Volume 7, Issue 8
Free Access

Variability in serpentinite mudflow mechanisms and sources: ODP drilling results on Mariana forearc seamounts

P. Fryer

P. Fryer

SOEST, University of Hawaii at Manoa, 1680 East-West Road,, Honolulu, Hawaii, 96822 USA

Search for more papers by this author
J. Gharib

J. Gharib

SOEST, University of Hawaii at Manoa, 1680 East-West Road,, Honolulu, Hawaii, 96822 USA

Search for more papers by this author
K. Ross

K. Ross

Department of Earth and Planetary Science, University of California Berkeley, 307 McCone Hall,, Berkeley, California, 94720 USA

Search for more papers by this author
I. Savov

I. Savov

Department of Terrestrial Magnetism, Carnegie Institution of Washington, 5241 Broad Branch Road, NW,, Washington, D. C., 20015-1305 USA

Search for more papers by this author
M. J. Mottl

M. J. Mottl

SOEST, University of Hawaii at Manoa, 1680 East-West Road,, Honolulu, Hawaii, 96822 USA

Search for more papers by this author
First published: 25 August 2006
Citations: 46

Abstract

Samples of metamorphic rock were collected from drilled holes on ODP Leg 195 and in piston/gravity cores collected with the Jason 2 remotely operated vehicle from S. Chamorro Seamount, a serpentinite mud volcano on the Mariana forearc. The recovered muds are approximately 90% serpentinite, including grit- to boulder-sized clasts of serpentinized peridotites, but also contain a wide variety of small fragments of metabasic rocks. These metabasic fragments include high-pressure, low-temperature rocks derived from the subduction zone. Other serpentinite seamounts also have yielded metabasic rock fragments as small clasts in the serpentinite mudflows, but none have as wide a variety of rock types as S. Chamorro Seamount. The sources of the rock clasts, both serpentinized peridotites and metabasic schists, vary with the eruptive episodes of the mud volcanoes. Swath mapping of S. Chamorro Seamount shows that a sector collapse of its southeastern flank has resulted in debris flows from the summit region of the seamount that have traveled more than 70 km eastward toward the trench. These debris flows, however, have a very different morphology from mudflows observed at the summit. High-resolution seafloor mapping of the summit shows both thin, presumably highly fluid- (or gas-) charged serpentinite mudflows and a relatively viscous protrusion that has formed the main summit knoll. By comparison, the summit of Conical Seamount drilled on ODP Leg 125 lacks a distinct summit knoll and has numerous, relatively thin and widespread mudflows covering the flanks of the edifice. The style of eruption at a given seamount probably varies with time and with the amount of fluid or gas incorporated in a given pulse of mud. The greater diversity of metabasic rocks at S. Chamorro Seamount may be a consequence of recycling of forearc materials through tectonic erosion and subduction in the southern part of the forearc.

1. Introduction

The Mariana convergent plate margin (Figure 1) is the “type” nonaccretionary subduction zone [Uyeda and Kanemori, 1979; Uyeda, 1982]. Thus the overriding plate has been in contact with the downgoing slab since the inception of subduction in Eocene time. Mud volcanism on the outer 100 km of the forearc has formed large edifices (up to 50 km in diameter and 2 km high) and provides a means by which we may study the processes of fluid-rock interactions in materials derived from shallow to intermediate depths (10–30 km) in the suprasubduction-zone environment [Mottl, 1992; Fryer et al., 1999; Mottl et al., 2003, 2004]. The muds that formed the forearc volcanoes most likely resulted from interaction of these deep-derived fluids with fault gouge and/or mylonized peridotite produced by movement along deep-seated faults in the lithosphere of the overriding plate [Fryer, 1992; Fryer and Fryer, 1987; Fryer et al., 1990a, 2000; Wessel et al., 1994].

Details are in the caption following the image
Shaded (illumination angle from NW) bathymetry map of the southern Mariana forearc contoured at 200 m intervals. The named seamounts are those from which samples have been collected. Only Conical Seamount and S. Chamorro Seamount have been drilled [after Fryer et al., 1999]).

The mudflow matrix is dominantly finely comminuted serpentinized peridotite of the forearc mantle (90–100%) with a small amount (<10%) of mafic components [Fryer et al., 1999; Fryer and Mottl, 1992; Lagabrielle et al., 1992; Savov et al., 2005a]. Rock clasts carried in the matrix can vary widely in composition and abundance relative to the matrix. They are dominantly serpentinized ultramafic rocks (harzburgite, minor dunite and lherzolite) from the supra-subduction zone mantle [Saboda et al., 1992; Parkinson and Pearce, 1998; Savov et al., 2005a], but some flow units contain a small percentage (<12%) of altered or metamorphosed mafic rocks (basalts and diabases), cherts, and possibly metasediments from the subducted plate [Fryer, 1992; Fryer et al., 1990b; Fryer and Mottl, 1992; Lagabrielle et al., 1992; Gharib et al., 2002; Johnson and Fryer, 1990; Johnson, 1992; Maekawa et al., 1993, 1995]. The mafic clasts recovered from Conical Seamount (ODP125) show a diverse provenance including supra-subduction zone (boninitic and arc tholeiitic) lavas, normal mid-ocean ridge basalts (MORB), ocean island basalts (OIB), alkalic basalts and transitional MORB [Johnson and Fryer, 1990; Johnson, 1992]. Analyses of those clasts provided a constraint on the temperature and pressure conditions of their metamorphism [Maekawa et al., 1993, 1995]. In this paper we combine our new data from shore-based analyses of metabasite schists recovered at S. Chamorro Seamount by drilling (ODP Leg 195), piston/gravity coring and from samples collected with the Jason 2 remotely operated vehicle (ROV) with data from preliminary onboard studies during Leg 195 to examine the range of mineral assemblages and its implications for the source of these materials.

