Geochemistry, Geophysics, Geosystems

Malaguana-Gadao Ridge: Identification and implications of a magma chamber reflector in the southern Mariana Trough

Authors


Abstract

[1] Six-channel seismic reflection data reveal a magma chamber reflector beneath the Malaguana-Gadao Ridge, the southernmost segment of the spreading center in the Mariana Trough. For most of its length the spreading center in this active back-arc basin is morphologically similar to slow spreading mid-ocean ridges, having a deep central graben flanked by a zone of abyssal hill fabric. This southernmost segment, however, has a broad, smooth cross section, lacks a deep central graben, and is thus similar in morphology to fast spreading ridges (e.g., the East Pacific Rise). We identify a magma chamber at 1.5 s two-way travel time below the crest of the ridge. Observations from remotely operated vehicles along the ridge reveal not only fresh pillows, lobate, and sheet lava flows but also an abundance of volcaniclastic debris and intense hydrothermal activity. These observations, together with the “fast spreading” morphology of the ridge, suggest that this segment has a considerably higher magma supply than is typical in the Mariana Trough. We suggest that the volcanic arc or enhanced melting of a hydrated mantle is supplying volatile-rich magma as evidenced by a highly negative coefficient of reflection, −0.42, for this MCR and the presence of evolved, highly vesicular lava and volcaniclastic materials. The southeastern Mariana back-arc basin spreading ridge does not compare readily with mechanical models for global mid-ocean ridge data sets because of marked asymmetry in both volcanism and deformation that may be the consequence of slab-related geometry in this part of the convergent margin system.

1. Introduction

[2] The Mariana convergent margin system is located at the eastern boundary of the Philippine Sea plate (Figure 1). Subduction of the Pacific plate beneath the Philippine Sea plate boundary began in the Eocene prior to about 50 Ma [e.g., Seno and Maruyama, 1984]. Stern and Bloomer [1992] and Macpherson and Hall [2001] suggested that the initial phase of subduction prompted widespread volcanism above the subduction zone. Initially this region may have been as much as 200 km wide, but by the Middle Oligocene (∼30 Ma) volcanoes of the arc had become localized along what is now referred to as the volcanic front [Kobayashi and Nakada, 1979]. From about 30 to 20 Ma a series of episodic arc building and rifting events occurred, forming from west to east, the Palau-Kyushu Ridge and Parece Vela Basin of the Philippine Sea (Figure 1), and the West Mariana Ridge [Okino et al., 1998]. Since 6 Ma extension has been concentrated in the eastern basin of the system, the Mariana Trough, and suprasubduction zone volcanism has been limited to the active Mariana volcanic arc (represented by the Mariana Ridge) and to the spreading center in the Mariana Trough.

Figure 1.

Location map for the area of this study showing major tectonic features and the convergence rate and direction between the Philippine Sea plate and the Pacific plate as determined by the GPS measurements of Kato et al. [2003]. Bathymetry has been extracted from the ETOPO-2 global topography data set [Smith and Sandwell, 1997].

[3] Extension of the Mariana Trough takes place by several different mechanisms along strike. At the northern end of the back-arc basin (north of 20°35′N) gravity data show thick crust that has been affected by an additional supply from island arc magma sources [Gribble et al., 1998; Kitada et al., 2006], but the morphology of the spreading axis suggests relatively slow spreading [Martinez et al., 1995; Baker et al., 1996]. Amagmatic rifting has produced a 5,500 m deep graben at ∼20°N that exposes abyssal peridotite [Stern et al., 1996, 1997; Ohara et al., 2002; Yamazaki et al., 2003]. Gravity data between 15°38′ to 20°35′N suggest variability between slow spreading with “plume-like mantle upwelling” under the axis to magma-starved rifting along the segments [Kitada et al., 2006]. From 14°22′ to 15°38′N gravity data [Kitada et al., 2006] indicate a very low magma supply, but south of 13°10′N the morphology is different from elsewhere along the spreading axis [Martinez et al., 2000; Hasegawa et al., 2000] and gravity data [Kitada et al., 2006] show a thicker crust (average 5800–6800 m), consistent with interpretations by Martinez et al. [2000], Fryer et al. [2003], and Taylor and Martinez [2003] that a greater magmatic supply exists along this southernmost spreading segment.

