From only two deeply drilled sites, it is clearly not possible to develop conclusive three-dimensional hydrologic models for the subseafloor PACMANUS hydrothermal system, yet valuable constraints have been gained from new measurements of physical properties and a variety of postleg isotopic research enabling assessment of fluid sources and mixing behaviors. In addition, aspects of research on precursor volcanic facies, alteration chemistry, and borehole imagery also relate to understanding fluid pathways.

Porosity, Permeability, Fracturing, and Fluid Flow

From measurements of porosity and intrinsic permeability on representative minicore samples from Sites 1188 and 1189, Christiansen and Iturrino (this volume) find porosity and permeability of totally altered rocks to be well correlated, both decreasing with depth from ~30% and ~10–16 m2, respectively, near the seafloor to ~15% and ~10–18 m2 below 300 mbsf. Although ranges overlap, the average permeability of PACMANUS samples is two to four orders of magnitude higher than other seafloor sites where this has been measured, most of which, however, are in basaltic crust with less pronounced alteration.

From X-ray CT studies, Iturrino et al. (this volume) and Ketcham and Iturrino (2005) examine processes associated with porosity and permeability development during alteration, noting that alteration may itself generate microfracturing and thereby increase fluid flow. These studies create new understandings of fluid behavior during hydrothermal alteration that could not be gained from investigating metamorphosed hydrothermal settings in ancient sequences. However, the similarity of ranges in core-scale permeability measured at Sites 1188 and 1189 indicate that other factors govern the contrast between low-temperature diffuse venting at Snowcap and high-temperature focused venting at Roman Ruins. Applying properties of seawater, a thermal gradient corresponding to borehole temperature measurements, and a permeability of 10–16 m2 in a simple linear model, Christiansen and Iturrino (this volume) calculate a diffusive fluid flow velocity through coherent rocks of <2 cm/yr. This is grossly deficient for explaining observed vent rates at Roman Ruins and most probably at Snowcap as well, so they conclude that focused fluid flow within open fractures is the predominant mechanism at both sites.

Bartetzko et al. (2006) provide electrical conductivity measurements on seawater-saturated minicores from a range of altered lithologies at Sites 1188 and 1189, concluding that porosity and pore space structure (tortuosity) are the main controls on conductivity, with connectivity of sulfide minerals important in some cases. Significantly, they observed that downhole geophysical logging during Leg 193 indicates substantially higher conductivities than those measured on the minicore samples. The logging profiles integrate much larger volumes and thus reinforce the importance of fractures dominating fluid flow regimes. From resistivity imagery of borehole walls, Iturrino and Bartetzko (2002) describe a predominance of subvertical fractures striking north-northeast, suggesting a possible tectonic control on fluid pathways.

Variations in the parental fabrics and precursor compositions of the altered sequence below Pual Ridge are unlikely to have a major influence on fluid flow patterns, except at the very early stages of hydrothermal alteration. Paulick and Herzig (2003) suggest that higher primary porosity and permeability of volcaniclastic horizons would focus fluid flow and facilitate higher fluid/rock ratios during alteration. Reflecting this, they measured significant silica loss from altered dacite clasts in volcaniclastic units, balanced by gains in the matrix, recalling Category Z alteration in the Stockwork Zone of Hole 1189B.

Strontium and Sulfur Isotope Ratios in Anhydrite

The use of strontium isotope ratios in sulfate minerals is a well-established tool for fingerprinting fluid sources and mixing in active seafloor hydrothermal systems (Tivey et al., 1995; Teagle et al., 1998). Analyses of 87Sr/86Sr in anhydrites filling veins, breccia matrixes, and vesicles from Sites 1188 and 1189 are presented by Roberts et al. (2003) and Bach et al. (this volume). Interpretations rely on knowing the 87Sr/86Sr ratios of seawater (0.7092), of fresh Pual Ridge lavas (0.7036 throughout the fractionation series), and of high-temperature end-member vent fluid at PACMANUS (0.7050, extrapolated from Douville et al., 1999).