We also present new bathymetry and high-resolution side-scan imagery data that bear on the nature of eruptive processes occurring at these serpentinite mud volcanoes. The detailed information from high-resolution side-scan sonar imagery provides a much better constraint on the nature of eruptive episodes at these seamounts and permits us to better relate flow morphology and changes in both fluid composition and mineralogy of deep-derived materials included in the mud matrix. Published data from pore fluids from the two drilled seamounts, and from several others we have sampled by piston and gravity coring, show a trend in composition across the forearc, and hence with depth to the subducting slab, that can be attributed to various dehydration and either decarbonation or carbonate dissolution reactions that occur in the slab as it descends [Fryer et al., 1999; Mottl et al., 2003, 2004]. The chlorinity of the pore fluids does not vary systematically with distance from the trench, however. Chlorinity values of the pore fluids from the seamounts [Mottl et al., 2004] may give us a clue as to the processes by which the mud volcanoes operate and suggest that the nature of eruptions may vary through time. The recent (2003) detailed seafloor mapping surveys of the summits of the active seamounts of the system present information critical to interpretation of the drilling results and help us to understand better the nature of serpentinite mud volcanism that forms these seamounts.

2. Forearc Deformation

Widespread deformation and faulting of the Mariana forearc controls the distribution of the serpentinite mud volcanoes. Regional seafloor mapping [Fryer et al., 1999, 2000; Fryer and Fryer, 1987; Wessel et al., 1994; Hussong and Fryer, 1981; Smoot, 1983; Fryer and Smoot, 1985; Stern and Smoot, 1998; Stern et al., 2003] and multichannel seismic surveys [Mrozowski et al., 1981; Mrozowski and Hayes, 1983; Oakley et al., 2002] demonstrate the interrelationship between faulting and mud volcanism in the outer half of the forearc. The Mariana forearc is in extension across strike as a consequence of plate roll-back [Moberly, 1971; Hsui and Youngquist, 1985; Fryer and Fryer, 1987] and along strike as the Mariana arc has increased in curvature with time [Fryer and Fryer, 1987; Wessel et al., 1994]. Active serpentinite mud volcanism is associated with this faulting in the Mariana region, but inactive serpentinite seamounts also occur on the northern Izu-Bonin section of the system [Fryer and Fryer, 1987]. The latter are confined to a continuous ridge of serpentinite within about 50 km of the trench [e.g., Taylor, 1992]. By contrast, the serpentinite seamounts of the Mariana forearc occur over a wider swath of the forearc and many have developed where fault trends intersect [Fryer et al., 1999, 2000]. Most of the complex faulting within the Mariana forearc is confined to the outer 100 km of the overriding plate. Fryer and Fryer [1987] suggested that the higher degree of deformation of the outer half of the forearc might be, in part, a consequence of a change from brittle deformation within the thin, cooler outer forearc toe region to ductile deformation of the inner half of the forearc wedge overriding the subducting plate. The fact that the pore fluids from all of the Mariana forearc seamounts sampled to date indicate a deep slab source implies that the fault-controlled pathways are connected to the decollement region [Fryer et al., 1999; Mottl et al., 2003, 2004] (Figure 2). The ductile deformation in serpentinite clasts [Lagabrielle et al., 1992], incipient blueschist minerals in metabasites [Maekawa et al., 1992, 1993, 1995], and abundant high-quality radiogenic and stable isotope data indicating strictly slab-related isotope fractionations [Savov, 2004; Benton et al., 2004; Savov et al., 2005b] also support a decollement-region origin for some of the muds. We observed the formation of chimney structures at active springs along fault scarps at the summits of some of the seamounts (Figure 3). Drilling on DSDP Leg 60 in the Mariana forearc showed faulting on all scales in the outer forearc Sites 458 and 459 [Hussong and Uyeda, 1981]. Multichannel seismic data [Mrozowski et al., 1981; Mrozowski and Hayes, 1983; Oakley et al., 2002] show a generally greater degree of faulting within the deeper sections of the forearc on approach to the trench.

Details are in the caption following the image
Schematic cross section of the Mariana forearc showing the relative positions of serpentinite seamounts with respect to hypothesized deep faults within the forearc wedge [after Fryer et al., 1999]).
Details are in the caption following the image
Jason2 photo of carbonate chimneys forming on Quaker Seamount.

There is a dramatic change in the width of the forearc from about 14°N southward, the latitude of S. Chamorro Seamount. North of that latitude, the forearc is almost uniformly 200 km wide. Between 14°N and 13°N the forearc southeast of Guam narrows to only about 130 km wide (see Figure 1). There are E-W fault trends just south of Guam, and N-S fault trends immediately west of Guam [34]. It has been suggested that these contribute to tectonic erosion of the forearc and the decrease in width [Fryer et al., 2003; Gvirtzman and Stern, 2004]. Dredges on the forearc fault scarps reveal mid-crustal level silicic plutonics [Fryer, 1998] as well as peridotites. Exposure of these silicic plutonic rocks suggests tectonic erosion. The possible fate of tectonically eroded materials is not known, but may be germane to understanding the diversity of the metabasic materials from S. Chamorro Seamount.

