How and at which thermal conditions the convergence between the Chinese Altai and East Junggar operated remain poorly understood. This issue is addressed in the current study by focusing on the timing and petrogenesis of syntectonic granite dykes from the representative areas of Fuyun (convergent front) and Kalasu-Aletai (Chinese Altai interior). It is shown that Fuyun and Kalasu-Aletai dykes are fractionated I- and S-type granites, with zircon and monazite U-Pb ages of 300−291 Ma and 281−265 Ma, respectively. Geochemically, the Fuyun dykes have lower contents of aluminous (ASI: 0.97−1.13) and light rare earth element-enriched features, while the Kalasu-Aletai dykes have ASI = 1.01−2.17 and show overall flat rare earth element patterns with tetrad effects. The Fuyun dykes exhibit less evolved Sr-Nd isotopic characteristics (87Sr/86Srinitial: 0.7039−0.7048, εNd(t): + 5.7 to + 6.1) with respect to those of the Kalasu-Aletai dykes (87Sr/86Srinitial: 0.6978−0.7183, εNd(t): −7.6 to +3.0). The Fuyun and Kalasu-Aletai dykes are geochemically compatible with isotopically less evolved East Junggar arc components and heterogeneous Ordovician wedge sediment of the Chinese Altai, respectively, implying genetic links. We propose that the late Paleozoic Chinese Altai−Junggar convergence created a local perturbation of weak mantle beneath the southern Chinese Altai, causing partial melting of the underthrusting East Junggar and the overriding Altai components successively. The resulting magmas were emplaced along northward propagating syn-tectonic tensional fractures perpendicular to the Chinese Altai−East Junggar deformation front that serves as an excellent indicator of the convergent-shortening process.

<p>The common view of melt transport in the continental crust involves an initial stage of percolation along grain boundaries, melt segregation into leucosomes and dykes, coalescence of small melt conduits into larger ones and quick nearly vertical melt flow leading to formation of plutons. An entirely different style of melt migration was described in the Bohemian Massif, eastern European Variscan belt. There, a sequence of metaigneous migmatites was described where veins are lacking, leucosomes are rare and relics of melt are spread along grain boundaries. Textural, geochemical and compositional variations in these rocks show that they formed due to equilibration with melt coming from an external source, and that pervasive flow along grain boundaries was the dominant mechanism of melt transport.</p><p>The question arises, at what conditions this style of melt transport can operate and what consequences the different styles of melt transport have on the crustal-scale tectonics. We address this question by means of a 2D crustal-scale model of two-phase flow using the code ASPECT (aspect.geodynamics.org). The system of pores through which the melt flows is not resolved in our model and it is described only by its permeability. A low permeability describes material with pores along grain boundaries while a high permeability corresponds to a system of leucosomes, dykes or cracks</p><p><span>For different material properties and thermal conditions we obtain different styles of melt migration and characteristics of the modeled crust. The melt can form a diffuse zone in the lower–middle crust, km-scale waves of high melt fraction gathering into sub-vertical channels, or a horizontal zone with high melt fraction in the middle crust. The lower crust is depleted and the middle crust is enriched in incompatible elements, and composition of the middle crust typically shows km-scale variations. The compositional variations are obtained even in the models with low permeability that corresponds to the melt percolation along grain boundaries, in agreement with the characteristics of the Bohemian migmatites.</span></p>