Sedimentary molybdenum and uranium: Improving proxies for deoxygenation in coastal depositional environments

Sedimentary molybdenum (Mo) and uranium (U) enrichments are widely used to reconstruct changes in bottom water oxygen conditions in aquatic environments. Until now, most studies using Mo and U have focused on restricted suboxic-euxinic basins and continental margin oxygen minimum zones (OMZs), leaving mildly reducing and oxic (but eutrophic) coastal depositional environments vastly understudied. Currently, it is unknown: (1) to what extent Mo and U enrichment factors (Mo-and U-EFs) can accurately reconstruct oxygen conditions in coastal sites experiencing mild deoxygenation, and (2) to what degree secondary (depositional environmental) factors impact Mo-and U-EFs. Here we investigate 18 coastal sites with varying bottom water redox conditions, which we define by means of five “ redox bins ” , ranging from persistently oxic to persistently euxinic, from a variety of depositional environments. Our results demonstrate that Mo-and U-EF-based redox proxies and sedimentary Mo and U contents can be used to differentiate bottom water oxygen concentration among a range of modern coastal depositional environments. This is underpinned by the contrasting EFs of Mo and U along the redox gradient, which shows a substantial difference of Mo-EFs between redox bins 3 – 5 (ir/ regularly suboxic – ir/regularly dysoxic – persistently oxic) and of U-EFs between redox bins 1 – 2 (persistently euxinic – ir/regularly euxinic). Surprisingly, we observe comparatively low redox proxy potential for U in environments of mild deoxygenation (redox bins 3 – 5). Further, we found that secondary factors can bias Mo-and U-EFs to such an extent that EFs do not reliably reflect bottom water redox conditions. We investigate the impact of limited Mo sedimentary sequestration in sulfidic depositional environments (i.e., the “ basin reservoir effect ” , equilibrium with FeMoS 4 ), Fe/Mn-(oxy)(hydr)oxide “ shuttling ” , oxidative dissolution, the sulfate methane transition zone in the sediment, sedimentation rate, and the local Al background on Mo-and U-EFs.


Introduction
Vast coastal areas are suffering from low "bottom water" dissolved oxygen (O 2 ) concentrations (e.g., Diaz and Rosenberg, 2008;Rabalais et al., 2010;Breitburg et al., 2018).Terminology and thresholds for quantifying O 2 depletion have been widely discussed in literature and they differ between biological and sedimentological studies (e.g., Tyson and Pearson, 1991;Diaz and Rosenberg, 1995;Levin et al., 2009;Naqvi et al., 2010).A recent review of redox thresholds for sedimentary environments (Algeo and Liu, 2020) suggests a four-stage scheme to classify O 2 depletion (Table 1), on which the redox classification used in this study is based on.The spread of coastal deoxygenation is strongly linked to human-induced non-climatic (e.g., eutrophication) and climatic impacts (e.g., temperature rise), which enhance organic matter production and degradation, and water mass stratification (Helly and Levin, 2004;Diaz and Rosenberg, 2008;Bindoff et al., 2019).Deoxygenation leads to a stepwise decrease in benthic fauna abundance, diversity and biomass (Diaz and Rosenberg, 1995).Progress in monitoring and research campaigns since the 1960s has identified globally around 700 hypoxic coastal areas, and an additional 230 areas that are at risk of developing hypoxia (Diaz et al., 2019).However, the estimated number of unreported hypoxic areas is assumed to exceed 700, due to regionally poor monitoring coverage, particularly in the tropical ocean (Altieri et al., 2017;Diaz et al., 2019).Even though, several areas show signs of recovery from hypoxia (Conley et al., 2011;Diaz et al., 2019), the pace at which low O 2 areas are spreading, and the severe impact on primary productivity, marine fauna, and energy and biogeochemical cycles, makes deoxygenation one of the most important drivers of change in coastal ecosystems today (Breitburg et al., 2018;Diaz et al., 2019;Bindoff et al., 2019).
At present, the spatial and temporal development of hypoxia in coastal waters is poorly understood due to incomplete environmental monitoring coverage and limited availability of existing data (Grégoire et al., 2021).Poor understanding of local depositional environments, and biogeochemical feedbacks involved in coastal marine O 2 dynamics (Keeling et al., 2010;Breitburg et al., 2018), also contribute to knowledge gaps concerning the relationship between O 2 supply, the burial efficiencies and benthic release of nutrients and major/minor elements, and redox-sensitive trace metal dynamics.New investments in global (Global Ocean Oxygen Database and Atlas GO 2 DAT, Grégoire et al., 2021) and European (European Marine Observation and Data Network EMODnet) marine observation networks and databases of O 2 conditions, as well as improved understanding of biogeochemical processes, aim to facilitate the development of strategies to remediate coastal hypoxia (e. g., Rabalais et al., 2010;Breitburg et al., 2018).
In this study, we test the applicability of existing Mo and U based redox proxies to reconstruct deoxygenation in coastal depositional environments.Here, these environments comprise marginal seas (Baltic Sea, and Black Sea, including their deeper sub-basins, and archipelagos), the German Bight, various types of fjords (along the Swedish west coast and on Svalbard) and a dammed former estuary (marine lake Grevelingen).To systematically test the applicability of Mo and U based redox proxies, we cover in the analysis coastal depositional environments capturing the full range of bottom water redox conditions, including besides strongly restricted, deep, euxinic sites, also human-impacted oxic and dysoxic coastal sites that are representative for wide areas of the coastal ocean and remain understudied in terms of trace metal redox proxies.We tackle common caveats encountered when interpreting sediment-based Mo and U redox proxies, which can bias their applicability.These include limited Mo sequestration in sulfidic depositional environments (Algeo and Lyons, 2006;van Helmond et al., 2018;Helz, 2021), particulate Fe-/Mn (oxy)(hydr)oxide "shuttling" (Crusius et al., 1996;Algeo and Tribovillard, 2009;Tribovillard et al., 2012;Scholz et al., 2013), re-oxygenation events (Cochran et al., 1986;Zheng et al., 2002a;McManus et al., 2005;Morford et al., 2009b;Jokinen et al., 2020), the depth and intensity of the sulfate-methane transition zone (SMTZ) in the sediment (the term "intensity" refers to rate of sulfide production, including through organoclastic sulfate reduction and sulfate-mediated anaerobic oxidation of methane) (Jokinen et al., 2020), sedimentation rate (Algeo and Maynard, 2004;Scott and Lyons, 2012;Noordmann et al., 2015;Liu and Algeo, 2020), and local detrital (Al) background (van der Weijden, 2002;Brumsack, 2006).The magnitude of these caveats and the extent to which they impact sediment-based Mo-and U-enrichment factors (Mo-EFs, U-EFs) in different redox conditions and depositional environments are poorly constrained.Therefore, we investigate multiple sites across a range of depositional environments.With our comprehensive data set, we show that by using a combination of Mo and U enrichments, it is possible to deconvolve past redox conditions in coastal settings for which monitoring data are lacking.

