Geochemical consequences of oxygen diffusion from the oceanic crust into overlying sediments and its significance for biogeochemical cycles based on sediments of the NE Pacific

Exchange of dissolved substances at the sediment–water interface provides an important link between the short–term and long–term geochemical cycles in the ocean. A second, as yet poorly 15 understood sediment–water exchange is supported by low–temperature circulation of seawater through the oceanic basement underneath the sediments. From the basement, upwards diffusing oxygen and other dissolved species modify the sediment whereas reaction products diffuse from the sediment down into the basement, where they are transported by the basement fluid and released to the ocean. Here, we investigate the impact of this ‘second’ route with respect to transport, release and consumption of 20 oxygen, nitrate, manganese, nickel, and cobalt on the basis of sediment cores retrieved from the Clarion Clipperton Zone (CCZ) in the equatorial Pacific Ocean. We show that in this abyssal ocean region characterised by low organic–carbon burial and sedimentation rates vast areas exist where the downward and upward directed diffusive fluxes of oxygen meet so that the sediments are oxic throughout. This is especially the case where sediments are thin, or in the proximity of faults. Oxygen 25 diffusing upward from the basaltic crust into the sediment contributes to the degradation of sedimentary organic matter. Where the sediments are entirely oxic, nitrate produced in the upper sediment by nitrification is lost both by upward diffusion into the bottom water and by downward diffusion into the

layers of the sediment. Hence, paleoceanographic interpretation of sedimentary layers should carefully consider such deep secondary modifications in order to prevent misinterpretation as primary signatures.

Introduction 35
The heat flux from the mantle through the lithosphere into the ocean is high where the ocean crust is young and sediment cover sparse such as at spreading zones, and the heat flux typically decreases with age as the crust cools (e.g., Williams and von Herzen, 1974;Sclater et al., 1980;Davis et al., 1999). Ventilation of the basaltic crust by seawater is an important aspect of this cooling and is predicted to occur in off-axis settings up to an age of 65 Ma (e.g., Stein and Stein, 1992;Fisher et al., 2003;Hutnak et al., 2006). Most of the ocean floor has cooled over long periods and conversely, most ventilation of the 40 basaltic crust occurs at low temperatures (Hasterok et al., 2011;Coogan and Gillis, 2018). Water percolating through the crust changes the chemistry of the basement and vice versa (Fisher and Wheat, 2010;Orcutt et al., 2013). Moreover, sediments overlying the basement may be modified by substances diffusing from the crust upwards such as oxygen whereas sedimentary products may diffuse down and interact with the crust (Ziebis et al., 2012;Mewes et al., 2016;Kuhn et al., 2017). Due to the vastness of the ocean floor, this low temperature ventilation may still have a large impact on the global 45 geochemical and biogeochemical cycles (e.g., Baker et al., 1991). This impact is however poorly constrained. This is to a large extent due to sediments accumulating on the basement, making it less accessible and due to poor sediment permeability, they progressively insulate the crust, both thermally and hydrodynamically (Anderson et al., 1979). As a result, water can enter or leave the basement only where sediment cover remains thin or absent such as at seamounts, ridges and faults. Finally, diffusion of dissolved species from the basement into the sediments and vice versa modifies the compositions 50 of both fluid and sediment and adds an extra aspect to the geochemical and biogeochemical impact of basement ventilation (e.g., Wheat and Mottl, 2000;Fisher et al., 2003;Wheat and Fisher, 2008;Kuhn et al., 2017).

