GEOCHEMICAL SINKS AND RECYCLING AT SUBDUCTION ZONES

Holes 801C and 1149D in the western Pacific Ocean provide well-described sections into Mesozoic fast-spreading crust, Layer 2A. This ocean crust has acted as a "sponge" and soaked up elements during alteration by warm hydrothermal fluids and, later, cold seawater. Much of this alteration occurs early in the history of the cooling of the plate as it moves away from the spreading center, but chemical exchange may continue for tens of millions of years and may be reactivated by later volcanic and hydrothermal activity or by fracturing during flexure of the crust into the trench. This crust is then transported into the subduction zone, where it dewaters and melts, providing fluids that lubricate the subduction system and drive melting in the mantle. These transformations to the downgoing plate ultimately lead to arc volcanism and mantle heterogeneity.

Geochemical alteration of the volcanic section in Hole 801C occurs in several discrete zones, associated with ocherous Si-Fe hydrothermal deposits and thick massive flows and breccias. These zones control the alteration pattern of crust and contrast with previous models for a gradual decrease downhole in the alteration of oceanic basement. The pattern of alteration at Site 801, controlled by local pathways for hydrothermal fluids, may be a feature of fast-spreading crust (Alt and Teagle, 2003).

It is a great challenge to calculate the bulk geochemical composition of such a heterogeneous section. The approach taken by Leg 185 scientists was to combine representative samples into physical composite mixtures, ultimately leading to a "super" composite that approximates the bulk composition of the site in a single sample powder (e.g., Staudigel et al., 1995; Kelley et al., 2003). Preparation of these composites was guided by the abundance of data on the proportions of alteration zones (e.g., alteration halos, numbers of veins, and types of vein fillings logged by shipboard scientists) (Plank, Ludden, Escutia, et al., 2000), alteration mineralogy (Alt and Teagle, 2003; Talbi and Honnorez, 2003), and reconstruction of core types and compositions from logging data (Barr et al., 2002; Révillon et al., 2002; Jarrard et al., 2003). The basic major and trace element compositions of these composites are described in Kelley et al. (2003), and below are descriptions of the many other studies that have used the Site 801 basement samples and composites to provide first-order constraints on the composition of old oceanic crust.

Rouxel et al. (2003a) conducted one of the first detailed studies of Fe isotope fractionation during oceanic crust alteration, using primarily Leg 185 material. Although the Site 801 composites show little variation in ⁵⁷Fe from average igneous rocks, different alteration domains show large shifts, with both positive and negative ⁵⁷Fe, up to a total range of 4. The proposed processes of isotopic fractionation include formation of secondary minerals, Fe oxidation, preferential leaching of Fe²⁺ and, possibly, microbial mediation. These results are among the first produced for this system and provide a reference for future studies of reservoirs of fresh and altered basalts in the modern Earth for comparisons with the ancient Earth, the moon, and terrestrial planets.

Alt (2003), Alt and Teagle (2003), and Jarrard (2003) used samples from Hole 801C to define the concentrations and stable isotopic ratios of the major anions of O, C, and Cl in ocean crust with respect to estimates from other deep ocean crust sections. The oxygen isotopic composition of Hole 801C secondary minerals records temperatures from 5° to 100°C, increasing downhole through the ~500-m basement section in Hole 801C (Alt and Teagle, 2003). The ¹³C composition of Hole 801C carbonates reflect incorporation of oxidized organic carbon, as well as ¹³C changes in seawater since the Jurassic (Alt and Teagle, 2003). One of the important differences in the basement alteration of Hole 801C relative to other drilled basement sections is the order-of-magnitude lower abundance of brown oxidation halos, probably due to smooth basement topography and high sedimentation rate restricting access of oxygenated seawater (Alt and Teagle, 2003). The "super" composite for Hole 801C also appears to have higher ¹⁸O than other upper oceanic crustal sections, even when corrected for ¹⁸O-rich sediments (Alt, 2003). These observations could lead to a more detailed understanding of the oxidation state and ¹⁸O composition of the suprasubduction zone mantle. Finally, Jarrard (2003) uses physical properties of Hole 801C sediments (Jarrard et al., 2003) in part to develop a strategy for calculating the subduction flux of H₂O, CO₂, Cl, and K worldwide.