Vein anhydrites to 70 mbsf in the upper cristobalite domain of Hole 1188A have distinctly radiogenic 87Sr/86Sr (0.7067–0.7086). Below 87 mbsf in the cristobalite domain and continuing through the underlying quartz domain to 364 mbsf, 87Sr/86Sr varies between 0.7050 and 0.7070 with no clear downhole trend (Fig. F11). In Hole 1189A there is a possibly coincidental difference between veins and breccias: anhydrites from veins span a similar range of bulk 87Sr/86Sr ratios (0.7052–0.7072) to the 87- to 364-mbsf section at Site 1188, whereas all but one of those from breccias are more radiogenic (87Sr/86Sr = 0.7073–0.7077) whether from the cristobalite or the quartz domains (Fig. F11). In the Stockwork Zone of Hole 1189B, however, breccia and vein anhydrites have overlapping 87Sr/86Sr ratios ranging from 0.7062 to 0.7076. In the Lower Sequence, anhydrite veins in cristobalite-bearing samples have 87Sr/86Sr ratios from 0.7069 to 0.7079, more radiogenic mostly than equivalents from quartz-bearing samples (0.7055–0.7071).

Roberts et al. (2003) and Bach et al. (this volume) ascribe the 87Sr/86Sr data to variable mixing between seawater and a high-temperature hydrothermal end-member (87Sr/86Sr = 0.7050) in the fluids that deposited the anhydrites. They calculate proportions of admixed seawater varying from 89% for the uppermost anhydrites (cristobalite domain) to an average of ~30–40 wt% in the quartz domains. Significant fluctuations around these values at both sites, however, imply a very heterogeneous mixing system, a conclusion supported by ion microprobe evidence of considerable 87Sr/86Sr variability in individual anhydrite veins (Craddock and Bach, 2004). Where evidence exists for timing (e.g., multiple vein fillings) the evolutionary trend is generally toward higher indicated seawater proportions. In the Lower Sequence of Hole 1189B, veins in cristobalite-bearing alteration assemblages formed from fluids with a higher proportion of admixed seawater than those from quartz-bearing intervals.

Paulick et al. (this volume) provide 87Sr/86Sr analyses of disseminated and veinlet anhydrite leached from a range of altered rocks at Sites 1188 and 1189, together with analyses of the corresponding bulk rocks from which anhydrite was removed. At both sites the majority of anhydrite-free altered rocks ("silicate" fractions in Fig. F11) have 87Sr/86Sr ratios (0.704–0.706) that span the inferred end-member vent fluid ratio and are more radiogenic than those of glassy Pual Ridge lavas, comparing closely with altered volcanic outcrops at the Snowcap seafloor (0.7044–0.7052; CSIRO data). The silicate fraction of altered rocks in the cristobalite domain of Hole 1188A, and a few at deeper levels, have even more radiogenic Sr (87Sr/86Sr as high as 0.7076). Variable degrees of exchange between igneous, hydrothermal, and seawater strontium have clearly occurred within the altered rocks, mostly with a net loss of total Sr (see Table T3). Disseminated anhydrites (the "leachates" of Fig. F11) at Site 1188 have 87Sr/86Sr ratios that correspond closely to those of silicate components of their host altered rocks, rather than to nearby vein or breccia anhydrites. At Site 1189 the reverse is the case (Fig. F11). If disseminated anhydrites at Site 1188 reflect permeation of the same fluids that deposited anhydrite veins, then there has been considerable exchange with the host rocks, but at Site 1189 they appear more directly cogenetic with vein or breccia anhydrites.