3. Deep-Derived Metabasic Clasts

The diversity of provenance of metabasic clasts from drill samples on Conical Seamount and the wide range of lithologies of metabasic schists recovered from S. Chamorro Seamount preclude their being derived solely from the overriding plate. The trace element signatures of the grit-sized schists recovered in the muds from ODP drilling and Piston/gravity coring on S. Chamorro Seamount indicate that they have MORB-like rather than sedimentary protoliths. They show Si/Mg ratios roughly between 2 and 9 (i.e., not from peridotites and unlikely from sediments), have B/Be ratios that average ∼360, have very low Th/Nd (ave. = 0.05) and very high Nb/Th (ave. = 11.2) [Savov et al., 2005a]. Sediments input to the Mariana subduction zone have Th/Nd = 0.12 and Nb/Th = 4 [Plank and Langmuir, 1998]. The cherts recovered from dredges in serpentinite mudflows in the vicinity of Conical Seamount contained Cretaceous fossils indicating a Pacific plate origin, and metabasic clast recovered from ODP Leg 125 drilling also likely derived from the subducted Pacific plate [Johnson and Fryer, 1990; Johnson et al., 1991; Johnson, 1992]. The high-pressure phases identified in some of the metabasic clasts indicate conditions deeper than that of extruded boninitic and arc tholeiitic lavas of the island arc [Maekawa et al., 1992, 1993]. The MORB and OIB samples must have derived from a subducted or trapped oceanic plate that had been buried to depths of between 15 to 30 km. Recovery of incipient blueschist facies materials [Maekawa et al., 1992, 1993; Fryer et al., 2000; Gharib et al., 2000] indicate high-pressure and low-temperature conditions typical of intermediate depth in the underlying subduction zone.

The mafic material contained in the serpentinite mud matrix of the mudflows is usually finely comminuted, but some are large enough to provide textural information. Most of the schist fragments from ODP Site 1200 and from our gravity/piston and Jason 2 ROV push cores are grit-sized (Figure 4). Shipboard descriptions include tremolite/chlorite schist (Sample 195-1200F-1H-4, 34–36 cm) and amphibole schists containing zoned crystals with sodic rims and calcic cores (Sample 195-1200F-1H-4, 34–36 cm) [Shipboard Scientific Party, 2002]. Sieved grit-sized lithics examined during ODP Leg 195 [Shipboard Scientific Party, 2002] are generally amphibole schists with chlorite and white micas and are uniformly very friable. Metamorphic assemblages in the metabasic clasts examined on shore consist of actinolite with sodic ampbibole rims (Figure 5a) or coexisting separate grains of calcic and sodic amphibole + chlorite + sphene, ± phengitic mica ± epdidote [Gharib et al., 2000]. Sodium-rich amphiboles range from magnesioriebeckite to barroisite/winchite in composition (Table 1). Metabasic clasts include types that show schistose textures (Figure 5b and, more obviously, Figure 5c) and others that crystallized with low directed stresses and have essentially random grain orientations (Figure 5d). The diversity of rock types from S. Chamorro Seamount recovered on Leg 195 and from our gravity/piston cores and ROV push cores is far greater than that recovered from Conical Seamount on Leg 125 [Fryer et al., 1990b] as can be seen by the range of amphibole compositions in samples analyzed thus far from each (Figure 6).