[4] The greater magma supply assumed for the southern spreading segment (the Malagana-Gadau Ridge) suggested to Kitada et al. [2006] that a “sheet-like” mantle upwelling takes place beneath the ridge. If so, one would expect magma chambers should exist along this segment, similar to those identified using seismic reflection data at the East Pacific Rise [Kent et al., 1990; Hussenoeder et al., 1996; Singh et al., 1998], the Juan de Fuca Ridge [Rohr et al., 1988], the Galapagos Ridge [Blacic et al., 2004], and at some spreading segments within back-arc basins (e.g., Valu Fa Ridge [Collier and Sinha, 1992; Harding et al., 2000] and the East Scotia Ridge [Livermore et al., 1997]). Recent estimates of the spreading rates by Kato et al. [2003], using global positioning system (GPS) measurements suggest a full rate of 45 mm/yr. Martinez et al. [2000] suggested that the inflated morphology resulted from entrainment of magma from the volcanic arc into the back-arc spreading center. The Malaguana-Gadao Ridge's anomalously high magma supply for its spreading rate would appear to place it in the same class of back-arc spreading ridges as the Valu Fa Ridge and Segment E2 and E9 of the East Scotia Ridge, which are spreading at slow to intermediate rates of 39–61 mm/yr [Zellmer and Taylor, 2001] and intermediate rates of 65–70 mm/yr [Livermore et al., 1997; Bruguier and Livermore, 2001], respectively. Thus both the interpretations by Martinez et al. [2000] and Kitada et al. [2006] would suggest that there should be magma chambers developed along the entire southern spreading segment.

[5] In this paper we present data from seismic surveys, swath-mapping sonar, and observations from remotely operated vehicles (ROVs) that have identified recent volcanic activity and active low- (10°C [Masuda et al., 2001]) to higher-temperature (>248°C [Wheat et al., 2003] and 220°C black smoker [Urabe et al., 2004]) hydrothermal venting on the ridge, and geophysical evidence for the presence of a single magma chamber reflector at one locality along the ridge. We summarize submersible and ROV observations on the active spreading segment and nearby off-axis volcanism. We also discuss the implications of previously published geochemical data for lavas collected in the study area and relate the topographic, seismic reflection, ROV observations, and geochemical data to processes associated with the generation of a magma chamber in one area of the southern Mariana Trough spreading axis as compared to models for global mid-ocean ridge data sets.

2. Data Acquisition

[6] Swath-mapping sonar data were acquired with both multibeam and side scan systems on several cruises from 1997 to 2003. The first survey [see Martinez et al., 2000; Fryer et al., 2003] used the HAWAII MR-1 towed side scan system [Rognstad, 1992] to collect both bathymetry and backscatter imagery from the R/V Moana Wave. This cruise also collected the six-channel seismic reflection data. The MR-1 system was used again in 2001 from the R/V Melville during a cruise that also collected SeaBeam 2000 multibeam data along the crest of the Malaguana-Gadao Ridge during transects between dredging locations and wax-coring sites. Finally, in 2003 a cruise aboard the R/V Thomas G. Thompson collected Hydrosweep and Simrad EM300 multibeam data for part of the ridge crest and nearby seamounts. Bathymetry data were merged (Figure 2) using the technique of Becker [2005], MB-System software [Caress and Chayes, 1995, 1996], and Generic Mapping Tools software [Wessel and Smith, 1991, 1998].

Figure 2.

