BACKGROUND AND OBJECTIVES

The western margin of the Pacific plate displays numerous convergent segments or subduction zones, most of which, over the past two decades, have been shown to exhibit seafloor hydrothermal activity at one or more sites in their vicinity (Fig. 1). The first place where hydrothermal "chimney" deposits and associated vent fauna were discovered, other than on a mid ocean spreading axis, was in the Manus Basin in the Bismarck Sea north of Papua New Guinea (Both et al., 1986). This was at a site now called Vienna Woods on the basaltic Manus spreading center, near the apex of a wedge of backarc oceanic crust (Fig. 2). By contrast, the Eastern Manus Basin has a more complex geological construction involving the creation of continental-type crust, and the Manus Basin accordingly shows closer affinities to ancient orebody settings. It contains the PACMANUS (Papua New Guinea-Australia-Canada-Manus) hydrothermal field, discovered in 1991 (Binns and Scott, 1993), where the host volcanic sequence is conspicuously siliceous. Now thoroughly surveyed at the seafloor surface, PACMANUS is the site where the first subsurface study of an active felsic-hosted convergent margin hydrothermal system was conducted during Leg 193. As anticipated, we found significant differences between this site and hydrothermal activity hosted by mafic volcanic rocks and sediments on divergent margins (seafloor spreading axes) previously drilled during Ocean Drilling Program (ODP) Leg 158 to the TAG hydrothermal area on the Mid-Atlantic Ridge and Legs 139 and 169 in the northeast Pacific, respectively. The differences are profoundly important in understanding chemical and energy fluxes in the global ocean, as well as for understanding mineral deposit geology.

The region that includes the Manus Basin is a highly mineralized sector of the Earth's crust. Notable hydrothermal ore deposits of Neogene to Quaternary age in Papua New Guinea include the Panguna porphyry copper-gold deposit on Bougainville Island, the Ladolam epithermal gold deposit in the Tabar-Lihir-Tanga-Feni island chain north of New Britain, and, on the mainland, the Porgera polygenetic gold deposit and the Ok Tedi and Grasberg porphyry copper-gold deposits. Combining these with modern mineralizing systems like PACMANUS, such fertility hints at an underlying cause, perhaps the presence of geochemically anomalous mantle lithosphere as the source of metals concentrated by hydrothermal processes in both continental and marine settings. The results of Leg 193 have added to our understanding of the processes involved in such ore formation. Although focused on the submarine environment, the knowledge gained will also have application to subaerial and transitional ore environments.

Regional Setting
The Manus Basin is a rapidly opening (~10 cm/yr) backarc basin set between opposed fossil and active subduction zones (Manus Trench and New Britain Trench, respectively) within a complex zone of oblique convergence between the major Indo-Australian and Pacific plates (Fig. 2). Adjacent to the now-inactive Manus Trench or its antecedent, volcanism above an Eocene–Oligocene intraoceanic subduction zone within the Pacific plate (or at its boundary with an oceanic portion of the Indo-Australian plate, now represented by the Solomon microplate) formed an island arc represented by exposures of this age on New Ireland, New Hanover, Manus, and parts of New Britain (e.g. Hohnen, 1978; Stewart and Sandy, 1988). Paleomagnetic measurements (Falvey and Pritchard, 1985) indicate that these islands have been relocated to their present positions by an imperfectly understood sequence of backarc developments (Exon and Marlow, 1988). During the late Miocene or Pliocene, when arrival of the Ontong Java Plateau blocked subduction at the Manus Trench, convergence switched to the New Britain Trench. Here the Cretaceous oceanic Solomon microplate is moving under what is now the South Bismarck microplate (a unit separated from the Pacific plate by more recent backarc processes). Above the north-dipping Wadati-Benioff Zone associated with the New Britain Trench, a chain of young arc volcanoes has formed along the concave northern side of New Britain (Bismarck or New Britain Arc; Johnson, 1976).