Sediment geochemistry.
The ocean floor is subject to a continuous rain of reduced material, notably organic matter (OM). Sedimentary life harvests the energy stored in this reduced material by exploiting and catalysing the redox reactions possible, giving priority to the 55 energetically most favourable redox pair (e.g., Froelich et al., 1979;Berner, 1980). In oxic sediments, the dominant redox process is aerobic respiration. Oxygen entering the sediment by diffusion is consumed on its way down and if sufficient OM is available it becomes exhausted resulting in an oxic/suboxic interface at depth. In a steady state situation, this interface keeps pace with sedimentation and its depth relative to the sediment surface remains constant (e.g., de Lange et al., 1983;Kasten et al., 2003). From this interface downwards successively energetically less favourable redox reactions follow, 60 provided sufficient electron donors remain (e.g., Canfield and Thamdrup, 2009;Kasten et al., 2003). However, due to successively reduced activation energies, oxidation of OM will be increasingly less complete (e.g., Tegelaar et al., 1989;https://doi.org/10.5194/bg-2021-112 Preprint. Discussion started: 30 April 2021 c Author(s) 2021. CC BY 4.0 License. Arnarson and Keil, 2007;Zonneveld et al., 2010). Dissolved redox products with remaining oxidation potential may diffuse to a zone with energetically more favourable redox reaction conditions and become further oxidised at the corresponding redox front (e.g., the oxic/suboxic interface). Oxidation of immobile components requiring a high activation energy slows 65 down or even stops, leading to long-term preservation as is the case of the more refractory OM below the oxic/suboxic interface (e.g., Arndt et al., 2013;Zonneveld et al., 2010).
Less input of reactive OM or lower sedimentation rates shift the balance between oxygen diffusion into the sediment and its benthic consumption deeper into the sediment. The result is increased oxygen exposure time of the sediments and a deeper oxic/suboxic interface, which can reach depths of up to several meters and more (Fischer et al., 2009;Rühlemann et al., 70 2011;Ziebis et al., 2012;Mewes et al., 2014;d`Hondt et al., 2015;Mewes et al., 2016;Volz et al., 2018). The higher degree of OM degradation results in a more refractory residue. Conversely, OM degradation rates decrease more rapidly and to a larger extent with distance into the sediment as compared to regions with higher productivity and/or sedimentation rates .
In sediment-filled basins of the western flank of the Mid-Atlantic ridge at "North Pond" and in the deep-sea sediments of 75 the Clarion Clipperton Zone (CCZ) oxygen was shown to not only diffuse down across the sediment-water interface but also upward into the basal sediments from the underlying oceanic basement (Ziebis et al., 2012;Mewes et al., 2016;Kuhn et al., 2017). For this upward diffusion both oxygen concentration profiles and redox dynamics involved differ from those near the sediment surface. Oxygen concentrations of the fluids circulating in the permeable crust were demonstrated to be lower than those in the bottom water overlying the sediment surface due to partial consumption as a consequence of oxidation of 80 reduced mineral phase contained in the crustal basalt (Fisher and Wheat, 2010;Orcutt et al., 2013;Mewes et al., 2016) The oxygen entering the sediment from below also encounters old, refractory OM and meets electron donors diffusing down from the redox-zones above. In this zone of upward diffusing oxygen, oxygen exposure time of OM increases downward and may be very long, up to equaling the age of the sediment involved in case this upward diffusive supply of oxygen never ceased. 85 As yet the upward diffusion of oxygen from the basement has only been documented for a few abyssal ocean environments and its impact on the sediment composition, OM preservation/burial and biogeochemical cycles is largely under-explored and poorly understood (Kuhn et al., 2017;Mewes et al., 2016;Ziebis et al., 2012). This is to a large extent a result of coring practice and/or list of parameters typically measured onboard ship. Mostly, coring devices such as piston and gravity corers are too short, or the sediment cover involved too thick to reach beyond the redox zones in the upper meters sediments and 90 reach the deep oxic redox zone induced by the upward diffusion of oxygen from seawater circulating within the basement.
Furthermore, although there is a long history of deeper drilling into the ocean sediments, and through them into the ocean crust (such as in the DSDP, ODP and IODP frameworks), the instances during which ex situ oxygen measurements were performed on these long sediment cores are extremely rare (e.g. Fischer et al., 2009;Ziebis et al., 2012;d'Hondt et al., 2013)  To overcome this problem, we went to the NE Pacific to the region of the CCZ in the framework of RV Sonne expedition 95 SO240 (Kuhn et al., 2015) and among other pore-water and solid-phase components performed ex situ oxygen measurements on whole-round gravity cores (Fig. 1). Sedimentation rates in the CCZ are low ranging between 0.2 and 1.2 cm/ka (Mewes et al., 2014;Volz et al., 2018) thus combining a relatively thin sediment cover with a long period of sedimentation and oxygen penetration from below. As such, the lower oxidation front had ample time to move up and to get into the reach of our coring devices. Sediments with such low sedimentation rates represent about 70% of the ocean floor 100 (Bowles et al., 2014;Mewes et al., 2016) and conversely we hypothesise that extensive and long oxidation must be a widespread phenomenon in abyssal ocean deposits. Therefore, understanding the processes associated with upward oxygen supply from the basement into overlying sediments may be crucial for our understanding of global biogeochemical cycles and for properly interpreting marine sedimentary records/archive. To shed more light on the exchange between the basaltic basement and the ocean as well as the role of the oceanic sediments herein, the oceanic basement and overlying sediments of 105 the abyssal plain in the CCZ have been investigated seismically, thermally and geochemically during and after RV Sonne cruise SO240 in the framework of the project FLUM (FLUM project Kuhn et al., 2015;Fig1).

Study area
The oceanic basement underlying the sediments between the Clarion and the Clipperton Fracture Zones has been formed from the late Cretaceous to the Miocene (Eittreim et al., 1992). The region hosts many seamounts and faults, which due to 110 their steep topography are often barren of sediment on their slopes (e.g., Rühlemann et al., 2011). This allows seawater to circulate through the oceanic crust to enter or leave (Fisher and Wheat, 2010;Kuhn et al., 2017;Mewes et al., 2016). Sediment accretion is slow (typically a few mm/ka) due to the great distance from land, relatively great water depth, a low carbon flux to the sea floor of 1.5-1.8 mg C org m -2 d -1 (Lutz et al., 2007) and a sea floor below the carbonate compensation depth. Meter-scale oxygen penetration into the sediment results in extensive organic matter mineralisation (TOC typically < 115 0.2%, and mostly below 0.4% at the surface) due to oxygen exposure times in the order of hundreds of thousands of years.
In this paper we concentrate on the impact dissolved substances migrating from the crust have on the overlying sediment and vice versa sediment interactions to better understand the impact of seawater ventilation of oceanic crusts on global biogeochemical cycles. We build upon recent insights on the low-temperature exchange of dissolved components between 120 the basement and the overlying sediment and role of the sedimentary record Mogollón et al., 2016;Mewes et al., 2014;Fisher and Wheat, 2010;Kuhn et al., 2017;Volz et al., 2018). We pay special attention to the upward diffusion of oxygen from the basement to obtain a better insight on its impact on sediment composition, preservation/burial of OM and its relevance for global biogeochemical cycles.

