GENERATION OF CRETACEOUS GRANITOIDS IN FIORDLAND, NEW ZEALAND: EXPERIMENTALLY TESTING THE ROLE OF PARTIAL MELTING OF BASALTIC UNDERPLATE

The genesis of adakite was previously proposed to occur by melting of the down-going lithosphere in a subduction zone. However, modeling and experimental studies show that such a scenario is unlikely unless the lithosphere is young, < 5Ma. Consequently, a second model was formulated in which partial melting of a basaltic underplate, in the presence of garnet, produces adakite-like melts (e.g. Sr/Y >40 or HISY). To test this hypothesis we are investigating natural HISY and HREE depleted rocks from the New Zealand arc.

The Separation Point Suite of Fiordland, New Zealand is HISY and depleted in HREE. The Median Suite is not HISY, but is HREE depleted. To test whether partial melting of a basaltic underplate formed these suites, piston-cylinder experiments have been conducted using natural, mafic dike materials collected from Fiordland. Experimental conditions are 1.4 GPa with temperatures of 800-1200°C, without added water. This pressure coincides with that of the base of the average global crustal thickness.

Major element data from partial melts show that granitoid compositions are produced from both an alkali basalt dike (SC5; hbl + plg + bt + czo) and amphibole-only dike (PM2, Fe-rich, Si-poor amph). SC5 partial melts encompass a broad spectrum of compositions, from granite at low temperatures to tonalite, at higher temperatures, and are controlled by changes in the mineral compositions and assemblages. On the other hand, partial melts derived from PM2 are only tonalitic. Using LA-ICPMS analyses, HISY, HREE-depleted melts were confirmed in SC5 experiments at 1.4 GPa and 975-1050°C. The solidus of SC5 at 1.4GPa is 825°C and the liquidus is 1100°C. With PM2, a plag-free starting material, HREE-depleted melts are produced below 1050°C, but all partial melt compositions have a LOSY signature. Melts that are HREE-depleted coexist only within the garnet-stability range (925°C to 1075°C), indicating that garnet is the phase controlling HREE signatures of the melt. The presence plagioclase in the starting material is an important control on HISY or LOSY signature. These experimental results provide evidence that partial melting of basaltic material in the deep crust is a possible source for the HREE depleted suites of Fiordland, both HISY and LOSY, dependent on the presence or absence of plagioclase in the source.