The present-day configuration of spreading segments, extensional faults, and obliquely oriented transform faults in the Manus Basin (Fig. 3) is well established by bathymetric, sidescan sonar, seismic reflection, gravity, and magnetic surveys (Taylor, 1979; Taylor et al., 1991), and by microseismicity (Eguchi et al., 1989), which defines left-lateral movement on the transform faults. In contrast to the wedge-shaped Manus spreading center, where new backarc oceanic crust has been forming since the 0.78-Ma Brunhes/Matuyama boundary (Martinez and Taylor, 1996), the rift zone of the Eastern Manus Basin lying between the islands of New Ireland and New Britain, and between two major transform faults (Djaul and Weitin Faults), is a pull-apart zone of distributed extension on mostly low-angle faults approximately normal to the transforms. Martinez and Taylor (1996) infer ~80 km of extension across a 150-km-wide rift zone, mostly concentrated in the bathymetrically deeper portion of thinned crust (Fig. 4), which is coincident with an isostatic gravity high. They argue that this amount of extension is equivalent to that accomplished by a combination of backarc spreading and microplate rotation in the central portion of the Manus Basin (Fig. 3). Bathymetry, gravity modeling, and reverse magnetization indicate that basement of the Eastern Manus Basin (called the Southeastern Rifts by Martinez and Taylor, 1996) is arc crust equivalent to the Eocene–Oligocene exposures on New Britain and New Ireland. Reflection seismic traverses across the Eastern Manus Basin (Fig. 5A; also Fig. 5B. Taylor and K.A.W. Crook, unpubl. data) show essentially undeformed graben and half-graben fills up to 0.3 s, equivalent to ~1 m.y. at current sedimentation rates. This is consistent with rifting in the Eastern Manus Basin covering a similar duration to spreading on the central Manus spreading center. The sediment fill is commonly tilted, denoting block rotation on listric master faults. The dredging of fault scarps, where seismic profiles indicate exposure of lower, more deformed sequences, has yielded fossiliferous calcareous mudstones and volcaniclastic sandstones ranging in age from early Miocene to the Pliocene/Pleistocene boundary. Although mainly of deeper marine origin, these are contemporaneous with the Miocene Lelet Limestone and Pliocene Rataman Formation, which overlie the Eocene–Oligocene Jaulu Volcanics of New Ireland (Stewart and Sandy, 1988), and with equivalent sequences on New Britain. Undated, mildly metamorphosed basalts dredged from inner nodal scarps near the active ends of the Djaul and Weitin transform faults may represent the presumed arc volcanic basement.

Built up on this nascent continental crust, and probably controlled by subtle, relatively recent changes in the extensional stress field, a series of high-standing neovolcanic edifices (eastern Manus volcanic zone; Binns and Scott, 1993) extends en echelon across the trend of the rift faults (Fig. 6). Because these edifices do not significantly disturb the negative regional magnetization derived from basement, they are considered to be superficial features (Martinez and Taylor, 1996). The neovolcanic edifices range from central eruptions of more mafic lavas (basalt and basaltic andesite) to linear ridges formed by fissure eruption of andesite, dacite, and rhyodacite. The westernmost volcanic feature of Figure 6 is a low axial ridge of basalts with compositions resembling mid-ocean-ridge basalt (MORB), set within a deep trough (Fig. 4). This is probably a failed spreading center. The other edifices, however, are variably but distinctly potassic. They have trace element and isotopic affinities comparable with the subaerial arc volcanoes of New Britain (Binns et al., 1996b; Woodhead and Johnson, 1993) rather than with the MORBs at the Manus spreading center (Woodhead et al., 1998), the Willaumez extensional transform fault (Taylor et al., 1994), and in the adjacent East Sherburne volcanic zone (R. Binns, unpubl. data) (Fig. 3). The eastern Manus volcanic zone appears to be a submarine segment of the New Britain arc, displaced from the main subaerial chain and erupted in the rifted backarc region.

The region is one of active seismicity, especially on transform faults. The major magnitude 8.2 earthquake of 16 November 2000 was typical. It occurred 25 km deep under the western active end of the Weitin Fault (Fig. 6), causing disruption in Rabaul and a consequent delay to commencement of drilling operations for Leg 193. Tregoning et al. (1998) measured 13 cm/yr of sinistral transcurrent movement on the Weitin fault, which is a major plate boundary.

The PACMANUS hydrothermal field targeted by Leg 193 is located near the crest of Pual Ridge, a 500- to 700-m-high felsic neovolcanic ridge with negligible sediment cover that trends northeast (Fig. 4, Fig. 6). This ridge is externally terraced and appears constructed of stacked subhorizontal lava flows 5–30 m thick, with negligible to minor sediment cover along the crest. Dacite and some rhyodacite block lavas with rough surface topography predominate, but there are also some smoother sheet flows and lobate flows of dacite (Waters et al., 1996). Consanguineous lobate flows of andesite occupy the lower reaches of Pual Ridge, whereas the 2200-m-deep valley to its east is floored by lobate flows of basaltic andesite (Fig. 6).

PACMANUS Hydrothermal Field
Isolated hydrothermal deposits are scattered for at least 13 km along the main crestal zone of Pual Ridge (Binns and Scott, 1993; Binns et al., 1995, 1996a, 1997a, 1997b). The more significant, active deposits are in the center of this zone between two low knolls on the ridge crest (Fig. 7). Lavas in this central area are exclusively dacitic to rhyodacitic (65–71 wt% SiO₂). Four principal fields of hydrothermal activity, including sulfide chimneys, and several smaller sites have been delineated and named. Much of the information cited below is currently unpublished and is derived from cruises listed in the caption of Figure 7.