Geochemical behavior of Mo and U
Molybdenum is present in seawater as dissolved molybdate (MoO 4 2− ; Mo(VI); Bertine and Turekian, 1973) with an average concentration of 110 nM.Molybdate behaves conservatively and has an oceanic residence time of ~440 thousand years (Miller et al., 2011).Under oxygenated conditions, Mo(VI) has little affinity for other components in the water column, with the exception of iron (Fe)-and manganese (Mn) (oxy)(hydr)oxides to which Mo can adsorb (Bertine and Turekian, 1973).Fe-and Mn (oxy)(hydr)oxides can efficiently shuttle Mo from the water column to the sediment surface (Berrang and Grill, 1974;Algeo and Lyons, 2006;Sulu-Gambari et al., 2017;Wagner et al., 2017).MoO 4 2− is released back to the ambient (pore) water once Fe-and Mn (oxy)(hydr)oxides dissolve under reducing conditions.Depending on the ambient H 2 S pore water concentration and availability of other host phases, MoO 4 2− can be (re)scavenged by a more refractory phase (Pilipchuk and Volkov, 1974;Helz et al., 1996;Algeo and Maynard, 2004;Helz and Vorlicek, 2019).
When approaching an aquatic H 2 S (H 2 S aq ) threshold value, i.e.
In non-euxinic conditions, these mechanisms are limited to sedimentary sulfide fronts associated with high rates of organoclastic sulfate reduction (e.g., Rickard and Luther, 2007;Jørgensen et al., 2019;
In contrast to Mo, authigenic U (U auth ) sequestered in sediments originates from diffusion across the sediment-water interface (SWI) regardless of the bottom water redox condition (Barnes and Cochran, 1991;Klinkhammer and Palmer, 1991;Algeo and Maynard, 2004;McManus et al., 2005), or from the deposition of Particulate Non-Lithogenic U (PNU) formed in surface waters, likely associated with particulate organic matter (Hirose and Sugimura, 1991;Zheng et al., 2002a).In sediments underlying an O 2 depleted water column, PNU can account for a substantial proportion of the total U auth pool, whereas in sediments underlying oxic waters, the PNU fraction is negligible (Zheng et al., 2002a).Despite being seemingly unresponsive to shuttling by Fe-and Mn (oxy)(hydr)oxide in the water column, according to several laboratory experiments U, in its oxidized U VI form, can be temporarily adsorbed to Fe-and Mn (oxy)(hydr)oxides in oxic surface environments (Brennecka et al., 2011;Singh et al., 2012;Dang et al., 2016).

Study sites and data collection
In our investigation, we use solid-phase geochemical data of 18  2 and S1.Modified from EEA (2019).
K.M. Paul et al. coastal depositional environments in Europe (Fig. 1, Table 2).The study sites were selected based on covering a range of bottom water O 2 concentrations, as well as distinct differences in their depositional environments, including geographical location, bathymetry, sedimentation rate, degree of eutrophication, and salinity (Fig. 1, Tables 2, S1, and S2).
For further analysis, the study locations were grouped into five "redox bins" according to ranges of bottom water O 2 conditions (Fig. 1, Table 2), broadly based on the redox classification by Algeo and Li (2020) for sedimentary environments (Table 1): persistently euxinic (redox bin 1), ir/regularly euxinic (redox bin 2), ir/regularly suboxic (redox bin 3), ir/regularly dysoxic (redox bin 4), and persistently oxic (redox bin 5).The term ir/regularly comprises locations with irregular or regular episodical (less than once per year), periodical (several times per year), or seasonal (each season) redox changes.The allocation of each site into redox bins was mainly based on available bottom water O 2 and hydrogen sulfide data (from either monitoring, or literature data).
For details regarding monitoring data availability, temporal coverage, and monitoring frequency we refer to Table 2.The period of deposition used to allocate sites to a given redox bin varies from one location to another depending on the core length (full core solid-phase profiles were considered in this study; Fig. S1).At four sites, the sediment data series was divided into two segments with different redox bins, because of distinct temporal shifts in water column O 2 conditions during the period of sediment deposition (Table 2).This led to a double listing of four study sites, giving a total number of 22.

Solid-phase analyses
Methods of solid-phase analyses are summarized below.For detailed sampling procedures we refer to the source literature (Table S1, S2).

Sediment sampling
Sediments for solid-phase analyses were recovered with various coring devices (i.e., multicorer, UWITEC or GEMAX™) and sectioned at 0.4-5 cm intervals.Sediment samples were treated under O 2 -free conditions (i.e., inside a N 2 or argon filled glove bag or box) to avoid oxidation of the samples.For further analyses (water and salt contents, porosity, total organic carbon (C org ) content, and total trace metal concentrations), sediment samples were (freeze)-dried, powdered, homogenized, weighed, and sub-sampled.

Determination of organic carbon content
The protocols for the determination of total C org contents were comparable for all sites.The protocols included decalcification, drying and subsequent thermal combustion using a CNS analyzer.Samples from Koverhar and Lappohja were not decalcified, due to expected low carbonate contents (~1.1 wt%; Virtasalo et al., 2005;Jilbert et al., 2018;Jokinen et al., 2020).Data were checked against in-house standards and internationally certified analytical references (such as IVA2, sulfamethazine, or GPA 2186A).Organic carbon contents were corrected for weight loss upon decalcification and salt content of the freeze-dried sediment.The average analytical uncertainty (relative standard deviation, RSD %) based on duplicates was <3% for all samples.For analytical uncertainties in the published data, we refer to the original publications (Table S2).

Total elemental contents
The protocols for the determination of sedimentary major and trace metal concentrations (Al, Mo, and U) were comparable for all sites.Aliquots of (freeze)-dried and powdered sediment samples were digested using a triple-acid (HClO 4 /HNO 3 /HF) digestion protocol, redissolved in HNO 3 (Mort et al., 2010;Jilbert and Slomp, 2013b;Zhang et al., 2016;Kraal et al., 2017;van Helmond et al., 2018;Lenstra et al., 2019;Bennett and Canfield, 2020;Jokinen et al., 2020).The sample residues were analyzed by ICP-MS for Mo and U and/or ICP-OES for Al (and in some cases for Mo, e.g., for Lappohja pockmark and Den Osse St. 1, 8/2012).For analytical uncertainties of the elements in the published data, we refer to the original publications (Table 3).The accuracy (recovery) of all new data was based on in-house standards and commercial sediment reference materials  and ranged between 91 and 106% (Mo), 89-109% (U), and 95-96% (Al), respectively.The precision (relative standard deviation, RSD %) based on sediment sample duplicates ranged between 3 and 5% (Mo), 2-5% (U), and 2-8% (Al), respectively.All elemental concentrations in the sediment were corrected for salt-dilution using the bottom water salinity and sediment porosity, assuming a solid-phase density of 2.65 g cm − 3 (Burdige, 2006).For a small set of samples of three sites within redox bins 1, 4, and 5 (n = 74), the average recovery for Al was 127%.This was considered to reflect instrumental drift, and therefore a 27% correction factor was applied to all samples in the series in question.The corrected Al values were validated for Gullmar Fjord and Koljö Fjord samples by previous analysis of samples from the same study sites (September 2018, this study), and for German Bight samples from estimates of nearby study sites (Hinrichs, 2001).

Authigenic trace metal concentrations and enrichment factors
Many trace elements are also present in the detrital fraction of sediments (Wignall and Myers, 1988), thus, only trace metal enrichments corrected for their detrital fraction (= authigenic enrichments) can be used to reconstruct the redox state at the time of deposition and provide insights on the enrichment pathway.Three widely accepted methods are applied to estimate authigenic trace metal enrichments: TM/Alratio (e.g., Bennett and Canfield, 2020); TM-EF (e.g., Calvert and Pedersen, 1993;Tribovillard et al., 2004); and the excess trace metal content (TM XS , e.g., Scholz et al., 2013;Jokinen et al., 2020).However, none of these methods is free of potential limitations (Reimann andDe Caritat, 2000, 2005;van der Weijden, 2002;Brumsack, 2006;Cole et al., 2017).A strong criticism of the normalization methods centers on their use of "average shale" (Turekian and Wedepohl, 1961), Post-Archean Australian shale (PAAS, Taylor andMcLennan, 1985, 1995), or upper continental crust (UCC, McLennan, 2001;Rudnick and Gao, 2014) as a substitute for local detrital background TM/Al values.For most elements enrichment factors are broadly similar between average shale, PAAS, and upper continental crust and average shale (Brumsack, 2006;Tribovillard et al., 2006;Tribovillard, 2020), implying that data from different studies using different reference material are comparable.Choosing a reference material that is close to local source material can decrease the error in the estimations.However, over-or underestimations of the true local detrital background can still occur, particularly if the composition of the detrital background has not been constant over time (Little et al., 2015;van der Weijden, 2002;Brumsack, 2006), or when it is low in Al (e.g. in carbonate rocks; van der Weijden, 2002).The use of true local background TM/Al content would be preferred over reference values.As these data were not available for our study sites, UCC values were used instead to objectively identify differences between a range of depositional redox environments with regards to their TM enrichments.
We chose the TM-EF as a primary trace metal redox proxy using UCC to increase the comparability with other studies.First, Mo and U contents were normalized against Al to distinguish the authigenic enrichment from the lithogenic background.Next, the sample ratios were divided by the average continental crust ratio of Mo/Al and U/Al to calculate the Mo and U enrichment factors (eq.1).
where TM sample is the measured Mo or U content, Al sample is the measured Al content, and TM/Al reference represents the global UCC ratio for Mo and U against Al.Reference values (Al: 8.15 wt%, Mo: 1.1 ppm, and U: 2.7 ppm, Table S3) were taken from Rudnick and Gao (2014), because these represent the most recent and complete UCC composition a Timescales for modern average bottom water redox conditions vary for each site, depending on available monitoring data.b Fårö Deep, Northern Gotland Basin, Bornholm Basin, and Lilla Värtan were divided into two core sections with different redox bins, due to major changes in bottom water redox conditions during the period covered by the sediment core (Fig. S1).The timing of these changes was estimated based on monitoring data.c Multiple campaigns from the same sampling sites with no significant redox bin variability were summarized as one data set.d Data from the three sites were combined.e Svalbard site SV1 (Bennett and Canfield, 2020) was excluded, because it is located in northern Norway and not on Svalbard.
f At time of sampling (Table S1). estimates.
For comparison to alternative approaches such as sample TM/Al, and TM XS , we additionally calculated the TM XS of each sample (eq.2) and provide both the raw data of Al, Mo, and U (Supplementary Material) and their concentrations in commonly used references materials (Table S3).
where TM sample is the trace metal content in the sediment sample, Al sample is the Al content in the sediment sample, and TM/Al reference is the ratio between the trace metal and aluminum in a reference material.
We followed the same approach as for the TM-EF calculation by using UCC background values.

