Volcanic Facies and Geochemical Stratigraphy

Preservation of volcanic structures and textures in altered rocks recovered below Sites 1188 and 1189 allowed reconstruction of the architecture of the upper two-thirds of Pual Ridge under PACMANUS despite the lack of deep penetration at the planned "background" location (Site 1190). Recognizable relict structures and fabrics include flow banding, autoclastic breccias, former glassy rinds (perlite), conspicuous to sparse vesicularity, phenocrysts, and hyaline, microlitic, or spherulitic groundmasses. Apart from spherulitic structure, all these fabrics are represented in fresh lavas dredged from Pual Ridge, although flow banding is rarely obvious.

Identified paleoseafloor horizons include flow-top breccias and gravelly to sandy volcaniclastic sediments, some containing fragments with differing mineral assemblages and textures. Perlite horizons may also denote former seafloor. A significant feature, subject to the constraint of poor recovery, is the total absence in core of foraminiferal hemipelagic sediment layers between apparent flow units. This suggests very rapid construction of Pual Ridge in view of high hemipelagic sedimentation rates in the region (Barash and Kuptsov, 1997, cite a uniform rate of 15.5 cm/k.y. over the past 16,000 yr for the central Manus Basin; rates from 14C dating of foraminifers in shallow sediment cores at four sites in the eastern Manus Basin average 23 cm/k.y.; J.B. Keene and R.A. Binns, unpubl. data).

No plutonic or distinctly hypabyssal igneous rocks were encountered in any drill hole. In the Lower Sequence of Hole 1189B a homogeneous interval of altered dacite (128–147 mbsf) with subvertically stretched vesicles was considered on board to be a possible dike (Shipboard Scientific Party, 2002c). Interpretation of resistivity imagery confirmed the presence of a discordant body (123–145 mbsf) with contacts inclined 42–66 from horizontal (Bartetzko et al., 2003). The geochemistry of this unit is consistent with a now-altered subvolcanic feeder for overlying lavas (Miller et al., this volume).

Paulick et al. (2004) delineate four main precursor volcanic facies in what were submarine, predominantly felsic eruptives below Snowcap and Roman Ruins, namely coherent lava, monomict breccia, polymict breccia, and volcaniclastic sediment. They interpret the coherent facies as the insulated central parts of individual flows and the autoclastic breccias and some volcaniclastic facies as the outer parts, and they suggest other volcaniclastic sediments were derived by gravity-driven mass transport. Noting greater abundance of coherent lavas at Site 1188 and of breccias at Site 1189, they propose that the former was close to an eruptive center whereas the latter was more distant. Assessments of geophysical logs and resistivity images of the borehole walls extend the volcanic facies analysis to intervals of zero or very poor core recovery and to uncored LWD Holes 1188B and 1189C (Bartetzko et al., 2003; Arnold et al., submitted [N1]). The results suggest small-scale lateral variability of structure within flow units below both sites. Throughout LWD Hole 1188B, for example, a higher proportion of brecciated lava is inferred than for the equivalent depth interval in cored Hole 1188A only ~5 m away at Snowcap. Interpreted facies units correlate poorly between Holes 1189A, 1189B, and 1189C at Roman Ruins despite their proximity (within 60 m).

Paulick et al. (2004) conclude that Pual Ridge was constructed from small-volume coherent and breccia flows, ranging from 5 to 40 m in thickness, with minor volcaniclastic horizons. Applying comparisons with exposed sections through shallow subaqueous rhyolitic lavas (Kano et al., 1991; McPhie et al., 1993), they propose inflationary growth as a common process, yielding domal topographies. This is not typical of dacite to rhyodacite eruptives at the PACMANUS seafloor, which range as thick as 30 m and several hundred meters in lateral extent, are highly fluid, and more closely resemble subaerial basaltic pahoehoe and aa flows in structure (Waters et al., 1996). The drilled volcanic sequence as at PACMANUS is attributed to three main phases of activity by Paulick et al. (2004), a conclusion greatly constrained by core recovery and restriction of observations to two sites not selected on volcanological grounds.

Fresh glassy lavas dredged along Pual Ridge and its flanks display a systematic increase in the ratio of Zr (in parts per million) to TiO2 (in weight percent) with fractionation from basaltic andesite to rhyodacite (see Fig. AF1 in the "Appendix"). Zirconium and Ti are widely considered immobile elements during alteration and metamorphism, and this is supported for Leg 193 altered rocks by consistent Zr/TiO2 ratios in zoned samples with varying alteration assemblages. The ratio can be used to delineate precursor compositions for altered rocks (see Table AT1 in the "Appendix").

On this basis, Figure F5 establishes a sequence of lavas below Sites 1188 and 1189 ranging from andesite through mafic and felsic dacite to rhyodacite, spanning most of the fractionation series represented by fresh seafloor lavas from Pual Ridge. Especially noteworthy are the closely comparable profiles in adjacent Holes 1189A and 1189B at Roman Ruins, with ~70 m of mafic dacite underlain by 40–50 m of rhyodacite, in turn underlain by a thin andesite, perhaps with a thin intermediate-SiO2 dacite intervening between the latter two. There is no equivalent correlation between Sites 1188 and 1189, although altered andesites occur at about the same level under both sites, taking into account the ~45-m difference in collar elevation. An abrupt change in wallrock precursors within the Stockwork Zone of Hole 1189B occurs from mafic dacite to rhyodacite somewhere between 50 and 60 mbsf. This does not correspond to a facies change at 76 mbsf interpreted by Bartetzko et al. (2003).

