Dilution and propagation of provenance trends recorded in detrital geochronology and geochemistry from central California (U.S.A.)

Clastic sedimentary deposits represent mixtures of contributions from source areas, permitting investigation of source-to-sink transport with tools such as mixture modelling. However, biases such as variations in zircon abundance (zircon fertility, C_Zr) in the source, complicate interpretations of sediment mixing, and therefore provenance tracking tools may misrepresent source area contributions. We use modern sediment of central California to understand the conditions in which sand (detrital zircon, DZ) and mud (geochemistry) provenance trends are propagated vs diluted.

Results show that where variations in lithology and C_Zr are minimal, DZ provenance trends effectively record sources and tributaries. However, in systems with significant lithologic variability (e.g., mixed ultra-mafic, volcanic, and plutonic), DZ signatures preferentially record lithologies characterized by high C_Zr or sand flux (Q_s). Consequently, some aerially extensive source regions are poorly represented, or entirely missing, from down-stream DZ records. Introduction of a tributary enriched in zircon can result in wholesale provenance dilution within an axial drainage. This makes DZ an excellent indicator of newly introduced (spatially or temporally) felsic source regions; however due to its susceptibility to dilution, one should use caution when interpreting the absence of a population.

We compare mixture modeling from seven tributary-trunk river combinations in the Central Valley to stream-gauge-derived predictions of relative bedload flux by correcting for C_Zr based on measurements of zirconium concentration. In all cases our median estimate is within an order of magnitude of the predicted relative bedload flux. However, uncertainty in the best mixture coefficients and the observed zirconium concentration yield a wide range of relative Q_s predictions, many of which are well outside our expected range. In addition, for a significant fraction of tributary-trunk combinations studied, mixture models do not provide a satisfactory description of the observed daughter. At best, results from this study suggest that mixture modeling of detrital geochronologic data coupled with a statistically robust quantification of source lithology (C_Zr) can provide a first order approximation of relative Q_s within drainages.