot say whether they were open in situ. In fact, the absence of low temperature alteration (clay minerals, for example) suggests that the cracks filled in situ, and were opened during sample recovery and/or sample preparation. Voids in these images represent about 1% of the olivine area; this small volume of very thin cracks is more than sufficient to reduce the effective elastic moduli of olivine in these samples to the observed values [e.g., Kuster and Toksöz, 1974; Cheng and Toksöz, 1979; Gangi and Carlson, 1996] in the unlikely event that the cracks are open in situ. If, as seems more likely, the cracks are not open in situ, the talc/serpentine crack fill will also affect the elastic moduli of the aggregate of olivine and crack fill, but to a much smaller degree.

[17] What remains to be determined is whether the degree of alteration that is required to explain the observed physical properties of the gabbro samples is consistent with the observed degree of alteration. We used a modeling procedure based on the Voigt average to estimate the effects of Mg content and alteration to serpentine plus magnetite on the properties of olivine in these samples, and thereby tested the hypothesis that these effects can account for the measured physical properties of the gabbro samples.

4. Modeling Procedure

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[18] The elastic properties of rocks are commonly estimated from the proportions of the mineral constituents and their known elastic properties using Voigt-Reuss-Hill (VRH) averaging scheme [Hill, 1952]. The inversion from which Carlson and Miller [2004] estimated the elastic moduli of pyroxene, plagioclase, amphibole and olivine in the gabbros was based on the VRH model. Our strategy here is to develop models to account for the low effective moduli of olivine by first subtracting the effective properties of olivine obtained from the inversion, then substituting the properties of Fo90, Fo73, or Fo73 plus serpentine and magnetite.

[19] Unfortunately, the VRH model cannot be used in this way because it is nonlinear. However, Carlson and Miller [2004] found that a linear inversion based on the Voigt volume average [Voigt, 1910] yields the essentially same values for the elastic moduli of plagioclase, pyroxene, amphibole, and olivine as the inversion based on the VRH model. These results are summarized in Table 2, and the estimated bulk and shear moduli of olivine are shown in Figure 2.

Table 2. Summary of Estimated Elastic Moduli
Mineralk (GPa)μ (GPa)
VoigtVRHVoigtVRH
Olivine104 ± 13108 ± 1648 ± 1053 ± 10
Pyroxene111 ± 6126 ± 854 ± 560 ± 5
Plagioclase73 ± 471 ± 338 ± 337 ± 2
Amphibole100 ± 787 ± 7 51 ± 644 ± 5

[20] Following Carlson and Miller [2004], we used the linear, Voigt volume average scheme to model the properties of the gabbro samples using different values for the moduli of olivine. The bulk or shear modulus of the rock, Mr was first computed from the measured P and S wave velocities measured at 200 MPa and the measured bulk densities. We then estimated the properties of the rock with different properties assigned to the olivine component. The modified average is

  • equation image

where αol is reported volume fraction of olivine, Mol is the modulus (or density) of olivine obtained from the inversion, and Mol is modulus or density that we substitute for the measured value.

[21] The substitution is made only for the properties of olivine. An important feature of this approach is that we retain the measured properties of the sample in Mr, which includes the effects of microcracks and small amounts of various mineral phases (e.g., oxides, sulfides, and alteration) that affect the properties of the rock, but occur in such low abundance that they cannot be accurately measured.

[22] Given the modified bulk and shear moduli, and bulk density of the sample, we computed the model density and shear and compressional wave velocities. We considered three cases: (1) olivine content as reported with Fo90 composition, (2) olivine content as reported with Fo73 composition, and (3) olivine consisting of an aggregate Fo73 plus crack fill.

[23] The elastic moduli we used for the models are summarized in Table 3 and shown in Figure 2. The properties of Fo90 and Fo73 are VRH average values from Carlson and Miller [2004]. An equilibrium reaction of olivine with water in the presence of sufficient oxygen produces about 6 moles of serpentine and 2 moles of magnetite (approximately 10:1 serpentine to magnetite by volume). We therefore modeled the crack fill as a mixture of 95% serpentine and 5% magnetite, and the model properties of the crack fill are taken to be the VRH average of serpentinite values from Carlson [2001] and magnetite reported by Aleksandrov and Ryzhova [1961].