Bathymetric map of the 1997 R/V Moana Wave survey area in the southern Mariana Trough including the Malaguana-Gadao Ridge. Contour interval is 250 m, and dashed lines show where six-channel seismic reflection data were acquired. White diamonds indicate locations of rock dredge sampling by Japanese scientists (see section 4). Inset shows detail of area where a seismic reflection profile shows a magma chamber reflector. Heavy red line A–A′ indicates the location of the seismic reflection profile shown in Figure 3, and red boxes K and J identify the locations of the dives by the ROVs Kaiko and Jason-2, respectively.

[7] In addition to collecting HAWAII-MR1 side scan data, the 1997 R/V Moana Wave survey collected digital six-channel seismic reflection data throughout the side scan survey region (Figure 2). The sonar ping repeat rate was synchronized with the seismic survey shot interval. The R/V Moana Wave towed two air guns with chamber sizes of 120 and 300 in3 (1.97 and 4.92 l). The streamer was configured with 6 channels at 25 m group interval, and the survey was run at 8.1 knots (15 km/h) with a firing interval of 9 s. This resulted in a shot spacing of 37.5 m thus yielding a twofold seismic coverage with 12.5 m CMP interval. In water depths greater than 5 km, however, it was necessary to increase the firing interval to 18 s in order to provide adequate time to record the side scan swaths, in which case seismic coverage was decreased to single fold at 12.5 m CMP interval. The data were digitized at a 2 ms interval and recorded in SEGY demultiplexed format (32-bit floating point) on Exabyte and DAT tape using a Sun-based seismic acquisition system. Data were processed using ProMAX software and processing included editing of bad traces, sorting, deconvolution, time-variant filtering, muting to water bottom, migration, and depth conversion. Signal-to-noise was improved by either summing adjacent traces or applying a running mix prior to migration, and the 12.5 m reflection point spacing allowed for migration of steep dips without spatial aliasing. Migration and depth conversion velocities were estimated from iterative migration, using a range of velocities. Migration results, however, were degraded because of the oblique crossing of topography.

3. Seismic Data Interpretation

[8] The side scan data provide high-resolution images and bathymetry of the southern Mariana back-arc and outer fore-arc regions and the seismic reflection data show sedimentary strata and some deeper structures. The survey pattern, however, was optimized to collect side scan imagery and consisted of east-west transects spaced 5 nautical miles (9.26 km) apart and oblique (∼65°) to the strike of the Malaguana-Gadao Ridge. The seismic reflection data show only a single possible magma chamber reflector despite 12 crossings of the ridge axis (see inset in Figure 2 and the seismic reflection profile in Figure 3a). The magma chamber reflector candidate is a high-amplitude reflector at a depth of 1.5 s two-way travel time (∼3.1–3.3 km) below the crest of the spreading ridge. This time is comparable to the magma chamber reflector found beneath the East Scotia Arc (∼1.4 s [Livermore et al., 1997]) and beneath the Valu Fa Ridge (∼1.6–1.8 s [Collier and Sinha, 1992]). The horizontal width is less than 550 m, which is much narrower than that observed at Valu Fa (0.6–4 km). No similar reflector appears on the parallel adjacent profiles ∼9 km to the north or south of this locality, nor did we identify any similar reflector in any other ridge-crossing profile (see Figure 2 for line locations). Therefore, we can only say we identified a single possible magma chamber reflector beneath the Malaguana-Gadao ridge and that it extends less than ∼18 km along the ridge axis.

Figure 3.

Seismic reflection data were collected aboard the R/V Moana Wave by a six-channel system in 1997. These data were filtered and processed with deconvolution and stacking (twofold) algorithms and F-K migration. Automatic gain control was not used in order to preserve relative amplitudes. (a) Seismic reflection profile (A–A′) across the ridge crest with the location of the magma chamber reflector (MCR) indicated. (b) Close-up of the MCR area. The small boxes B and C indicate the locations of the traces used for polarity analysis. Detail of individual traces (stacked only) showing the (c) seafloor and (d) MCR. The traces showing the MCR are displayed with their polarity reversed to compare them with the seafloor reflector. The phase angle for both sets has been rotated through −120° to aid interpretation.