Material and Methods 125
The oceanic crust in the region of investigation between 11.5°N, 116.3°W and 13.1°N, 119.6°W (Fig.1) is of middle to late Eocene age (17-21 Ma) ( Barckhausen et al., 2013). Sediment cover is between 0 and 90 meters thick (Kuhn et al., 2015; Table 1). Sediments have been recovered during RV Sonne cruise SO240 (Kuhn et al., 2015) by means of box-, multiple-, gravity-and piston corer (Table 1; Fig 1). Cores consist generally of stiff and compact brown clays with depending on the MnO 2 contents lighter brown (less MnO 2 ) and darker (more MnO 2 ) sections (Kuhn et al., 2015). Of these, cores 22KL, 130 42SL, 72SL, 81SL and 96SL were taken close by or above a fault, cores 5SL, 35SL, 51SL, 53SL and 117SL were taken close to a seamount, core 69SL was taken near both a fault and a seamount and cores 9KL, 58SL, 65SL and 108SL were taken near neither a fault nor a seamount.

Oxygen analyses
Pore-water oxygen in the retrieved sediments was determined ex situ onboard using amperometric Clark-type oxygen 135 sensors according to the procedure described by Ziebis et al. (2012), Mewes et al. (2014Mewes et al. ( , 2016 and Volz et al. (2018). For conversion of the measured mV to mmol/l we used Eq. (1):

140
where V = sensor voltage at depth (V d ), under O 2 saturation (V sat ) and in argon (V ar ). O 2(sol) is the oxygen solubility at the given temperature (measured), salinity (measured) and pressure conditions (atmospheric pressure) was calculated using the equation of García and Gordon (1992). Measurements were performed in an acclimatised room kept at 4 to 6°C.

Pore-water sampling and analyses
Upon recovery of the sediment cores, pore water was sampled using rhizons according to the procedure described by 145 Seeberg-Elverfeldt et al. (2005) with a typical average pore diameter of 0.1 µm (Mewes et al., 2014). For NO 3 analyses a QuAAtro Continuous Segmented Flow Analyser was used (Seal Analytical). Pore-water Mn 2+ Ni 2+ and Co 2+ were determined after sample acidification with distilled HNO 3 by inductively coupled plasma optical emission spectrometry (ICP-OES; IRIS Intrepid ICP-OES Spectrometer, Thermo Elemental) using the NIST2702 standard as reference and following the procedures described by (Mewes et al., 2014). For solid-phase element analysis, 50 mg of freeze-dried and homogenised bulk sediment were digested with a mixture of HF (0.5 ml, 40%, suprapur), HNO 3 (3 ml, 65%) and HCl (1.5ml, 30%, suprapur) at 220°C using a Mars Express microwave system (CEM) as described by Nöthen and Kasten (2011). Major and trace elements have been quantified in the laboratories of the Alfred Wegener Institute Helmholtz Center for Polar and Marine Research (AWI) in Bremerhaven and the Jacobs 155 University Bremen using inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma optical emission spectrometry (ICP-OES), and at the Federal Institute for Geosciences and Natural Resources (BGR) in Hanover by quantitative XRF (X-ray Fluorescence Spectroscopy) (Rietveld, 1967). Cross calibration, achieved by quantifying the same elements with ICP-MS, ICP-OES and XRF for a few cores showed that the XRF measurements reliably reproduce the changes in the elemental composition of the sediment. 160 The total carbon content (TC), the content of organic carbon (C org ) and the total sulphur (TS) content were determined using LECO CS 230 Carbon-Sulfur-Analyzer at the BGR. The samples were burned in a high-frequency-oven at 1800°-2200°C and the carbon and sulphur contents were determined using infrared-detection. Untreated samples were used to determine TC and TS. Another sample aliquot was treated with 2N HCl at 80°C to remove carbonate-carbon and subsequently measured for C org The difference between TC and C org equals to carbonate-carbon. 165 Porosity has been calculated from the water content of samples with a known volume, corrected for salinity and using a bulk mineral density of 2.53 g cm -3 measured on core 22KL.

