The Hawaiian plume delivers lavas representing at least three different magma-source components, as is evident from a wealth of published radiogenic isotope data. All these source components have been shown to be chemically fractionated and none represents a primitive-mantle reservoir; therefore it has been argued that this plume contains significant amounts of recycled oceanic crust (Hofmann, 1986).
We have determined trace element abundances in basalts from historical and prehistoric Kilauea and Mauna Loa eruptions and from the Hawaiian Scientific Drilling Project (HSDP), which yielded a total length of about 1000 m of drill core from Mauna Loa and Mauna Kea, spanning eruption ages of several hundred thousand years. Here we concentrate on the very highly incompatible elements whose relative abundances in Hawaiian basalts are very nearly identical to source abundances. The HSDP samples, analyzed by spark source mass spectrometry, have (primitive mantle-normalized) ratios of (Th/Ba)n ª (Th/La)n ª (Th/Nb)n ª 0.5 and (Th/U)n ª 0.7 to 0.8. This results in a very distinctive pattern of Th and U depletion relative to the abundances of Rb, Ba, Nb, and La in both volcanoes. Similar patterns are found in published data from other volcanoes of the Island of Hawaii.
High-precision isotope dilution (TIMS) data from Mauna Loa and Kilauea also show the Th-U depletion, as well as more subtle positive Eu anomalies and distinctive positive Sr anomalies. The Sr anomaly is especially prominent and unambiguous in Mauna Loa tholeiites, which have normalized Sr abundances greater than those of all the REE and the highly incompatible elements. These positive anomalies cannot be explained by the presence of cumulus plagioclase in the magma itself, because most of the lavas have olivine as the only phenocryst phase. The isotope dilution data, after correction for olivine fractionation or addition, also show that Yb is essentially constant and identical (standard deviation ¾ 3 %) in all Mauna Loa and Kilauea magmas. This requires the presence of a mineral with a high partition coefficient for Yb, namely garnet, to "buffer" the abundance of Yb in the partial melts.
The new oxygen isotope data on olivine phenocrysts
(see companion abstract by Eiler et al., 1996) yield low d18O values that are correlated with 206Pb/204Pb and other
radiogenic isotope ratios and point to a hydrothermally altered source component which is restricted to "Kea-type" volcanoes such as Mauna Kea, Kilauea and Kohala, and is isotopically indistinguishable from present-day Pacific crust (Eiler et al., 1996).
The special chemical characteristics distinguishing most, perhaps all, Hawaiian tholeiites, namely relative Th-U depletion and Sr-Eu excess, are also found in layered ophiolitic gabbros containing cumulus plagioclase, e.g. the Samail ophiolite (Pallister and Knight, 1981) and the Gabal Gerf Ophiolite (Zimmer et al., 1995), thus suggesting the presence of a gabbroic component in the source. However, the requirement for garnet indicates that the "gabbroic" source component contains garnet rather than plagioclase and is thus located in the mantle. This further suggests that it is actually part of the plume itself and is therefore recycled. On the other hand, the oxygen and radiogenic isotope data suggest that the 18O-depleted source component is derived form the interaction of plume-derived magma with the present-day lower oceanic crust (Eiler et al., 1996).
The somewhat paradoxical conclusion is that all tholeiitic Hawaiian magmas are to a significant extent derived from (former or present) gabbroic source components, but that one of these (the Kea-type component) most likely resides in the present day Pacific crust, whereas the others represent much older, recycled oceanic crust. A possible alternative interpretation would be that the Kea-type component is also recycled but nevertheless matches the radiogenic and (probable) oxygen isotopic composition of the present-day oceanic crust. This option is possible if the former oceanic crust had an approximately flat REE pattern and a U/Pb ratio corresponding to m × 8 to 10, i.e. chemical characteristics of an E-type oceanic crust.
Although the suggestion of a gabbroic source component, in both its recycled and its present-day version, is clearly speculative, there is in fact no other known geochemical reservoir that possesses the distinctive source chemistry which appears to be characteristic of all tholeiitic lavas on the Island of Hawaii. Also, none of the documented metasomatic processes in the mantle produces the appropriate magma source chemistry.
Eiler, J.M., Farley, J.W., Valley, J.W., Hofmann, A.W. & Stolper, E.M., J. Conf. Abs. 1, 156 (1996).
Hofmann, A.W., Chem. Geol. 57, 17-30 (1986).
Pallister, J.S. & Knight, R.J., J. Geophys. Res. 86, 2673-2697 (1981).
Zimmer, M., Kröner, A., Jochum, K.P., Reischmann, T. & Todt, W., Chem. Geol. 123, 29-51 (1995).