[9] Because a magma chamber has lower acoustic impedance than surrounding rock, a magma chamber reflection should display a reversed polarity. Detail of the magma chamber reflector (MCR) is shown in Figure 3. The bottom reflections (Figure 3c) show normal polarity, whereas the traces we interpret to be the MCR (Figure 3d) show reversed polarity. The apparent repetition of reflectors in these detailed views result from oscillations of the bubble pulse source. Because the polarity of the reversed traces match the polarity of the bottom reflector, the MCR candidate represents a boundary between a zone of higher acoustic impedance above a zone of lower acoustic impedance. This observation is consistent with its representing the top of a magma body.

[10] We compare the MCR with similar reflectors observed in other spreading centers, by calculating its reflection coefficient following the method of Collier and Sinha [1992]. The line on which the MCR was identified does not contain a seafloor multiple (Figure 3a), we estimated amplitudes of the bottom reflector and its multiple from the next survey line to the south where the ridge crest was shallower. We also used the same values for stacking velocity of the MCR (Vrms = 2.2 km/s), and transmission loss correction factor (L = 1.23) as Collier and Sinha [1992] (after Morton and Sleep [1985]), and we assumed a seawater velocity, V1, of 1.5 km/s. We then calculated the value of the coefficient of reflection for the MCR, R, on the 33 traces that show it (Figure 4). All values were calculated from data within stacked lines without further processing. This yields a mean value for R of −0.42 with a standard deviation of 0.14. When compared with magma chamber reflectors identified beneath other seafloor spreading axes, R for the Malaguana-Gadao Ridge MCR is comparable to those determined for the Valu Fa Ridge [Collier and Sinha, 1992].

Figure 4.

Histogram of values for the reflection coefficient calculated for the magma chamber reflectors on the 33 traces that show them. These values have a mean and median of −0.42 and a mode of −0.40. The standard deviation of these calculated values is 0.14.

4. Submersible Observations of Associated Ridge Volcanism

[11] A deep tow video camera survey was carried out along the ridge crest from 12°56′N to 13°08′N [Mitsuzawa et al., 2000]. The video images indicate that the volcanic activity is greatest in the vicinity of the MCR, between 13°04′N to 13°08′N. Hydrothermal activity was first observed on the ridge crest during this survey at about 13°05′N, 13°06′N and 13°07′N [Mitsuzawa et al., 2000; Masuda et al., 2001]. Temperature and Eh anomalies were observed in the seawater column when the video camera passed 13°06′N [Mitsuzawa et al., 2000]. Fresh volcanic rocks ranging from basalt and andesite to rhyolite (H. Masuda, personal communication, 2004) were dredged from 13°08′N. These had abundant surface glass and a strong sulfide smell. Following the deep tow survey a Kaiko ROV dive 164 [Masuda et al., 2001] took place between 13°05.8′ and 13°06.3′N and observed hydrothermal chimney structures, pit craters, two lava lakes, white flocculant matter suspended in the water throughout the dive area, and white smoker vent activity.

[12] A Jason-2 ROV lowering (J2-42) at the crest of the Malaguana-Gadao ridge 12°57.214′N, 143°37.147′E in a water depth of ∼2860 m (see Figure 2) in 2003 from the R/V Melville revealed similar volcanic products and active, higher-temperature hydrothermal sites. This location is ∼15 km SW of the MCR locality, but revealed similar lava types and volcanic structures, as those observed with Kaiko on dive 164, but the hydrothermal systems have higher-temperature (∼248°C [Wheat et al., 2003]) hydrothermal vents. Relief on the ridge axis in the Jason-2 lowering area is less than 20 m. The dive location was chosen on the basis of a water column turbidity maximum identified with towed instruments [Embley et al., 2004]. There were also elevated turbidity readings both north (close to the area of the MCR) and south of the Jason-2 lowering [Baker et al., 2005]. The high backscatter of the ridge on side scan imagery [Fryer et al., 1998; Martinez et al., 2000; Fryer et al., 2003] is consistent with this ridge being active volcanically along most of its length.

