BACKGROUND
Crustal Recycling
at the Izu-Mariana Margin
There are several reasons why the Izu-Mariana margin (Fig. 2) is ideal for
studying subduction recycling. Significant progress has already
been made on many parts of the flux equation. Forearc sites of
fluid outflow (serpentine seamounts) have already been drilled
(Leg 125; Fryer, 1992), as have most of the sedimentary
components being subducted at the Mariana Trench (Leg 129;
Lancelot, Larson, et al. 1990). These volcanic arcs and backarcs
are among the best characterized of intraoceanic convergent
margins, both in space and time (Legs 125 and 126; Gill et al.,
1994; Arculus et al., 1995; Elliott et al., 1997; Ikeda and
Yuasa, 1989; Stern et al., 1990; Tatsumi et al., 1992; Woodhead
and Fraser, 1985), and the problem is simplified here because the
upper crust is oceanic, so upper crustal contamination is
minimized. Sediment accretion in the forearc is nonexistent
(Taylor, 1992), so sediment subduction is complete. Despite the
simple oceanic setting and the shared plate margin, there are
still clear geochemical differences between the Mariana and Izu
Arcs (e.g., in Pb isotopes and elemental enrichments; Figs. 3A and 3B). The divergence of
compositions between the volcanics of these two oceanic arcs
provides the simplest test for how the composition of the
subducting crust affects them. The key missing information is the
composition of the incoming crustal sections, particularly the
(altered) basement sections.
Existing
Crustal Inventory at the Izu-Mariana Margin
Of the eight sites drilled in the seafloor immediately east
of the Izu-Bonin Trench (not including the Shatsky Rise sites),
only one penetrated basement, Site 197 (Leg 20; Fig. 2). No sediments,
however, and only 1 m of basalt were recovered. Recovery was poor
at most of the other sites: less than 25 m of sediment was
recovered at all sites drilled during Leg 20 (Sites 194-198), and
the recovered material was dominantly pelagic clays above
resistant cherts. Although more was recovered at Sites 52 (45 m)
and 578 (165 m), drilled during Deep Sea Drilling Project (DSDP)
Legs 6 and 86, respectively, coring again was halted by chert
layers, leaving hundreds of meters of unsampled sediment below.
Prior drilling has provided us with many samples of the upper 50
to 150 m of the pelagic clay and ash unit, but almost nothing of
the units below, including the oceanic crust.
* The main goal of Izu-Bonin coring is to sample all
sedimentary units and the upper alteration zone (~300 m) of the
oceanic crust below.
The success
rate in coring sediments and basement to the south, seaward of
the Marianas, was just about as poor as that experienced to the
north, until Leg 129, when three complete sections (ODP Sites
800-802) were sampled through the cherts to "basement."
During Leg 129, sedimentary units were well sampled at Sites
800-802, but normal oceanic crust was sampled at only one site,
Hole 801C. At the other two sites, off-axis Cretaceous sills and
flows were encountered as "basement." The crustal
inventory at the Mariana Trench includes (from top to bottom):
pelagic clay, chert, and radiolarite (+ chalk), Cretaceous
volcaniclastic turbidites, radiolarite, off-axis Cretaceous
intrusives and extrusives, and Jurassic oceanic crust. Leg 129
coring and prior DSDP efforts provided adequate samples of the
sedimentary units being subducted at the Mariana Trench
(providing estimates of chemical fluxes with better than 30%
precision for most elements [Plank and Langmuir, 1998]). However,
the only sample of Jurassic oceanic crust, which must comprise
the largest mass of crustal material being subducted at the
Mariana Trench, comes from the lowest 63 m in Hole 801C (of the
~135 m total penetration into basement in Holes 801B and C, only
the lower 63 m of drilling recovered 43 m of normal, Jurassic
tholeiitic oceanic crust).
* Thus, the main goal of Mariana coring is to provide a more
complete sampling of the upper alteration zones in the Jurassic
seafloor, which constitutes a significant part of the budget for
many geochemical tracers of the subduction process, including H2O,
CO2, Rb, and U.
Existing
Crustal Mass Balance for the Marianas
With information in hand, it is possible to calculate many of
the input and output fluxes for a few chemical components through
the Marianas subduction zone. We consider here a preliminary flux
balance for H2O (Fig. 4). The sediment
input is fairly well constrained by previous coring during Leg
129 (Sites 800-802), and by the extensive chemical analyses of
the recovered material (Karl et al., 1992; Karpoff, 1992; Lees et
al., 1992; France-Lanord et al., 1992) as well as the geochemical
logs for the different holes (Pratson et al., 1992; Fisher et
al., 1992). As a result, H2O flux estimates for
sediments from Sites 800 and 801 are quite consistent with one
another (within 15%). The other crustal input flux is the
subducting oceanic crust, which is very poorly constrained
because of a lack of significant penetration into the mid-ocean
ridge (MOR) basement in this area (63 m in Hole 801C). The
geochemical budget of elements in the oceanic crust has two
sources: primary igneous and secondary alteration. The primary
igneous composition is fairly well constrained, based on
extensive sampling of modern mid-ocean ridge basalt (MORB) and on
the relatively unaltered samples recovered from Hole 801C. The
secondary alteration fluxes are virtually unknown, however, and
can only be estimated from various other regions, compilations,
and assumptions: the average global H2O flux in
Peacock (1990), alteration studies of DSDP Hole 504B (Leg 69; Alt
et al., 1986) and DSDP Sites 417/418 (Legs 51, 52, and 53;
Staudigel et al., 1995), and assuming 10% interpillow material in
Hole 801C (Castillo et al., 1992b). These estimates show that the
alteration fluxes may be large, but are poorly known. The
applicability of existing data (obtained for slow-spreading old
crust at Sites 417 and 418 and medium-spreading young crust in
Hole 504B) to the crust seaward of the Mariana Trench (old-fast
spreading) remains to be seen and is, in fact, a major goal of
Leg 185.