Interpretation of TM-EFs
Commonly, TM-EFs are interpreted using arbitrary threshold values.An EF >1 indicates an enrichment relative to the crustal background, whereas an EF < 1 indicates a depletion (Tribovillard et al., 2004).Algeo and Tribovillard (2009) further differentiated between a detectable authigenic enrichment (EF > 3) and a substantial enrichment (EF > 10).In our study, we move away from arbitrary threshold values and instead look at the total ranges of the TM-EF at each site or within one redox bin.In each analysis, we used all data points from a given core (core lengths range between 0 and 50 cm) and considered the range of values using violin plots.We applied the same approach to sedimentary C org contents, which are often considered alongside TM-EF data as an independent proxy for redox conditions (e.g., Bennett and Canfield, 2020).
We further test the use of combined Mo-EF vs. U-EF ratio data as a simple tool to assess controls on TM enrichment mechanisms (Algeo and Tribovillard, 2009;Tribovillard et al., 2012;van Helmond et al., 2018).Algeo and Tribovillard (2009) differentiated three key controls that impact on relative enrichments of the two elements: (A) benthic redox variations (average bottom water redox conditions and degree of variability; for the purpose of our study, we divided these into A1.and A2), (B) operation of a particulate Fe/Mn (oxy)(hydr)oxide "shuttle", and (C) changes in the water mass chemistry related to the degree of water mass restriction (Fig. 2).According to the conceptual model, minor bottom water redox variations (A1) are reflected first in U-EF, while a variation towards increased organic matter degradation and release of sulfide into the pore waters results in a coincident increase in both Mo-EF and U-EF, leading to a steepening of the gradient of Mo-EF vs. U-EF (A2).Controls B and C largely affect Mo-EF.The operation of a particulate "shuttle" (control B) can increase the transport of Mo from the water column to the SWI, favoring increased Mo-EF over U-EF even in mildly reducing systems.In strongly sulfidic and (semi)-restricted basins Mo sequestration may be limited by a combination of 1) insufficient hydrological resupply of Mo to the basin and simultaneous accelerated removal of Mo through the geochemical switch (the so-called "basin reservoir effect", Algeo, 2004, Algeo andLyons, 2006), and 2) equilibrium of FeMoS 4 with Mo diss at high sulfide concentrations (Helz, 2021).Both effects may result in a limitation of the maximum attainable Mo-EF, while U-EFs   and Tribovillard (2009).The four black arrows and letters A1, A2., B, and C refer to the key enrichment controls: A1. and A2.bottom water redox variations (mild redox variations between oxic and dysoxic + increase in organic matter degradation and release of sulfide to the pore-and bottom waters), B operation of a particulate Fe/Mn (oxy)(hydr)oxide "shuttle", and C limited Mo sedimentary sequestration in sulfidic depositional environments (comprising the "basin reservoir effect" and equilibrium of FeMoS 4 with Mo diss at high sulfide concentrations (Helz, 2021).Uranium is not affected by such mechanisms, and therefore continues to enrich in the sediment.The three diagonal lines represent multiples (0.3, 1, and 3) of the Mo:U ratio of present-day seawater (SW) converted to an average weight ratio of 3.1 for the purpose of comparison with sediment Mo:U weight ratios (Tribovillard et al., 2012).
K.M. Paul et al. continue to rise.

Statistics: Assessing variability and uncertainty in the dataset
Data variability and uncertainty between redox bin pairs, displayed as violin plots, were assessed based on both the spread of data using the Median Absolute Deviation (MAD, e.g., Pham-Gia and Hung, 2001), and 0.99 confidence intervals (CIs) of the mean using the bootstrapping resampling method (e.g., Carpenter and Bithell, 2000).Briefly, the narrower the range of the MAD and CI of the mean, the more reliable a generalized pattern can be deduced from the data.Both parameters were calculated in R (ggplot) using the stat_summary() function combined with the hmisc package.These methods were chosen, since the dataset is largely non-normally distributed and includes small subsample sizes (Rowland et al., 2021).

Organic carbon contents
Sedimentary C org contents are the highest among redox bin 1 sites (median: 8.2 wt%, Fig. 3A), and the lowest among redox bin 4 sites (median: 3.1 wt%).The smallest spreads (MAD) of C org contents are observed for redox bins 2 and 3 (1.3 and 1.4 wt%), respectively.C org ranges of redox bin 3 also have the highest certainty with a 0.99 CI of the mean of 0.5, compared to the largest uncertainty within redox bin 1 with a 0.99 CI of the mean of 1.4.The higher uncertainty in redox bin 1 coincides with a large data spread (MAD: 3.3 wt%), only the MAD of redox bin 5 is higher (3.4 wt%).Redox bins 1-3 show a considerable difference between each other (no overlap of the MADs), the differences between redox bins 3 and 5 are very weak, as the MADs show considerable overlap.Strikingly, instead of an expected decrease in C org contents from bins 3 to 5 (increase in aerobic C org degradation), C org contents increase from redox bin 4 to 5 (medians: 3.1, 4.5 wt%), and are larger in redox bin 5 than in redox bin 3 (medians: 4.5 vs. 3.5 wt%).Given the inability to differentiate between redox bins 3-5, C org contents alone do not accurately describe bottom water O 2 concentrations in our dataset.

