MAGMA EVOLUTION AND GENESIS

Variations in the major- and trace-element geochemistry, taken together with the observed phenocryst phases in these rocks, can help constrain the differentiation history of the Alboran volcanic rocks. The decrease in Al₂O₃, Sr, and CaO with increasing SiO₂ and degree of differentiation, as well as the negative Eu anomaly present in most samples (especially the most evolved: dacites and rhyolites), indicate fractionation of plagioclase, which was the major phenocrystic phase found in all rock types. The presence of clinopyroxene in the basalts through andesites, and the decreases in CaO/Al₂O₃, CaO, and Cr are consistent with clinopyroxene fractionation, whereas the decreases in MgO and Co most likely reflect olivine fractionation in the more mafic rocks. The decreases in FeO*, TiO₂, and V with decreasing SiO₂ and presence of titanomagnetite phenocrysts in all rock types illustrate the importance of titanomagnetite as a fractionating phase. Removal of ~5% titanomagnetite is necessary to explain the decrease in V from basalt to rhyolite. Finally, the decrease in P₂O₅ and the depletion in the MREE reflect fractionation of apatite and possibly amphibole. In summary, fractionation of plagioclase, clinopyroxene, titanomagnetite, olivine, apatite and amphibole can explain most of the observed variations in the major and compatible trace elements. Nevertheless, the presence of xenocrysts and the large range in Pb concentration in the dacites and rhyolites of 1.5 to 66 ppm (only three samples, however, have Pb > 15.5 ppm) suggest that crustal contamination was also an important process affecting the chemistry in at least the more evolved magmas.

The geochemistry of the volcanic rocks from the Alboran places important constraints on their origins. The calc-alkaline character of most of these rocks, which range from basalt through rhyolite, is a typical feature of subduction zone volcanism. Furthermore, andesites and dacites occur almost exclusively in island arcs or along active continental margins. The trace-element systematics provide further strong support for a subduction origin of the Alboran volcanic rocks. These include (1) the spiked incompatible-element patterns, (2) troughs at Nb, Ta, and Ti on incompatible-element diagrams (Fig. 9A), (3) high Ba/La, Ba/(Nb, Ta), K/(Nb, Ta) ratios, and (4) low (Nb, Ta)/REE, Nb/U, and Ce/Pb ratios. Because of the high solubility of mobile elements (e.g., Rb, Ba, U, K, Sr, Pb) in hydrous fluids at mantle temperatures, these elements are transported from the slab to the overlying asthenospheric wedge, resulting in enrichment of the wedge (and the melts derived from the wedge) in these elements (e.g., Gill, 1981). Melts of sediments can also enrich the wedge in these elements. Such enrichment processes can explain the spikes in the mobile elements on multi-element diagrams (e.g., see the LREE-depleted basalts in Fig. 9A), as well as high ratios in most samples of mobile elements to immobile elements, such as Nb and Ta. The troughs on the incompatible-element diagrams at Nb and Ta can reflect the low abundances of these elements in sediments and/or the retention of these elements in the source by a residual phase such as titanite, rutile, or possibly phlogopite (e.g., Gill, 1981; Wilson, 1989; Ionov and Hofmann, 1995).

Although contamination or mixing of MORB melts with continental crustal material can generate similar incompatible-element characteristics (e.g., Wilson, 1989) to those observed in some of the basaltic samples, crustal assimilation (during ascent) cannot alone explain the incompatible-element chemistry of the Alboran volcanic rocks. For example, the enriched basaltic samples extend to higher compositions of many elements (e.g., K₂O, Ba, Sr, Pb, Th, U, LREEs) than generally found in MORB or the continental crust (Hofmann, 1988; Taylor and McLennan, 1985). Furthermore, crustal assimilation during fractional crystallization (AFC) cannot explain the inverse correlations of SiO₂ with Nb/Zr and Nd/Sm in the basaltic samples, because with increasing differentiation (as reflected by increasing SiO₂), both ratios should increase, not decrease. The variation in Ta/Nd and Sm/Nd ratios in the basaltic samples are also inconsistent with crustal contamination (Fig. 10). The extreme depletion in highly incompatible, immobile elements observed in some of the LREE-depleted basalts are primarily found in subduction environments and are believed to form by high degrees of melting in the presence of water and/or as a result of progressive depletion of the source (e.g., Stolper and Newman, 1994). Finally, we note the striking similarity between the incompatible-element abundances of the LREE-enriched Alboran basaltic rocks and those from basaltic rocks in the active Aeolian Arc, also in the western Mediterranean (Fig. 9A). We conclude that the major- and trace-element compositions of the volcanic rocks provide evidence that subduction was active in the Alboran from 6 to at least 12 Ma.

