This paper gives the results of a set of laboratory experiments designed to analyse the petrological implications of mantle wedge plumes—large buoyant structures predicted by thermomechanical numerical modelling of subduction zones. A particular design of layered capsule was used to simulate the complex multilayer formed by intense flow within the mantle wedge as predicted by numerical models. A basaltic [mid-ocean ridge basalt (MORB)-derived amphibolite] component was sandwiched between two adjacent layers of a sedimentary (Bt-rich metagreywacke) component. Conditions were fixed at temperatures of 1000–1200°C at pressures of 1·5–2·0 GPa. Our results suggest that significant volumes of hybrid, Cordilleran-type granodioritic magmas can be generated by sub-lithospheric partial melting of a mechanically mixed source. Partial melting of the end-members is not buffered, forming granitic (melting of metasediment) and trondhjemitic (melting of MORB) melts in high-variance assemblages Melt + Grt + Pl and Melt + Grt + Cpx, respectively. However, the composition of melts formed from partial melting of metasediment–MORB mélanges is buffered for sediment-to-MORB ratios ranging from 3:1 to 1:3, producing liquids of granodiorite to tonalite composition along a cotectic with the lower-variance phase assemblage Melt + Grt + Cpx + Pl. Our model explains the geochemical and isotopic characteristics of Cordilleran batholiths. In particular, it accounts for the observed decoupling between major element and isotopic compositions. Large variations in isotopic ratios can be inherited from a compositionally heterogeneous source; however, major element compositions are more strongly dependent on the temperature of melting rather than on the composition of the source.

We have studied experimentally the melting relationships of the Ollo de Sapo gneiss (OSG), an important crustal protolith for the Iberian leucogranites, of possible volcanoclastic origin. The results of this study are compared with previously determined PTt paths, allowing us to interpret the mechanisms of melting and granitoid production during the Hercynian orogenic cycle. Phase relationships determined in fluid-absent experiments indicate that the OSG is a fertile source for peraluminous leucogranites. The slope of the fluid-absent solidus is strongly controlled by the breakdown of Ms in the presence of Qtz, Pl and Kfs. This solidus curve has a positive slope ranging from dP/dT = 30 bar/°C at low P (<6 kbar) to dP/dT = 70 bar/°C at higher P (6–15 kbar). The relationships between the Ms vapour-absent solidus and the PTt metamorphic paths in different sectors of the Iberian massif have two important implications: (1) melt productivity is strongly favoured at low P; (2) anatexis in the Iberian massif probably took place by decompression associated with crustal thinning and extension. These results are in agreement with the relationships between granite production and tectonic deformation phases observed in the Iberian massif. Our results emphasize that anatexis is a process that is strongly controlled both by the phase relationships of the crustal protoliths and by the thermal structure of the continental crust. Consequently, one must be careful when assigning potential crustal protoliths to particular granite associations exclusively on the basis of geochemical comparisons.