Results
The oxygen profiles of the cores analysed which are oxic throughout are highly asymmetric. From core top to bottom, oxygen shows a strong decrease in the upper 30 centimetres and below a less pronounced decrease with depth leading to an 170 interval of minimum oxygen concentrations between 2-3 m (Fig. 2). From there in most of these cores, a linear re-increase in oxygen concentrations is observed towards the bottom of the cores (51, 53SL, 117SL, 35SL, 81SL, 69SL, 96SL, Fig. 2).
Core 5SL does not entirely follow these general patterns but shows an S-shaped oxygen profile (Fig. 2) we explain by assuming non-equilibrium from recent mass transport adding sediment at the location of this core, and/or by lateral transport of oxygen and for that reason we do not take this core into further consideration. 175 At two sites, these upper and lower oxic zones are separated by an intermediate suboxic interval as indicated by the presence of Mn 2+ in the pore-water (Fig. 3, 22KL, 42SL). There are five cores where the suboxic zone continues to the core bottom (9KL, 58SL, 65SL, 72SL 108SL). Of these seven cores that are partly suboxic, none has been retrieved close to a seamount, and none except for 22KL and 42SL has been retrieved from above a fault. Except for 58SL, all cores which are partly suboxic show convex pore-water Mn 2+ profiles whereby the profile from the upper oxic/suboxic interface downward is 180 https://doi.org/10.5194/bg-2021-112 Preprint. Discussion started: 30 April 2021 c Author(s) 2021. CC BY 4.0 License. steeper than the upward profile from the lower (extrapolated) oxic/suboxic interface. For core 22KL pore-water oxygen concentrations decrease linearly with depth. For site 58SL a downward decrease in Mn 2+ concentration is absent and concentrations increase to the core bottom.
In all but two cores pore-water NO 3 concentrations decrease downward from the core top (Figs 2, 3). For cores 22KL and 35SL the maximum is not at the sediment surface but, slightly below in the top meter and from this maximum the downward 185 decrease commences (Figs 2, 3). In the cores with a suboxic zone the downward decrease in NO 3 has a concave shape in the upper part, which is followed in the suboxic zone by a less steep linear decrease (9KL, 42SL, 58SL) or linear increase (22KL). The entirely oxic cores (35SL, 53SL, 117SL, 96SL) all display a linear downward decrease in NO 3 -. Extrapolation of the NO 3 concentrations to the sediment-basement interface (see Table 1 for sediment thicknesses) provides values between 43 and 48 µmol/l at the interface for all oxic cores and for cores 22KL and 108SL. Core 5SL, and the remaining 190 cores with a suboxic interval (lower extrapolated NO 3 concentrations) do not follow this general pattern.
Pore-water trace metals that co-vary in their concentrations with Mn are Co and to a lesser extent Ni (Fig. 4). In combination with the solid-phase data of these elements, we see from top to bottom (if present, Fig. 4) an oxic zone (with for 22KL clearly decreasing TOC values), without free pore-water Mn and Co and with low pore-water Ni and elevated solid-phase Mn, Co, and Ni values, often with considerable scatter. Below this, a 'transition zone' (Fig. 4) is present with 195 low oxygen, (and for core 22KL TOC low and constant) and solid-phase Mn, Co, Ni at the same level as in the oxic zone directly above whereas in the pore-water these elements mostly steeply decrease downward. Below this 'transition zone', solid-phase trace metal values are low and pore-water values are high. Below this, there is a sharp transition to oxic conditions without a transition zone. Like in the upper oxic zone, pore-water Mn and Co are below detection limit and dissolved Ni is low again, while the solid-phase values of all three elements increase (Fig. 4). Dissolved Fe has not been 200 detected in any of the pore-waters, neither in the oxic nor in the anoxic sections. Solid-phase manganese for core 22KL shows an interval with low values in the suboxic interval.

Discussion
For most oceanic sediments oxygen is high at the sediment-water interface and steeply decreases to zero at depth (e.g., Wenzhöfer and Gludd, 2002) due to mineralisation of organic matter (OM). The oxygen profile is concave since oxygen is 205 consumed over its entire oxic length (e.g., Froelich et al., 1979;Cai, and Sayles, 1996;Wenzhöfer and Gludd, 2002). The cores analysed in this study show the typical features of abyssal, carbon-starved oceans with sufficient deep-water ventilation. OM oxygen exposure times are long, OM degradation extensive, the remaining OM refractory and conversely sedimentary oxygen consumption rates low. As expected, we observe oxygen penetrating several meters into the sediments (see also e.g. Fischer et al., 2009). Despite this, for some cores, sufficient degradable OM remains available to exhaust 210 oxygen at depth. Below the oxic zone and with increasing sediment depth, we observe zones of denitrification and manganese reduction. (Froelich et al., 1979;Berner, 1980;Mogollón et al., 2016). However, iron reduction and sulphate reduction have not been observed due to the low content and reactivity of buried OM.