The incompatible-element data for the Alboran basaltic rocks can also be used to constrain the residual mineralogy of the sources of both groups. The LREE-depleted basalts have heavy (H)REE patterns with near zero to slightly positive slopes and (Sm/Yb)_N, (Gd/Yb)_N, and (Dy/Yb)_N 1 (Fig. 9A), indicating that garnet was not a residual phase. The enrichment or peaks in mobile elements (Rb, U, K, Pb, and Sr) vs. the depletion or troughs in immobile elements (Th, HFSE, and REE) could reflect high degrees of melting of sources metasomatized by hydrous fluids from the subducting slab (e.g., Stolper and Newman, 1994). A high degree of melting (because of high water content) will result in an apparent depletion (i.e., dilution) in immobile elements, which are not enriched by the hydrous fluids. The mobile elements are, however, continually enriched in the wedge by new fluid fluxes. In conclusion, these basalts are consistent with derivation from metasomatized MORB-type material in the mantle wedge above a subducting slab.

The steep negative slopes of the HREE patterns and (Sm/Yb)_N, (Gd/Yb)_N, and (Dy/Yb)_N > 1 (Fig. 9A) indicate that garnet may have been a residual phase during melting to form the LREE-enriched basalts. Although garnet is only stable at depths >75-80 km (pressures in excess of 2.3-2.5 GPa) in peridotite, it can be stable to depths as shallow as 40 km in pyroxenite (Hirschmann and Stolper, 1996). The negative anomalies on incompatible-element diagrams in Rb, Ba, K, Zr, Ti (and possibly Nb and Ta) and the lack of a pronounced Pb anomaly (in contrast to the LREE-depleted basalts; see Fig. 9A) most likely reflect the presence of residual phlogopite (± amphibole) in the source, which can be enriched in all of these elements (Ionov and Hofmann, 1995, and references therein). The LREE-enriched basalt, therefore, could have been derived from a phlogopite-bearing garnet-pyroxenite or -peridotite source within the mantle wedge above the subducting slab.

There are two possible interpretations for enrichment in Th, LREE, and MREE in the LREE-enriched basalts. First, they could reflect derivation from an enriched OIB- or plume-type of source. Using seismic tomography and isotope geochemistry, it has been shown that large-scale upwelling of plume material has been occurring throughout the Cenozoic beneath the eastern Atlantic, western Mediterranean, and western and central Europe (Hoernle et al., 1995). The trace-element pattern for a basalt sample from Gomera (Canary Islands) with the Sr, Nd, and Pb isotopic composition of this plume end-member (LVC) is shown in Figure 9A (S. Dorn and K. Hoernle, unpubl. data). The close similarity of the incompatible-element composition of this and other samples from the aforementioned areas with those from high ²³⁸U/²⁰⁴Pb (HIMU)-type ocean islands, such as Saint Helena, suggest an origin of the LVC plume component from recycled oceanic crust. Recycled ocean crust is likely to be present in the mantle as garnet pyroxenite. As is commonly observed in peridotite massifs (believed to be obducted portions of the upper mantle) in northern Morocco (Beni Boussera) and southern Spain (Ronda), pyroxenite layers (with and without garnet) commonly occur within a peridotite matrix. Therefore, the OIB-type, LREE-enriched end-member is likely to occur as garnet pyroxenite layers within depleted (MORB-type) peridotite. It has been shown in regions of upwelling mantle, such as beneath the nearby Canary Islands, that zones of both OIB- and MORB-type material can occur over distances of <100 km (Hoernle et al., 1991; Hoernle and Schmincke, 1993). Subduction of ocean crust beneath the Alboran resulted in metasomatism of both MORB and OIB components within the mantle wedge by hydrous fluids, which resulted in the stabilization of phlogopite in the garnet-pyroxenite component. Both garnet and phlogopite remained in the residuum during melting to generate the LREE-enriched basalts.

Alternatively, the trace-element characteristics of the LREE-enriched basalts could reflect the presence of sediments in the source, as has also been proposed for the Aeolian Arc Volcanic Rocks with similar chemistry (Fig. 9; Ellam et al., 1989). If this is the case, then the steep HREE patterns do not necessarily require the presence of garnet in the source. The presence of phlogopite (± amphibole) in the residuum of the Alboran basalts may reflect their derivation from lithospheric mantle sources, in contrast to derivation of the Aeolian Arc Basalts from asthenospheric sources. Isotope data are crucial for placing further constraints on the source compositions for both basalt groups and for distinguishing between involvement of sediments or OIB-type material.