Oxygen Isotope Constraints on the Sources of Hawaiian Volcanism

John M. Eiler California Institute of Technology, Pasadena, CA 91125, USA


Kenneth A. Farley California Institute of Technology, Pasadena, CA 91125, USA

John W. Valley University of Wisconsin, Madison, WI 53706, USA

Albrecht W. Hofmann Max-Planck-Institut für Chemie, Postfach 3060, 55020 Mainz, Germany

Edward M. Stolper California Institute of Technology, Pasadena, CA 91125, USA

Hawaiian lavas are frequently lower in d18O than typical MORB. However, studies of whole rocks and glasses do not demonstrate any correlations between oxygen and other isotope ratios or other coherent variations in oxygen isotope ratio between or within Hawaiian volcanoes. This limitation has left the source of 18O depletion unconstrained, with hypotheses including: low d18O in some or all Hawaiian plume components, low values of d18O in plume sources in general, or isotopic fractionation during melting. Because analyses of Hawaiian lavas make up a large fraction of published analyses of OIBs, these conclusions have been generalized to apply to many or all of the sources of hotspot volcanism (Harmon and Hoefs, 1995).

We have measured oxygen isotope ratios in olivine phenocrysts from 88 Hawaiian lavas by laser fluorination. Results indicate that the total range is oxygen isotope ratios among Hawaiian lavas (1 ”) is less than indicated by studies of whole rocks and glasses, yet reveal unprecedented correlations between d18O and radiogenic isotope ratios. Low values of d18O in olivine phenocrysts (down to 4.6 ”) are uniquely associated with relatively radiogenic Pb, low 3He/4He ratios and "depleted" Sr and Nd isotope signatures in their host lavas, and are preferentially sampled in lavas from the so-called Kea trend volcanoes (Kilauea, Mauna Kea, Kohala, Haleakala). In addition, we find that lavas having radiogenic Sr and non-radiogenic Pb contain olivine higher in d18O (~5.6 ”) than typical upper mantle values (d18Ool = 5.2±0.2 ”) (Mattey et al., 1994). Other isotopic end members to Hawaiian lavas (e.g. high 3He/4He and post erosional lavas) are found to have d18O values within the normal upper mantle range.

The characteristics we find correlated with low d18O have been previously identified with an end member to the isotopic variability of Hawaiian lavas variously called the "Kilauea", "lithospheric" or "Kea" end member (the last of which is adopted by this study) (Stille et al., 1986). The only major silicate reservoir which is known to be commonly lower in d18O than typical mantle is the lower oceanic crust (i.e. layer 3 gabbros), which becomes 18O depleted during high temperature exchange with seawater near actively spreading ridges (Gregory and Taylor, 1981). Our data therefore indicate that the "Kea" end member is derived from a gabbroic or basaltic source. While we cannot disprove that this is a recycled component of the Hawaiian plume, the radiogenic isotope characteristics of the "Kea" end member closely resemble Pacific MORB, including that recovered from drilling immediately south of the Hawaiian islands. The combination of low d18O and MORB-like radiogenic isotope ratios in the "Kea" end member strongly suggests that it is derived from the lower portions of the local Pacific crust. Supporting evidence for this hypothesis includes several geochemical properties of Hawaiian lavas that suggest crustal contamination (e.g D/H ratios, B abundances and isotope ratios, Os isotope ratios) (Chaussidon and Marty, 1995; Martin et al., 1994). Geophysical and petrologic evidence indicates that extensive ponding and olivine crystallization occurs at the base of the oceanic crust beneath Hawaiian lavas, and therefore this hypothesis is also physically reasonable (Clague, 1987). Given the distinction in oxygen isotope ratio between Loa and Kea trend volcanoes, this conclusion implies a fundamental difference in the magma supply and plumbing systems of these two trends.

d18O values equal to average mantle in the post erosional source are consistent with the common interpretation that such lavas are derived from the local upper mantle. Similarly, typical upper mantle values in an end member characterized by high 3He/4He and moderately depleted Sr and Nd isotope ratios (the "Loihi" end member) are consistent with this plume component being derived from a source which is overwhelmingly made up of mantle peridotite. In contrast, high d18O Hawaiian lavas with "enriched" isotopic compositions (the "Koolau", end member) are indicated to be derived from a source containing several % of recycled sediment.


Chaussidon, M. & Marty, B., Science 269, 383-386 (1995).

Clague, D.A., Bull. Volcanology 49, 577-587 (1987).

Gregory, R.T. & Taylor, H.P.J., J. Geophys. Res. 86, 2737-2755 (1981).

Harmon, R.S. & Hoefs, J., Contr. Min. Petr. 120, 95-114 (1995).

Martin, C.E., Carlson, R.W., Shirey, S.B., Frey, F.A. & Chen, C.Y., Earth Planet. Sci. Lett. 128, 287-301 (1994).

Mattey, D., Lowry, D. & Macpherson, C., Earth Planet. Sci. Lett.128, 231-241 (1994).

Stille, P., Unruh, D.M. & Tatsumoto, M., Geochim. Cosmochim. Acta 50, 2303-2319 (1986).