Roman Ruins (1693–1710 m below sea level [mbsl] and 150 m across) contains many close packed simple columnar chimneys that stand as high as 20 m and some complex multispired chimneys with numerous conduits. Commonly, these coalesce into wall-like constructions oriented north-south. Many chimneys are broken (seismic effects?) and some show later regrowth. Fallen chimneys form a 10-m-high pediment for the active structures including black smokers and diffuser-style chimneys that emit clear fluid. A smaller, deeper (1730–1740 mbsl) field to the north, Rogers Ruins, is linked to Roman Ruins by a zone of Fe oxyhydroxide deposits. Numerous small occurrences of the latter and of Mn oxides are common throughout the PACMANUS field.

Satanic Mills (1708–1720 mbsl and 200 m across) is an equivalent-sized field of rather more scattered chimneys and spires marked by clouds of dark particulate-rich hydrothermal fluid ("black smoke"). Both black to gray smokers and structures that are vigorously venting clear fluid are in close proximity. East of this field are north-south oriented fissures in dacite, encrusted with fauna, that are emitting clear fluid and are interpreted as juvenile vents soon to become smoker fields. To its south, Satanic Mills is linked by a zone of altered dacite with diffuse venting and scattered Fe-Mn oxide deposits to the smaller Marker 14 field, which, at 1745 mbsl depth, is the deepest hydrothermal site so far recognized at PACMANUS. Deflections of bathymetric contours beyond both the Roman Ruins-Rogers Ruins and Satanic Mills-Marker 14 lines suggest that both fields are located on north-northwest-trending fracture zones.

The Tsukushi field (1680–1686 mbsl) at the southwestern end of PACMANUS contains numerous actively venting chimneys as high as 30 m, many very slender but some as much as 10 m in diameter. This area was traversed by bottom camera in 1993 and by a Shinkai-6500 submersible dive in 1995 without sighting chimneys. However, chimneys were discovered during a 1996 Shinkai-2000 submersible dive, and additional large chimneys were present in 1998. Hence, it might be very young. Fe oxyhydroxide and Fe-Mn oxide crusts are common in the zone that extends northeast from Tsukushi.

Snowcap Knoll (1654–1670 mbsl), the other major active hydrothermal site at PACMANUS, is very different in character. It occupies the crest and the flanks of a 10- to 15-m-high hill, 100 m x 200 m in size, bounded on its eastern side by a north-northeast-striking fault scarp 60–80 m high. Outcrops of altered dacite-rhyodacite lava and hyaloclastite predominate, locally covered with patches of both sandy sediment and metalliferous hemipelagic ooze (only millimeters thick). Gravity corer and grab operations revealed the sand to be altered lava disaggregated by bioturbation or hydrothermal fragmentation. Typical alteration assemblages at Snowcap Knoll are dominated by cristobalite, with lesser natroalunite, diaspore, and illite-montmorillonite with traces of pyrite, marcasite, chalcopyrite, enargite, and formerly molten globules of native sulfur. These reflect interaction at relatively low temperature between dacites and a highly acid, relatively oxidized hydrothermal fluid ("advanced argillic alteration"), indicating that SO₂-bearing magmatic components were present in the fluid.

Diffuse low-temperature venting (6°C, compared with 3°C in ambient seawater) is extensive across the gently undulating to flat crest of Snowcap Knoll. More intense shimmer occurs at the edges of the occasional Mn-encrusted outcrop of altered dacite. The diffuse vent sites are marked by white surficial patches that probably include both microbial mat and methane hydrate deposits. Around the southwestern fringes of Snowcap Knoll are several small fields of actively smoking and inactive chimneys aligned in north-south trending clusters.

Fluid temperatures measured at the orifices of black or gray smokers and sulfide chimneys venting clear fluid are comparable for the Satanic Mills, Roman Ruins, and Tsukushi fields, ranging between 220° and 276°C. End-member vent fluids are very acid (pH = 2.5–3.5), show high K/Ca ratios reflecting equilibration with dacite wallrocks, are high in Mn and Fe relative to mid-ocean-ridge fluids, and have variable salinities (Gamo et al., 1996; Auzende et al., 1996; Charlou et al., 1996). The variable salinities imply subsurface phase separation, meaning hydrothermal temperatures exceed 350°C at indeterminate depths below the chimney fields. This is supported by mineralogical evidence of multiple fluid compositions in chimney assemblages (Parr et al., 1996). End-member gas compositions of 20–40 mM CO₂, 20–40 µM CH₄, and R/RA(He) = 7.4 denote significant contribution to the hydrothermal fluids from arc-type magmatic sources (Ishibashi et al., 1996). Douville et al. (1999) ascribe unusually high fluorine contents in the fluids to magmatic sources. Temperatures of 40° to 73°C have been measured in
the shimmering clear fluid emitted from Fe oxyhydroxide deposits in the Tsukushi-Snowcap Knoll zone.