Pore-water oxygen
The presence of a second, upward directed diffusive flux of oxygen in many of the sediments investigated in our study and in 215 some others suggests that oxygen is supplied from low-temperature seawater ventilation of the underlying basement. This seems to be a widespread phenomenon (Wheat and Fisher, 2008;Fisher and Wheat, 2010;Ziebis et al., 2012;Orcutt et al., 2013;Mewes et al., 2016;Kuhn et al., 2017). The interaction between the upward and downward directed oxygen profiles through time shapes the pore-water and solid-phase element profiles in the investigated sediments. Two types of geochemical settings have been encountered in the investigated sediments. One type is represented by entirely oxic 220 sediments with concave oxygen profiles (Fig. 2) and the other by cores with a suboxic zone of which some show below this a second oxic zone where oxygen concentrations increase towards the core bottom (Fig. 3). A compilation of the extension of the upward and downward pore-water oxygen and manganese profiles (Fig. 5) shows that the oxygen pore-water profiles from the core top downward are restricted to the upper three meters. The relatively constant thickness of the upper oxygen pore-water profile (mostly 2-3 m), for the oxic cores suggests that the combination of OM flux to the sediment and benthic 225 mineralisation, oxygen diffusion and sedimentation rate is relatively constant so that the system would turn into suboxic conditions somewhere below the upper 2-3 m. Although sedimentation rates in the area vary (Müller and Mangini, 1980;Mewes et al., 2016;Mewes et al., 2014;Volz et al., 2018), they are significantly lower than the rate at which oxygen diffuses into the sediments and thus have no significant influence on the oxygen penetration depth. This would imply that the thickness of the sediment cover and the diffusive oxygen delivery from the basement into the overlying sediments, i.e., the 230 steepness of the upward oxygen profile in the end determines if, and to what extent, a suboxic zone is present in the sediment.
In case oxygen consumption exceeds supply from above and below, two oxic/suboxic fronts are formed with the intermediate suboxic zone having free Mn 2+ in pore-water. In a steady state situation, both fronts are metastable equilibria between diffusion of reduced compounds produced in the suboxic zone to the oxidation front and consumption and diffusion 235 of oxygen in the oxic zones. Whereas in the upper oxic zone the concave oxygen profile indicates that oxygen is consumed over the entire oxic zone due to the presence of more reactive OM close to the sediment surface, in the lower oxic zone the linear oxygen profiles indicate that oxygen consumption along the profile is insignificant compared to that at the front (Fig.   3). In combination with sediment thickness, this linearity allows us to extrapolate the profiles to the basement and calculate basement oxygen concentrations for cores 35, 69 and 96 (using the seismic-based estimates of sediment thicknesses; Table  240 https://doi.org/10.5194/bg-2021-112 Preprint. Discussion started: 30 April 2021 c Author(s) 2021. CC BY 4.0 License. 1). The thus obtained values (SL35, 66µmol/l @ 18m; 69SL, 204µmol/l @ 70m; 96SL, 25 µmol/l @ 15m) are considerably different, an observation also made for North Pond at the Mid-Atlantic Ridge (Ziebis et al., 2012;Orcutt et al., 2013). The differences may be caused e.g. by differences in fluid velocities, fluid path-length in the basement in combination with consumption of oxygen by reduced compounds in the basement or proximity to subsurface faults (Kuhn et al., 2017).
The influence of the sediment thickness on pore-water oxygen profiles is evident from the cores from areas with a thin 245 sediment cover, 96SL, 5SL, 51SL, 53SL 117SL which are all entirely oxic (  1) whereby core 81SL has been taken above a fault in contrast to core 108SL. Considering the different lengths of the 250 oxygen diffusion paths from below, we suggest that subsurface faults may facilitate upward oxygen diffusion, thus explaining the absence of a suboxic zone for core 81SL as already suggested by Kuhn et al. (2017). The presence of a fault may also explain why oxygen penetrates further upwards at site 72SL (near fault) compared to 65SL (no fault) despite both cores having been taken close together (Fig. 1) from sediments with similar thickness (72 and 71 m respectively; Fig 5). The presence of an underlaying fault may also explain why 69SL is entirely oxic, despite the thickness of the sediment cover. 255 Finally, it may explain why in core 22SL the lower oxic/suboxic interface is located at a shallower sediment depth than in core 9SL, despite the thicker sediment cover for core 22KL. Considering the differences between core 72SL (partly suboxic) and core 69SL (entirely oxic) which are both taken above a fault we need to take into account that the degree of facilitated diffusion by subsurface faults may differ so that sediments may differ in the degree to which they are suboxic. Especially in combination with extension, pore volumes may increase, so that it is conceivable that faults increase sediment permeability. 260 For these cores (> 10 m thickness), fluid advection may be neglected due to the low permeability of these sediments (Spinelli et al., 2004;Kuhn et al., 2017).
Apart from sediment thickness and the effect of faults, the oxygen concentration at the sediment-basement interface is a third variable determining the oxygen profile into the overlying sediments. This third variable depends a.o. on the flow rate, length of way of the fluids in the basement, the amount of reducible compounds in the basin, and basement temperature, 265 whereby the latter two variables also depend on basement age (e.g., Coogan and Gillis 2018). Near sites where oxygen-rich seawater enters the basement, such as near larger seamounts, oxygen concentrations at the sediment-basement interface will be highest (e.g., Fischer et al., 2003;Wheat et al., 2004). Since within the basement oxygen is consumed, and oxygen diffuses from the basement into the sediment, further away from these sites of inflow, oxygen concentrations at the sediment-basement interface will be lower. This decrease in oxygen concentration per unit distance from the site of 270 entrance will be faster if flow rates through the basement are low. Over time, the reactivity of the basement decreases as does temperature, and both may lead to a reduction of oxygen loss per unit distance through the sediment, provided a https://doi.org/10.5194/bg-2021-112 Preprint. Discussion started: 30 April 2021 c Author(s) 2021. CC BY 4.0 License. simultaneous reduction in fluid flow rates doesn't compensate this. The oxygen diffusion is not necessarily vertical only. For instance, 69SL is taken close to a ridge system with a steep slope and sediment cover bay be sufficiently thin (Fig. 1) to allow fluid venting from the basement into the ocean. As such it is conceivable that lateral oxygen diffusion, perpendicular 275 to this venting, contributes to the upward oxygen profile for 69SL resulting in sediment columns in proximity of these venting systems that may be entirely oxic, possibly overruling the effect of sediment thickness.
The only core which in its pore-water Mn 2+ profile shows no downward decrease and as such no signs of influence from below has been taken from the thickest sediment cover encountered (82 m) and as such the sediment is 10 m thicker than for the other sites with thick sediments (65SL, 69SL, 72SL). Moreover, there are no faults or seamounts nearby. So, of all cores 280 this one has the best prerequisites for having a suboxic zone that extends deep into the sediment. We therefore suppose that also for this location upward oxygen diffusion modifies the sediment but we are not able to see this since the pore-water Mn 2+ maximum and subsequent downward decrease are found below the depth of the retrieved sediments/bottom of the core.

