There is general consensus that the distribution of incompatible trace elements in island arc volcanics largely reflects the chemical composition of the underlying mantle wedge. The initially depleted mantle is fertilized by a trace element flux from the subducted slab, which generates high LILE/HFSE and LREE/HREE ratios. However, there is disagreement on whether the metasomatizing agent is a melt or an aqueous fluid. The Batan xenoliths discussed here originate from calc-alkaline basaltic andesites from Mt. Iraya, Batan Island, North Luzon-Taiwan Arc. The xenoliths, most of which are harzburgites, comprise olivine, pyroxene, spinel, and various amounts of secondary minerals such as phlogopite and amphiboles. Since mantle xenoliths from subduction zones are extremely rare, these rocks have gained broad attention (Maury et al., 1992; Vidal et al., 1989; Fourcade et al., 1994; Schiano et al., 1995). Nevertheless, some authors favour melt-induced, others aqueous-fluid-induced metasomatism of the mantle wedge. In this contribution we present additional trace element data determined by ICPMS on bulk xenoliths and mineral separates. Based on these data we describe in detail the trace element signature imposed on the depleted mantle by the slab-derived metasomatizing agent, and show that this agent was a melt rather than an aqueous fluid.
Primitive-mantle normalized trace element patterns of bulk xenoliths show similar characteristics, irrespective of the presence or absence of metasomatic minerals such as phlogopite. There is no specific trace element signature that clearly distinguishes modal from cryptic metasomatism. The patterns decrease from Cs to the MREE and increase from the MREE to the HREE. Deviations from a smooth trend exist as positive anomalies for Pb and Cu, and as positive or negative anomalies for Sr, Zr, and Hf. Patterns for mineral separates (ol, opx, cpx, am, phlog) are more variable than those for bulk xenoliths owing to differences in the partition coefficients. However, all separates show high LILE/HFSE and high LREE/MREE ratios, and deviations from the general trend occur for Th, U, Pb, Sr, Zr, Hf, and Cu. Hence, the basic features of the trace element distribution in the mineral separates are the same as those in bulk xenoliths.
The most reliable information regarding the nature of the metasomatizing agent are provided by the phlog, ol, and opx separates, since the former is directly of metasomatic origin, and the latter two are neoblasts recrystallized during the metasomatic event. All these separates show positive anomalies for Pb, Sr, Zr, and Hf, which in ol and phlog are accompanied by positive Cu anomalies. The positive Pb and Cu anomalies may be related to sulphide droplets, and the high LILE/HFSE and LREE/MREE ratios and positive Sr, Zr, and Hf anomalies to melt inclusions, the presence of which in all minerals was reported previously (Schiano et al., 1995). That Pb and LREE are hosted by different phases is further supported by the variation of Ce/Pb ratios which vary between 1.1 and 3.1 in bulk xenoliths (ratios typical of island arc basalts) but between 0.06 and 22 in mineral separates. Recrystallized minerals such as ol and opx show Ce/Pb of 2, whereas phlog separates show ratios between 0.06 and 3.3 due to varying LREE content. The fact that the REE content in the different phlog separates varies over two orders of magnitude may suggest that even in phlogopite the (L)REE are not incorporated into the crystal lattice, but that they are hosted by inclusions.
As is discussed elsewhere (Bau, 1996a, 1996b), Y/Ho and Zr/Hf ratios may allow to distinguish melt- from aqueous-fluid-induced modification of the Zr-Hf-Y-REE distribution in geological samples, since the Y-Ho and Zr-Hf twin pairs, respectively, are fractionated in aqueous media but remain tightly coupled in pure melt systems. Y and Ho in the ol and phlog separates, and Zr and Hf in the ol, phlog, and opx separates are of metasomatic origin. These separates display both Y/Ho and Zr/Hf ratios that are close to those of chondrites, suggesting that these elements where transported with and deposited from a melt rather than an aqueous fluid. This is further supported by the absence of Eu anomalies (Bau and Knittel, 1993). Note, however, that this does not necessarily exclude that other (especially LIL) elements are transported from the subducted slab by aqueous fluids.
Bau, M., J. Conf. Abs. 1, (1996a)
Bau, M., Contrib. Mineral. Petrol. (in press) (1996b).
Bau, M. & Knittel, U., Chem. Geol. 105, 233-251 (1993).
Fourcade S. et al., Chem. Geol. 114, 199-215 (1994).
Maury, R.C. et al., Nature 360, 661-663 (1992).
Schiano, P. et al., Nature 377, 595-600 (1995).
Vidal, P. et al., Geology 17, 1115-1118 (1989).