Mo and U enrichment factors 4.2.1. Trace metal enrichments as a function of bottom water redox conditions
Sedimentary Mo-EF and U-EF patterns broadly resemble the C org enrichment patterns, following the expected trend governed by bottom water redox changes (Fig. 3B and C).Highest EFs coincide with the most restricted and O 2 depleted sites of redox bin 1 (Mo-EF median: 103.7;U-EF median: 13.17), while lowest EFs coincide with the least restricted and O 2 depleted sites of redox bin 5 (Mo-EF median: 1.3; U-EF median: 1.8).However, each redox bin varies in the overall spread and CI of the mean for both EFs.The largest spread is observed for both EFs in redox bin 1 (Mo-EF MAD: 279.3;U-EF MAD: 29.0), coinciding with the largest 0.99 CI of the mean (Mo-EF: 46.3; U-EF: 6.02).In contrast, the smallest spread is observed in redox bin 5 (Mo-EF MAD: 1.9; U-EF: 0.9) with the highest certainty of 0.4 and 0.2 for Mo-EF and U-EF, expressed as 0.99 CIs of the mean.Overall, both the spread and confidence intervals are smaller across U-EF redox bins (MADs: 1.8-3.5;CIs: 0.2-0.4;excluding redox bin 1) than for Mo-EF redox bins (MADs: 1.3-38.4;CIs: 0.4-12.7;excluding redox bin 1).
The two elements show noticeable differences in their ability to reflect contrasts between the redox bins.Mo-EF resolves the contrasts between the more mildly reducing redox bins 3-5 (medians: 16.9, 3.2, 1.3), but the contrast is less strong between the more reducing redox bins 1-2, due to the overlap in the spread of the data of bins 1 and 2. Several samples of redox bin 4 and 5 sites plot below the average UCC background value (Mo/Al UCC : 0.13 ppm/ wt%, horizontal dashed line, Fig. 3B).For U-EF, the strength of redox sensitivity is reverse to that of Mo-EF.While U-EF is very sensitive among redox bins 1-3 (medians: 13.2, 3.5, 1.8), between redox bins 3-5 no difference can be found, as the medians and MADs are within a similar range.Similar to Mo-EF, but Fig. 3. A) Organic carbon content (wt%) for the different redox bins (Table 2) plotted as violin plots (log 10 scale).Each violin plot represents the data distribution of each redox bin.The black cross and vertical black solid line with two whiskers within each violin plot denotes the mean with a 0.99 confidence interval (CI).The median absolute deviation (MAD) -a measure of the variability of non-normal distributed datais illustrated by an empty vertical crossbar, divided by a horizontal black solid line, which represents the median.Numbering of the redox bins follows the scheme introduced in section 3.1: persistently euxinic (redox bin 1), ir/ regularly euxinic (redox bin 2), ir/regularly suboxic (redox bin 3), ir/regularly dysoxic (redox bin 4), and persistently oxic (redox bin 5). to a much lesser degree, selected samples from redox bin 5 sites have lower U auth enrichments than the UCC background value (0.33 ppm/ wt %).In summary, both TM-EFs are stronger in reflecting bottom water redox changes than C org .However, the large spread and confidence intervals across the redox bins indicate that additional controls likely play important roles in determining the magnitude of enrichment at any given location.

Station scale comparision of Mo-and U-EFs
To investigate the different sensitivity of Mo-and U-EF among the redox bins, all sites were plotted individually in order of decreasing mean Mo-and U-EFs.Broadly, the decreasing EFs correspond to changes in the redox bin classification of the sites from redox bin 1-5 (Fig. 4).Redox bin 1 sites have the highest EFs and redox bin 5 sites the lowest.However, there are several distinct anomalies in which the order of EF ranking deviates from that predicted by the redox bin, hence accounting for part of the observed overlap between the bins in the original analysis (Fig. 3B and C).We highlighted four examples of sites where either the Mo-EF or U-EF, or both, appear anomalous with respect to the redox bin classification (site numbers 4, 7, 8, and 10 in Fig. 4).Although this selection is partly arbitrary, since overlaps in TM-EFs between redox bins are reflected in multiple sites, the selection serves to illustrate key secondary factors influencing enrichments as described below.Site 4 has lower Mo-EFs compared to the other three sites of the same redox bin.The same observation is true for U-EF, although the difference is less distinct.Sites 7, 8, and 10 show a strong deviation from the expected U-EF ranking, but plot quite normally in Mo-EF.Additionally, we note that seven (Mo-EF) and two (U-EF) sites in redox bins 4 and 5 show data ranges that intercept with the average UCC background value; one such site has Mo-EFs considerably lower than the UCC (16.Black Sea NW shelf ).The anomalous patterns observed in the station-scale comparison confirm that additional site-specific controls beyond redox alone must affect the observed trace metal enrichments.

Mo-and U-EF covariation patterns
To assess the secondary factors controlling TM-EF, we begin by considering Mo-and U-EF covariation patterns according to the conceptual model of Algeo and Tribovillard (2009) (Figs. 2 and 5).For this approach, all sites were plotted individually to increase the interpretability.The results show that all three Mo-and U-EF covariation patterns in the conceptual model may be discerned in our study sites.Generally, redox bin 5 sites are only slightly enriched in Mo and U and show the characteristic steepening of the gradient of Mo-EF vs. U-EF above U-EF = 2 (Mo-and U-EF covariation patterns A1 and A2).Redox bin 4 sites are typically moderately enriched in Mo-EF, but a notable scatter is observed in Mo-EF vs. U-EF ratio, due to the variable effects of patterns A2 (increasing organic matter degradation and release of sulfide into the pore water) and B (activity of a particulate Fe/Mn (oxy)(hydr)oxide "shuttle") at different sites.Redox bin 3, which is represented by only two sites, plots at the higher end of the Mo-EF vs. U-EF gradient corresponding to covariation pattern B. Sites of redox bins 2 and 1 display the flattening of the Mo-EF vs. U-EF gradient corresponding to covariation pattern C, ceased Mo sequestration from sulfidic depositional environments, with redox bin 1 sites clustered towards the more extreme end of this gradient.

Bottom water redox conditions as a primary control on Mo-and U-EFs
Both Mo-EF and U-EFs show contrasting behavior along the redox gradient, with Mo-EFs being a stronger proxy in mildly reducing depositional environments (redox bins 3-5), and U-EFs in stronger reducing depositional environments (redox bins 1-2).Nevertheless, both trace metals have in common that they show the largest enrichment at most reducing sites (redox bins 1 and 2, Fig. 3 B and C).This pattern is reproduced at the station scale (Fig. 4), with strongly reducing sites (redox bins 1 and 2) and mildly and non-reducing sites (redox bins 4 and 5) clearly differentiable.Therefore, we infer that the dominant control of Mo-and U-EFs at our coastal sites from a range of depositional environments is exerted by bottom water redox conditions.Our findings are thus in line with studies focusing on (strongly) restricted basins and continental margin OMZs in the modern ocean (McManus et al., 2005;Algeo and Lyons, 2006;van Helmond et al., 2018;Liu and Algeo, 2020;Bennett and Canfield, 2020).
In detail, however, our results demonstrate that both elements' EFs have considerable limitations in reflecting the whole range of smallscale redox variability (e.g., between redox bins 1 and 2, or 4 and 5).Mo-EFs show substantial differences between redox bins 3-5, but it is weaker in differentiating redox bins 1 and 2 (Figs. 3, 4, and 5).By contrast, U-EFs appear more suitable to detect variability between redox Fig. 4. Sedimentary A) Mo and B) U enrichment factors (EFs) among the study sites (log 10 scale).For each element, the sites are arranged according to the order of their means, starting with the highest.The color of each box (site) corresponds with its redox bin (Fig. 3, Table 2).For some sites the total ranges vary between the two elements, due to different amounts of data points available (e.g., Lappohja, and Arkona, Table 2).Fårö Deep, Northern Gotland Basin, Bornholm Basin, and Lilla Värtan are listed twice: as the upper core section, covering the last <30-40 years, and as the lower core section, covering the preceding >30-40 years.The numbers refer to the sites that plot anomalously (see text).For detailed description of the violin plots see Fig. 3   bins 1 and 2, but weaker in differentiating redox bins 4 and 5.The stronger differentiation potential of U-EF between redox bins 1 and 2 stands in contrast to a recent finding of Bennett and Canfield (2020), who found no evidence for U enrichment patterns that could reliably differentiate strongly reducing sites from less-reducing ones.A simple explanation for this deviation could be that these authors compared two very distinct types of depositional environments within the two redox bins (enclosed euxinic basins and open continental margin OMZs), while our sites are typically more comparable to each other with regards to their bathymetry within the same two redox bins.More striking, however, is the weaker differentiation potential of U compared to Mo between redox bins 4-5 because it stands in contrast to what is known based on their geochemical behavior.Based on existing knowledge, Mo should more reliable than U as redox proxy under (strongly) reducing bottom water redox conditions, due to the presence of hydrogen sulfide (e.g., Helz et al., 1996;Zheng et al., 2000), while U should be more sensitive than Mo under mildly reducing bottom water redox conditions, because U reduction occurs at higher reduction-oxidation potentials, likely governed by microbial activity (e.g., Zheng et al., 2002b;Lovley et al., 1991Lovley et al., , 1993)).Our contrasting findings suggest that certain sitespecific or element-specific secondary factors are obscuring the initial redox signal.In this context, it should be reemphasized that C org (Fig. 3A) shows a similarly striking (reverse) enrichment pattern to that of U-EF (Fig. 3B), which might suggest that U and C org are affected by (a) similar factor(s) or they are directly, or indirectly, linked in their enrichment pathway.Next, a selection of possible site-specific or element-specific secondary factors will be discussed.