5. Discussion

[13] The Malaguana-Gadao Ridge has an “inflated” morphology despite its intermediate spreading rate [Martinez et al., 2000]. The recovery of arc magmas [Pearce et al., 2005] and our discovery of an MCR support the interpretation that this ridge segment taps not only a back-arc basin magma source, but also the nearby arc source. The spreading rate of the Malaguana-Gadao Ridge remains enigmatic although marine magnetic anomaly data and GPS measurements indicate asymmetric spreading, provided there have been no ridge jumps. Martinez et al. [2000] could only constrain Brunhes chron full-spreading rate to less than 64 mm/yr from their magnetic data collected on the eastern side of the Ridge. Magnetic data collected from the western side of the Ridge, however, give a half spreading rate of 35 mm/yr (N. Seama, personal communication, 2004), and would yield a full rate of 70 mm/yr assuming symmetric spreading. Kato et al. [2003] used GPS measurements to determine that the southern Mariana Trough is currently opening (E–W) with a full spreading rate of 45 mm/yr. The magnetic- and GPS-derived spreading rates together suggest that the southern Mariana Trough is spreading asymmetrically as it does in the central and northern parts of the Trough [Deschamps and Fujiwara, 2003; Kitada et al., 2006]. Such asymmetry would continue the trend from the central and northern portions of the Trough and would also be consistent with maintaining the position of the spreading center at the eastern edge of the back-arc region, proximal to the magmatic arc source region.

[14] The proximity of submarine arc volcanoes [Fryer, 1995; Fryer et al., 1998] and their magma source region likely provides the surplus magma required to produce the ridge's inflated morphology and the MCR. A similar correlation of arc volcanoes proximal to an inflated spreading segment has been observed for the Valu Fa Ridge in the southern Lau Basin [Collier and Sinha, 1992; Wiedicke and Collier, 1993; Martinez and Stern, 2009]. The fact that we do not observe MCRs on any of our other ridge crossings may indicate that deep-sourced arc magma is entrained into the spreading axis directly from discrete arc magma sources of limited extent that feed only portions of the en echelon ridge segments [Baker et al., 2005] that make up the Malaguana-Gadau Ridge. Alternatively, as suggested for the Lau Basin [Martinez and Taylor, 2002] and later for the Mariana Trough as well [Taylor and Martinez, 2003], fluids released from the descending slab of the Pacific plate may have lowered the melt solidus of the mantle beneath the spreading axis; thus decompression of hydrated mantle results in enhanced melting relative to melting of “dry” mantle [Stolper and Newman, 1994]. This would be more consistent with the “sheet-like” dike feeder system envisioned by Kitada et al. [2006]. If this were the case, however, we would have expected to see more MCRs along the 12 crossing of the ridge during our survey. A multichannel seismic survey along the length of the ridge axis would be one way to quantify the distribution of magma chambers beneath it, but distinguishing between these two mechanisms of magma genesis would require a detailed trace element geochemical study of the erupted lavas. Either mechanism, however, could produce evolved lavas with high volatile contents to explain the R values we observe. The presence of both evolved lavas as well as basaltic ones from the MCR area of the ridge (H. Masuda, personal communication, 2004) is consistent with this.

[15] The Malaguana-Gadao Ridge is not the first back arc spreading segment to be identified with some of the aforementioned characteristics. Segment E2 of the East Scotia Ridge has a similar morphology and MCR [Leat et al., 2000] to the Malaguana-Gadao Ridge, and erupts basalts with 0.77 to 1.02 wt % H2O [Fretzdorff et al., 2002], as the result of mantle flow bringing enriched mantle below the spreading center [Livermore et al., 1997]. Likewise, the Valu Fa ridge in the southern Lau Basin also has an inflated morphology and an MCR. It erupts andesitic basalts, andesites, and rhyolites [Frenzel et al., 1990]. These lavas have high volatile contents, averaging 1.23 wt % H2O and 0.03 wt % CO2 [Vallier et al., 1991]. We would also expect evolved rocks with elevated volatile contents from Malaguana-Gadao Ridge and the observations by Masuda et al. [2001] from the region near the MCR suggest this is likely.