Details are in the caption following the image
Photomicrograph of one fragment of metabasic rock from S. Chamorro Seamount.
Details are in the caption following the image
Images are element maps that have been falsely colored relative to elemental concentration as indicated for each: (a) ODP Leg 195, Site 1200F, 004H, 01WR, 130.0–140.0, S. Chamorro Seamount. An amphibole-chlorite schist; red is aluminum content, green is calcium content, and blue is sodium content. The amphiboles show growth in two phases, initially a more calcic tremolite phase in light green followed by a more sodic winchite phase in blue. The red mineral is chlorite, and the green mineral is sphene. (b) ODP Leg 195, Site 1200D, 007H, 01WR, 120.0–130.0, S. Chamorro Seamount. This schist is around 2 cm in diameter and consists of amphiboles, phengitic muscovite, chlorite, and sphene. Red is potassium content in this image and illustrates the phengite. Green is calcium content, and blue is sodium content; they illustrate the amphiboles. They range from barroisite in green to magnesioriebeckite in blue. (c) Gravity Core #52, S. Chamorro Seamount. In this image, red is aluminum content, green is calcium content, and blue is sodium content. The amphiboles in this schist are predominantly sodic (magnesioriebeckite) in composition and are blue, although some regions near their cores are more calcic and show up a little greener. Epidote is in yellow, chlorite is in red, and sphene is in green. (d) Gravity Core #52, S. Chamorro Seamount. In this image, red is aluminum content, green is calcium content, and blue is sodium content. The amphiboles are barroisites and magnesioriebeckites, illustrated by the blue-green to blue colors. The brighter and the darker red minerals are higher and lower aluminum chlorites, respectively. The red-orange minerals surrounded by the darker red chlorite are garnets. The bright green mineral is apatite. The black minerals are titano-magnetites.
Details are in the caption following the image
Amphibole compositional plots used to classify amphiboles after Leake et al. [1997]. The structural formulae of the amphiboles were calculated using the 13ECNK method, wherein 13 cations are calculated per 23 oxygens, excluding Ca, Na, and K [after Robinson et al., 1982]. Plotted against each other are the magnesium over magnesium plus ferrous iron ratio versus the silicon cations per 23 oxygens in the structural formula. Amphiboles tend to be ferrous iron-poor and silica-rich. In some cases the amphiboles had a relatively more calcic core, surrounded by a more sodic rim. The amphiboles analyzed from S. Chamorro tend to be more sodic than those recovered and analyzed by Maekawa et al. [1992]. These values are for individual analyses and not averages from discrete samples; therefore some of these values include multiple points analyzed from the same amphibole crystal.
Table 1. Representative Microprobe Analyses of Amphiboles in Small Metabasic Clasts, S. Chamorro Seamounta
Location Mineral GC52 -1 mgrb GC52 -1 mgrb GC52 -1 w GC52 -1 mghb/br 1200D mgrb 1200D br 1200D mghb 1200E mgrb 1200F w 1200F tr 1200F mghb 1200F br 1200F mghb/br 1200F mgrb
SiO2 55.17 55.6 54.11 47.44 52.9 44.82 52.09 55.49 55.86 55.01 51.2 55.88 50.74 55.97
TiO2 0.08 0.26 0.09 0.14 0.09 0.42 0.15 0.04 0.04 0.04 0.15 0.02 0.12 0.11
Al2O3 2.63 3.44 2.69 9.14 2.03 10.9 4.76 2.48 2.44 2.69 6.05 1.19 6.4 2.09
Cr2O3 0.06 0.1 0.02 0 - - - 0.01 0.19 0.17 0.37 0.07 0.26 0.26
Fe2O3(c) 13.32 11.66 6.63 12.11 12.34 9.12 7.48 14.06 10.47 5.81 8.47 6.75 7.46 13.54
FeO(c) 7.21 5.79 5.17 0 6.61 2.38 0 2.42 2.9 1.33 0 1.8 0.92 2.95
MnO 0.32 0.33 0.29 1.58 0.56 0.84 0.65 0.64 0.45 0.64 0.7 0.53 0.82 0.28
MgO 10.43 11.77 16.63 15.11 11.36 13.92 18.93 13.77 15.35 19.51 18.23 19.1 17.82 13.42
CaO 0.96 1.55 9.55 10.08 2.86 9.21 10.72 3.07 4.75 10.8 10.79 9.5 10.83 2.1
Na2O 6.77 6.56 2.06 1.84 5.64 2.92 1.32 5.43 4.81 1.45 1.25 2.12 1.82 5.9
K2O 0 0.03 0.07 0.16 0.03 0.37 0.11 0.05 0.06 0.06 0.07 0.05 0.07 0.02
H2O(c) 2.08 2.11 2.12 2.1 2.02 2.04 2.12 2.12 2.13 2.16 2.14 2.14 2.13 2.11
Sum Ox% 99.03 99.19 99.43 99.7 96.43 96.92 98.34 99.59 99.46 99.67 99.42 99.15 99.39 98.76
Si 7.948 7.91 7.67 6.758 7.853 6.6 7.356 7.83 7.853 7.652 7.184 7.824 7.154 7.942
Ti 0.009 0.027 0.009 0.015 0.01 0.046 0.016 0.004 0.004 0.004 0.016 0.002 0.013 0.012
Al/Al IV 0.052 0.09 0.33 1.242 0.147 1.4 0.644 0.17 0.147 0.348 0.816 0.176 0.846 0.058
Al VI 0.395 0.488 0.12 0.293 0.209 0.491 0.148 0.243 0.257 0.093 0.185 0.021 0.217 0.291
Cr 0.007 0.012 0.003 0.001 - - - 0.001 0.021 0.019 0.041 0.008 0.029 0.029
Fe3+ 1.444 1.248 0.708 1.298 1.378 1.01 0.795 1.493 1.107 0.609 0.894 0.711 0.791 1.446
Fe2+ 0.868 0.689 0.613 0 0.82 0.293 0 0.285 0.341 0.155 0 0.211 0.108 0.351
Mn2+ 0.038 0.04 0.034 0.19 0.071 0.105 0.078 0.076 0.053 0.075 0.084 0.063 0.098 0.033
Mg 2.239 2.496 3.513 3.208 2.513 3.055 3.985 2.897 3.217 4.045 3.813 3.985 3.744 2.838
Ca 0.149 0.236 1.451 1.538 0.455 1.452 1.622 0.465 0.716 1.61 1.622 1.425 1.636 0.32
Na 1.891 1.811 0.566 0.507 1.624 0.833 0.362 1.487 1.312 0.39 0.34 0.575 0.497 1.624
K 0.001 0.005 0.013 0.029 0.005 0.07 0.02 0.009 0.011 0.011 0.012 0.008 0.013 0.004
OH 2 2 2 2 2 2 2 2 2 2 2 2 2 2
Sum Cat# 17.04 17.05 17.03 17.08 17.09 17.36 17.03 16.96 17.04 17.01 17.01 17.01 17.15 16.95
XMg 0.72 0.78 0.85 1.00 0.75 0.91 1.00 0.91 0.90 0.96 1.00 0.95 0.97 0.89
Si 7.95 7.91 7.67 6.76 7.85 6.60 7.36 7.83 7.85 7.65 7.18 7.82 7.15 7.94
Mg/(Mg+Fe2+) 0.72 0.78 0.85 1.00 0.75 0.91 1.00 0.91 0.90 0.96 1.00 0.95 0.97 0.89
Ca 0.15 0.24 1.45 1.54 0.46 1.45 1.62 0.47 0.72 1.61 1.62 1.43 1.64 0.32
Na 1.89 1.81 0.57 0.51 1.62 0.83 0.36 1.49 1.31 0.39 0.34 0.58 0.50 1.62
Bsite 2.04 2.05 2.02 2.05 2.08 2.29 1.98 1.95 2.03 2.00 1.96 2.00 2.13 1.94
Fe3+ 1.44 1.25 0.71 1.30 1.38 1.01 0.80 1.49 1.11 0.61 0.89 0.71 0.79 1.45
Fe2+ 0.87 0.69 0.61 0.00 0.82 0.29 0.00 0.29 0.34 0.16 0.00 0.21 0.11 0.35
Total Fe 2.31 1.94 1.32 1.30 2.20 1.30 0.80 1.78 1.45 0.76 0.89 0.92 0.90 1.80
Al VI - Fe3b −1.05 −0.76 −0.59 −1.01 −1.17 −0.52 −0.65 −1.25 −0.85 −0.52 −0.71 −0.69 −0.57 −1.16
  • a GC52-1, sample taken from 35–39 cmbsf in gravity core 52 (R/V Thompson, 1997); 1200D, ODP195-1200D; 1200E, ODP195-1200E; 1200F, ODP195-1200F; mgrb, magnesoriebechite; w, winchite; mghb/br, magnesiohornblende/barosite; br, barosite; mghb, magnesiohornblende; tr, tremolite.
  • b If Al VI > Fe3, then Glaucophane.

