Which minerals control the Nd-Hf-Sr-Pb isotopic compositions of river sediments?

River sediments naturally sample and average large areas of eroded continental crust. They are ideal targets not only for provenance studies based on isotopic compositions, but also to establish average continental crust isotopic values. However, in large fluvial systems, mineral sorting processes significantly modify the mineralogy, and thus the geochemistry of the transported sediments. We still do not know, in any quantitative way, to what extent mineral sorting affects and fractionates the isotopic compositions of river sediments. Here, we focus on this issue and try to decipher the role of each mineral species in the bulk isotopic compositions of bedloads and suspended loads sampled at the outflow of the Ganga River that drains the Himalayan mountain range. We analyzed Nd, Hf, Sr and Pb isotopic compositions as well as trace element contents of a large number of pure mineral fractions (K-feldspar, plagioclase, muscovite, biotite, magnetite, zircon, titanite, apatite, monazite/allanite, amphibole, epidote, garnet, carbonate and clay) separated from bedload sediments. We combine these data with mineral proportions typical of the Ganga sediments to perform Monte-carlo simulations that quantify the contribution of individual mineral species to the Nd, Hf, Sr and Pb isotopic budgets of bedloads and suspended loads. We show that the isotopic systematic of river sediments is entirely buffered by very few minerals. Despite their extremely low proportions in sediments, zircon and monazite/allanite control Hf and Nd isotopes, respectively. Feldspars, epidote and carbonate buffer the Sr isotopic budget while clay, feldspars and heavy minerals dominate Pb isotopes. We also demonstrate that the observed difference in Hf, Sr and Pb isotopic compositions between bedloads and suspended loads reflects their different mineral proportions. Our findings highlight the need to be very careful about the choice of isotopic compositions measured on sediments when used as source proxies. In addition, we anticipate that fractionation of the isotopic systems continues at the river/ocean interface to deliver to the deep ocean sediments that are not necessarily similar to their crustal precursors, creating a long-term and systematic bias between the compositions of crustal sources and oceanic sediments.