A conservative interpretation of the immobile element geochemistry is that 12 compositional units (flows or flow sequences), averaging 31 m but ranging from a few meters to 60 m in thickness, are present between the seabed and 372 mbsf under Snowcap. At Roman Ruins, four units averaging ~30 m thick are indicated in Hole 1189A and six units averaging 28 m thick in Hole 1189B. Allowing for the many constraints, average flow thicknesses derived from facies analysis and geochemistry are comparable. The two approaches to understanding the volcanic architecture of Pual Ridge have not yet been integrated. Although Paulick et al. (2004) question application of the term "layer cake" to the sequence, it is clear that Pual Ridge is not constructed by inflation of only a few high-profile domes. Submersible observations and dredge samples suggest that individual flows may extend further than inferred in the models based on facies correlations between Site 1188 and Site 1189.

Cap of Unaltered Felsic Lavas

Fresh or relatively unaltered volcanic rocks, all coherent and significantly vesicular, were intersected immediately below the seafloor at the four Leg 193 drill sites. Their thickness exceeds penetration depth (17 and 20 m, respectively) at Sites 1190 and 1191. At Snowcap, changes in resistivity and gamma profiles at ~30 mbsf in LWD Hole 1188B (Bartetzko et al., 2003) define a fresh/altered boundary similar to that cored at ~35 mbsf in Hole 1188A. Below chimneys at Roman Ruins, Hole 1189A recovered ~0.2 m of fresh vesicular dacite samples before passing into altered rock. Casing obscured the immediate subseafloor in Hole 1189B, and LWD Hole 1189C appears entirely altered (Bartetzko et al., 2003).

The fresh lava caps at these sites range from mafic dacite to rhyodacite in composition and represent different lava flows. They display similar fractionation behaviors in major and trace elements (Miller et al., this volume) to seabed lavas dredged from Pual Ridge and vicinity (Binns et al., 2002b). The topmost fresh lavas intersected in cores compare closely with dredged seafloor exposures at Sites 1188, 1189, and 1191 (Fig. F3). Occasional rhyodacites dredged near Roman Ruins are absent from Hole 1189A and so must represent a younger flow or flow sequence relative to the dominant 64.9 wt% SiO2 outcrops of that vicinity. The youngest "Tsukushi group" of glassy lavas (67.5 wt% SiO2; open squares in Fig. F3) was not intersected in Hole 1188A, although it occurs overflowing altered outcrops in the vicinity.

Shipboard porosity measurements on samples from the fresh lava cap in Holes 1188A, 1190B, and 1191A range from 0.4% to 11.5%, less porous than underlying altered rocks. The measured core-scale permeabilities of samples from Hole 1188A and Hole 1191A (7.0 x 10–15 and 1.8 x 10–16 m2, respectively), however, fall above or in the high range for altered rocks (Christiansen and Iturrino, this volume). Following X-ray computed tomography (CT) studies of the Hole 1188A specimen, Ketcham and Iturrino (2005) consider the core-scale permeability measurement unreliable because the sample was too small and state that whereas vesicles provide porosity they are disconnected and the rock is effectively impermeable. Formation Micro Scanner (FMS) resistivity imagery was not conducted for any intersection of fresh volcanic rocks; hence, any differences in fracture patterns or intensity compared to altered rock intersections cannot be assessed. Whereas intuitively the fresh lavas directly below the seafloor appear to constitute an impervious cap to the hydrothermal system (otherwise venting would be far more widespread?), this remains unconfirmed by hard data.

An issue of importance to modeling the evolution of the PACMANUS hydrothermal system is whether the fresh lavas in Holes 1188A and 1189A pass abruptly or gradually into altered rocks below. Might they, for example, overlie the alteration system unconformably (as the "Tsukushi group" lavas do at Snowcap), or were they present prior to alteration to become preserved as relatively impervious caps to the system? If the fresh lavas were indeed erupted after formation of the drilled alteration system, the likelihood of repeated cycling between hydrothermal and volcanic events during the construction of Pual Ridge becomes enhanced and major uncertainties would be introduced regarding relationships between the present-day chimney-forming hydrothermal fluids at PACMANUS and those responsible for the underlying zone of extensive alteration and limited mineralization revealed by Leg 193 drilling.

Low core recoveries prevent a conclusive answer to the question. Based on changes in megascopic appearance and relative distribution of opaline silica and cristobalite, progressive downhole grading from largely unaltered lavas to pervasively altered rocks was inferred on board ship in both Hole 1188A (Shipboard Scientific Party, 2002b) and Hole 1189A (Shipboard Scientific Party, 2002c). Critical samples covering the transition, however, require detailed study to fully exclude any possibility, considered remote, of an "alteration unconformity." At both sites the transition evidently occurs within a few meters or less.