Pore-water nitrate
Usually in oxic sediments, nitrification of ammonium and diffusive loss of nitrate to the bottom water results in a 285 nitrate maximum in the subsurface (Hensen et al., 2006). We find this back in the multicores (not shown) where values increase from 36 µmol/l at the sediment surface to reach maxima between 40 and 60 µmol/l at 10 to 50 cm depth. The absence of such a subsurface nitrate maximum for all piston-and gravity-cores but 22KL and 35SL (Fig. 2) is explained by the loss of surface sediments as a result of the core recovery procedure. As OM mineralisation slows down with increasing sediment depth so does nitrate production. Furthermore, nitrate is diffusively lost to the overlying ocean bottom water (e.g., 290 Hensen and Zabel 2000, Hensen et al., 2006), and to the deeper sediment. The subsurface pore-water nitrate maximum in the upper oxic zone of the cores reflects the balance between these three processes. For the entirely oxic cores, deeper in the sediment the linearly downward decreasing nitrate concentrations (Fig. 2) suggest the profiles to be driven by diffusion only, whereby nitrate is delivered to the basement fluid. Whereas the MUC core-tops indicate nitrate maximum concentrations of 35-40 µmol/l for ocean bottom waters (e.g., Hensen et al., 2006) The cores with a suboxic zone (Fig. 3) have a more complex nitrate chemistry. From the nitrate maximum in the upper oxic zone well into the suboxic zone, the curvature of the profile becomes increasingly less (as also observed in some cores by e.g. Mewes et al., 2014;Mogollón et al., 2016 for this region), indicating a decreasing contribution of aerobic nitrification 305 and suboxic heterotrophic nitrification with depth. Extrapolation of the nitrate profile of core 108SL down to the anticipated depth of the basement-sediment interface provides values estimated between 42 and 47 µmol/l, which agrees with the values inferred from the entirely oxic cores. As such, the sediments represented by core 108SL are a source of nitrate for the basement fluid. For core 22KL, nitrate values reach below those inferred for the basement fluid. Here, the change from a gentle decrease to a linear increase at 9 m core depth in core 22KL marks the point where upward nitrate diffusion from the 310 basement equals downward diffusion and nitrate consumption. Linear extrapolation of this downward increase provides for the basement-sediment interface 44 µmol/l, which again agrees well with the values estimated from the entirely oxic cores/sites. However, in contrast to the oxic cores and core 108SL, the sediments represented by 22SL are a sink for basement-fluid nitrate. For the remaining cores with a suboxic zone (9KL, 42SL, 58SL, 65SL, and 72SL), denitrification is even more substantial and minimum nitrate values always reach below those inferred for the basement fluid. We infer for the 315 sediments below these cores a nitrate profile analogous to that seen in 22KL. However, whereas the nitrate minimum occurs within 22KL, we infer that these other cores did not reach sufficiently deep reach the nitrate minimum. From this minimum, (anticipated to be located below the maximum core depth), we infer a linear downward nitrate increase towards the basement -suggesting that the basement fluid represents a nitrate source.
There appears to be a clear linear relationship between the minimum pore-water nitrate concentration measured and the 320 maximum pore-water manganese concentration (Fig. 6), which roughly suggests that for each µmol/l increase in manganese, nitrate decreases by also one µmol/l and this suggests a coupling between the manganese and nitrate cycles. The ratio is very different from the manganese mediated denitrification where 2.5 manganese are needed to consume one nitrate to produce 0.5 nitrogen molecule (e.g., Mogollón et al., 2016). We note that such an opposite relation can be induced by two processes: (1) Oxidation by O 2 increases NO 3 -(from NH 4 + ) and decreases Mn 2+ (producing MnO 2 ) whereas (2) OM degradation 325 consumes NO 3 -(producing NH 4 + ) and produces Mn 2+ (from MnO 2 ) (see e.g., Mogollón et al., 2016). Apparently, the longterm dynamic equilibria between these processes result in the observed net relation between the minimum pore-water nitrate concentration measured and the maximum pore-water manganese concentration in the sediments of this region.
Thus, we observe that some sediments deliver nitrate to the basement whereas others take up nitrate from the basement. We also observe that the inferred nitrate values in the basement are higher than those in the ocean bottom waters, which implies 330 that after entering the basement, the basement fluids get enriched in nitrate. We infer, therefore, that overall, the nitrate uptake from the basement by the sediment is less than the nitrate delivery to the basement by the sediment so that the basement gets enriched in nitrate relative to the ocean bottom waters. This nitrate enrichment can also be observed for the Cocos Ridge (Wheat and Fisher, 2008).