4. Bathymetric and Side-Scan Sonar Survey Data

Detailed mapping of the summit of S. Chamorro Seamount in 2003 using the DSL-120 high-resolution, deep-towed side-scan sonar system (operated by the National Deep Submergence Facility at Wood Hole Oceanographic Institution) shows that the knoll at the summit of S. Chamorro Seamount is about 2500 m in diameter and about 240 m high (Figure 7). The knoll itself has a small (∼220-m-diameter) central, oval-shaped summit region that is flat to slightly depressed inside the rim of the oval (Figure 8). The active springs where communities of mussels, snails and tubeworms live [Fryer and Mottl, 1997], supported by a microbial community within the upper 20 or so meters of serpentinite muds [Mottl et al., 2003], lie along a set of fractures that mark the northwestern edge of the flat summit region. The drill holes 1200 A, B, E and F lie close to this set of fractures (Figure 8). Hole 1200C (a cased and CORKed borehole observatory site) is located 125 m north and 25 m west of 1200A and 1200D is located 85 m north and 25 west of 1200A. Holes 1200C and D are located in the summit source region of a relatively thin (maximum thickness ∼20 m thick), but well-defined small flow feature on the north flank of the summit knoll (Figure 7). To the west of the summit depression there is a series of smaller flow lobes that drain westward down-slope (Figure 7). These flow lobes do not perturb the local contours of the summit knoll and thus cannot be more than a few meters thick. Previous Shinkai 6500 dives [Fryer, 1996a] and Jason 2 lowerings in the vicinity of these holes in March–April of 2003 show the surfaces to be devoid of sediment. This indicates that the flows were emplaced recently. Cores recovered from Hole 1200E in the vicinity of the source of these flows show variations in color and texture of the serpentinite matrix muds that change over a matter of tens of centimeters or less (Figure 9). These suggest separate thin flow units. Formation of such thin flows is consistent with the interpretation that occasional small-volume, relatively fluid flows overtop the summit depression on this seamount.

Details are in the caption following the image
(top) DSL120 side-scan image of the summit region of S. Chamorro Seamount and (bottom) imagery superimposed on shaded, color bathymetry. Contour interval is 20 m. Dashed red lines outline the more fluid mudflow lobes emanating from the summit region.
Details are in the caption following the image
Close-up DSL120 side-scan image of the relative positions of Site 1200 holes superimposed on bathymetry data contoured at 5 m intervals. Red dashed lines indicate the position of lineaments tracing the pattern of fractures that have formed at the NW edge of the summit knoll. Seeps are located near these lineaments, and DSDP Holes 1200A, 1200B, 1200E, and 1200F were drilled near these seeps.
Details are in the caption following the image
Serpentinite muds recovered from ODP Hole 1200E. Note color variations indicating separate mudflow units and the thinness of the units (top of the core is to lower right and the bottom of the core (CC, core catcher) is to upper left).

The imagery from SeaMARC II side-scan surveys of Conical Seamount [Fryer, 1992; Fryer et al., 1990a, 1990b] show high-backscatter, sinuous and narrow, flow features on all flanks (Figure 10). The debris flows from sector collapse of the southeastern flank of S. Chamorro Seamount that were described by Fryer et al. [2000] are, by contrast, neither sinuous nor narrow, but form a broad band that extends down the inner trench slope for more than 70 km. Where observed by Shinkai 6500 dives these S. Chamorro debris flows lack an enclosing matrix for the clasts, and are more akin to debris or rock avalanches [Fryer, 1996a]. The flows on the eastern flank of Conical Seamount travel for more than 18 km from the summit of the seamount toward the deep graben at the base of the eastern flank of the seamount. Alvin dives on these flows in 1987 showed them to be devoid of sediment cover and to expose a rough surface of hummocky yellow-green serpentinite mud with enclosed clasts of various serpentinized ultramafic rock [Fryer et al., 1990a, 1990b; Fryer, 1992].