Depositional environmental factors as secondary controls of Mo-and U-EFs
Anomalies to the expected trends with variable redox conditions are a relatively common observation in sedimentary trace metal studies (Scott and Lyons, 2012;Bennett and Canfield, 2020).However, past studies that have investigated such anomalies have typically focused on a narrow selection of factors, namely depositional environments, and redox conditions (e.g., Algeo and Tribovillard, 2009;Liu and Algeo, 2020).Here, we systematically test the influence of six secondary factors controlling Mo-and U-EFs, while also considering the whole range of redox conditions and depositional environments.We identified four sites that have Mo-EF and/or U-EFs anomalous to their assigned redox bin (Fig. 4, station numbers 4, 7, 8, and 10), either by showing over-or under-enrichment.Below we discuss how depositional environmental factors can potentially bias either the Mo-EF, U-EF, or both.

Limited Mo sequestration in sulfidic depositional environments
Limited Mo sequestration in sulfidic depositional environments (pattern C, Fig. 5) can be explained by two processes.Once conditions become sulfidic, Mo sequestration may be limited by insufficient hydrological re-supply of Mo to the basin (the so-called "basin reservoir effect", Algeo, 2004, Algeo andLyons, 2006).Alternatively, precipitation of Mo may be limited by the equilibrium of FeMoS 4 with Mo diss at high sulfide concentrations (Helz, 2021).In contrast, authigenic U enrichment under sulfidic redox conditions is not determined by such solubility effects, but rather by the activity of iron and sulfate reducing bacteria (Lovley et al., 1991(Lovley et al., , 1993)).Uranium also shows generally lower sensitivity to the basin reservoir effect (Algeo and Maynard, 2008;Tribovillard et al., 2012;van Helmond et al., 2018).We interpret pattern "C" of the Mo-EF vs. U-EF covariation patterns (Fig. 2) to indicate a combination of the above effects, whereby the enrichment of Mo is limited with respect to U in the most sulfidic settings (Fig. 5).Sites that show this enrichment pattern are typically subject to severe deoxygenation in response to strong water mass stratification, and restricted water mass exchange, which gradually depletes the pool of sequesterable Mo (Algeo and Lyons, 2006;Tribovillard et al., 2006;van Helmond et al., 2018).Three of the four sites in redox bin 1 show a clear signal of such effects with limited Mo enrichment in comparison to U (Fig. 5).Using is cross-plots of unnormalized sedimentary Mo, and C org enrichments (as per Algeo and Lyons, 2006;Algeo and Rowe, 2012;Tribovillard et al., 2012;van Helmond et al., 2018), we show that all four sites in redox bin 1 show signs of a considerable sedimentary Mo depletion relative to C org (Fig. S7A).Notably, the anomalous site 4. Skurusundet (Fig. S7A) shows the lowest Mo/C org of any location in the dataset, implying the strongest sedimentary Mo depletion of the redox bin 1 sites.Based on our knowledge, this may be the first described shallow, low salinity (Table 2), and strongly euxinic depositional environment for which limitation effects on Mo sequestration have been identified.Moreover, we postulate that the anomalously low U enrichments at the same site could be explained by water column depletion of U diss (i.e.basin reservoir effect), a phenomenon previously described for the strongly euxinic Kyllaren Fjord, Norway (Noordmann et al., 2015).

The particulate Fe and Mn (oxy)(hydr)oxide shuttle
The particulate Fe and Mn (oxy)(hydr)oxide shuttle (pattern B, Fig. 5) can enhance the rate of sequestration of Mo to sediments due to sorption of Mo diss onto particle surfaces (Algeo and Tribovillard, 2009;Jilbert and Slomp, 2013b;Scholz et al., 2014;Sulu-Gambari et al., 2017;Lenstra et al., 2019).Once Fe and Mn (oxy)(hydr)oxides are dissolved below the oxic-suboxic redox boundary, Mo is released to the pore water or water column.Depending on the availability of more refractory host phases, Mo diss is either (re)sequestered, e.g., by FeS, or diffuses upward to overlying oxic water.The efficiency of this shuttle mechanism is dependent on the amount of Fe and Mn input from the water column or through remobilization by bioturbation, bioirrigation, or pore water acidification by cable bacteria (Seitaj et al., 2015;Sulu-Gambari et al., 2016;Hermans et al., 2019a;Lenstra et al., 2019), and the vertical distance between Fe and Mn (oxy)(hydr)oxide formation and reductive dissolution (Sulu-Gambari et al., 2017).Although Fe-and Mn (oxy) (hydr)oxides can temporarily sequester U under oxic conditions in the solid phase, they do not play an essential role in the particle transport of U or serve as long term host phases for U (Morford et al., 2005;Wang et al., 2013;Jokinen et al., 2020).Therefore, in principle, active particulate shuttling will result in a high ratio of Mo-EF to U-EF (Fig. 2).Of our sites, this pattern can be identified for 8. Den Osse basin deep and 10.Gullmar Fjord (Fig. 5).These sites are from different redox bins, but are both seasonally stratified (Filipsson and Nordberg, 2004;Hagens et al., 2015), and show evidence for strong Mn oxide re-cycling in the water column, potentially enhancing sequestration of Mo (Sulu-Gambari et al., 2017;Goldberg et al., 2012).This typical enrichment pattern can explain why 8. Den Osse Basin deep shows anomalously low U-EFs and high Mo-EFs in the station-scale comparison (8, Fig. 4).Despite elevated Mo-EFs with respect to U-EFs, this pattern is less clear for 10.Gullmar Fjord because both elements' EFs are close to UCC values (Fig. 4).Generally, our results show that the Mo-EF and U-EF covariation model developed by Algeo and Tribovillard (2009) is applicable to identify the operation of the particulate shuttle at coastal sites.With respect to 10. Gullmar Fjord, seemingly, at least one other depositional environmental factor must be influencing Mo auth and U auth sequestration, which requires further investigation.

The depth and intensity of the SMTZ in the sediment
The combination of low-salinity (low sulfate concentration) and high labile organic matter loading in response to anthropogenic eutrophication (Fleming-Lehtinen et al., 2015) have led to an upwards migration of the SMTZ during the late 20th century to the uppermost 10-20 cm in many coastal areas of the northern Baltic Sea (e.g., Bothnian Sea and Gulf of Finland) (Slomp et al., 2013;Jilbert et al., 2018;Myllykangas et al., 2020).In such systems, a shallow and compacted SMTZ may catalyze the uptake of Mo and U even under oxic bottom water conditions, which may impact their applicability as redox proxies.To investigate the potential impact of the depth and intensity of the SMTZ depth as a secondary factor controlling Mo-and U-EFs, we propose the use of the ratio between C org (C) and salinity (Sal), C/Sal hereafter.The approach is based on the reasoning that the depth of the SMTZ is largely controlled by the loading of labile organic matter (Slomp et al., 2013;Egger et al., 2015) and the ambient salinity, which controls the sulfate concentration (Jilbert et al., 2018).Since C org and bottom water salinity are more routinely measured in sediment studies than porewater sulfide and methane profiles, C/Sal provides a useful proxy to estimate relative SMTZ depth between sites.Using the C/Sal value of the 12. Koverhar site (redox bin 5) as a reference, due to its known shallow SMTZ (Fig. 6, upper row; Jokinen et al., 2020), we tested whether sites with anomalous enrichments may be influenced by contrasting depths of the SMTZ.The redox bin 4 sites 7. Lilla Värtan and 10.Gullmar Fjord (Fig. 4) appear to fit this description; 7. Lilla Värtan U-EFs are anomalously high, while 10.Gullmar Fjord U-EFs are anomalously low, and their position in the U-EF range corresponds well with their C/Sal ratios.7. Lilla Värtan has a high C/Sal ratiocorrelating well with its shallow SMTZ (Dalcin Martins et al., 2022) which could explain higher U-EFs compared to other redox bin 4 sites in response to increased authigenic sequestration of U within the sulfide front.By contrast, 10.Gullmar Fjord has a very low C/Sal ratioindicating a deep SMTZyielding exceptionally low Mo-EF and U-EF for a redox bin 4 site, likely due to minimal sulfate reduction in the upper sediment column (Goldberg et al., 2012).
Besides the ability to explain anomalously high or low EFs, the C/Sal ratio may also help to explain the remaining uncertainty in the lower sensitivity of U-EF between redox bins 3-5.Three sites of the redox bin 5 sites (Koverhar, Lappohja inactive pockmark , and Sandöfjärden) have C/Sal ratios indicating a shallow SMTZ, which is agreement with previous studies (Jokinen et al., 2020;Dalcin Martins et al., 2022).In addition, the sites have the highest Mo-and U-EFs among the oxic sites (Fig. 4).Through removing these three sites from our original data set, we were able to improve the sensitivity of U and Mo in the lower redox range considerably (Fig. 7C and B), compared to the original Fig. 3.This finding illustrates that enhanced Mo auth and U auth sequestration above the background are not restricted to low bottom water redox conditions, but can also be indicative of eutrophied, but oxic coastal depositional environments.Our approach had a similarly positive effect for C org , in that the initial trend of increasing C org contents from redox bins 3-5 was reversed and now is in line with the expected trend of decreasing C org preservation at rising O 2 concentration (Fig. 7A).Consequently, the combination of a short water column, low salinity, high organic matter input, and high sedimentation ratesall common characteristics of a near coastal depositional environment (as present in the Baltic Sea) -is likely the cause of the high C org contents at sites with oxygenated water columns, which sufficiently explains the higher C org contents in redox bin 5 than in redox bins 3 and 4 (Fig. 3A).