[16] Rock samples recovered from the MCR area have mainly andesitic [Mitsuzawa et al., 2000] to rhyolitic compositions (H. Masuda, personal communication, 2004). The facts that the lavas are highly vesicular, and that spatter cones are present [Mitsuzawa et al., 2000], suggests that the lavas from this portion of the ridge do have a gas-rich magma source. Some of the highly vesicular fragments are similar to the frothy lava described as “volcanic mousse” cored in the Sumisu Rift on ODP Leg 126 drilling Site 791 (units 11–14) [Taylor et al., 1990].

[17] Modeling efforts to interrelate ridge topography, magmatic supply rate, faulting, crustal thickness and lithosphere/asthenosphere rheology at mid-ocean ridge systems [e.g., Wang and Cochran, 1995; Buck et al., 2005; Behn and Ito, 2008; Ito and Behn, 2008] provide a context for discussion of factors that may influence development of the Malaguana-Gadau Ridge. Wang and Cochran [1995] showed that spreading ridges with a high magma supply rate lack the “bull's-eye” pattern in mantle Bouguer anomalies (MBA) that are typical of slow spreading (and thus low-magma supply) ridges. The axial MBA for high-supply ridges also has a generally low gradient. The gravity data of Kitada et al. [2006] along the Malaguana-Gadau Ridge show both of these characteristics. Furthermore, the MBA is greater in the vicinity of the MCR than elsewhere along the Malaguana-Gadau Ridge. Kitada et al. [2006] suggest average crustal thickness of between 5800 and 6800 m for this part of the back-arc basin. This is consistent with models indicating efficient along-axis melt transport, so we should expect magma chamber reflectors along more of the ridge than we have so far observed. An along-strike multichannel seismic survey might show such reflectors have greater extent than we were able to resolve.

[18] The faulting pattern along the Malaguana-Gadau Ridge does not match that postulated by models of Buck et al. [2005] for normal ridges. Faults to the southeast side of the ridge have greater throw than those to the northwest side (see Figure 2). Buck et al. [2005] assume symmetric spreading for their models and likely this is not the case in the southern Mariana back-arc region because of unusually greater extensional tectonics in the southeastern corner of the Mariana system. Fryer et al. [2003] presented bathymetric and side scan data that show two characteristics of the southeastern Mariana back-arc basin related to this deformation that differ significantly from typical mid-ocean ridge spreading environments. A submarine ridge composed of older arc volcanoes lies about 25 km SE of the Malaguana-Gadau Ridge (Figure 2). Side scan imagery [Fryer et al., 2003] shows that some of these (in the extreme northeast and southwest) have been overprinted by recent volcanism. The imagery also shows that the seafloor between the Malaguana-Gadau Ridge and the volcanic ridge has been largely resurfaced by recent lava flows and small volcanic cones. Volcanism most likely was rejuvenated along NW–SE striking fracture/fault systems formed when the fore-arc region of the SE Mariana system was deformed by adjustment to a tear in the downgoing plate that prompted widespread extension of the fore-arc region [see Fryer et al., 2003; Gvirtzman and Stern, 2004]. The question, then, is whether a ridge, such as the Malaguana-Gadau, fundamentally behaves like a typical mid-ocean ridge in a back-arc setting and conforms with recently published mechanical models. The ridge morphology [Martinez et al., 2000] and gravity data [Kitada et al., 2006] suggest a high magma supply rate and a thickened crust. The presence of the old volcanic ridge, and the fact that the volcanic centers along it are spaced far closer than they are elsewhere along the Mariana arc [Embley et al., 2004], suggests that arc volcanism is voluminous in this general area. The facts that the seafloor between the spreading axis and the old volcanic ridge has been resurfaced and that some of the volcanic centers are rejuvenated also suggests a higher than average magma supply for this area [Fryer et al., 2003]. All of these factors point to a ridge that should behave as expected for high–magma supply ridges according to models of Buck et al. [2005], Behn and Ito [2008], and Ito and Behn [2008]. However, there are some unique characteristics of the ridge that reflect its unusual tectonic setting.