Details are in the caption following the image
SeaMARC II side-scan imagery of Conical Seamount showing rough-surfaced, unsedimented thin, lobate flows on all flanks. Dark areas, high-backscatter; light areas, low backscatter (smooth sedimented seafloor). White stripes are nadir directions of tracks (no data collected).

5. Discussion

5.1. Source of Muds for Episodic Eruptions

Fault gouge and/or mylonitized, serpentinized peridotite formed by deep forearc faulting most likely provides the source for the dominant serpentinite matrix and clasts in the mudflows [Fryer, 1992; Lagabrielle et al., 1992; Savov et al., 2005a]. The faulting also provides long-lived conduits as pathways for rise of the muds. The footprint of a given seismic event on a fault underlying a seamount may determine the size and depth of the source region from which the muds of an individual protrusive episode are derived. Savov et al. [2005a] showed that the composition of the mud matrix of a flow is similar to the ultramafic clasts from a given flow unit, suggesting that the matrix material derives from the included clasts. The material that makes up the mudflow unit during a given protrusive event is thus essentially self-derived. Some flow units on Conical Seamount contain no trace of mafic components whereas others do [Fryer and Mottl, 1992; Lagabrielle et al., 1992]. Differences between flow units can be accounted for by inclusion of varying amounts of “exotic materials” (possibly materials derived from a deeper source such as the decollement region) carried up with the serpentinite mud. This would help to explain the variability of the matrix compositions within separate flow units recovered from flank drill Sites 778 and 779 on Conical Seamount as described by Fryer and Mottl [1992] and Lagabrielle et al. [1992].

An interesting possible connection between the serpentinite seamounts and underlying structures has recently been revealed by multichannel seismic transects across the Izu forearc [Kamiya and Kobayashi, 2000; Kamimura et al., 2002]. These show a zone of low-velocity material underlying the outer forearc. A low-velocity zone beneath the Mariana forearc was reported by Fryer et al. [1985], who interpreted it as a region of serpentinized forearc mantle overlying the decollement, but they did not consider this a zone of upwelling serpentinite from the decollement region that might “feed” the seamounts above and trenchward of it. Kamiya and Kobayashi [2000] also suggest that the low-velocity zone in the Izu-Bonin forearc is serpentinite, but note that it is a continuous feature with the serpentinite ridge in the outer 50 km of the Izu forearc. They interpret this zone in the Izu system as the “root” (and thus the source) of the serpentinite rising to form the ridge and accompanying seamounts. This would imply that the serpentinite that forms the seamounts might be derived from depths greater than the decollement surface vertically beneath the seamount. The systematic trend in pore fluid composition across the forearc identified by Mottl et al. [2003, 2004] is inconsistent with the suggestion that the serpentinite might follow a long path from downdip, such as that suggested by the results of the Izu-Bonin geophysical investigations. Studies of pore fluids from gravity and piston cores and from push cores taken with the Jason 2 ROV [Fryer et al., 1999; Mottl et al., 2003, 2004] have shown that there is a distinct pattern of across-forearc changes in composition of fluids derived from the subducting slab. The changes in composition reflect proposed dehydration reactions taking place as pressure increases with depth to the slab. Of course, the conduits under the seamounts are not likely to be vertical for their entire length, but the fact that the variations in pore fluids are so regular supports the idea that the source region closely underlies the seamounts and that the fluids do not travel laterally updip for long distances or come from very different depths beneath edifices that lie at about the same distance from the trench axis.

The composition of the serpentinite mud matrix material and of the included serpentinized ultramafic clasts indicate that the serpentinite derives solely from the supra-subduction zone mantle (not from the mantle of the subducting slab) [Savov, 2004; Benton et al., 2004]. The serpentinite comes from regions affected by movement on fault zones underlying the individual seamounts and is most likely episodically mobilized during seismic events that prompt surficial protrusion and form major mudflow lobes on the flanks of the seamounts. It is clear from the high-resolution imagery and the results of pore fluid studies that fluid release can be sustained for longer periods than the actual mud protrusion and gives rise to a variety of chimney structures at active seep sites near the summit conduit regions of some of the seamounts (Figure 3), and possibly also along fault traces on the seamount flanks or seafloor near the edifices [Fryer et al., 2000].

5.2. Variability of Mud Compositions

The major and trace element composition of the muds and schists indicate that these metabasic fragments have mixed with the dominant serpentinite material that comprises the mud matrix and altered its composition [Savov et al., 2005a]. Smear slide studies of both drill samples from both Conical Seamount (Leg 125) and S. Chamorro Seamount (Leg 195) show diverse lithic and crystal fragments incorporated in the muds to varying degrees in different flow units down-hole [Fryer and Mottl, 1992; Lagabrielle et al., 1992]. Our studies of the mud matrix from both the drilled samples and from our piston/gravity and ROV push cores confirm the presence of mafic materials [Fryer and Mottl, 1992; Lagabrielle et al., 1992; Savov et al., 2005a; Gharib et al., 2002] and show that the mudflows that form S. Chamorro Seamount have a greater variation in composition than those of Conical Seamount.

