Radiogenic strontium isotopes are routinely used in provenance studies, but their application to sediments is often complicated by various grain size and weathering effects, which can influence measured 87Sr/86Sr ratios. Here, we report Sr isotopic data for a large number of sediment samples (n = 61) from the world's largest rivers and other river catchments draining particular geological and climatic settings; using both clay-rich (<4 μm) and silt-size (4-63 μm) detrital fractions to re-examine the factors controlling their distribution in sediments. In agreement with previous studies, the detrital material transported by world rivers defines general inverse Nd-Sr isotope relationships, which provide further empirical evidence for the utility of radiogenic Sr isotopes in sediment provenance studies. In a novel departure, however, we show that the 87Sr/86Sr difference between paired clay- and silt-size fractions (Δ87Sr/86Sr Clay-Silt) relates to the degree of chemical alteration of river sediments, as inferred from various relationships with weathering indices, such as the CIA, WIP and Na2O/Al2O3. The weathering dependence of Δ87Sr/86Sr Clay-Silt appears to be mainly controlled by temperature. In sub-Arctic and temperate regions, river sediments systematically display positive Δ87Sr/86Sr Clay-Silt indicative of preferential alteration of biotite in soils. In contrast, in sub-tropical watersheds characterized by mean annual temperatures >20 °C, intense feldspar weathering leads to the preferential incorporation of unradiogenic Sr into secondary clay minerals; a process which results in negative Δ87Sr/86Sr Clay-Silt values. In addition to climate forcing, the degree of size-dependent Sr isotope decoupling is also shown to be dependent on the type of weathering regime in watersheds, being more pronounced in low-elevation environments (< 2000 m), where transport-limited conditions and the presence of thick soil sequences can be associated with intense silicate weathering, than in high mountain regions (> 4000 m) dominated by kinetically-limited weathering regimes.

While further studies will be required to test the validity of these conclusions at the local scale of weathering profiles, these findings suggest that combined Sr isotopic analyses of separate size fractions could be used as a new weathering proxy in sediment records, ideally complementing the conventional use of radiogenic Sr isotopes as provenance tracers. Finally, our results are also used to re-assess the mean Sr flux and 87Sr/86Sr composition of the suspended sediment exported to the ocean yearly, yielding a global flux-weighted average of 0.7160, identical to that proposed earlier in the seminal study of Goldstein and Jacobsen (1988).