Table 3. Model Elastic Moduli
Materialk (GPa)μ (GPa)ρ (kg m–3)Reference
  • a

    95% serpentinite and 5% magnetite.

Fo90128783340Carlson and Miller [2004]
Fo73129723540Carlson and Miller [2004]
Serpentinite44142480Carlson [2001]
Magnetite161.591.45200Aleksandrov and Ryzhova [1961]
Crack filla47.816.22620VRH average

[24] Modeling the properties of the gabbro samples containing Fo90 and Fo73 is a straightforward substitution using equation (1), but in a model that includes the crack fill material, the degree of alteration is a variable that can be chosen to match the model to the measured properties of the rock

  • equation image

where βcf is the volume fraction of crack fill (i.e., the degree of alteration of olivine to serpentine plus magnetite), Mcf is the modulus or density of the crack fill material, and Mol is the modulus or density of Fo73.

[25] We did not model the degree of alteration by obtaining a formal “best fit.” Instead, we adjusted the degree of alteration by trial and error to match the model P wave velocities to the measured P wave velocities [Carlson and Miller, 2004] of several samples chosen at random. In each case we found good agreement between the model and measured P wave velocities for 11% alteration. Accordingly, we then used 11% alteration to model the P and S wave velocities and the bulk density of the entire the data set.

[26] The models are illustrated in Figures 46. Model P and S wave velocities are compared with the measured velocities in Figure 4, model densities are compared with measured densities in Figure 5, and Figure 6 illustrates the variation of P wave velocity with reported olivine content.

image

Figure 4. Model velocity versus measured velocity. The Fo90 model yields much higher velocities than those observed, Fo73 velocities are closer to the measured velocities but systematically higher, and a nearly exact match to both P and S wave velocities is obtained for 11% alteration of olivine to serpentine and magnetite. Measured velocities are those reported by Carlson and Miller [2004].

Download figure to PowerPoint

image

Figure 5. Model versus measured wet bulk densities. There is good agreement between the measured Fo73 and Fo73 + alteration models, though the models overestimate the sample densities by as much as 0.1 Mg m−3. On average the difference increases with increasing density from about 0.03 to about 0.05 Mg m−3. Dashed lines are best fits.

Download figure to PowerPoint

image

Figure 6. Measured and model P wave velocity versus reported olivine content. Dashed lines represent best fits. The Fo90 model shows a strong correlation of P wave velocity with olivine content, but no correlation is observed for olivine with a crustal Mg content or if the reported olivine is composed of Fo73 plus alteration.

Download figure to PowerPoint

[27] The Fo90 model leads to P and S wave velocities that are both highly variable and systematically higher than the measured velocities (Figures 4a and 4b). In accordance with conventional wisdom, the model P wave velocities increase markedly with olivine content, as shown in Figure 6. If the olivine in the gabbros had Mg content typical of mantle olivine, we would see the expected very strong dependence of P wave velocity on olivine content in oceanic gabbros.

[28] Fo73 model shows significantly better agreement with the measured velocities (Figures 4a and 4b), though the model velocities are systematically somewhat higher than the measured values, as are the model bulk densities (Figure 5). Significantly, the Fo73 model P wave velocities show no variation with olivine content (Figure 6); the lower Mg content of the olivine grains in the gabbros is sufficient to explain the fact that higher olivine concentrations do not cause higher seismic velocities in these oceanic gabbros, but does not account for the lower measured velocities.

[29] The model in which the reported olivine consists of Fo73 with 11% alteration to talc/serpentine produces a nearly exact match to the measured P wave velocities (Figure 4a). There is some mismatch to the S wave velocities (Figure 4b), particularly for samples with measured velocities less than about 3.8 km/s. Model densities average 0.03 to 0.05 Mg m−3 higher than the measured densities. Thus a model in which the reported olivine is composed of Fo73 plus 11% alteration of olivine to serpentine plus a small amount of magnetite accounts well for the measured properties of the laboratory samples. The agreement between the model values and the measured values arises from two conditions. One is that the misfit results from the fact that the elastic moduli of Fo90 are systematically too high; the other is that there is that there is apparently little variation of the effective moduli of the aggregate grains. Consequently, the misfit is proportional to the apparent olivine content. For that reason, we are able to match the P wave velocities simply by using lower bulk and shear moduli.