Barites and rare early formed anhydrites in PACMANUS chimneys have a limited range of 87Sr/86Sr from 0.7046 to 0.7054. Anhydrite more often occurs as a late-stage chimney mineral in transgressive veinlets or cavity fillings where 87Sr/86Sr varies from 0.7069 to 0.7090 (Binns et al., 2002b; R. A. Binns et al., unpubl. data; Kim et al., 2004). In shallow CONDRILL semimassive sulfides (<2 mbsf), early formed barites (87Sr/86Sr = 0.7046–0.7059), mid-stage anhydrite (87Sr/86Sr = 0.7076), and late-stage anhydrites (87Sr/86Sr = 0.7068–0.7080) define progressive dilution of hydrothermal fluids by seawater during continuing dilation and mineral deposition. Pulses of fluid with higher "hydrothermal" component followed by progressive dilution with seawater appear a common aspect of the PACMANUS system generally.

Sulfur isotope characteristics of anhydrite differ between Snowcap and Roman Ruins (Roberts et al., 2003). Apart from two veins in the uppermost cristobalite domain (+21.6 and +21.7), anhydrite 34S values at Site 1188 range from +18.1 to +21.1, below that of seawater (21.0; Rees et al., 1978). Anhydrite 34S at Site 1189 mostly ranges from +20 to +22.3 with the majority above the seawater value. Roberts et al. (2003) ascribe the contrast to disproportionation of a magmatic SO2 component at Snowcap and partial reduction of seawater sulfate by wallrock ferrous iron at Roman Ruins.

Sulfur Isotope Ratios in Pyrite

Conventional sulfur isotope analyses of hand-picked grains (Roberts et al., 2001; R.A. Binns and A.S. Andrew, unpubl. data) and laser ablation microanalyses of ultra-thin sections (R.A. Binns and J.B. Gemmell, unpubl. data) yield comparable results for coarser pyrites from veins, vesicles, breccia matrixes, and semimassive sulfides from both Site 1188 and Site 1189. The overall range in 34S, from –0.7 to +5.2 (mostly +2 to +4), overlaps that of sulfides from PACMANUS chimneys (–2.1 to +7.5; average at Snowcap = +0.3, Satanic Mills = +0.4, and Roman Ruins = +3.0; Gemmell et al., 1996; Binns et al., 2002b).

By contrast, finely disseminated pyrites in altered volcanic rocks at Site 1188 (kernels and transitional zones) range to higher values of 34S (+1.0 to +13.7). Those in altered wallrock clasts within breccias from the Stockwork Zone of Hole 1189B range from +0.3 to +10.9. The more elevated 34S values suggest substantial contributions from seawater sulfate reduced by ferrous iron contained within altered or altering volcanic rocks.

Fluid Flow Models

Adopting the concept of single hydrothermal events at each site, Roberts et al. (2003) use strontium and sulfur isotope data for vein, breccia, and vesicle anhydrites to propose separate upwelling zones below each drilled vent field and describe contrasted fluid evolutions at Snowcap and Roman Ruins involving sulfate-rich low-pH and sulfate-poor high-pH fluids, respectively. At each site they envisage introduction of seawater from local shallow circulation systems, whereas a deeper circulation and mixing system surrounds the entire PACMANUS site. A genetic connection between anhydrite veins and wallrock alteration is assumed, and to account for the presence of pyrophyllite at Snowcap only, they introduce to this system a component of magmatic volatiles from depth from which SO2 disproportionation causes acid sulfate conditions.

Lackschewitz et al. (2004) develop a different model to explain their oxygen isotope temperature profiles for clay minerals and quartz, particularly the apparently higher temperatures evident in the upper pyrophyllite interval of Hole 1188A and also in the Stockwork Zone of Hole 1189B. This model involves advective heating of the Pual Ridge sequence and progressive cooling through admixture of seawater within separate high-temperature fluid upwelling zones located to the sides of each of Sites 1188 and 1189. The upwelling zones spread laterally on encountering the fresh lava cappings in each case. At both sites drill holes penetrated the fringes of the ponded fluid bodies, but at Snowcap Hole 1188F continued to drill through a small-scale descending eddy beside the main upwelling zone.

