Abstract Leucogranites are a characteristic feature of collisional orogens. Their generation is intimately related to crustal thickening and the active deformation and metamorphism of metapelites. Data from Proterozoic to present day orogenic belts show that collisional leucogranites (CLGs) are peraluminous, with muscovite, biotite and tourmaline as characteristic minerals. Isotopic ratios uniquely identify the metapelitic sequences in which CLGs occur as sources. Organic material in pelitic sources results in f O 2 in CLGs that is usually below the fayalite–magnetite–quartz buffer. Most CLGs form under vapour-poor conditions with melting involving a peritectic breakdown of muscovite. The low concentrations of Mg, Fe and Ti that characterize CLGs are largely related to biotite–melt equilibria in the source rocks. Concentrations of Zr, Th and rare earth elements are lower than expected from zircon and monazite saturation models because these minerals often remain enclosed in residual biotite during melting. Melting involving muscovite may limit the temperatures achieved in the source regions. A lack of nearby mantle heat sources in thick collisional orogens has led to thermal models for the generation of CLGs that involve flux melting, or large amounts of radiogenic heat generation, or decompression melting or shear heating, the last one emphasizing the link of leucogranites and their sources to crustal-scale shear zone systems.

Leucogranites are typical products of collisional orogenies. They are found in orogenic terranes of different ages, including the Proterozoic Trans-Hudson orogen, as exemplified in the Black Hills, South Dakota, and the Appalachian orogen in Maine, both in the USA, and the ongoing Himalayan orogen. Characteristics of these collisional leucogranites show that they were derived from predominantly pelitic sources at the veining stages of deformation and metamorphism in upper plates of thickened crusts. Once generated, the leucogranite magmas ascended as dykes and were emplaced within shallower parts of their source sequences. In these orogenic belts, there was a strong connection between deformation, metamorphism and granite generation. However, the heat sources needed for partial melting of the source rocks remain controversial. Lack of evidence for significant intrusion of mafic magmas necessary to cause melting of upper plate source rocks suggests that leucogranite generation in collisional orogens is mainly a crustal process. The present authors evaluate five types of thermal models which have previously been proposed for generating leucogranites in collisional orogens. The first, a thickened crust with exponentially decaying distribution of heat-producing radioactive isotopes with depth, has been shown to be insufficient for heating the upper crust to melting conditions. Four other models capable of raising the crustal temperatures sufficiently to initiate partial melting of metapelites in thickened crust include: (1) thick sequences of sedimentary rocks with high amounts of internal radioactive heat production; (2) decompression melting; (3) thinning of mantle lithosphere; and (4) shear-heating. The authors show that, for reasonable boundary conditions, shear-heating along crustal-scale shear zones is the most viable process to induce melting in upper plates of collisional orogens where pelitic source lithologies are usually located. The shear-heating model directly links partial melting to the deformation and metamorphism that typically precede leucogranite generation.

In order to elucidate how mineralogy and composition of crustal sources influences production of leucogranite magmas, we modelled the potential fertility of a sequence of metapelites and metagraywackes from the Black Hills, South Dakota, U.S.A., using a least-squares mixing approach. Rocks analogous to the Black Hills schists were the sources of the Harney Peak leucogranite. Both muscovite and biotite fluid-absent melting reactions (MM and BM, respectively) were investigated. Using the Harney Peak Granite composition as the melt analogue and mineral compositions from the schists for mixing calculations, it is shown that MM of metapelites would lead to highly variable residue mineralogy in the investigated samples. The average residue includes 36 wt.% biotite, 32 wt.% quartz, 12wt.% plagioclase, 8 wt.% K-feldspar, 9 wt.% sillimanite and 2 wt.% garnet. Melt production ranges from 5% to 23% with an average of 14%. It is limited by the amount of H 2 O that must be in the melt at the conditions of melting, relative to the amount that is in muscovite in the source rocks. Plagioclase-rich metagraywackes contain little to no muscovite, thus MM cannot occur in them. Although BM is continuous over a wide temperature range, for the purposes of modelling melting at 975°C and 10kbar was chosen. The temperature is near the terminal stability of biotite, thus the calculations give near-maximum melt production. At this temperature, the mineralogy of the model residues from both metapelites and metagraywackes is dominated by garnet. The potential melt production in the metapelites ranges from 0% to 58% with an average of 32%. It is limited by the availability of plagioclase in the source rocks. Potential melt production in the metagraywackes ranges from 9% to 37% with an average of 23%. At the chosen conditions of melting, melt production is limited by the available K in biotite, although at lower temperatures, the available H 2 O limits melt production. The total potential melt production (MM + BM) in the metapelites is higher because they have on average a low normative An/Ab ratio (0.14) that approaches the ratio in the leucogranites (0.04). The paragonite component in muscovite significantly contributes to the low ratio in the metapelites. The higher ration (0.27) in the metagraywackes is defined by the feldspar composition. Using the calculated melt fractions and residue mineralogies, we modelled the concentrations of Rb, Sr and Ba in the melts, as these elements are important indicators of melt-generating processes. The results indicate that both Sr and Ba are likely to be heterogeneous in extracted melt batches and will be depleted in partial melts relative to their pelitic sources, irrespective of whether the melting is fluid-absent or fluid-present.