[30] As a last step in our analysis, we used a stochastic procedure to estimate the elastic moduli of the altered olivine for comparison with the effective elastic moduli of olivine in the samples obtained by inversion using the VRH and Voigt models [Carlson and Miller, 2004]. We assumed that the degree of alteration has a lognormal distribution with a mean of 11%. The mean of the lognormal distribution depends on both the mean and standard deviation of the underlying normal distribution. For that reason, we adjusted the mean and standard deviation of the underlying normal distribution to obtain a match between the measured and observed P wave velocities; a nearly exact match was achieved with mean of 11% alteration and a standard deviation of 7%. The lognormal distribution is shown in Figure 7. The bulk and shear moduli of the altered olivine (i.e., an aggregate of Fo73 and 11 ± 7% alteration to serpentine plus magnetite) are 117 ± 8 GPa and 60 ± 6 GPa, respectively. These values are in good statistical agreement with the effective moduli of olivine in the samples (see Table 2 and Figure 2) reported by Carlson and Miller [2004].

image

Figure 7. Lognormal distribution of the degree of alteration of Fo73 to talc/serpentine plus magnetite with a mean of 11% and standard deviation of 7% that yields the best match to the measured P wave velocities in gabbro samples.

Download figure to PowerPoint

5. Discussion and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[31] The objectives of this study were to determine whether the effects of low Mg content and/or cracks that are characteristic of olivine grains can explain why P and S wave velocities do not exhibit the widely expected increase with increasing olivine content in oceanic gabbro samples. The Mg content of olivine in these rocks averages less than Fo75 [Pederson et al., 1996; Carlson and Miller, 2004; Wilson et al., 2006], and optical, SEM and electron microprobe analyses indicate that the “olivine” is actually an aggregate of olivine and 13 ± 4% crack fill consisting of 13% talc and/or serpentine plus 1% magnetite and 1% tremolite.

[32] Voigt average models based on the measured properties samples and modal analyses of gabbro samples demonstrate that there is a strong, positive correlation between P wave velocities and olivine content if the olivine is Fo90, but there is no correlation if the olivine is modeled as Fo73. Because of their low Mg content, olivine grains that are typical of the lower crust does not influence its seismic velocity. Adding 11% alteration of olivine to serpentine plus magnetite in the form of crack fill to the Fo73 model yields a nearly exact match between the model and measured P and S wave velocities. The fact that the crack fill consists largely of talc and/or serpentine indicates that these minerals formed by hydrothermal alteration in young lower crust; hence our results apply to oceanic gabbros in situ, as well as to the laboratory samples. We are thus able to conclude that seismic velocities in lower oceanic crust produced at high and intermediate spreading rates are controlled principally by the degree of hydrothermal alteration. While a detailed analysis or interpretation of the seismic structure of the crust in terms of its state of alteration is beyond the scope of this paper, we note our results serve to strengthen the inference of previous studies [Salisbury and Christensen, 1978; Spudich and Orcutt, 1980; Carlson and Miller, 2004] that a gabbroic lower oceanic crust must be altered throughout, and suggests that it is possible to interpret seismic structure models in terms of the degree of alteration.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

[33] Ray Guillemette contributed significantly to the microprobe analyses. Helpful reviews were provided by Lisa Gilbert, Roy Wilkens, and Damon Teagle. This work was funded by NSF grant OCE-0221250 and by the United States Science Support Program.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Hypotheses
  5. 3. Alteration Products
  6. 4. Modeling Procedure
  7. 5. Discussion and Conclusions
  8. Acknowledgments
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
ggge1453-sup-0001-t01.txtplain text document0KTab-delimited Table 1.
ggge1453-sup-0002-t02.txtplain text document0KTab-delimited Table 2.
ggge1453-sup-0003-t03.txtplain text document0KTab-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.