Publications

Thirteen dredge hauls from the active Sumisu and Torishima rift grabens west of the Izu-Bonin arc at about 30°–31°N, 140°E, recovered a suite of tholeiitic basalts to sodic rhyolites. Volcanism occurs along tensional faults within and bounding the rift grabens and along the transfer zones between adjoining rift segments. The Sumisu and Torishima rift lavas differ significantly from the lavas of the adjacent arc volcanic centers in having lowerAl2O3/Na2O,Ba/Zr,V/Ti, andBa/Ce and higher abundances of the rare earth elements. The rift lavas also have characteristics of backarc basin basalts, in that they are enriched in Al2O3, and depleted in total iron and TiO2 relative to mid-ocean ridge basalt, characteristics which are consistent with a higher water content in the source. Thus, the model of a progressive change in backarc basin basalt composition from arc-like to mid-ocean ridge-like, as a function of evolution of the basin, as has been suggested from many backarc regions, is not generally applicable. The comparison of the Sumisu and Torishima rift lavas with Mariana backarc basin lavas indicates that backarc basin basalts differ in composition from one basin to another. The comparison of these backarc basin suites with mid-ocean ridge suites from similar axial depths indicates that the overall control over the spectrum of backarc basin basalt compositions may be different extents of melting of the mantle. The Sumisu and Torishima rift lavas formed by a slightly higher extent of melting than the Mariana backarc basin basalts, a phenomenon which is related to the depth of the ridge segments. Furthermore, the data suggest a systematically higher extent of melting in the arc lavas than in the backarc lavas for both of these arc/backarc systems, consistent with a greater flux of water beneath the arc than beneath the backarc region.

Bimodal volcanism, normal faulting, rapid sedimentation, and hydrothermal circulation characterize the rifting of the Izu-Bonin arc at 31°N. Analysis of the zigzag pattern, in plan view, of the normal faults that bound Sumisu Rift indicates that the extension direction (080° ± 10°) is orthogonal to the regional trend of the volcanic front. Normal faults divide the rift into an inner rift on the arc side, which is the locus for maximum subsidence and sedimentation, and an outer rift further west. Transfer zones that link opposing master faults and/or rift flank uplifts further subdivide the rift into three segments along strike. Volcanism is concentrated along the ENE-trending transfer zone which separates the northern and central rift segments. The differential motion across the zone is accommodated by interdigitating north-trending normal faults rather than by ENE-trending oblique-slip faults. Volcanism in the outer rift has built 50–700 m high edifices without summit craters whereas in the inner rift it has formed two multi-vent en echelon ridges (the largest is 600 m high and 16 km long). The volcanism is dominantly basaltic, with compositions reflecting mantle sources little influenced by arc components. An elongate rhyolite dome and low-temperature hydrothermal deposits occur at the en echelon step in the larger ridge, which is located at the intersection of the transfer zone with the inner rift. The chimneys, veins, and crusts are composed of silica, barite and iron oxide, and are of similar composition to the ferruginous chert that mantles the Kuroko deposits. A 1.2-km transect of seven alvin heat flow measurements at 30°48.5′N showed that the inner-rift-bounding faults may serve as water recharge zones, but that they are not necessarily areas of focussed hydrothermal outflow, which instead occurs through the thick basin sediments. The rift basin and arc margin sediments are probably dominated by permeable rhyolitic pumice and ash erupted from submarine arc calderas such as Sumisu and South Sumisu volcanoes.

Beryllium is an incompatible trace element that closely parallels neodymium in its geochemical behavior. Be analyses conducted on well-characterized oceanic and arc volcanic rock suites, as well as on marine sediments, suggest a bulk solid/liquid distribution coefficient of 0.03–0.06 for melting of the mantle and crystallization of basalts. The Be/Nd ratio for many volcanic rocks from diverse tectonic environments is approximately .05, similar to the ratio in chondrites. Be data for samples from volcanic arcs show that there are significant variations in 10Be/9Be among different arcs, and that variations in 10Be are not due to variations in Be concentration alone. For at least one volcano (Bogoslof), the 10Be/9Be ratio is constant for samples that vary by a factor of three in both their Be and 10Be concentrations, suggesting that 10Be is an inherited magmatic signature and not simply a result of contamination near the surface. In addition, the Be, Nd and Pb isotope systems for this volcano are all consistent with a model in which small amounts of sediment were incorporated into the Bogoslof source region—provided the mantle wedge has the isotopic characteristics of depleted MORB. Since 10Be exists only in the uppermost tens of meters of oceanic sediments, the data suggest an efficient return flux of sediment to the mantle at subduction zones.