[19] Not only is the ridge topography highly asymmetric, but volcanism on the southeast side of the ridge is also resurfacing a large region of the seafloor (up to 25 km from the ridge). The only place that high-temperature black smokers have been discovered (at 2 and 5 km from the spreading axis) is in some of these small volcanic centers southeast of the ridge. Martinez and Taylor [2002] suggested that hydrous conditions in the mantle wedge near the volcanic front of the southern Mariana back arc could explain the Malaguana-Gadau Ridge morphology. Other possibilities are that if the interpretation that a tear exists in the subducting plate to the east of the ridge segment is correct [Fryer et al., 2003], it is possible that Pacific mantle may protrude around the edge of this tear and provide an additional source of heat for the southern fore-arc and back-arc regions. Similarly, Gvirtzman and Stern [2004] also suggested that as the slab to the west of the area of the tear is much shallower, there may be asthenospheric flow from beneath the lower edge of the slab replenishing the magma source region. Trace element and isotopic analyses underway on samples collected from the southern back-arc region will be needed to assess the validity of these hypotheses.

6. Summary

[20] Six-channel seismic reflection data indicate that a magma chamber reflector (MCR) lies 1.5 s TWT beneath the Malaguana-Gadao Ridge. The Malaguana-Gadao Ridge, like other spreading segments within the Mariana Trough, spreads asymmetrically and if it lacks ridge jumps in its past, it may accrete new crust to the west up to several times as rapidly as it does to the east. The magma chamber reflector has a high amplitude and a highly negative coefficient of reflection, comparable to other MCRs identified in back-arc basin settings, consistent with the presence of volatile-rich magma. This segment is volcanically active with a range of lava compositions from basaltic to rhyolitic, and various morphologies, as well as numerous hydrothermal sites. The highest-temperature sites thus discovered occur off ridge to the southeast of the spreading axis. Though it has a slow to intermediate spreading rate, the Malaguana-Gadao Ridge is most likely morphologically similar to a fast spreading ridge because it receives a greater supply of magma than other segments along the Mariana back-arc spreading axis. The volcanic arc magmas that supply the ridge may either originate because of hydration of the melt source region by slab-derived fluids [Martinez and Taylor, 2002] or the region may be hotter in general because of incursion of Pacific mantle asthenosphere as a consequence the geometry of the subducting plate [Fryer et al., 2003; Gvirtzman and Stern, 2004] in this part of the Mariana convergent margin system.

Acknowledgments

[21] We wish to thank S. Poulis, S. Stahl, and D. Gravat for their assistance with the operation of seismic reflection gear and Z. Zhao for his help with processing the seismic reflection data. We would also like to thank B. Appelgate and the Hawai'i Mapping Research Group for their outstanding operation of the HAWAII MR-1 side scan sonar system, and we would also like to thank the members of WHOI's Deep Submergence Laboratory for their smooth and efficient deployment of the Jason-2 ROV. Additionally, we would like to thank the officers and crew of the R/V Thomas G. Thompson, the R/V Moana Wave, and the R/V Melville for their very capable and professional operations. We gratefully acknowledge A. Deschamps, J. Collier, C. Beier, R. D. Larter, and R. J. Stern for their thorough and constructive reviews and T. Becker for editorial suggestions. We are especially grateful for discussions with and the contribution of information on unpublished geochemical data from H. Masuda for sample compositions from the Kaiko dive 164 area. This study was funded through grants OCE-9633501, OCE-9907063, and OCE 0002584 from the National Science Foundation. This is SOEST contribution 7854 and HIGP contribution 1838.