Post-depositional remobilization
The impact of post-depositional remobilization of labile Mo and U mineral phases on the interpretation of sedimentary Mo-EF and U-EF is currently unknown.Generally, Mo can easily be remobilized upon reductive dissolution of labile Fe/Mn (oxy)(hydr)oxides (Sulu-Gambari et al., 2017;Dellwig et al., 2021), and oxidative dissolution of labile Mo-Fe-S phases (Hermans et al., 2019b).Consequently, Mo might escape into the water column, if redox conditions are insufficient (low sulfide levels) to fixate Mo at the surface sediments (Erickson and Helz, 2000).However, oxidative dissolution of refractory authigenic Mo phases (e.g., FeMoS 4 ) under natural conditions is highly unlikely due to irreversible K.M. Paul et al. mineral formation kinetics (Erickson and Helz, 2000;Vorlicek et al., 2018;Jokinen et al., 2020;Helz, 2021).Similarly, post-depositional remobilization of U is dependent on the type of authigenic U species.Principally, the sedimentary U pool is thought to consist of a mixture of refractory uraninite and labile monomeric non-uraninite U IV species strongly associated with organic matter (Bernier-Latmani et al., 2010;Sharp et al., 2011;Bone et al., 2017).Non-uraninite U IV can easily be oxidized due to weak U-U bonds (Alessi et al., 2012).Therefore, study sites subject to sediment re-oxygenation in response to inflow events (Cochran et al., 1986;Shaw et al., 1994;Zheng et al., 2002a), bioturbation (Zheng et al., 2002b;Morford et al., 2009a), or bioirrigation (Martin and Sayles, 1987) could have lower Mo-and U-EFs than expected.These criteria theoretically apply to redox bins 2-5, although benthic macrofaunal activity is highly unlikely for redox bin 2 sites.For sites rates of high Fe/Mn (oxy)(hydr)oxide refluxing (8.+ 9.: Den Osse Basin, and 10.Gullmar Fjord), we expect that any effluxed Mo diss, is efficiently returned to the sediments after trapping by oxides in the water column.Since both Den Osse Basin sites (8.: redox bin 3; 9.: redox bin 4) have Mo-EFs >10, this scenario is indeed likely.In contrast, U-EFs at these sites appear to be depressed, suggesting loss by re-oxidation.Redox bin 5 sites are most prone to post-depositional remobilization due to oxygenated bottom waters, and generally higher macrofaunal densities (e.g., Hermans et al., 2019a).However, among these sites we found a moderate to large scatter in U-EFs (1-3) and particularly Mo-EFs (〈1− 10), respectively.Generally, this speaks for sufficient Mo auth and U auth preservation for the vast majority of redox bin 5 sites, even if postdepositional re-mobilization occurs.Targeted site-specific studies are required to determine the actual impact of post-depositional re-mobilization for Mo-and U-EF at each site.

Detrital dilution
Detrital dilution of sedimentary Mo auth and U auth signals through high sedimentation rates has repeatedly been suggested as a critical factor for the applicability of Mo-and U-based redox proxies (Lyons and Kashgarian, 2005;Scott and Lyons, 2012;Noordmann et al., 2015;Liu and Algeo, 2020).Dilution effects caused by high sedimentation rates are suggested to lead to lower authigenic enrichments of both U (Anderson and Fleisher, 1991) and Mo (Lyons and Kashgarian, 2005).If true, this implies an inverse relationship between sedimentation rate and Mo-and U-EF values for sites with similar redox conditions.However, we found no clear evidence for such relationships in our data (Fig. 6, middle row).The sites within each redox bin show no systematic Fig. 6.Selection of three depositional environmental factors and their impact on A) Mo-EFs and B) U-EFs among the study sites (log 10 scale).Upper row: C org /salinity (C/Sal) ratio.The horizontal dashed line denotes the average C/Sal ratio at the reference site 12. Koverhar.The numbers represent the "anomalous sites" (see Fig. 4 caption).Middle row: whole core or core segment average sedimentation rates.Bottom row: whole core or segment Al concentrations (wt%).The arrow highlights the outlier site 3. Black Sea NW Abyssal Plain .The horizontal dashed line indicates the Al UCC value (8.15, Rudnick and Gao (2014)."Fluffy layer" data are not shown.For absolute values, and references see Table S2.
K.M. Paul et al. ordering towards lower EFs for higher sedimentation rate sites, implying that such effects, if present, are generally obscured by other variability.We note that sedimentation rates among our study sites range between ~0.1 cm yr − 1 in the Black Sea (Wijsman et al., 2002;Lenstra et al., 2019) and ~ 2 cm yr − 1 in the Den Osse Basin deep (Malkin et al., 2014).This range is considerably higher than the euxinic Black Sea sites with "high" (ca.0.1 cm yr − 1 ) and "fairly high" (0.01 cm yr − 1 ) values reported in previous studies into the impact of sedimentation rates on TM-EFs (Algeo and Lyons, 2006;Scott and Lyons, 2012) and thus the conclusions of these studies may not be directly applicable to our data.

Detrital Al background content
The absolute TM-EF can be influenced by site specific detrital Al contents in such a way that a lower Al background content will lead to unrealistically high authigenic enrichments, or vice versa (Brumsack, 2006).To detect whether there are sites within this study showing considerably deviating local Al enrichments than the UCC Al background and thereby causing anomalous TM-EFs, we arranged the local sedimentary Al contents in the order of the Mo-and U-EFs (Fig. 6, lower row).Most sites have an offset to the UCC Al value of 10-30%.Five sites have an offset of >30%, of which one (3.Black Sea NW Abyssal plain ) has a 400% offset (black arrow, Fig. 6, lower row).This is the deepest of all studied sites (> 2000 m) and farthest from shore.Sediments in the pelagic Black Sea are typically starved of detrital material (low Al contents) and enriched in calcium carbonate (Pilipchuk and Volkov, 1974;Kraal et al., 2017), which likely yield an overestimation of the EFs (Little et al., 2015;Bura-Nakić et al., 2020).In the context of redox-sensitive trace metal studies, pelagic Black Sea sediments have commonly been normalized against global background values of PAAS or UCC (Algeo and Tribovillard, 2009;Algeo and Li, 2020;Yano et al., 2020), likely due to the lack of local background estimates, limited data availability of Al, Mo, and U contents in carbonates (Turekian and Wedepohl, 1961), and improved comparability with previous studies.Despite this discrepancy, since the majority of sites only show minor deviation from the UCC Al background, we infer that the effect of over-or underestimation, due to varying Al contents, is negligible among our study sites.

