Slab and Sediment Melting during Subduction Initiation: Mantle Plagiogranites from the Oman Ophiolite

Granitoid dykes up to several hundred metres wide and 2 km long are found in depleted harzburgites in the mantle section of the Oman ophiolite. They vary in composition from tonalite to potassic granite and are generally more potassic than the crustal plagiogranites found within the sheeted dyke complex higher up within the ophiolite stratigraphy. Some granites are strongly peraluminous and contain garnet and andalusite. They are geochemically variable, some with REE that are relatively unfractionated ((La/Yb)_n= 3.5-6.0, flat middle to heavy REE, steep light REE) to those which are highly fractionated ((La/Yb)_n= 28-220). On primitive-mantle normalised plots some have very high concentrations of fluid-mobile elements – Cs, Rb, Th, U and Pb. Few have significant Ta-Nb anomalies. On the Ca-Fe-Mg-Ti discrimination diagram of Patino Douce (J. Petrol., 1999) whole-rock compositions define a spectrum between felsic-pelite derived melts and amphibolite-derived melts.

There is a chemical similarity between the least REE fractionated plagiogranites (generally tonalites and granodiorites) and melts of an amphibolitic parent. This is supported by the occurrence of mafic xenoliths in some dykes, the presence of hornblende and highly calcic cores (up to An₈₅) in some plagioclase grains. Trace element modelling using Oman Geotimes lavas as the starting composition indicates that melting took place in the garnet stability field, although enrichment in the melt in Cs, Rb, Ba and Pb suggests that there was another component present in addition to the mafic parent. Other plagiogranites (trondhjemites and granites) have a strongly peraluminous chemistry and mineralogy and geochemical similarities with the Himalayan leucogranites implying that they were derived from a sedimentary protolith. These mantle plagiogranites are more prevalent in the northern outcrops of the ophiolite.

The volume of granitoid melt and the depth of melting preclude their derivation from the sole of the ophiolite, rather they were derived during subduction by the partial melting of the slab and associated sediment and emplaced into the overlying mantle wedge. Current subduction-initiation models for supra-subduction ophiolites should integrate this process into their thinking.