A very high thermal gradient of 15°C/m was measured at a sediment pocket on Snowcap Knoll adjacent to a 6°C shimmering water zone. Fluids collected near this by a funnel sampler are close to seawater in composition but are enriched in Mn, Fe, and Al. All outcrops of altered dacite at Snowcap Knoll are encrusted by Fe-Mn oxide.

Chimneys collected from Roman Ruins and Satanic Mills are comparatively rich in precious metals (average Au = 15 ppm; average Ag = 320 ppm) and are composed predominantly of chalcopyrite and sphalerite, with subsidiary pyrite, bornite, tennantite, galena, and dufreynosite (Scott and Binns, 1995; Parr et al., 1996). Barite is the principal gangue, but anhydrite substitutes in some samples. Chimneys at Roman Ruins typically contain less Cu than those at Satanic Mills. Fewer samples have been recovered from Tsukushi and the southwestern side of Snowcap Knoll, but these are virtually devoid of Cu and Au and contain more Pb and Ag. Their gangue includes appreciable amorphous SiO₂ as well as barite.

PACMANUS chimneys have elevated contents of "magmatophile" trace elements (e.g., As, Sb, In, Tl, and Te). Sulfur isotope ratios near zero per mil ³⁴S (Gemmell et al., 1995; 1996) indicate a larger magmatic-sourced component than found at mid-ocean-ridge hydrothermal sites and mature backarc spreading axes. Direct evidence for the importance of magmatic fluids is found in the Cu + Zn-rich gas-filled cavities within glass melt inclusions in the phenocrysts of Pual Ridge andesites (Yang and Scott, 1996), as well as in the gas compositions of collected vent fluids (see above).

The PACMANUS hydrothermal field supports an exceptionally abundant vent macrofauna broadly similar to those of other southwest Pacific sites (Hashimoto et al., 1999), dependent on chemosynthetic microbes. At Snowcap Knoll, dredged samples of altered dacite possess microscopic tube worms (unidentified species) along internal hairline fractures. These, and their presumed symbiotic microbes, were the first indication of the presence of a subsurface biosphere that was also investigated during Leg 193.

Scientific Objectives
The overall aims of Leg 193 were to delineate the subsurface volcanic architecture, the structural and hydrologic characteristics, the deep-seated mineralization and alteration patterns, and the microbial activity of the PACMANUS hydrothermal field. From these data and subsequent laboratory analyses of samples and structural data, we planned to pursue the following specific scientific objectives.

1. Quantify the manner in which fluids and metals derived from underlying magmatic sources, and from leaching of wallrocks by circulated seawater, respectively, have combined within the PACMANUS hydrothermal system. This would be approached by applying geochemical and isotopic modeling to the vertical and lateral variations in hydrothermal alteration styles and sulfide mineral occurrences established by the drilling.
   
2. Evaluate the mechanisms of subsurface mineral precipitation, including comparison of exhalative and subhalative mineralizing processes, assessing the consequences of fluid phase separation, and seeking explanations for the elevated contents of copper, zinc, silver, and gold in massive sulfide chimneys at the PACMANUS seafloor.
   
3. Delineate probable fluid pathways within the system and establish a hydrologic model by measuring and interpreting variations in physical properties and fracture patterns of fresh and altered bedrocks.
   
4. Test whether the volcanic construction of Pual Ridge is a simple "layer cake," with potential older exhalative or subhalative massive sulfide horizons concealed by younger lavas or, alternatively, whether inflation of the volcanic edifice by lava domes or shallow intrusions is the predominant process in this submarine felsic volcanic environment.
   
5. Develop a petrogenetic model for Pual Ridge igneous rocks and seek evidence pertaining to the nature of the possible underlying source for magmatic components in the hydrothermal fluids.
   
6. By combining the above models, develop an integrated understanding of coupling between volcanological, structural, and hydrothermal phenomena in the PACMANUS system for comparison with equivalent hydrothermal phenomena at mid-ocean ridges and to provide a new basis for interpreting ancient ore environments.
   
7. Establish the nature, extent, and habitat controls of microbial activity within the hydrothermal system and interpret the differences encountered in diversity and biomass in terms of nutrient supplies and environmental habitats in the context of the geochemical and hydrologic understanding of the total hydrothermal system.