Pore-water manganese and iron 335
While in sediments with sufficient reactive OM the manganese reduction zone is followed by iron reduction, there is no evidence of this latter process in the cores investigated -i.e. the abundance of electron donors in the form of OM is too low to induce Fe reduction. In all cases where dissolved manganese is present in the pore-water, we observe a maximum followed by a decrease with depth (Fig. 3). However, in those cases where we can follow this decrease further to zero, it is followed by oxygenated sediments and not by free iron. Consequently, we infer that in the complete region of this study 340 insufficient reductive power is available to induce iron reduction.

Nickel and cobalt in pore water and solid phase
Both pore-water and solid-phase nickel and cobalt in the suboxic cores tend to behave similar as/to manganese (Fig. 4), which agrees with the assumption that both metals are preferentially associated with the Mn oxide phase in the sediment (Aplin and Cronan, 1985;Koschinsky et al., 2001). In the suboxic zone, Mn oxide reduction also releases dissolved Ni 2+ and 345 Co 2+ into the pore-water. Especially cobalt behaves similar to manganese, both clearly showing the 'transition zone' between the upper oxic zone and the suboxic zone (upper front). We call this a 'transition zone' since these elements are still present in the pore-water but already show elevated solid-phase contents. Nickel is much less bound to the oxidation fronts and seems to remain more mobile in the pore waters at higher oxygen levels than is the case for manganese and cobalt. The reason for this may be oxidation of pore-water Co 2+ to Co 3+ when bound to the Mn oxide phase in the sediment, which is a 350 much more efficient scavenging process compared to simple adsorption, such as for Ni 2+ (Murray and Dillard, 1979).
Experiments with manganese oxide rich sediments also showed that the bonding of Ni is on average weaker and more easily reversible than for Co, and oxidation of Co during sorption on Mn oxides was confirmed (Kay et al., 2001).

Stability of the upper and lower redox fronts
The absence of a 'transition zone' for cobalt and manganese between the suboxic zone and the lower oxic zone (lower front) 355 illustrates that the upper front is much more dynamic than the lower front. The lower front is the result of a very long route of oxygen transport from the ocean into and through the basement. As such, changes in ocean water oxygen arrive at the lower front delayed and with lower amplitude and as a consequence of oxidation of reduced mineral phases in the crust also with lower concentrations. As a result, the lower front is much more stable and probably foremost moves slowly but steadily upward, oxidising the OM residue left behind when the upper front moved upwards with ongoing sedimentation. The oxygen 360 from the upper front has a much shorter route from its oceanic source and is thus much more subject to changes in deepwater oxygen content and/or marine production.
Considering the above, we may add geochemical details to the conceptual model of low-temperature circulation/diffusion of seawater/fluids and their components through ocean crust and overlying sediments that was presented by Fisher and Wheat 365 (2010) (Fig. 7). The classical model of a succession of redox zones from the ocean floor into the sediment has to be extended by adding a similar reversed zonation scheme from the basement-sediment interface into the sediment for those regions where basement fluids are oxic. However, there are some marked differences between the redox succession from the sediment surface downwards, and that from the basement and lower/overlying sediments upwards.
At the sediment surface the OM is fresh, highly reactive and subject to aerobic degradation first (Mewes et al., 2014;370 Mogollón et al., 2016). With ongoing sedimentation and concurrent upward movement of the oxic zone, progressively less energetically favourable redox processes follow. Here, the reaction time is often too short for a given redox process to consume all available substrate. Reduced species that can't be oxidised by the next, energetically less favourable redox pair will remain, resulting in a surplus of reduced material that becomes fossilised at depth.
In contrast to this, oxygen penetrating from the basement into the sediment encounters the oldest sediment first. Compared to 375 the sediment surface, reaction time is long and may equal sediment age. Here, not oxygen exposure time is limiting the degree to which the substrate (OM) can be oxidised but the reactivity of the substrate. From the basement, the successively energetically less favourable redox zones follow each other upward so that at a particular depth, with time, progressively stronger oxidants may be expected, each consuming the fossilised reduced species (such as OM) that were left by the upwards migrating redox zones near the sediment surface. It should be noted that the fossilisation process may have changed 380 the reactivity and degradability of the fossilised species over time (e.g., in sediments where sulphate reduction occurs (not in our sediments), natural sulphurisation may have cross-linked residual organic matter reducing its degradability by microorganisms (Zonneveld et al., 2010)). The order of redox processes upwards thus mirrors that from the sediment-water interface downwards. Considering the large amount of time available, consumption in the lower oxic zone (from the basement upwards) should be negligible so that the oxygen profile will approach linearity up to the oxic/anoxic interface 385 above. In sediments that are carbon-starved and have virtually no oxidisable organic material left in the solid phase. Thus, the front position is primarily a balance between the downward migration/burial of oxidisable species and upward oxygen flux from the basement. This is what we observe at site 22KL. Here, the downward directed profile of Mn, Co, Ni and other reduced species become linear already in the suboxic zone and meet a counter-directed linear upward oxygen diffusion profile. If oxygen supply from below increases, the oxygen flux from below may meet the upper oxic zone. Thus, in the 390 CCZ, the entirely oxic cores display a configuration where the balance between supply of reduced species to the sediment, oxygen supply and sedimentation rate is such that a suboxic zone is absent as is a pool of reduced fossilised material. Here If in such entirely oxic sediments the upward oxygen flux becomes less, the oxygen minimum and the downward directed 395 oxygen diffusion will extend downward where it will meet increasingly refractory carbon since oxygen exposure time by the lower oxygen profile increases towards the basement. Where the residual carbon is very refractory, this downward profile will become quasilinear. Despite very low oxygen consumption in such sediments, if they are sufficiently thick, the system may become suboxic eventually. Since this suboxic zone is formed where previously the oxygen from below had exhaustively removed electron donors. Manganese re-mobilisation may be very slow relative to diffusion, or even absent 400 due to lack of electron donors and thus, we expect to encounter a suboxic sediment where porewater manganese is undetectable, a situation that may apply to core 81SL (Fig. 2).