The radiogenic isotopes of Sr are reset by the circulation of seawater (e.g., Staudigel et al., 1995; Alt and Teagle, 2003), and this effect is well identified in fluid transfer as contributing a radiogenic isotope signature to arc lavas during dewatering at subduction zones (e.g., Elliot et al., 1997). The detailed sampling approach has permitted quantitative estimates of radiogenic isotope compositions of Sr, Pb, Nd, Hf, and Os of the altered oceanic crust being subducted into the Izu-Mariana system. These results have been used to quantify the different components involved in the generation of arc magmas (Hauff et al., 2003; Marini and Chauvel, 2001).

One of the most significant geochemical transfers in low-temperature alteration of oceanic crust is the uptake of U from seawater and its consequent disruption to the coupled ²³⁸U-²⁰⁶Pb, ²³⁵U-²⁰⁷Pb, and ²³²Th-²⁰⁸Pb isotopic systems (Hart and Staudigel, 1989). One of the major achievements of Leg 185 was to define the U and Pb content of altered oceanic crust (Révillon et al., 2002; Kelley et al., 2003) (Fig. F6A) and thus provide a realistic estimate for modeling subduction fluxes (Fig. F6B) (Kelley et al., 2005). Alteration in the upper 500 m of the oceanic crust leads to more than a fourfold increase in U/Pb, and if crust of this composition were subducted and incorporated into the mantle over Earth history, it would lead to mantle compositions that are never observed (Hart and Staudigel, 1989; Kelley et al., 2005). Thus, the subduction zone must further modify subducting slab compositions, and indeed, mass balance calculations of oceanic crust input with Marianas arc and backarc output predict shallow loss of 45%–75% of the slab Pb to the arc and deeper loss of 20%–40% of slab U to the backarc, with the net effect producing a source capable of evolving to compositions typical of the modern oceanic mantle (Kelley et al., 2005). These studies on Hole 801C sediments have thus quantified for the first time a whole-Earth geochemical cycle that has long been argued as critical in the development of mantle heterogeneity (e.g., Hofmann and White, 1982).

Studies are still in progress on the isotopic and elemental distribution of Li and B in the ocean crust (Valentine et al., 2001). Boron, and probably to a lesser extent Li, are dewatered at an early stage of subduction and are commonly enriched in fluids emanating from forearc regions (e.g., Benton et al., 2004). Quantification and modeling of these lower-temperature slab fluxes require an understanding of their distribution in the altered oceanic crust.

Many of the accomplishments of Leg 185 have stemmed from the attempt to use the same sample suite for all geochemical analyses. Such a practice should serve as a model for future IODP efforts. The composite sample suite from Leg 185 is already a valuable reference for the geochemical community, and samples remain for further study.

One of the primary goals of drilling during Leg 185 was to test whether chemical contrasts between the Marianas and Izu volcanic arcs derive in part from different input fluxes to the respective trenches. Different basement sections may contribute some variability (Hauff et al., 2003), but the different sedimentary sections create the largest chemical contrast in input fluxes along the margin. Site 1149 was chosen primarily to provide such a reference site for sediment recycling at the Izu-Bonin margin.

Subduction recycling studies are impossible without attacking systems on both sides of the trench. At the same time that Leg 185 postcruise activities focused on oceanic input to the trench (also see the contents of the G³ special volume on this theme; Table T2), many complementary studies have been carried out to characterize the composition of Izu-Bonin-Mariana (IBM) volcanic output through time, including new results from the submarine portions of the arcs (Hochstaedter et al., 2000, 2001; Ishizuka et al., 2003), as well as detailed marine ash records (Straub, 2003; Straub et al., 2004; Bryant et al., 1999, 2003). Site 1149 also preserves a well-dated volcanic ash record (Escutia et al., this volume), some of which derives from the Izu arc (based on preliminary laser inductively coupled plasma–mass spectrometry [ICP-MS] analyses of ash shards, T. Plank and Mordick, unpubl. data). The uppermost 120 m at Site 1149 contains 94 discrete ash layers, some light and some dark (Escutia et al., this volume; Sager and Escutia, 2005), much like material recovered in the Izu forearc basin (Fujioca et al., 1992). Nonetheless, layers at Site 1149 may have been windblown from multiple volcanic sources, and future geochemical analyses will resolve the provenance of the ash record. Such arc ash also contributes significantly to the subducted input flux. Discrete ash layers make up a minimum of 5% of the total sediment deposited over the last 7 m.y. (Sager and Escutia et al., 2005), whereas dispersed ash may make up as much as 35%–50% (based on Al- and Nb-based calculations in Plank, Ludden, Escutia, et al., 2000). Thus, Site 1149 is uniquely situated to provide both an input and an output record to the Izu subduction recycling problem.

