stage of this island arc when shallower hornblende-bearing plutonic rocks were buried to depths exceeding 25-30 km and heated to temperatures above c. 900 C. Available experimental data on dehydration-melting of amphibolitic sources imply that thickening of oceanic arcs to depths >30 km (equivalent to c. 1 . 0 GPa), together with the hot geotherms now postulated for lower island arc crust, should cause dehydration-melting of amphibole-bearing plutonic rocks generating dense garnet granulitic roots in island arcs. Dehydration-melting of hornblende-bearing plutonic rocks may, hence, be a common intracrustal chemical and physical differentiation process in island arcs and a natural consequence of their maturation, leading to the addition of granitic partial melts to the middle-upper arc crust and formation of dense, unstable garnet granulite roots in the lower arc crust. Addition of LREE-enriched granitic melts produced by this process to the middle-upper island arc crust may drive its basaltic composition toward that of andesite, affording a plausible solution to the 'arc paradox' of formation of andesitic continental-like crust in island arc settings.

Detailed geological mapping in the Drosh-Shishi area in southern Chitral (NW Pakistan) was combined with high-precision U-Pb dating on zircons to constrain the timing of magmatism and associated deformation/metamorphic events related to the Kohistan-Karakoram convergence and collision. Our new ages indicate that the Mesozoic to Tertiary magmatic history of this region is influenced by long-lived melt generation above an active subduction zone. Dated intrusive rocks range in age from 130 to 39 million years, indicating that subduction-related magmatism continued after the Kohistan-Karakoram and the India-Asia collisions. Initial hafnium isotope ratios were measured on the dated zircons to constrain the type of melt source of the dated plutons. The data reveal the different nature of partly coeval magmatism in these units, i.e. continental arc magmatism in the Karakoram (ca. 130-104 Ma) and arc magmatism on the Kohistan side (112-39 Ma). Intrusions within the suture zone can be clearly traced to be Karakoram-derived on the basis of initial Hf isotopic compositions. Granite dykes crosscutting the Kohistan units have sampled an underlying, old continental basement of Gondwana affinity. The geochronological evidence presented in this paper is consistent with Cretaceous subduction beneath the Karakoram Terrane. The related calc-alkaline magmatism seems to have stopped at about 100 Ma. Granite dykes on the Kohistan side show that the magmatic and tectonic history of the KarakoramKohistan Suture Zone continued to the Eocene. This long tectono-metamorphic and magmatic activity in the arc plates was likely due to complex and few million year long interplays between subduction and thrusting events in the forearc, within-arc and back-arc regions between two active subduction zones.

No abstract

We present major and trace element analyses and U-Pb zircon intrusion ages from I-type granitoids sampled along a crustal transect in the vicinity of the Chilas gabbronorite of the Kohistan paleo-arc. The aim is to investigate the roles of fractional crystallization of mantlederived melts and partial melting of lower crustal amphibolites to produce the magmatic upper crust of an island arc. The analyzed samples span a wide calc-alkaline compositional range (diorite-tonalite-granodiorite-granite) and have typical subduction-related trace element signatures. Their intrusion ages (75.1 ± 4.5-42.1 ± 4.4 Ma) are younger than the Chilas Complex (*85 Ma). The new results indicate, in conjunction with literature data, that granitoid formation in the Kohistan arc was a continuous rather than punctuated process. Field observations and the presence of inherited zircons indicate the importance of assimilation processes. Field relations, petrographic observations and major and trace element compositions of the granitoid indicate the importance of amphibole fractionation for their origin. It is concluded that granitoids in the Kohistan arc are derivative products of mantle derived melts that evolved through amphibole-dominated fractionation and intra crustal assimilation.

If the net flux to the island arc crust is primitive arc basalt, the evolved composition of most arc magmas entails the formation of complementary thick ultramafic keels at the root of the island arc crust. Dunite, wehrlite, and Cr-rich pyroxenite from the Jijal complex, constituting the Moho transition zone of the Kohistan paleo–island arc (northern Pakistan), are often mentioned as an example of high-pressure cumulates formed by intracrustal fractionation of mantle-derived melts, which were later extracted to form the overlying mafic crust. Here we show that calculated liquids for Jijal pyroxenites-wehrlites are strongly rare earth element (REE) depleted and display flat or convex-upward REE patterns. These patterns are typical of boninites and are therefore unlike those of the overlying mafic crust that have higher REE concentrations and are derived from light rare earth element (LREE)–enriched melts similar to island arc basalt. This observation, along with the lower 208Pb/204Pb and 206Pb/204Pb ratios of Jijal pyroxenites-wehrlites relative to gabbros, rejects the hypothesis that gabbros and ultramafic rocks derive from a common melt via crystal fractionation. In the 208Pb/204Pb versus 206Pb/204Pb diagram, ultramafic rocks and gabbros lie on the same positive correlation, suggesting that their sources share a common enriched mantle 2 (EM2) signature but with a major depleted component contribution for the ultramafic rocks. These data are consistent with a scenario whereby the Jijal ultramafic section represents a Moho transition zone formed via melt-rock reaction between subarc mantle and incoming melt isotopically akin to Jijal gabbroic rocks. The lack in the Kohistan arc of cogenetic ultramafic cumulates complementary to the evolved mafic plutonic rocks implies either (1) that a substantial volume of such ultramafic cumulates was delaminated or torn out by subcrustal mantle flow from the base of the arc crust in extraordinarily short time scales (0.10–0.35 cm/yr), or (2) that the net flux to the Kohistan arc crust was more evolved than primitive arc basalt