Description: In this study we combine for the first time silicon (Si) isotope compositions of small mixed diatom species (δ30SibSiO2) and of large handpicked mono-generic (i.e. genus = Coscinodiscus) diatom samples (δ30SiCoscino) with diatom assemblages extracted from marine sediments in the Peruvian upwelling region in order to constrain present and past silicate utilisation. The extension of a previous core-top data set from the Peruvian shelf demonstrates that δ30SiCoscino values record near-complete Si utilisation, as these are similar to the isotopic composition of the subsurface source waters feeding the upwelling. In contrast, the δ30SibSiO2 of small mixed diatom species increase southward along the shelf as well as towards the shore. We attribute highest δ30SibSiO2 values partly to transient iron limitation but primarily to the gradual increase of Si isotope fractionation within the seasonal diatom succession, which are mainly recorded by small diatom species during intense bloom events. In contrast, lower δ30SibSiO2 values are related to initial Si isotope utilisation during periods of weak upwelling, when low Si(OH)4 concentrations do not permit intense blooms and small diatom species record substantially lower δ30Si signatures. As such, we propose that the intensity of the upwelling can be deduced from the offset between δ30SibSiO2 and δ30SiCoscino (Δ30Sicoscino–bSiO2), which is low for strong upwelling conditions and high for prevailing weak upwelling. We apply the information extracted from surface sediments to generate a record of the present-day main upwelling region covering the past 17,700 years and find that this location has also been characterized by a persistent offset (Δ30Sicoscino–bSiO2). By comparison with the diatom assemblages we show that the coastal upwelling system changed markedly between weak and strong upwelling conditions. In addition, our model calculations to quantify species-specific Si isotope fractionation effects based on the diatom assemblages indicate an overall minor influence that cannot explain the high amplitude in the measured δ30SibSiO2 record.

Description: The cause of massive blooms of Ethmodiscus rex laminated diatom mats (LDMs) in the eastern Philippine Sea (EPS) during the Last Glacial Maximum (LGM) remains uncertain. In order to better understand the mechanism of formation of E. rex LDMs from the perspective of dissolved silicon (DSi) utilization, we determined the silicon isotopic composition of single E. rex diatom frustules (δ30SiE. rex) from two sediment cores in the Parece Vela Basin of the EPS. In the study cores, δ30SiE. rex varies from −1.23‰ to −0.83‰ (average −1.04‰), a range that is atypical of marine diatom δ30Si and that corresponds to the lower limit of reported diatom δ30Si values of any age. A binary mixing model (upwelled silicon versus eolian silicon) accounting for silicon isotopic fractionation during DSi uptake by diatoms was constructed. The binary mixing model demonstrates that E. rex dominantly utilized DSi from eolian sources (i.e., Asian dust) with only minor contributions from upwelled seawater sources (i.e., advected from Subantarctic Mode Water, Antarctic Intermediate Water, or North Pacific Intermediate Water). E. rex utilized only ~24% of available DSi, indicating that surface waters of the EPS were eutrophic with respect to silicon during the LGM. Our results suggest that giant diatoms did not always use a buoyancy strategy to obtain nutrients from the deep nutrient pool, thus revising previously proposed models for the formation of E. rex LDMs.