LAUSR

The ~ 570 Ma old Khantaishir ophiolite is built by up to 4 km harzburgitic mantle with abundant pyroxenites and dunites followed by ~ 2 km of hornblende-gabbros and gabbronorites and by a ~ 2.5 km thick volcanic unit composed of a dyke + sill complex capped by pillow lavas and some volcanoclastics. The volcanics are mainly basaltic andesites and andesites (or boninites) with an average of 58.2 ± 1.0 wt% SiO2, XMg = 0.61 ± 0.03 (XMg = molar MgO/(MgO + FeOtot), TiO2 = 0.4 ± 0.1 wt% and CaO = 7.5 ± 0.6 wt% (errors as 2σ). Normalized trace element patterns show positive anomalies for Pb and Sr, a negative Nb-anomaly, large ion lithophile elements (LILE) concentrations between N- and E-MORB and distinctly depleted HREE. These characteristics indicate that the Khantaishir volcanics were derived from a refractory mantle source modified by a moderate slab-component, similar to boninites erupted along the Izu-Bonin-Mariana subduction system and to the Troodos and Betts Cove ophiolites. Most strikingly and despite almost complete outcrops over 260 km2, there is no remnant of any pre-existing MORB crust, suggesting that the magmatic suite of this ophiolite formed on completely denudated mantle, most likely upon subduction initiation. The architecture of this 4–5 km thick early arc crust resembles oceanic crust formed at mid ocean ridges, but lacks a sheeted dyke complex; volcanic edifices are not observed. Nevertheless, low melting pressures combined with moderate H2O-contents resulted in high-Si primitive melts, in abundant hornblende-gabbros and in a fast enrichment in bulk SiO2. Fractional crystallization modeling starting from the observed primitive melts (56.6 wt% SiO2) suggests that 25 wt% pyroxene + plagioclase fractionation is sufficient to form the average Khantaishir volcanic crust. Most of the fractionation happened in the mantle, the observed pyroxenite lenses and layers in and at the top of the harzburgites account for the required cumulate volumes. Finally, the multiply documented occurrence of highly depleted boninites during subduction initiation suggests a causal relationship of subduction initiation and highly depleted mantle. Possibly, a discontinuity between dense fertile and buoyant depleted mantle contributes to the sinking of the future dense subducting plate, while the buoyant depleted mantle of the future overriding plate forms the infant mantle wedge.