Quantifying the loss of continental crust into the mantle from volume/mass balance calculations in modern collisional mountains

• Around ∼20–50% of crust shortening during continental collision is unaccounted for by thickening, extrusion and erosion. • The missing crust has been recycled into the mantle. • Crust lost to the mantle exceeds surficial erosion by 2 times in Himalaya-Tibet, Zagros and probably Alps. • The imbalance during the India-Asia collision is due to delamination of dense crustal roots. Reworking and recycling of continental crust, through processes such as erosion and delamination, are essential geological mechanisms that not only shape the topography of continents but also influence the composition of the continental crust and mantle. Continent-continent collisions are crucial settings to study these processes, as they primarily involve the thickening and uplift of the existing crust, with little new crustal addition compared with ocean-continent convergent plate boundaries. In this study, we investigate the three modern collisional systems that formed the Himalaya-Tibetan Plateau, the European Alps, and Zagros in central Asia, and quantify the amount of crust lost into the mantle by comparing the shortened crustal volume with the present-day preserved thickened crust, laterally extruded crust and surficial eroded crust. We find that crustal loss into the mantle accounts for at least 30% of the shortened crust, which exceeds the crust lost by surficial erosion by at least a factor of 2 in the Himalaya-Tibetan Plateau and Zagros. The volume of crust lost into the mantle during the formation of the Alps lies between 15% and 50%, depending on the values assumed for the pre-collisional crustal thickness and the volume of eroded crust. For the Himalaya-Tibetan Plateau, our calculated crustal loss corresponds to an elevation increase of ∼ 2 km, which can be explained by delamination of thick, eclogitised lower crustal roots in the late Oligocene, consistent with the distribution of shoshonitic-adakitic magmatism in southern Lhasa. This phase of rapid uplift, which followed the removal of dense lower lithosphere, corresponds with monsoon intensification in southern Asia. Furthermore, extending the concept of crustal loss to ancient mountain belts that occurred during the past cycles of supermountain formation, we propose that detachment of lower crustal roots can explain the trace element and isotopic characteristics of exotic crustal components in some plume-related mantle melts, ultimately linking mountain-building and mantle heterogeneity on a multi-million-year timescale.