Stable Silicon Isotopes Uncover a Mineralogical Control on the Benthic Silicon

Biogeochemical cycling of silicon (Si) in the Barents Sea is under considerable pressure from physical and chemical changes, including dramatic warming and sea ice retreat, together with a decline in dissolved silicic acid (DSi) concentrations of Atlantic inflow waters since 1990. Moreover, further expansion of the Atlantic realm (termed ‘Atlantification’) is expected to shift phytoplankton community compositions away from diatom-dominated spring blooms in favour of Atlantic flagellate species (coccolithophore-dominated). The changes in pelagic primary production will alter the composition of the material comprising the depositional flux, which will subsequently influence the recycling processes at and within the seafloor. In this study we assess the predominant controls on the early diagenetic cycling of Si, a key nutrient in marine ecosystems, by combining stable isotopic analysis of pore water DSi and of operationally defined reactive pools of the solid phase. We show that low biogenic silica (BSi) contents (0.39-0.52 wt% or 92-185 μmol g dry wt−1) drive correspondingly low asymptotic Preprint submitted to Geochimica et Cosmochimica Acta November 12, 2021 concentrations of pore water DSi (∼100 μM). However, while these surface sediments appear almost devoid of BSi, we present evidence for the rapid recycling of bloom derived BSi that generates striking transient peaks in sediment pore water [DSi], which is a feature that is subject to future shifts in phytoplankton community compositions. Using a simple mass balance calculation we show that the pore water DSi pool is supplemented by a lithogenic Si source (LSi), while our sediment pore water Si isotopic profiles also uncover a coupling of the iron (Fe) and Si cycles. This has previously been observed in lower latitude marine sediment systems and thus provides further support for a redox influence on oceanic pore water DSi. We suggest that a high LSi:BSi ratio and apparent Fe (oxyhydr)oxide influence could lead to a degree of stability in the annual background benthic flux of DSi despite the pressures on pelagic phytoplankton communities. Coupled with supporting isotopic evidence for the precipitation of authigenic clays in Barents Sea sediment cores, our observations have implications for the sink vs recycling terms in the regional Si budget.

plate (sampling resolution of 0.5 cm intervals from 0-2 cm below seafloor (cmbsf), 1 cm from 98 2 cmbsf), which were then stored at -20 o C. For the dissolved phase, the overlying core top 99 water was collected first, after which pore water samples were extracted with Rhizon filters 100 attached to 30 mL plastic syringes, using spacers to create a vacuum (sampling resolution 101 of 1 cm from 0.5-2.5 cmbsf, 2 cm to 20.5 cmbsf, 5 cm to 35.5 cmbsf). Pore waters were then 102 acidified with Romil UpA HCl and stored at 4 o C.

103
For sediment pore water element concentration analysis, pore waters were collected from 104 three separate Multicorer deployments at each station and year (Fig. 2). One of the replicate 105 deployments for each year at B13, B14 and B15 were also sampled for Si isotopic analysis 106 (Fig. 3). These three stations span the three main hydrographic domains of the Barents Sea and diluted with Milli-Q.

182
The pre-concentrated sea and pore water samples, filtered solid phase leachates and 183 reference standards were all passed through cation exchange columns, following Georg et al.

236
The flux magnitude uncertainties were obtained from the error on the gradient of the linear 237 regression (Fig. 4).
, where θ represents sediment tortuosity, φ is porosity in the surface sediment, D sed is the is 246 the diffusion coefficient of DSi in seawater (D sw ) corrected for tortuosity (Boudreau, 1996) (Table 1).
, where D sw is in cm 2 s −1 , T in kelvin and η in poises (g cm −1 s −1 ). BSi contents were measured in the surface sediment interval (0-0.5 cmbsf) across the 293 three sites for 2019 samples, which ranged from 0.39-0.52 wt% (92-185 µmol g dry wt −1 ),        Table   390 2). There is little change in δ 30 Si DSi−Inc across the incubation at B15, although the two sam-  (Table 2; Table 3). This discrepancy is most apparent across the incubation period to determine the theoretical composition of BSi (equation 6). 406 We find that the dissolving phase would require a composition of +4.5, +2.7 and +1.9‰ where CT refers to the core top water and f represents the mixing fraction between the two phases. δ 30 Si mix was calculated across a range of f values.

431
The pore water isotopic data do not fall on the calculated mixing lines plotted in Fig. 6A ranging from 94-306 µM (Fig. 2), which is much lower than the theoretical solubility of  released into the DSi pool (Fig. 3). Through a simple mass balance, akin to equation 6, we   in the NaOH leachates of B13, B14 and B15 (Fig. S2). These values are higher than the  a fractionation factor of -1.18‰ or lower (Fig. 6B).

520
To summarise, the benthic Si cycle of the Barents Sea cannot be characterised as a   This increase in the δ 30 Si DSi−P W is unlikely to be caused by the dissolution of a solid phase, as the δ 30 Si DSi−P W at 3.5 cmbsf at the three stations increases to higher values than that measured in the operational pools, especially at B15 (Fig. 5B). Additionally, dissolution 541 would not resolve the relative shift from 0.5 cmbsf to 3.5 cmbsf observed at B14 and B15 542 (Fig. 3), which requires enrichment in the heavier isotope downcore.

Evidence for a redox influence on the benthic Si cycle
Below 3.5 cmbsf at B13 and B14 and below 10.5 cmbsf at B15, we see an enrichment in 569 the lighter isotope downcore across all cruise years (Fig. 3) in addition to a general trend 570 towards increased [DSi] towards the base of the cores at B13 and B14 (Fig. 2), albeit at  The ubiquitous presence and desorption of Si from this Fe phase at the three stations could 593 explain the 28 Si enrichment we observe across the oxic-anoxic boundaries, as well as the Examination of the [Fe] pore water profiles of the same sampling stations indicates that the light isotope enrichment occurs at a similar depth interval to where Fe appears in the pore 597 water phase (Fig. 3). This is consistent with a change in redox state to anoxic conditions, with the different depths of the redox boundaries found at the two sites, which is shallower 607 at B13 than at B15 (Fig. 3). 608 It has previously been suggested for sediments of the Greenland Shelf that the reductive  asymptotic form (Fig. 3, Table 2).

642
Oceanic frontal zones are highly dynamic and the PF (B14, Fig. 1 to the 2018 cruise (Fig. 7). In 2018 the MIZ in the Barents Sea retreated more rapidly and Si budget, as the conversion of terrestrially-derived LSi to AuSi represents a true sink term, 714 as opposed to a recycling term, which is the case when AuSi precipitation reflects a solid 715 phase conversion from BSi. 716 We show that fresh BSi derived from pelagic phytoplankton blooms is rapidly recycled in