Paulick and Bach (2006) suggest that abundance of fragmental volcanic facies below Roman Ruins favors development of secondary circulation cells dominated by seawater to the sides of a constrained conduit with focused flow of high-temperature vent fluid. For Snowcap they propose that the predominant coherent lava flows hinder drawdown of seawater in secondary cells and that relatively less modified hydrothermal fluids flowing in a nonfocused manner become ponded directly below the unaltered and impermeable dacite cap, causing pyrophyllite-bearing acid sulfate alteration. In a different explanation for opposed venting characteristics between the two sites, Vanko et al. (2004) suggest that an anhydrite-cemented shell surrounding the Roman Ruins upflow zone isolated it from seawater ingress.

Looking at finer details, Roberts et al. (2003) and Vanko et al. (2004) draw attention to discrepancies between temperatures indicated by fluid inclusion studies and those calculated for isenthalpic mixing of cold seawater and a hypothetical 250C hydrothermal fluid in ratios indicated by the Sr isotope data. Various scenarios of conductive heating of seawater or conductive cooling of the ascending high-temperature hydrothermal fluid prior to their mixing, advanced to explain relationships at the respective sites, may need reassessment following recent measurements of much higher temperatures at PACMANUS vents (as high as 356C; Seewald et al., 2006; Tivey et al., 2006).

Such models rely on selective data sets and would likely change if many more holes were to be drilled at PACMANUS. They overlook the issue of phase separation and uncertainties in the relative timing of different alteration styles or between alteration, veining, and mineralization. Nevertheless, the models clearly demonstrate considerable complexity of fluid pathways within the subseafloor hydrothermal system at PACMANUS, and processes such as heating by advective fluid circulation. They point the way toward new approaches to interpreting ancient massive sulfide ore bodies and their environments.

Source of Chimney Metals: The Role of Magmatic Volatiles

Figure F12 is a cartoon originally drawn to illustrate the conceptual background and objectives of Leg 193 in its precruise Scientific Prospectus (www-odp.tamu.edu/publications/prosp/193_prs/193toc.html). In a broad sense it still illustrates the general fluid flow pattern believed to occur below Snowcap, ignoring the numerous complexities in detail arising as described above from the drilling and subsequent research, such as advective thermal and chemical changes associated with small-scale fluid cells. Excess fluid pressures indicated by volume expansion during pervasive alteration, evidence for phase separation or boiling at least within vein systems, and presence of an almost continuous impervious capping of unaltered lava below the seabed constitute unforeseen outcomes of the leg that collectively form the basis of a "PACMANUS pressure cooker" model in which episodic breaching of the cap leads to development of individual chimney fields at the seafloor. In the precruise hypothesis, magma-derived metal-rich fluids rising from a subjacent intrusion have become increasingly mixed at all levels with circulating seawater. That mixing happens in the higher drilled part of Pual Ridge is unequivocally established by results presented above. That it also happens at deeper levels than explored by drilling is required by the more radiogenic nature of the modeled high-temperature end-member fluid of the upper system (87Sr/86Sr = 0.7050) relative to the highly constrained Sr isotope ratio of Pual Ridge glassy lavas (87Sr/86Sr = 0.7036), provided the intrusion and lavas are cogenetic (87Sr/86Sr varies little across the various neovolcanic edifices of the eastern Manus Basin) and, of course, depending on the confidence assigned to the magmatic fluid concept itself. It is therefore appropriate to conclude the geological part of this synthesis with a review of the evidence favoring that concept and its implied importance in governing the high Cu, Zn, Ag, and Au tenors of PACMANUS chimneys.