Secondary depositional environmental factors
Our results show that both Mo-and U-EF have limitations compromising their universal use as recorders of bottom water redox changes.These limitations are clearly linked to the influence of secondary Fig. 7. Modified version of Fig. 3 to illustrate the impact of oxic sites with a shallow SMTZ (high C/Sal ratio) on the applicability of A) C org , and B) Mo-EFs, and C) U-EFs.After the removal of sites Koverhar, Lappohja inactive pockmark , and Sandöfjärden (see text for details), the differences between redox bins 4 and 5 have become more distinct, in particularly for U-EF (cf.Fig. 3).See Fig. 3 caption for details.

Table 4
A selection of depositional environmental factors impacting sedimentary Mo-and U-EF.Their impact power is indicated by symbols: --strong negative impact (considerably lower enrichments than expected); ++ strong positive impact (considerably higher enrichments than expected); − negative impact (lower enrichments); + positive impact (higher enrichments); o no/ undetectable impact; +/− negative or positive impact (under− /overestimation); ?indication of impact power, but with uncertainty.depositional environmental factors.When applying Mo-and U-EFs to reconstruct redox variability in a new study location, it may be challenging to identify each of the specific depositional environmental factors, assess the degree to which they impact Mo-and U-EFs, and finally, to correct for their influence on the data.Some depositional environmental factors can be easily identified (e.g., those reflected in the Moand U-EF covariation model established by Algeo and Tribovillard, 2009).As a result of the large heterogeneity among depositional environments, determination of most secondary factors requires additional data collection and analysis, such as water column trace metal concentrations, elemental mass accumulation rates, sediment grain size distributions, bioturbation and bioirrigation potentials, diffusive fluxes, and sediment trace metal speciation.A summary of potential depositional environmental factors that could impact Mo-and U-EFs is given in Table 4.This is intended as a guide to the careful interpretation of data from new study locations in coastal depositional environments.
Based on our data, we suggest that Mo is a more reliable redox proxy within redox bins 3-5 compared to U, likely because of a stronger sequestration potential in the sediment, particularly when H 2 S is present in the pore water, and a lower oxidative dissolution potential.In contrast, we attribute the stronger reliability of U within redox bins 1 and 2 to the relative immunity of U to the processes limiting Mo sequestration in euxinic basins.

Data quality and definition of redox bins
A potential limitation to our approach is that the interpretations depend on the quality of the monitoring data used to assign the sites to a given redox bin.Although many of these sites have been intensively monitored for dissolved O 2 concentrations, the frequency of sampling, length of data series and maximum depth of water sampling are variable (Table 2).We use whole sediment core profiles with a range of temporal coverage that does not necessarily match that of the monitoring data when assessing the Mo-EF and U-EF ranges of each site (Fig. 1).It should also be noted that the sediment profiles show site-and element-specific differences and fluctuations in the Mo and U enrichments over time.These fluctuations can be indicative of either periodic variability in the bottom water redox conditions or of secondary depositional environmental factors as described above.A further concern in dynamic coastal settings is that a significant fraction of sedimentary material can be deposited from lateral transport (e.g., Jokinen et al., 2015), thus potentially obscuring the primary signal of redox conditions recorded in Mo-and U-EFs.Further investigations, such as assessment of grain size distributions combined with trace metal speciation may help to disentangle the reasons for within-site variability and to improve the understanding of Mo and U sequestrations in heterogeneous coastal sediments.
The approach of allocating sites to redox bins is also not free of limitation.In our dataset, each redox bin contains at least three different depositional environments with regards to bottom water salinity, temperature, sedimentation rate, and water depth (except for redox bin 3, which contains two sites, dominated by one site consisting of data from two different sampling years and seasons).As demonstrated, these factors have implications for the observed ranges of Mo-and U-EFs (Fig. 3B  and C).Particularly, we identified the impact of a shallow SMTZ (high C/Sal ratio) in coastal depositional environments with an oxygenated water column as a key influence dictating the ability to differentiate between ir/regularly dysoxic and permanently oxic depositional environments.However, the statistical comparison of the violin plots demonstrates that when all sites are considered together, contrasts between redox bins are well reflected by Mo-EF, U-EF, or both, confirming the overall applicability of these proxies in coastal depositional environments.As a result, artefacts resulting from summarizing multiple depositional environments thus manifest as outliers around the expected trend controlled by redox conditions.

Normalizing TM enrichments against Al and the crustal background
The practice of normalizing TM enrichments against crustal background values is a topic of ongoing debate, due to potential overestimations of the authigenic signals.Several authors instead support the use of either "unnormalized" EFs, simply using the sample TM/Al ratio (Bennett and Canfield, 2020), or TM XS using UCC and estimates of the local lithogenic background ratio in cases where TM XS yields negative values (Jokinen et al., 2020).We compared the sequence of sites according to TM-EF values with the corresponding order according to TM/Al ratio and TM XS (Figs.S5 + 6).The order of sites remained unchanged when using TM/Al ratios, due to the comparable calculation, and changed only slightly without improving the redox sequence when using TM XS .A general issue that arises from using Al and crustal background values concerns depositional environments that are dominated by carbonates (marine or detrital), resulting in considerably lower Al contents compared to siliciclastic-dominated depositional environments.Consequently, if the chemical composition of the sediments largely varies across the study sites, then the resulting EFs become less comparable to each other, due to likely overestimations of the EFs at sites with lower Al (higher carbonate) contents.Certainly, more accurate results could be gathered from local background values, but these are often difficult to determine.Alternatively, average carbonate standards could provide a simpler solution (Steinmann et al., 2020).Although such carbonate standards exist (sedimentary carbonate rocks, deep-sea carbonates, Turekian and Wedepohl, 1961), the trace metal data are incomplete and subject to error.Essentially, as long as sites have a comparable chemical composition, this includes carbonate-rich sediments (e.g., Deng et al., 2020), normalization against the crustal background remains a valid tool.Among our study sites the chemical composition is relatively similar, except for the carbonate rich Black Sea NW Abyssal Plain site, which makes it principally less comparable to the other sites.These comparability issues arise both when using "unnormalized" EFs (sample TM/Al) and EFs using crustal background values.The reason is that while the particulate and sedimentary Mo and U concentrations of the river loads draining into the Black Sea are relatively close to UCC values (closer than to shale), Al contents are more similar to those of average deep-sea carbonates (Turekian and Wedepohl, 1961;Yigiterhan and Murray, 2008).Calculating EFs Black Sea NW Abyssal Plain site using UCC values for Mo and U, and deep-sea carbonate values for Al resulted in a considerably poorer differentiation of the redox bins 1 and 2-5 for both elements.Using unnormalized trace metal enrichments in this case might be the most suitable alternative for the Black Sea, but it might cause complications for sites with a larger detrital component of Mo and U.For the purpose of our study, focusing on mildly to non-reducing coastal depositional environments, largely with a heterogenous siliciclastic detrital background mixed, we found that using the UCC was the best approach.

