A Mantle-derived Origin for Mauritian Trachytes

Trachytes are typically interpreted in terms of extreme fractional crystallization from basaltic magmas. Data from Mauritius suggest otherwise. Here, intrusive, nepheline-bearing trachytes are associated with Older Series basalts (9.0-4.7 Ma), as confirmed by a U-Pb zircon age of 6.8 Ma. Trachyte mineralogy is dominated by low-Ca alkali feldspar with variable Na/(Na+K), lesser highNa nepheline, aegirine-rich clinopyroxene, titanomagnetite and accessory zircon and apatite. A few samples contain xenocrysts of anorthoclase, Al-and Ti-rich clinopyroxene and kaersutite; these phases show prominent reaction rims that approach typical trachyte mineral compositions. Trachyte major elements cluster at similar to 63 wt % SiO2 and Na2O +K2O similar to 12 wt %, but reconstructed pre-alteration compositions suggest that they were originally mainly phonolites, and form a prominent Daly Gap when plotted with the basalts. Incompatible trace elements are enriched in all trachytes, except for Ba, Sr and Eu, which show prominent negative anomalies. Rare earth element patterns have variable abundances, prominent negative Eu anomalies and shapes that differ markedly from those of the basalts. Initial eNd values cluster at+4.03 +/- 6 0.15 (n = 13), near the lower end of the range for basalts (epsilon(Nd) = +3.70 to+5.75), but the initial Sr-87/Sr-86 is highly variable (I-Sr = 0.70408-0.71034) compared with the relatively constant ISr of 0.70411 +/- 19 for the basalts. Fractional crystallization models, using the PELE and MELTS algorithms, starting with a primitive, nephelinenormative Mauritian basalt parent (P = 1 kbar, fO(2) = QFM-3, where QFM is quartz-fayalite-magnetite buffer) fail, because when plagioclase joins olivine in the crystallizing assemblage, successive liquids become depleted in Al2O3, do not produce nepheline, and do not approach phonolitic or trachytic compositions. Similar results are obtained using more alkaline parent liquids including basanite; however, such compositions are not observed anywhere amongst the Older Series basalts. Plutonic xenoliths from Mauritius do not fill the Daly Gap as in some other occurrences (e. g. Hawaii, Pantelleria, Azores). Fractional crystallization was not the operative process that produced Mauritian trachytes. Likewise, liquid immiscibility is excluded because the compositions do not fall at the ends of known miscibility gaps. What remains as plausible is some type of partial melting process, although the source cannot be Precambrian continental crust, as suggested to exist under Mauritius, because such material should not yield nepheline-bearing melts, and would not account for the Sr-Nd isotopic compositions. Small amounts of contamination with such continental crust, however, can account for the Sr-Nd isotopic compositions of the trachytes. Partial melting of basaltic volcanic rocks or extant gabbroic bodies, either from the oceanic crust or from Re ' union plume-related magmas, should yield quartz-saturated melts different from the critically undersaturated Mauritian trachytes. A remaining possibility is that the trachytes represent direct, smalldegree partial melts of fertile, perhaps metasomatized mantle; we suggest that the xenocryst assemblage present in some samples might represent fragments of this source. A mantle source is supported by the presence of trachytic and phonolitic glasses in many mantle xenoliths, and experimental results show that low-degree alkaline melts can be produced from mantle peridotites even under anhydrous conditions. If some feldspar is left behind as a residual phase, this would account for the negative Ba, Sr and Eu anomalies observed in Mauritian trachytes. These considerations may also apply to other trachyte and phonolite occurrences worldwide.