LAUSR

Abstract Continental crust in average exhibits a similar composition in both major and trace elements to andesites along active continental margins. For this reason, andesitic magmatism above oceanic subduction zones is considered to have played a key role in the generation of continental crust along convergent plate boundaries. With respect to the origin of andesites themselves, however, there is still a hot debate on how they have acquired their geochemcial compositions. The debate is mainly centralized on the relative contributions of crustal contamination, magma differentiation and source mixing, which reaches an impasse in the past decades. The essential reason for this kind of debates is that these three mechanisms only can account for some of the geochemical observations for andesites, leading to insufficient discrimination among them. Nevertheless, the geochemical features of andesites are primarily controled from early to late by the composition of their source rocks in addition to partial melting and magma differentiation processes. If source mixing and partial melting processes in the early stage of andesite magmatism can account for the first-order geochemical features of andesites, there is no need to invoke the late processes of magma differentiation and crustal contamination for andesite petrogenesis. This is illustrated by quantitative forward modeling of the geochemical data for Quaternary andesites from the Andean arc in South America based on an integrated interpretation of these data. The modeling has run with four steps from early to late: (1) dehydration of the subducting oceanic crust at forearc depths; (2) partial melting of the subducting terrigenous sediment and altered oceanic basalt at subarc depths to produce hydrous felsic melts; (3) the generation of basaltic metasomatites (e.g., Si-excess pyroxenite) in the mantle wedge through reaction of the mantle wedge peridotite with large amounts of the hydrous felsic melts; (4) the production of andesitic melts by partial melting of the basaltic metasomatites. The results not only testify the hypothesis that the trace element and radiogenic isotope compositions of andesites can be directly produced by the source mixing and mantle melting but also demonstrate that partial melting of the basaltic metasomatites can reproduce the lithochemical composition of andesites. The compositional variations of Andean andesites within a single volcanic zone and among different volcanic zones can be explained by incorporating different amounts of heterogeneous hydrous felsic melts into their mantle sources, followed by different degree of partial melting under different pressures and temperatures. Therefore, the source mixing and partial melting processes at subarc depths can account for the first-order geochemical features of Andean andesites. In this regard, it may be not necessary for andesite petrogenesis to invoke the significant contributions from the processes of magma differentiation and crustal contamination.