Arc volcanoes occur at convergent margins with a wide range in subduction parameters, and variations in these parameters might be expected to lead to variations in the chemistry of magmas parental to arcs. Major element analyses from approximately 100 volcanic centers within 30 arcs, normalized to 6% MgO to minimize the effects of crystal fractionation, display wide variations. Na2O and CaO at 6% MgO (Na6.0 and Ca6.0) correlate remarkably well with the thickness of the overlying crust. These systematics are consistent with two possible models. In the first model, the crust behaves as a chemical filter; where the crust is thick, magmas crystallize at higher pressure and interact more extensively with the arc crust. Modeling of high-pressure crystallization and assimilation, however, does not reproduce the associated variations in Na6.0 and Ca6.0 without calling upon complicated combinations of fractionating phases and assimilants. In the second model, crustal thickness determines the height of the mantle column available for melting beneath arc volcanoes. If melting begins beneath arcs at similar depths, then the column of mantle that undergoes decompression melting is much shorter beneath the thickest arc crust. The shorter mantle column for arcs built on thick crust will lead to smaller extents of melting in the mantle, and hence higher Na6.0 and lower Ca6.0 in the parental magmas. Modeling shows that variations in the extent of melting in the mantle can easily account for the associated variations in Ca6.0 and Na6.0. The abundances of the other major elements at 6% MgO do not correlate well with crustal thickness, or any other subduction parameter. Co-variation of some of these other major elements (e.g., Si6.0 and Fe6.0) within individual arcs suggests that they are strongly influenced by local crustal level processes that obscure partial melting systematics. Correction for the crustal processes improves the relationship between Na6.0 and Ca6.0 that is so readily explained by partial melting. The extents of melting in the mantle beneath arc volcanoes estimated from the ranges in Na6.0 and Ca6.0 are remarkably similar to those estimated beneath mid-ocean ridges. This observation provides further evidence that the mantle wedge, and not the slab, melts beneath arc volcanic fronts.

Regional averages of the major element chemistry of ocean ridge basalts, corrected for low-pressure fractionation, correlate with regional averages of axial depth for the global system of ocean ridges, including hot spots, cold spots, and back arc basins, as well as “normal” ocean ridges. Quantitative consideration of the variations of each major element during melting of the mantle suggests that the global major element variations can be accounted for by ∼8–20% melting of the mantle at associated mean pressures of 5–16 kbar. The lowest extents of melting occur at shallowest depths in the mantle and are associated with the deepest ocean ridges. Calculated mean primary magmas show a range in composition from 10 to 15 wt % MgO, and the primary magma compositions correlate with depth. Data for Sm, Yb, Sc, and Ni are consistent with the major elements, but highly incompatible elements show more complicated behavior. In addition, some hot spots have anomalous chemistry, suggesting major element heterogeneity. Thermal modeling of mantle ascending adiabatically beneath the ridge is consistent with the chemical data and melting calculations, provided the melt is tapped from throughout the ascending mantle column. The thermal modeling independently predicts the observed relationships among basalt chemistry, ridge depth, and crustal thickness resulting from temperature variations in the mantle. Beneath the shallowest and deepest ridge axes, temperature differences of approximately 250°C in the subsolidus mantle are required to account for the global systematics.

Lithium is a moderately incompatible trace element in magmatic systems. High precision analyses for lithium conducted on well characterized suites of MORB and ocean island basalts suggest a bulk distribution coefficient of 0.25−0.35 and behavior which is similar to Yb during low pressure fractionation and V during melting, as long as garnet is not an important residual phase. Data for peridotites and basalts suggest a mantle lithium content of about 1.9 ppm and show that significant concentrations of lithium reside in olivine and orthopyroxene, resulting in unusual inter-mineral partitioning of Li and complex relationships between lithium and other incompatible trace elements. The lithium abundances of arc basalts are similar to those of MORB, but their Li/Yb ratios are considerably higher. The high Li/Yb suggests the addition of a Li-rich component to arc sources; relatively low Yb abundances are consistent with the derivation of some arc magmas by larger extents of melting or from a more depleted source than MORB. Although Li is enriched at arcs, K is enriched more, leading to elevated K/Li ratios in arc volcanics. The high K/Li and relatively low La/Yb of primitive arc basalts requires either incorporation of altered ocean crust into arc magma sources, or selective removal of K and Li from subducted sediments. Bulk incorporation of sediments alone does not explain the Li systematics. Data from primitive MORB indicate a relatively low (3–4 ppm) Li content for new oceanic crust. Thus, the Li flux from the ocean crust is probably <1 × 1011 g/yr, and the oceanic crust may not be an important net source in the oceanic budget of lithium.