Use of unnormalized trace metal threshold values
We further assessed the applicability of threshold TM content values to differentiate our defined redox bins.It has become a well-established approach to use a "universal" threshold value of unnormalized 25 ppm Mo as a baseline to differentiate non-euxinic from intermittently euxinic/ euxinic (paleo) redox environments, with data commonly displayed as histograms (Scott and Lyons, 2012;Dahl et al., 2013;Sweere et al., 2016;Scholz et al., 2017).As for Mo-and U-EFs, such thresholds must be applied with care due to the impact of secondary depositional environmental factors but provide a useful guide.
The majority of the samples of our sites in redox bins 1 and 2 have Mo contents between 10 and 50 ppm, with ~20% of the samples >50 ppm (Fig. 8A), which is generally in agreement with the findings by Scott and Lyons (2012).The relatively high occurrence of samples with Mo < 25 ppm in redox bins 1 and 2, however, leads to an overlap with samples from redox bins 3 and 4. The clearest differentiation in our dataset are samples where Mo >100 ppm, exclusively representing redox bin 1 (persistently euxinic).Contents of Mo >5 ppm are indicative of O 2 K.M. Paul et al. depletion corresponding to at least redox bin 3 (ir/regularly suboxic).Values of <5 ppm are observed only in redox bins 4 and 5 (ir/regularly dysoxic to persistently oxic).Differentiation between redox bins 4 and 5 is improved once the three oxic sites with a shallow SMTZ (high C/Sal ratio) are removed (gray shaded bars, Fig. 8A).A small overlap remains for values <1 ppm, however the majority of sites belong to redox bin 5. Overall, the histograms for Mo resemble Mo-EF violin plots in which redox bins 3-5 are clearly distinguishable in the frequency distribution.
For comparison, we followed the same threshold approach for U (Fig. 8B).In general, U enrichments of 5-10 ppm largely represent redox bin 2 sites, and values >10 ppm represent redox bin 1 sites, supporting the suggestion from the U-EF violin plot that U is a better redox proxy for differentiating between euxinic and ir/regularly euxinic environments.However, values <10 and < 5 cannot further distinguish between redox bins 3-5.As in the case for Mo-EF, this circumstance improves considerably once the three redox bin 5 sites are removed (gray shaded bars, Fig. 8B).In turn, redox bin 5 sites are more clearly recognizable by values <2 ppm.This finding supports our hypothesis that U, expressed as normalized EF or as unnormalized enrichments, is unable to reliably record bottom water redox changes in the range of oxic to ir/regularly suboxic, which contradicts the commonly accepted view of U as a more sensitive bottom water redox proxy than Mo.High C org degradation and sulfide production rates, promoting higher U enrichments in oxic sites, may strongly weaken the sensitivity of U in more mildly reducing conditions.Despite showing similar discrepancies when using U-EFs, it must be noted that it is generally not advised to use unnormalized U contents due to having a larger detrital component compared to Mo, which dilutes the authigenic signal (here: 18 sites average U detrital = 46.4% of total U, Mo detrital = 25.7% of total Mo; Adams and Weaver, 1958;Cole et al., 2017).Ultimately, the total range within each redox bin provides more information on the impact of secondary factors, and mechanisms of Mo and U enrichments, than threshold values do.

Conclusions
Our study demonstrates that a combination of Mo-and U-Enrichment Factor (EF)-based redox proxies, supported by unnormalized molybdenum (Mo) and uranium (U) concentrations, allows differentiation of a variety of modern coastal depositional environments with regards to their bottom water O 2 conditions.In contrast to the common understanding, we found that Mo, rather than U, is the more sensitive and reliable bottom water redox proxy in mildly reducing depositional environments.At more severe levels of deoxygenation, U rather than Mo, is the more reliable proxy, owing to the relative insensitivity of U to factors limiting sedimentary enrichment in sulfidic depositional environments (i.e., "basin reservoir effect" and saturation limits).Through consideration of several secondary depositional environmental factors, we identified cases where the element-specific chemistry, shuttling of Mo with Fe (oxy)(hydr)oxides, depth of the sulfide front in the sediment, or a combination of these factors may obscure the relationship between redox conditions and Mo and U enrichments.We propose a scheme for assessing the potential impact of such secondary factors for Mo-and Ubased redox proxy studies.Further studies should also address the potential role of grain size and sedimentary sources in influencing Mo-and U-EFs.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Scott and Lyons (2012) and Dahl et al., 2013).Data for samples from "fluffy" layers are not shown in this figure .K.M. Paul et al. Geology 615 (2023)

Fig. 2 .
Fig. 2. Covariation patterns between Mo-EF and U-EF, modified fromAlgeo and Tribovillard (2009).The four black arrows and letters A1, A2., B, and C refer to the key enrichment controls: A1. and A2.bottom water redox variations (mild redox variations between oxic and dysoxic + increase in organic matter degradation and release of sulfide to the pore-and bottom waters), B operation of a particulate Fe/Mn (oxy)(hydr)oxide "shuttle", and C limited Mo sedimentary sequestration in sulfidic depositional environments (comprising the "basin reservoir effect" and equilibrium of FeMoS 4 with Mo diss at high sulfide concentrations(Helz, 2021).Uranium is not affected by such mechanisms, and therefore continues to enrich in the sediment.The three diagonal lines represent multiples (0.3, 1, and 3) of the Mo:U ratio of present-day seawater (SW) converted to an average weight ratio of 3.1 for the purpose of comparison with sediment Mo:U weight ratios(Tribovillard et al., 2012).
Fig.3.A) Organic carbon content (wt%) for the different redox bins (Table2) plotted as violin plots (log 10 scale).Each violin plot represents the data distribution of each redox bin.The black cross and vertical black solid line with two whiskers within each violin plot denotes the mean with a 0.99 confidence interval (CI).The median absolute deviation (MAD) -a measure of the variability of non-normal distributed datais illustrated by an empty vertical crossbar, divided by a horizontal black solid line, which represents the median.Numbering of the redox bins follows the scheme introduced in section 3.1: persistently euxinic (redox bin 1), ir/ regularly euxinic (redox bin 2), ir/regularly suboxic (redox bin 3), ir/regularly dysoxic (redox bin 4), and persistently oxic (redox bin 5).B) Sedimentary Mo and C) U enrichment factors (EFs) across the redox bins (log 10 scale).The horizontal dashed lines indicate the TM-EF value of 1, which is the threshold for authigenic Mo and U enrichments.Values below 1 indicate non-authigenic Mo and U concentrations.Data from organic-rich "fluffy" layers at the sediment surface of sites in redox bins 1-3 are not shown on this figure.Otherwise, all data points are shown.The total number of data points for each element and redox bin is given at top of each panel.
Fig. 4.Sedimentary A) Mo and B) U enrichment factors (EFs) among the study sites (log 10 scale).For each element, the sites are arranged according to the order of their means, starting with the highest.The color of each box (site) corresponds with its redox bin (Fig.3, Table2).For some sites the total ranges vary between the two elements, due to different amounts of data points available (e.g., Lappohja, and Arkona, Table2).Fårö Deep, Northern Gotland Basin, Bornholm Basin, and Lilla Värtan are listed twice: as the upper core section, covering the last <30-40 years, and as the lower core section, covering the preceding >30-40 years.The numbers refer to the sites that plot anomalously (see text).For detailed description of the violin plots see Fig.3caption).The horizontal dashed lines indicate the threshold for authigenic Mo or U enrichment (see Fig. 3 caption).Data from organic-rich "fluffy" layers are not shown on this figure.
K.M.Paul et al.

Fig. 5 .
Fig. 5. Mo-and U-EF covariation patterns among the study sites based on whole core sediment samples (concept after Algeo and Tribovillard, 2009).The legend to the left shows all study sites with their corresponding bottom water redox color.The double arrow on the right shows the color coding of the redox bins 1-5 (Table 2)."Fluffy layer" samples are shown on this figure and indicated by dashed lines around the symbols.The two gray shaded boxes highlight the sites that have lower than UCC values for Mo and U.For detailed description of the lines and arrows, see Fig. 2 caption.

Fig. 8 .
Fig. 8. Occurrence (histogram) distribution of unnormalized A) Mo and B) U concentrations for the different redox bins.The redox bins are indicated by their color and assigned redox bin numbers to the right of each panel.The vertical gray solid lines mark a selection of threshold values as defined inScott and Lyons (2012) andDahl et al., 2013).Data for samples from "fluffy" layers are not shown in this figure.

Table 2
List of study sites categorized in redox bins (1-5) based on their modern average bottom water redox condition, estimated based on monitoring data and literature.Physical and chemical bottom water (BW) data (temperature, salinity, O 2 , H 2 S) at the time of sampling is provided.BDL = Below Detection Limit.Further details about the study sites and monitoring data are provided in TableS1and the listed references.

Table 3
References for the sources of Mo and U data for all study sites.
a References refer to both core sections of Fårö Deep, Northern Gotland Basin, Bornholm Basin, and Lilla Värtan stated in Table2.b Only Mo data were available for the two campaigns.
121203 the Netherlands Organisation for Scientific Research (NWO; Vici-grant number 865.13.005), and the European Research Council (ERC Synergy Marix grant number 854088).This work was carried out under the program of the Netherlands Earth System Science Centre (NESSC), financially supported by the Ministry of Education, Culture and Science (OCW).HLF acknowledges support by the Swedish Research Council VR (grant number 2017-04190).