The matrix of the mudflows contains individual amphibole and phyllosilicate grains. We suggest that these derive from incipient blueschist clasts dismembered by a combination of expansion, as a consequence of pressure release during rise from depth, and dynamic interaction within the matrix as the mud rises within the conduit. The rate of ascent of these materials is unknown, although profiles of pore fluid composition suggest seepage speeds for the fluids (at the sampling sites and at the time of sampling) range from a few cm/yr to 48 cm/yr (maximum from Big Blue Seamount) [Mottl et al., 2004]. At springs it may be much more rapid. At Site 1200 the fluids are rising at a rate of at least 3 cm/yr faster than the muds [Shipboard Scientific Party, 2002]. We cannot provide a rate for the rise of the serpentinite mud, though it would not rise more rapidly than the fluids. Unfortunately, there is as yet no data on physical properties of these muds to help determine possible rates of rise nor any reliable data on the age of flows from seamount flank drilling. The drilling on ODP Leg 125 was exclusively rotary drilling, which resulted in considerable drilling disturbance and low recovery of the muds. The best information we have so far, regarding the relationship of the clasts to the mud matrix of the flows, is from the APC and XCB cores taken at ODP Site 1200 and the gravity/piston cores from near the summit knoll of S. Chamorro Seamount.

The epidote-bearing schist (Figure 5c) is of particular interest in the suite of metabasic rocks brought up in these serpentinite mud volcanoes. This sample shows the strongest metamorphic fabric among samples that we have studied thus far, and it contains an assemblage that permits us to place constraints on the conditions of metamorphism occurring at or near the decollement. The coexistence of epidote and magnesiorebeckite/barroisite amphibole, with the compositions that we have observed, indicates that this assemblage was formed at approximately 0.4–0.50 GPa and 250–300°C [Maruyama et al., 1986; Enami et al., 2004; Evans, 1990]. This result suggests somewhat higher temperatures than have been estimated for the IBM arc [Peacock, 1992, 1996]. A complete presentation of mineral compositions and metamorphic petrology of these metabasic schists is beyond the scope of this paper and will be presented elsewhere. Previous work has called on shear-stress heating to produce warmer temperatures at the slab-hanging wall interface, but we suggest that another source of elevated temperature could be heat released by the exothermic reactions [e.g., Allen and Seyfried, 2004] that produce serpentine in the near-slab mantle of the overriding plate.

Most clasts show a lack of evidence of retrograde metamorphism, which may argue for a relatively rapid rise of the mud [Fryer et al., 1999, 2000]. Thermodynamic arguments indicate that retrograde reactions are normally very sluggish or nonexistent for most metamorphic rock upon uplift and exposure and thus the lack of retrograde metamorphism in serpentinite mud materials might not seem surprising. The conditions under which the Mariana forearc serpentinite muds and their entrained schists are emplaced, however, are unusual by comparison with the typical slow uplift and erosional exposure normally assumed for metamorphic rocks of exotic terranes. We contend that clasts within the serpentinite muds are rising in a relatively small conduit region of an active mud volcano. Thus the materials are in contact with slab-derived and chemically active fluids in a dynamic paste or slurry of actively serpentinizing fragments of peridotite. An environment such as this would be far more likely to support conditions necessary for metamorphic back-reactions. The fact that all the single metabasic crystals observed to date and nearly all the schists within the serpentinite matrix lack evidence of retrograde processes argues strongly for a relatively rapid rise from the source depths. Additional support for a rapid rise comes from studies of diverse Li isotope fractionation patterns in serpentinite clasts from Conical Seamount [Benton et al., 2004]. The 6/7Li isotopes apparently diffuse extremely rapidly even within individual non-Li mineral grains (e.g., olivine); thus observed isotope disequilibrium fractionation within otherwise similar serpentine minerals (from +10 per mil to −6 per mil [Benton et al., 2004]) may also reveal not only diverse depth of origin (i.e., different slab fluid compositions), but possibly rapid emplacement.

The greater diversity of metabasic schists recovered from S. Chamorro Seamount. on ODP Leg 195 and the fact that debris flows from this seamount reach nearly to the trench [Fryer et al., 1999, 2000] suggests the possibility of recycling of forearc material within the subduction zone and incorporation within the mudflow source regions at depth. To determine whether the metabasic clasts may have derived from a trapped fragment of Philippine Sea plate, recycled forearc/arc massif, or from the subducting Pacific plate will require further analysis, particularly Hf and Nd isotopic work.

5.3. Control Over Eruptive Style

Until recently the eruptive processes of these seamounts were largely unknown. It was assumed that the mudflows forming these large edifices protruded from the summit area at a long-lived central conduit and slowly covered the flanks over time through repeated episodes of protrusion between periods of quiescence [Fryer, 1992]. DSL-120 data for the summit of S. Chamorro Seamount shows that the seamount may undergo episodes of greater fluid flux than measured today and therefore eruption events that involve lower viscosity muds than those that formed the summit knoll (Figure 7). These events may create multiple small-volume, thin mudflows in addition to the larger-volume, higher viscosity protrusions. It is possible that between episodes of activity the summit region deflates as fluids alone escape at seeps near the edge of the vent.

