A multi-proxy perspective on stream sediment provenance and anthropogenic impacts in an ore-exploited, geologically complex river basin

Stream sediments integrate the geochemical and mineralogical signals of entire drainage basins, reflecting both bedrock lithology and anthropogenic inputs. This study applies knowledge-driven Sequential Binary Partitions (SBPs) of isometric log-ratio (ilr)-transformed data to unravel relationships among geochemistry, mineralogy, and mass-specific magnetic susceptibility (χ), with an approach tested on a representative river basin (i.e., Ombrone Grossetano River Basin, Southern Tuscany, Italy), lithologically diverse and impacted by past mining activities. The method aims to improve the understanding of natural and anthropogenic influences on sediment composition. Geochemical data (X-ray fluorescence) were transformed through ilr-balances, whereas mineralogical compositions (X-ray diffraction), lithological information, and χ measurements were used to support the interpretation of sediment sources. Strong correlations between ilr-balances and mineral phases confirm that the geochemical signals reliably reflect sediment mineralogy and the basin's lithological variability. Comparisons between χ and ilr-balances detect localized anthropogenic inputs, notably Fe-enrichments linked to historical mining in a specific sub-basin (Farma-Merse). The combined use of compositional data analysis, mineralogical constraints and magnetic susceptibility provides a transferable framework to investigate sediment provenance and anthropogenic legacy in complex river systems. Beyond the case study, the results highlight the importance of considering whole-composition geochemical relationships to correctly interpret element distributions in fluvial sediments.