The Hawaiian Islands have become a natural laboratory for multidisciplinary research on mantle plumes and their surface expression. Through collaborative efforts, the Re-Os isotope system is being systematically applied to subaerial, submarine, and drilled stratigraphic suites of lavas from individual Hawaiian volcanoes. The isotopic systematics of Os have provided an especially clear picture of the mechanisms of magma transport from the Hawaiian plume, and in so doing have provided new constraints on models of Hawaiian mantle composition proposed from other geochemical studies.
The good correlations of Os concentrations with MgO, Ni, and Cr in Hawaiian lavas (and all OIB) clearly indicate the compatible behavior of Os and implicate olivine as a major controlling phase. Small amounts of reaction with the oceanic mantle lithosphere will buffer the Os isotopic composition of the reacting magma at depleted mantle values (187Os/188Os = 0.1200-0.1275) without significantly affecting Sr, Nd, and Pb; as a result, even minor magma-lithosphere exchange will decouple Os from the incompatible isotopic tracers. However, the strong correlations between Os and other isotopes categorically rule out any model involving significant reaction between migrating magmas and the oceanic mantle lithosphere, and require that magma pathways through the lithosphere beneath Hawaii are either open channels, or are saturated in plume-derived Os (and thus limited in scale). With the exception of lavas clearly effected by crustal contamination, the melt migration constraints indicate that the isotopic heterogeneity observed in Hawaiian shield-stage lavas originates entirely from beneath the lithosphere.
Shield-stage Hawaiian lavas can be explained by three components, exemplified by extremes in the data from Koolau (enriched), Loihi-Kilauea (depleted, high 3He/4He), and a low-3He/4He depleted component (probably local asthenospheric upper mantle) observed in late shield-stage lavas at Kohala, Haleakala and Mauna Kea. Available isotopic data reaffirm a compositional distinction between volcanoes located on the "Kea" (e.g. Mauna Kea, Kilauea, Kohala) and "Loa" (e.g. Mauna Loa, Kahoolawe, Koolau) structural trends. In Os-Sr-He isotope space, Kea-trend volcanoes display consistently lower 3He/4He and 187Os/188Os for a given 87Sr/86Sr, consistent with a significant proportion of depleted upper mantle. The Kea-Loa isotopic differences extend to trace and major elements as well.
These spatially and chemically distinct groups of volcanoes suggest that the Hawaiian plume is compositionally zoned. Major and trace element characteristics of Kea-trend lavas (higher pressure and lower degree of melting than Loa-trend) suggest that Kea-trend volcanoes sample the periphery of the Hawaiian plume, while Loa-trend volcanoes are generated closer to the axis of the plume. The inferred thermal and isotopic zonation is precisely what is predicted from the fluid dynamic analysis of mantle plumes undergoing thermal entrainment.
Examination of the present Hawaiian geochemical database reveals that isotope ratios are correlated with SiO2, FeO, and CaO/Al2O3 in shield-stage lavas from all Hawaiian volcanoes which have been analyzed in sufficient detail (n=8). The isotope-major element correlations suggest variations in the bulk composition of the material which makes up the Hawaiian plume. Only Loihi lavas have compositions that could be expected from melting typical mantle lherzolite (at pressures of at least 4 GPa). Alternatively, Kea-trend lavas (as well as Loihi lavas) could be generated at lower pressures (3 GPa) by melting peridotite enriched in FeO. The isotopically-enriched Loa-trend lavas (Koolau and late Mauna Loa lavas) are the most enigmatic; high fractionation-corrected SiO2 contents (49-52%) and low FeO contents (10-11% FeO) suggest shallow melting (<60 km), but the oceanic lithosphere is in the way and these lavas clearly retain a strong garnet signature. Strong correlations of Os isotopes with SiO2 (positive) and CaO/Al2O3 (negative) indicate a substantial component (at least 10-20%, possibly more) of recycled crustal material in the Koolau endmember, a portion of which is likely sedimentary. The isotopic scatter observed at Koolau suggests that this crustal component is probably not well mixed but may be preserved as lithologically-distinct entities (eclogite), unlike other hotspots with large amounts of recycled crust as inferred from radiogenic Os isotope signatures (HIMU islands Mangaia and Tubuai).