On a much larger scale there is evidence to suggest that Conical Seamount (ODP Leg 125) has experienced more voluminous fluid–charged eruption events in the relatively recent past. The side-scan sonar data from Conical Seamount, indicating long, narrow flow features, suggests relatively low viscosity flows. None of the other seamounts imaged to date [Fryer et al., 2000] have similar backscatter character. The fact that the serpentinite mudflow surfaces observed in Alvin dives in 1987 [Fryer et al., 1990a] and Jason 2 lowerings in 2003, on Conical Seamount are unsedimented suggests recent eruption. Geochemical data from the pore fluids in these flows indicate derivation by dehydration reactions in the slab [Mottl, 1992]. The pore fluids are thus freshened relative to seawater. The low values for chlorinity of the pore fluids from Conical Seamount (234 mmol/kg) by comparison with those for S. Chamorro Seamount pore fluids (510 mmol/kg) [Mottl et al., 2003] are consistent with more highly fluid-charged eruptions on Conical Seamount.

There may be two reasons for a greater fluid content of the recent mudflows on Conical Seamount. The flows could have formed by eruption of muds with a larger volume of slab-derived fluids. This would account for both the flow morphology and the lower chlorinity content in pore fluids within the muds. Alternatively, gas-hydrate destabilization may play a role. A high methane content in pore fluids was detected at Site 789 on ODP Leg 125 [Fryer et al., 1990b]. If there are gas-hydrate layers within the seamount (methane produced by serpentinization), it is possible that there may be periodic gas-hydrate destabilization events prompted by phenomena such as flank mass-wasting events, a change in temperature of rising fluids, or local seismic disturbance. Destablilization of gas-hydrates could explain a freshening of the fluids, but could not of itself explain the ratios of minor elements ascribed to the slab origin of the fluids [Mottl, 1992]. We conclude that, by whatever mechanism, a larger volume of freshened, slab-derived fluids accompanied the recent eruptions from Conical Seamount. We also suggest that the smaller apparently less viscous mudflows observed on DSL120 imagery from the summit of S. Chamorro Seamount also reflect a higher fluid content.

It is important to note that there may be a temporal variability in style of eruption as well as in composition of the muds. If the style is indeed a function of the fluid (or possible gas) content of the muds, the style may vary on much shorter timescales than does that of the composition of muds involved in a given eruptive episode. The highest rate for fluid egress at the Mariana seamounts studied to date is 48 cm/yr (Big Blue Seamount, Jason2 push core results [Mottl et al., 2003, 2004]). If the mud rises more slowly than fluids or gas it could become more viscous as the eruption proceeds.

6. Conclusions

The origin of the serpentinite muds that form these seamounts is complex. The sources of the muds may vary from seamount to seamount and within each seamount they vary with time. Studies of the composition of the serpentinized peridotite that comprises the serpentinite muds show that it derives exclusively from the forearc mantle wedge. Likewise, the serpentinized peridotites that are enclosed in the muds show supra-subduction-zone characteristics [Parkinson and Pearce, 1998; Savov et al., 2005b]. Savov et al. [2005a] show that the mud matrix is geochemically identical to the ultramafic clasts except where “exotic materials” have been added to the matrix through the incorporation of lithic clasts and crystals derived dominantly from metamorphosed mafic rocks. The source of the metabasite clasts is variable. Some boninitic and island arc basaltic materials recovered in drilling on ODP Leg 125 at Conical Seamount must have come from the shallow arc crust after the onset of arc volcanism. The MORB, ocean island basalts, alkalic basalts and other oceanic plate-derived materials (including Cretaceous cherts) must have come from either trapped Philippine Sea plate or from the subducted Pacific plate [Fryer, 1996b; Pearce et al., 1999]. The greater diversity of metabasic schists recovered from S. Chamorro Seamount on ODP Leg 195 and the fact that debris flows from this seamount reach nearly to the trench [Fryer et al., 1990a, 1999] suggests the possibility of recycling of forearc material within the subduction zone (and possibly back up the conduit of the mud volcano?). To determine whether the metabasic clasts may have derived from a trapped fragment of Philippine Sea plate, the recycled forearc/arc massif, or from the subducting Pacific plate will require further isotopic analysis.

It is likely that protrusion events that are accompanied by large-volume flushing episodes with a high volume (low chlorinity) of slab-derived fluids would be more apt to contain deep-derived metabasic materials from the subduction zone. Presumably, such events would generate a lower-viscosity mud and thus could transport materials to the surface from depth more rapidly. Such events might more readily explain the paucity of retrograde metamorphic effects observed in the high-pressure low-temperature schists. More detailed sampling and analysis of potential mudflow units and further deep-sea drilling on the flanks of the seamounts would help to test this hypothesis.

Acknowledgments

We thank the captains and the crews of the JOIDES Resolution (ODP Legs 125 and 195) and the R/V Thompson (2003) for their expert assistance during the surveys. We are grateful to M. Heinz and the Deep Submergence Operations Group of the National Deep Submergence Facility at WHOI for their expert assistance in operations of the DSL120 side-scan sonar and Jason2/Medea ROV systems during the cruise in 2003. We particularly thank N. Becker for his renderings of the bathymetric maps presented in this paper and F. Martinez for the perspective map for Conical Seamount. The manuscript was improved by helpful suggestions from reviewers J. Beard and V. Salters. This work was carried out under support from the U.S. Science Support Program of the Ocean Drilling Program grant F001312 and NSF grant OCE-0002584 (2003 cruise). This is Hawaii Institute of Geophysics and Planetology contribution 1427 and SOEST contribution 6730.