At a regional scale, the combination of Legs 185 and 129 have given us an unprecedented understanding of subducted sediment fluxes along >1000 km of plate boundary at the IBM margin. Key in this endeavor are the seismic correlations that permit interpolation between widely spaced drill sites (Abrams, this volume). In contrast to the sediments subducting at the Marianas Trench (based on Sites 801 and 800), sediments in the Izu-Bonin Trench lack a mid-Cretaceous volcaniclastic section and contain more siliceous and carbonate-rich biogenic material due to a longer residence of the site beneath zones of high biological productivity. The more northerly latitude of Site 1149 leads to its Cenozoic passage into the westerly wind belt, resulting in a significant section of eolian dust of similar composition to Asian loess, as well as arc ash, both of which are largely absent from Site 801. As biogenic sediments are impoverished in most trace elements of interest to subduction cycling (Plank and Ludden, 1992; Plank and Langmuir, 1998), the most significant chemical difference is the abundance of eolian dust and ash at Site 1149 (Izu) vs. ocean-island volcaniclastics at Site 801 (Marianas).

The impact of such differences in sediment input have now been looked for and found in corresponding volcanic output. Eolian-dominated (for Izu) and volcaniclastic-dominated (for Marianas) sediments lead to different Pb isotopic sediment mixing end-members for the two arcs (Fig. F7A). For example, the entire Mariana arc is shifted to higher ²⁰⁶Pb/²⁰⁴Pb, which is largely due to the high ²⁰⁶Pb/²⁰⁴Pb volcaniclastics subducted there, combined with different components from the basement and mantle (Hauff et al., 2003). Another useful tracer in the recycling problem is Th/La, which is uniquely high in continental detritus (e.g., eolian dust) (see Fig. F5A) and is efficiently recycled through subduction zones (Plank, 2005). The IBM arcs mix toward sediment with progressively higher Th/La (Marianas to Izu to Honshu) as the seafloor enters the westerly wind belt and receives increasing amounts of high-Th/La eolian dust to the north (Fig. F7B). The IBM system thus illustrates for the first time how latitudinal variations in sedimentary stratigraphy can be linked to latitudinal variations in arc geochemistry on a regional scale (~1500 km). This experiment in the spatial coherence of sediment-arc recycling complements another recent experiment finding such coherence through time at the Nicaragua margin (Plank et al., 2002b).

Although there has been considerable success in tracing regional sediment inputs to IBM arc outputs, there remains a problem in the absolute mass fluxes involved. The most trenchward line of volcanoes that define the Izu arc (the "volcanic front") is highly deficient in the concentration of slab tracers, such as Th, relative to the Marianas arc and the rest of the global arc array (Fig. F8A). Results of Leg 185 demonstrate clearly that this deficiency in the Izu volcanic front is not caused by a deficiency in the sediment input flux to the Izu Trench. The mass flux delivered by Site 1149 is broadly comparable to that delivered by Site 801 for K, Th, and U, which are robustly calculated from downhole natural gamma logging data (Fig. F8A). Thus, while the Th deficiency indicates that there is a clear sediment contribution at the Izu volcanic front, the amount it is too low. However, submarine lavas erupted 100–150 km behind the Izu volcanic front (Hochstaedter et al., 2001) have the appropriate composition for the sediment input fluxes (Plank and Kelley, 2001; Straub, 2003). This suggests that the Izu subduction factory has a delayed delivery system, with most sedimentary slab material missing the main arc but feeding the backarc region. Both the Izu and Mariana arcs are consistent with subducted sedimentary components derived from deeper than 175 km in these cold slabs but the near vertical slab beneath the Marianas, allowing multiple components to contribute to the volcanic front (Fig. F8B). Drilling along the IBM margin has thus revealed new dynamics within the subduction factory relating to dip and thermal structure of the downgoing plate, which can be tested at other margins.