| Abstract: | Calc-alkalic chemical trends characteristic of arc volcanic rocks mainly result from three mechanisms which act additively: (1) fractional crystallization involving separation of titanomagnetite; (2) selective concentration of plagioclase phenocrysts and selective depletion of titanomagnetite phenocryst compared with the actually fractionated proportion; and (3) mixing of magmas on continuous fractionation trends. The association of calc-alkalic and tholeiitic trends in a single composite volcano may not represent different fractional crystallization processes or different chemistries of primary magmas, but the calc-alkalic chemical trend can be considered as a mixing trend resulting from mixing of various magmas on associated tholeiitic chemical trends. Chemical variations of most arc volcanic rocks, including calc-alkalic ones, can accordingly be essentially accounted for by the low-pressure fractional crystallization of phenocrystic phases from primary basaltic magmas.Crystallization sequences of arc magmas which are strongly dependent on water content in magmas are deduced from the phenocryst assemblages. The crystallization sequence changes laterally across-arc, suggesting increasing water contents in magmas toward the back-arc side, as is also seen for other incompatible elements such as K and Rb. Systematic differences in the characteristic crystallization sequence are also observed among arcs, roughly correlating with the crustal thickness. Water content in magma, like other incompatible elements, tends to increase with increasing crustal thickness. The variation of incompatible elements including water roughly represents that of the degree of partial melting of the upper mantle, which is broadly controlled by the crustal thickness.The variation of water content indicates that arc magmas are not saturated with water during differentiation to late differentiates such as dacite or rhyolite. This strongly constrains the maximum water contents in primary basaltic magma, at most 2.5 wt.%. This value suggests that magma generation beneath arcs is dependent on dry solidus of peridotite. Diapiric uprise of the hot deeper mantle and associated adiabatic decompression would be necessary for mantle peridotite to attain the temperature as high as dry solidus. Diapirs that begin to rise from the subduction zone may stop at or near the crust-mantle boundary because of the surrounding density change, and their degree of partial melting is roughly controlled by their stopped depth assuming their similar temperature. Across-arc variation is also explained by the stopped depth of diapirs, but is not controlled by crustal thickness. |