Unique to the seafloor seaward of the Mariana Trench is an overprint of Cretaceous basement flows and sills. There are two sources of uncertainty in estimating this flux: the thickness of the Cretaceous "basement" layer and its chemical composition. Calculations based on sonobuoy velocities, reflection data, and drilling results from Leg 129 indicate a 100- to 400-m-thick layer of massive Cretaceous basalt, and possibly some interbedded sediments, overlying Jurassic oceanic crust (Abrams et al., 1993). Although this is not the case for water, the total flux of many elements depends critically on whether this Cretaceous basalt is alkalic (as for Site 800 basalts and various seamounts of the Pigafetta Basin [PB]) or tholeiitic (as for Site 802 basalts of the East Mariana Basin [EMB]). Although plate trajectories (Fig. 2) indicate that the seafloor subducting beneath the Marianas is largely the tholeiitic EMB, we consider both EMB (tholeiitic) and PB (alkalic) type basalts in estimating the water flux into the subduction zone (Fig. 4). Both estimates yield small water input fluxes relative to the sediment and altered Jurassic MORB.
The first measurable outputs from the subduction zone are the forearc fluids, which have shown to be freshened and from a subducted source (Mottl, 1992). It is currently difficult to estimate rates based on the fluid flow itself. We, therefore, use a model based on the total (maximum) water generated during clay mineral breakdown in the subducted sediments (Plank et al., 1994). This calculation is model dependent, but further study of the nature of these fluids will help to identify the actual dehydration reactions that are occurring with depth during subduction. Figure 4 shows that the water outputs to the forearc may be a significant fraction of the sediment input. Magmatic outputs to the volcanic arc and backarc are determined from the chemical composition of arc and backarc basalts (assuming 5.7 and 1.25 wt% H2O above MORB background, respectively; Stolper and Newman, 1994) and from magmatic addition rates. The magmatic arc water flux is the largest of the crustal outputs from the subduction zone.
These
preliminary calculations provide some initial insights into the
flux balance in subduction zones and reveal where the major
uncertainties lie. If we ignore the igneous MORB and Cretaceous
basalt contributions as no net gain from the mantle perspective,
then the continental water inputs and outputs appear to be
remarkably closely balanced across the subduction zone. The
balance hinges critically, however, on the magnitude of the
basement alteration fluxes. Present estimates are poor, and the
actual flux balance could still go either way.
* Drilling through the upper oceanic crust subducting beneath
the Marianas can dramatically improve one key flux in the mass
balance equation‹the alteration flux.
Existing
Crustal Mass Balance for Izu-Bonin
Because neither the sediments nor the altered oceanic crust
seaward of the Izu Trench have been sampled to any significant
extent, our ability to determine mass balance is much more
limited here. However, we can make some predictions about the
crustal inputs to the Izu Trench based on the Izu volcanic
output. Izu basalts record almost half the K or Ba enrichment of
Marianas basalts (Fig. 3B), whereas sediment
mass fluxes into the two trenches are roughly comparable, or even
greater, at Izu (600 m of sediment into Izu vs. 400 m into the
Mariana Trench). Thus, Izu sediments should be much poorer in K
and Ba than Marianas sediments. One way to explain this would be
to replace the volcaniclastic sections in the Marianas sediment
columns with cherts, which are barren of K and may be very poor
in Ba (Karl et al., 1992). This makes some sense given what we
know about the history of sedimentation in this part of the
ocean‹the Cretaceous overprint east of the Marianas may be
absent to the north, east of Izu (Fig. 2), where the
seafloor spent a longer time on average beneath equatorial zones
of high biologic productivity (Fig. 5), possibly leading
to greater sections of chert and/or carbonates. Drilling the
seafloor east of Izu can directly test these predictions.
Sediment layers are fairly uniform throughout the region,
reflecting uniform pelagic sedimentation. Thus, a single hole
should give us a fairly representative sampling of sediments
being subducted at the Izu Trench. If the extra thickness of
sediments off Izu is not dominantly barren cherts, this means
that much of this sediment does not contribute to arc magmas,
either because it is underplated (we can see that it is not
accreted), or because it fails to dewater or melt beneath the
arc. Thus, by drilling and sampling the crustal inputs, we can
learn more about the process of sediment subduction and
recycling.
The geochemical differences between the Marianas and Izu arc volcanics could also be related to the chemical composition of the altered oceanic crust. The K or water contents in the altered basaltic sections may vary regionally, possibly explaining regional variations in K and the extent of melting reflected in Marianas and Izu lavas. This can be tested by drilling the upper oxidative alteration zone, which contains most of the alkali budget in the oceanic crust, in both regions.
Finally, some of the differences between the Izu and Marianas lavas may have nothing to do with subducted input and may be explained by more enriched mantle beneath the Marianas. Evidence for enriched mantle in the region comes from enriched shoshonites at the adjacent volcanic arc (Bloomer et al., 1989; Lin et al., 1989). Although drilling cannot test whether enriched mantle exists beneath the Marianas, it can make invoking it unnecessary.