Conclusions
Analyses of the concentrations of oxygen, nitrate, manganese, nickel and cobalt in pore waters and the solid-phase of sediments from the CCZ have been summarised in a conceptual graph (Fig. 8) and show: 405 1. An upper oxygen profile due to oxygen penetrating the sediments from the ocean bottom water downwards and a lower oxygen profile due to oxygen penetrating from the basement upwards.
2. The length of the upper profile (2-3 m) to be relatively constant, the length of the lower profile to be very variable; the latter as a result of facilitated penetration near faults, differences in basement conditions such as fluid flow rate, path length and basement oxygen consumption. 410 3. Where the upper and lower oxygen profiles meet, sediments are entirely oxic and organic carbon mineralisation is exhaustive. Nitrate production is larger than denitrification so that the sediments are a source of nitrate for the ocean bottom water and basement fluid. 4. Where the upper and lower oxygen profiles do not meet, they are separated by an intermediate suboxic zone with free pore-water manganese and cobalt. Here denitrification continues and may be in excess of nitrate production so that the 415 basement fluid delivers nitrate to the sediment. 5. Extrapolation of nitrate profiles indicates that the nitrate concentrations at the basement-sediment interface are relatively constant (43-48 µmol/l) and higher than in the ocean bottom waters and we conclude that overall more nitrate goes from the sediment to the basement (see conclusion 3) than vice versa (see conclusion 4).
6. The behaviour of the upper and lower oxic/suboxic transitions is very different. The upper oxic/anoxic redox boundary is 420 at the end of a curved oxygen profile. The redox boundary moves up with sedimentation, leaving unoxidised OM behind in the manganese reduction zone below. Moreover, due to the relatively short distance between bottom water and the front, the position of the upper front is sensitive to changes in bottom water oxygen content and as such relatively unstable. The lower oxic/anoxic redox boundary is at the end of a linear oxygen profile, and it moves up as a result of lacking (sufficiently labile) oxidisable material further down. It leaves behind an 'energetic desert' without further oxidation potential. The long 425 transport route of bottom water oxygen, via basement fluid (advection) into the sediment and to the lower oxic/suboxic transition (diffusion), strongly buffers amplitude and frequency of changes in oxygen content, so that the lower transition is much more stable than the upper front. 7. Over time, manganese, nickel and cobalt diffuse from the suboxic zone to the upper and lower redox fronts/transitions where they are oxidised and precipitate. At the upper transition they add to the manganese, cobalt and nickel that were 430 deposited by sedimentation. With this transition moving upward with sedimentation, these elements re-enter the suboxic zone, remobilise and move up to the oxic/suboxic transition again, leading to a continuous increasing in solid-phase contents of these elements near the redox front. With the lower transition moving upwards, these elements remain fixed in the oxidised sediment below, resulting in a permanent loss of these elements from the suboxic zone.
8. Since in the region between the Clarion and Clipperton Fracture Zones sediment cover is generally thin and sedimentation 435 rates are generally low, we are mostly able to observe the upper oxygen profile and accompanying redox zone as well as in reverse the lower redox zone and oxygen profile, each with its associated set of phenomena such as nitrate diffusion into the basement fluid. In many other ocean regions sediments are thicker and productivity higher so that the lower oxygen profile will be below the reach of the gravity-or piston-coring device due to a more complete redox zonation reaching deeper into the sediment and drilling is needed to reach the successive redox zones as well as the reversed redox zonation followed by 440 the deep oxygen profile to the basement (provided sufficient oxygen is provided by the basement). We are just beginning to understand the impact of this deep redox zonation on the biogeochemical cycles of carbon and nitrogen as well as the distribution of carbon and other elements (such as manganese and associated metals such as nickel and cobalt and, in organically richer sediments, iron and sulphur) in the deep sediments. Fortunately, deep drilling (e.g., through international consortia like DSDP, ODP and IODP) logistically and financially enable access to these deeper sediments to investigate to 445 what extent theories and hypotheses erected on the basis of shallower settings are applicable for the majority of the ocean floor. Considering the potential area involved this warrants a substantial research effort. 9. Our study has pointed out that not only diffusive fluxes in the top part of deep-sea sediments modify the geochemical composition over time, but also diffusive fluxes from the basement into the bottom layers of the sediment. Hence, paleoceanographic interpretation of sedimentary layers should carefully consider such secondary modifications in order to 450 prevent misinterpretation as primary signatures.

Code and Data Availability
Data are available at https://doi.pangaea.de/10.1594/PANGAEA.xxxxxx. A doi has been applied but has not yet been issued.