Fluid bubbles in melt inclusions within phenocrysts from lavas at various eastern Manus Basin volcanic edifices including Pual Ridge contain high levels (weight percent range) of Cu and Zn (Yang and Scott, 1996). Treating these bubbles as a proxy for the behavior of large magma bodies, Yang and Scott (1996, 2002, 2005) extensively investigate degassing of lavas or devolatilization of subjacent magma bodies as feasible mechanisms for introducing magma-derived metals into seawater-dominated hydrothermal fluids. In a separate study, abrupt decreases in Cu and Au contents at the basaltic andesite stage of fractionation in east Manus lavas recorded by Sun et al. (2004) are attributed to their removal into exsolved fluids as a consequence of redox effects associated with initiation of magnetite precipitation. The process is more likely to apply in water-rich, relatively oxidized island arc magmas as distinct from those associated with mid-ocean or backarc spreading. Only small volumes of exsolved magmatic fluids need adding to dominant seawater-derived hydrothermal fluids in order to sufficiently enrich the overall mixture in the metals that ultimately deposit in seafloor chimneys.

Adding to earlier geochemical and isotopic interpretations for PACMANUS chimneys and vent fluids (see "Geological Setting"), Leg 193 has contributed further indirect evidence for involvement of magmatic fluids in the hydrothermal system. At Snowcap, sulfur isotope ratios in vein and breccia anhydrites and the presence of pyrophyllite-bearing intervals of hydrothermal alteration are interpreted as consequences of SO2 disproportionation and thus of magmatic volatiles in the hydrothermal fluid (Roberts et al., 2003). Relatively light sulfur isotope ratios in pyrites from veins and semimassive sulfides bear the same connotation, although relative to other sites (e.g., Lau Basin; Herzig et al., 1998) there may be a higher proportion of reduced seawater sulfate at PACMANUS. The nature of rare earth element (REE) patterns in anhydrites from Snowcap implies input of magmatic hydrofluoric acid as well as SO2 (Bach et al., 2003).

The alternative to a magmatic source for base and precious metals is leaching during alteration of subseafloor sequences traversed by seawater-dominated hydrothermal fluids. Below Snowcap, some altered volcanic rocks are depleted in Cu, Zn, and Pb relative to their precursors, so this is also a feasible process. However, except for Zn and Cd, the overall abundances of chalcophile elements (including "magmatophile" trace elements) in altered wallrocks, particularly at Roman Ruins, discount significant contribution of these elements to chimney-forming hydrothermal fluids during alteration of the volcanic sequence drilled during Leg 193. Barium—by far the dominant gangue element of chimneys at Roman Ruins—is also conspicuously enriched in underlying altered rocks at Site 1189. Derivation of Cu and associated elements by alteration at deeper levels including basement can not be discounted, but for As and Sb in particular the enrichments in both chimneys and overall in the drilled subseafloor sequence appear difficult to explain in this way. Slight differences in Pb isotope ratios of glassy Pual Ridge lavas and subseafloor or chimney sulfides at Roman Ruins suggest that Pb in the latter is derived in part from deep-seated reaction of hydrothermal fluids with radiogenic basement (Binns, this volume).

Had we drilled deeper, or if, as initially expected, the vertical sequence from intrusion to high-temperature reaction zone to upper crustal hydrothermal system were more condensed than proved the case, more conclusive evidence for the magmatic component might have been gained. Among the initial objectives of Leg 193 we also hoped to find explanations for the particularly high Cu, Zn, Ag, and Au contents of PACMANUS chimneys. In the event, scarcity of subsurface mineralization in the cores recovered has prevented assessment of potential processes such as phase separation or "zone refining." Subseafloor precipitation, predominantly of pyrite in altered rocks, veins, and breccias, may have depleted the hydrothermal fluids in Fe. In the semimassive sulfide at 31 mbsf in Hole 1189B and similar material from shallow CONDRILL cores, pyrite is the first-formed sulfide and is followed by chalcopyrite then sphalerite, suggesting the possibility of a "fluid fractionation" process leading to Cu and Zn enrichment (Binns, 2005). Such a process is an alternative to "zone refining" that involves dissolution, upward transport, and reprecipitation of elements like Cu, but it is unlikely to be uniquely operative at PACMANUS.